Molecular Cell, Vol. 18, 283–293, April 29, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.molcel.2005.03.027

AMP-Activated Protein Kinase Inducesa p53-Dependent Metabolic Checkpoint

Russell G. Jones,1 David R. Plas,1 Sara Kubek,1

Monica Buzzai,1 James Mu,2 Yang Xu,4

Morris J. Birnbaum,2 and Craig B. Thompson1,3,*1Abramson Family Cancer Research InstituteDepartment of Cancer Biology2Howard Hughes Medical Institute3Department of MedicineUniversity of PennsylvaniaPhiladelphia, Pennsylvania 191044Division of Biological SciencesUniversity of California, San DiegoLa Jolla, California 92093

Summary

Replicative cell division is an energetically demand-ing process that can be executed only if cells havesufficient metabolic resources to support a doublingof cell mass. Here we show that proliferating mamma-lian cells have a cell-cycle checkpoint that respondsto glucose availability. The glucose-dependent check-point occurs at the G1/S boundary and is regulated byAMP-activated protein kinase (AMPK). This cell-cyclearrest occurs despite continued amino acid availabil-ity and active mTOR. AMPK activation induces phos-phorylation of p53 on serine 15, and this phosphory-lation is required to initiate AMPK-dependent cell-cycle arrest. AMPK-induced p53 activation promotescellular survival in response to glucose deprivation,and cells that have undergone a p53-dependent meta-bolic arrest can rapidly reenter the cell cycle uponglucose restoration. However, persistent activation ofAMPK leads to accelerated p53-dependent cellular se-nescence. Thus, AMPK is a cell-intrinsic regulator ofthe cell cycle that coordinates cellular proliferationwith carbon source availability.

Introduction

Mammalian cells depend on extracellular cues for theregulation of growth, proliferation, differentiation, andsurvival. External signals provided by growth factorsregulate mammalian cell homeostasis in part by activa-ting signaling cascades that regulate nutrient uptake,thus matching cellular demand to the maintenance ofbioenergetics. The insulin-like growth factor (IGF)/insu-lin receptor pathway in particular supports cellulargrowth by controlling the ability of a cell to take up nu-trients, including glucose (for review, see Plas et al.,2002). Mutations in components of this pathway, nota-bly activation of phosphatidylinositol 3#-kinase (PI3K)or Akt or loss of tumor suppressors PTEN, TSC1, orTSC2, renders cells able to take up nutrients in a cell-autonomous fashion, allowing cells access to the nutri-ents needed to support the bioenergetic demands ofincreased growth and proliferation (Shamji et al., 2003).

*Correspondence: craig@mail.med.upenn.edu

Deregulation of nutrient uptake may be a key step ena-bling transformation.

In addition to extracellular signals, intracellular nutri-ent levels may also regulate cell growth and prolifera-tion by modulating the activity of nutrient-regulatedkinases (Wilson and Roach, 2002). During times of nu-trient or environmental stress, mammalian cells senseintrinsic energy levels and attempt to restore bio-energetic homeostasis through compensatory changesin the regulation of intermediate metabolism. In re-sponse to increases in the ratio of ADP to ATP, the en-zymatic activity of adenylate kinase converts two mole-cules of ADP into ATP and AMP, thus restoring aneffective ATP:ADP ratio sufficient to maintain energy-dependent cellular processes and generating AMP asa by-product. Since a rise in AMP is a consequence ofa declining ATP:ADP ratio, the ratio of AMP to ATP isan indicator of overall energy charge. AMP stimulates apathway of energy conservation by activating the AMP-activated protein kinase (AMPK) (for review, see Hardieet al., 1998). AMPK is regulated in vivo by the build-upof cellular AMP and phosphorylation by an upstreamcomplex containing the tumor suppressor LKB1 (Haw-ley et al., 2003; Woods et al., 2003). Once triggered,AMPK activates catabolic pathways and inhibits ana-bolic pathways through direct phosphorylation of meta-bolic enzymes or by inducing alterations in gene ex-pression. The net result of this AMP-induced signalingpathway is the engagement of ATP-generating path-ways and inhibition of ATP-consuming processes (Car-ling, 2004). Since AMPK activation is directly linked tothe energy status of the cell, it is often considered tobe a cellular fuel gauge. In addition to its metabolic ef-fects, AMP accumulation has also been suggested, di-rectly or indirectly, to have effects on cell growth andproliferation. Treatment of cells with adenine analogsthat mimic AMP have been reported to induce cell-cycle arrest (Imamura et al., 2001). Recent work hasuncovered a link between AMPK and TSC2 in the inhibi-tion of translation during energetic stress (Inoki et al.,2003), indicating that AMPK may serve to link intracel-lular energy levels to the regulation of protein synthesis.

Limitation of low-molecular-weight nutrients wasnoted to inhibit cell proliferation over 30 years ago (Hol-ley and Kiernan, 1974; Pardee, 1974), although themechanisms underlying this process have remainedunclear. The simplest model to explain this phenome-non is one of substrate limitation—cells simply cannotgrow or proliferate when metabolic building blocks arein short supply. Alternatively, nutrient levels may tethercellular metabolism to endogenous pathways of cell-cycle control as a means to maintain cell viability whencells lack sufficient nutrients to support proliferation. Toinvestigate whether specific forms of nutrient availabil-ity are coupled to cell-cycle control, we studied whetheror not depletion of specific defined nutrients can in-duce cell-cycle arrest under conditions where the re-maining nutrients are sufficient to maintain cell survivaland cell growth. In this work, we demonstrate that glu-cose availability directly regulates cell proliferation and

Molecular Cell284

that the metabolic sensor AMPK is a mediator of this bwprocess. Cells treated with low glucose or transduced

with a constitutively active form of AMPK arrest in thepG1 phase of the cell cycle. Induction of this arrest is

p53 dependent, and p53 activation is required to pro- 2imote cell survival under conditions of glucose depriva-

tion. The p53-dependent metabolic arrest in response 1bto glucose deprivation is rapidly reversible upon readdi-

tion of glucose. In contrast, p53-deficient cells fail to eoarrest in response to AMPK activation and continue to

proliferate under low-glucose conditions, leading to a lMrapid decline in cell viability. Activation of AMPK leads

to phosphorylation of p53 at serine 15 (serine 18 of BAmouse p53), and this phosphorylation event is essential

for mediating the effects of AMPK on p53-dependent cTcell-cycle arrest. These findings indicate that AMPK is

a cell-intrinsic metabolic sensor that couples glucose Asavailability to the cell-cycle machinery of a proliferating

cell. This discovery highlights a previously unappreci-cated relationship between glucose metabolism and cell

proliferation and implicates p53 as an essential compo- tanent of metabolic cell-cycle control.(fResultsapGlucose Limitation Restricts

Cell-Cycle Progression eWhether the availability and metabolism of specific nu-trients contributes to control of the mammalian cell cy- G

ocle is not known. Here we investigated the effects ofglucose limitation on cell proliferation in otherwise O

lcomplete medium. Primary mouse embryonic fibro-blasts (MEFs) cultured in medium containing 1 mM glu- d

acose accumulated at a lower rate than cells incubatedin full-glucose media (25 mM glucose) (Figure 1A). c

lWhen serum-deprived cells were induced to reenter thecell cycle, their ability to traverse the G /S boundary t

Figure 1. Glucose Limitation Induces a G1

Cell-Cycle Arrest

(A) Glucose limitation blocks cell prolifera-tion. Growth curve of primary MEFs culturedfor 6 days under conditions of 25 or 1 mMglucose. The data presented are the mean ±the standard deviation (SD) of three indepen-dent experiments.(B) Glucose limitation inhibits cell-cycle en-try. MEFs synchronized in G0/G1 by culturein 0.1% serum were induced to enter Sphase by addition of media containing 10%serum and various concentrations of glu-cose. The percentage of cells in the S or G1

phase was measured by BrdU/PI staining at24 hr. The data presented are the mean ± SDfor triplicate samples.(C) Activation of AMPK in MEFs in responseto low glucose. MEFs were treated for 2 hrin the presence of 25 or 1 mM glucose. Celllysates were analyzed for phospho-T172AMPK, phospho-S79 ACC, total AMPK, andactin levels by Western blot.(D) Growth curve of MEFs in the presence of

high-glucose media (25 mM) containing 0.5 mM AICAR. The data represent the mean ± SD for triplicate samples.(E) AICAR induces proliferative arrest at the G1/S cell-cycle checkpoint. MEFs were synchronized at G0/G1, then induced to enter S phase byaddition of complete media (10% serum, 25 mM glucose) containing the indicated concentrations of AICAR. S or G1 phase cells weredetermined at 24 hr with BrdU/PI staining. The data represent the mean ± SD.

1

ecame progressively impaired as extracellular glucoseas reduced (Figure 1B).Proliferating mammalian cells use glucose as their

rimary fuel source for ATP production (Zong et al.,004). Glucose availability is monitored in lower organ-

sms by the bioenergetic sensor AMPK (Carlson et al.,981). Decreases in extracellular glucose have alsoeen shown to activate AMPK in mammalian cells (Saltt al., 1998). To test whether AMPK is activated at levelsf extracellular glucose that impair mammalian cell pro-

iferation, we measured the level of AMPK activation inEFs cultured under low-glucose conditions (1 mM).oth the phosphorylation status and the activity ofMPK increased as cells were cultured in reduced glu-ose (Figure 1C). AMPK became phosphorylated onhr-172, and increased Ser-79 phosphorylation of theMPK substrate acetyl CoA carboxylase (ACC) was ob-erved when cells were cultured at 1 mM glucose.As an initial test of the role of AMPK activation in the

ellular response to decreased glucose availability, wereated cells in glucose-replete medium with the AMPKctivator 5-aminoimidazole-4-carboxamide ribosideAICAR). Even in 25 mM glucose, AICAR-treated cellsailed to accumulate (Figure 1D). G0/G1-arrested MEFslso exhibited a dose-dependent block in the G1 to Srogression when stimulated with serum in the pres-nce of AICAR (Figure 1E).

lucose Restriction Induces Activationf a Reversible Cell-Cycle Checkpointne mechanism that can limit cell accumulation in pro-

iferating cultures is the induction of apoptotic celleath. Glucose depletion has been reported to inducepoptosis through activation of the proapoptotic mole-ule Bax (Rathmell et al., 2003). However, significant

evels of apoptosis were not observed at concentra-ions of extracellular glucose that induce a near com-

AMPK Regulates p53-Dependent Cell-Cycle Arrest285

plete cell-cycle arrest. Apoptotic cell death was ob-served only when cells were transferred to mediumcompletely deficient in glucose (Figure 2A).

Recent evidence has demonstrated that ATP deple-tion can suppress cell growth by inhibiting proteintranslation, a process initiated by an AMPK-TSC1-

Figure 2. Glucose Limitation Invokes a Re-versible Cell-Cycle Checkpoint

(A) Cell viability is maintained under low-glu-cose conditions. G0/G1-synchronized MEFswere induced to proliferate by addition ofmedia containing 10% serum and variousconcentrations of glucose. Cell viability wasmeasured 24 hr following addition of serumby PI exclusion. The data presented are themean ± SD.(B) Low glucose promotes an mTOR-depen-dent increase in cell size. MEFs were cul-tured in the presence of low glucose (0.5mM) for 24 hr, and cell size was measuredby forward scatter (FSC) of G1 phase cells.Cell size of MEFs grown in full glucose (25mM) is indicated in gray and test conditionsin white. mTOR-specific growth was mea-sured by culturing cells in the presence of 20nM rapamycin (RAPA). A representative ex-periment is shown. Cell proliferation wasmeasured by BrdU incorporation, and thepercentage of BrdU+ cells (mean ± SD) wasdetermined by flow cytometry.(C) AICAR elicits dose-dependent effects oncell size. MEFs were cultured in the presenceof low (0.5 mM) or high (2.0 mM) AICAR inmedium containing full glucose (25 mM) for24 hr, and cell size was determined by for-ward scatter (FSC) of G1 phase cells. Cellsize of MEFs grown in the absence of AICAR(control, 25 mM glucose) is indicated in grayand test conditions in white. mTOR-specificgrowth was measured by culturing cells inthe presence of 20 nM rapamycin (RAPA).(D) Proliferative arrest induced by glucoselimitation is reversible. MEFs were grown for24 hr in media containing 25 or 1 mM glu-cose, washed with glucose-free media, thencultured in complete media containing 25mM glucose. Cells undergoing active DNAsynthesis before and after the switch to 25mM glucose conditions were measured byBrdU incorporation. The percentage ofBrdU+ cells is quantified in the upper rightcorner of each panel.

TSC2-dependent pathway (Inoki et al., 2003). A re-duced translation rate could lead to prolongation of theG1 phase of the cell cycle by reducing the rate of cellgrowth that normally accompanies proliferation. To as-sess whether glucose depletion alone affects cellgrowth, we measured the cell size of MEFs cultured

Molecular Cell286

under low-glucose conditions. Surprisingly, when MEFswere switched into medium containing low glucose for24 hr, they exhibited a larger cell size than cells main-tained in standard medium, even though they un-derwent a cell-cycle arrest (Figure 2B).

To test whether the size increase induced by glucoselimitation was regulated by the activity of mTOR, weanalyzed the growth of glucose-restricted cells in thepresence of the TOR inhibitor rapamycin. Treatmentwith rapamycin completely suppressed the size in-crease observed when cells were cultured in low glu-cose in the presence of a full complement of amino

Facids (Figure 2B). Similar results were observed when(cells were treated with 0.5 mM AICAR in the presencewof complete medium (Figure 2C). At this level of AICAR,aa G1/S arrest was observed while rapamycin-sensitiveo

cell growth was maintained. However, at higher doses aof AICAR, the cell-cycle arrest was associated with a (progressive inhibition of protein translation and a loss (

Bof the rapamycin-sensitive increase in cell size in theeG1-arrested cells (Figure 2C). Together, these resultsBdemonstrate that low extracellular glucose can induce

a cell-cycle arrest under conditions where TOR remainsactive and continues to stimulate cell growth. A

Mammalian cells can also respond to cellular stresses cby initiating specific cell-cycle checkpoints that serve cto maintain viability and functional integrity until the tstress is corrected or removed. To assess whether tthe proliferative checkpoint triggered by low-glucose pstress was reversible, MEFs arrested by glucose re- astriction were assessed for their ability to initiate prolif- ieration upon the readdition of glucose. Within 24 hr of aincreasing glucose levels from 1 mM to 25 mM, cells cwere able to fully recover and maintain rates of DNAsynthesis comparable to cells grown continuously un- ider full-glucose conditions (Figure 2D). c

uAMPK Activity Is Sufficient to Induce lCell-Cycle Arrest dTo address whether AMPK activity is sufficient to medi- late cell-cycle arrest under normal growth conditions, swe expressed a constitutively active mutant of the pAMPKα2 catalytic subunit (CA-AMPK) in primary MEFs gand assessed their proliferative capacity. Similar to ccells exposed to AICAR or low glucose, MEFs express- cing CA-AMPK were impaired in their ability to prolifer- gate as judged by growth curves (Figure 3A). This impair- tment correlated with a reduced percentage of cells in S cphase as measured by BrdU incorporation (Figure 3B). w

cAMPK Requires p53 to Induce Cell-Cycle ArrestThe results described to this point suggest that activa- t

gtion of AMPK, either through pharmacological or ge-netic means, can induce a reversible cell-cycle arrest. r

TTo assess whether AMPK activation is required to in-duce cell-cycle arrest under low-glucose conditions, w

rAMPK activity was blocked using a dominant-negativeform of AMPK (DN-AMPK) (Mu et al., 2001). MEFs in- a

tfected with a retroviral vector expressing DN-AMPKdisplayed a 3-fold higher rate of S phase entry when a

wgrown under low-glucose conditions than control-infected cells (Figure 4A). A similar result was observed r

twhen MEFs expressing DN-AMPK were treated with

igure 3. AMPK Activity Is Sufficient to Induce Cell-Cycle Arrest

A) AMPK activity inhibits cell proliferation. MEFs were infectedith an empty vector (Vec) or a vector expressing constitutivelyctive AMPK (CA-AMPK), and cell accumulation was determinedver time. Data are mean ± SD of triplicate samples. After day 3,ll differences in cell counts are statistically significant (p < 0.01).

B) AMPK activation prevents S phase entry. MEFs infected as inA) were pulsed with 10 �M BrdU for 1 hr and then analyzed forrdU incorporation. The percentage of BrdU− and BrdU+ cells arexpressed as mean ± SD for triplicate samples. The difference inrdU incorporation is statistically significant (p < 0.001).

ICAR (data not shown). The effect of AMPK was spe-ific to low-glucose stress, as DN-AMPK-expressingells arrested to the same extent as controls whenreated with γ radiation (Figure 4A). MEFs deficient forhe tumor suppressor p53 were included in these ex-eriments as a negative control for γ-radiation-inducedrrest. As expected, p53-deficient MEFs failed to arrest

n response to γ radiation. However, p53-deficient MEFslso displayed resistance to arrest under low-glucoseonditions (Figure 4A).p53 is a central cell-cycle checkpoint protein that can

nitiate cell-cycle arrest at either the G1/S or G2/Mheckpoint in response to cellular stress. The contin-ed ability of p53-deficient MEFs to proliferate under

ow-glucose conditions suggested that p53 might me-iate the checkpoint response induced by glucose

imitation. To assess the role of p53 in the glucose-sen-itive checkpoint response, we examined the ability of53-deficient MEFs to bypass the G1/S checkpoint trig-ered by glucose restriction. Unlike their wild-typeounterparts, p53−/− MEFs were unable to initiate theell-cycle arrest normally induced by low extracellularlucose and readily traversed the G1/S boundary underhese conditions (Figure 4B). Proliferation of p53-defi-ient MEFs was suppressed to levels comparable toild-type cells only under conditions of complete glu-ose withdrawal.The observation that p53−/− MEFs have a defect in

heir ability to undergo cell-cycle arrest in response tolucose restriction raised the possibility that p53 wasequired for the G1 arrest induced by AMPK activation.o address this possibility, MEFs lacking p53 or theirild-type counterparts were infected with a control ret-

ovirus (Vec) or a retrovirus encoding constitutivelyctive AMPK (CA-AMPK), and their proliferation in cul-ure was measured. Wild-type cells displayed a prolifer-tive disadvantage when expressing active AMPK,hile p53−/− MEFs expressing AMPK accumulated at a

ate comparable to cells infected with the control vec-or (Figure 4C). p53−/− MEFs infected with CA-AMPK

AMPK Regulates p53-Dependent Cell-Cycle Arrest287

Figure 4. AMPK Requires p53 to Induce Cell-Cycle Arrest

(A) DN-AMPK prevents cell-cycle arrest in-duced by low glucose but not DNA damage.MEFs infected with control vector (Vec) orvector expressing dominant-negative AMPK(DN-AMPK) were incubated in low glucose(0.5 mM), and BrdU incorporation at 24 hrwas measured. For arrest induced by DNAdamage, MEFs were treated with 5.5 Gy of γradiation (γ-Rad), and BrdU incorporationwas measured after 16 hr. Uninfected p53−/−

MEFs were included to control for stress-induced cell-cycle arrest. All values are cal-culated as a percentage of untreated con-trols and presented as mean ± SD.(B) Loss of p53 disrupts the G1 arrest check-point induced by glucose restriction. Wild-type or p53−/− MEFs synchronized in G0/G1

by culture in 0.1% serum were induced toenter S phase by addition of media contain-ing 10% serum, 65 �M BrdU, and variousconcentrations of glucose. Cells that haveprogressed into S phase after 24 hr were de-termined by BrdU incorporation (mean ± SD).

(C) Growth curves of p53+/+ and p53−/− MEFs infected with control retrovirus (Vec) or active AMPK (CA-AMPK). The data represent the mean ± SDfor triplicate samples.(D) p53−/− MEFs are resistant to AICAR-induced arrest. G0/G1-synchronized wild-type or p53−/− MEFs were induced to enter S phase byaddition of complete medium in the presence (+) or absence (−) of 0.5 mM AICAR. Cells were treated continuously with 65 �M BrdU, and thepercentage of BrdU+ cells (mean ± SD) was measured after 24 hr.

readily entered S phase as measured by BrdU incorpo-ration, while their wild-type counterparts showed a dis-tinct G1 arrest with more than a 70%–75% reduction inthe number of cells in S phase (data not shown). Toindependently confirm that this decreased proliferativerate was due to activation of a G1/S cell-cycle check-point, the ability of serum-deprived p53+/+ or p53−/−

MEFs to reenter S phase upon serum readdition in thepresence of AICAR was measured. Less than one-thirdof cells expressing wild-type p53 could initiate DNAsynthesis in the presence of AICAR as measured byBrdU incorporation (Figure 4D). In contrast, AICAR hadno effect on the ability of p53−/− MEFs to enter S phase,with approximately 90% of all p53−/− cells scoring posi-tive for BrdU incorporation.

p53 Promotes Cell Viability in Responseto Glucose DeprivationAn essential feature of an adaptive cell-cycle check-point is to enhance the ability of cells to survive an im-posed insult. However, previous studies have sug-gested that p53 activation induces apoptosis. Toinvestigate whether p53 activation affects the ability ofcells to survive glucose deprivation, the viability ofp53+/+ and p53−/− MEFs cultured in the absence of glu-cose was determined. Surprisingly, cells expressingwild-type p53 displayed a survival advantage over p53-deficient cells when deprived of glucose (Figure 5,closed bars). Moreover, p53+/+ MEFs cultured first un-der low-glucose conditions displayed a greater resis-tance to apoptosis upon removal of glucose, whilep53−/− MEFs exposed first to low-glucose conditionswere not protected from apoptosis induced by com-plete depletion of extracellular glucose (Figure 5, openbars). Together, these data indicate that glucose restric-

cycle arrest, the possibility that AMPK may modulate

Figure 5. p53 Maintains Cell Viability in Response to Glucose Depri-vation

p53+/+ or p53−/− MEFs were precultured in medium containingeither 25 mM (closed bars) or 0.5 mM (open bars) glucose andthen transferred to glucose-free medium containing 10% dialyzedserum. Cell viability was measured 24 hr later by PI exclusion. Via-bility was greater than 94% for all the starting populations. Thedata are presented as mean ± SD. Statistically significant differ-ences as calculated by paired Student’s t test are indicated (n.s.,no significant difference).

tion promotes an adaptive response that enhances theability of cells to survive further glucose deprivation.p53 is required to mediate this adaptive response.

AMPK Promotes p53 Activity throughPhosphorylation at Ser-15Given the dependence of AMPK on p53 to induce cell-

Molecular Cell288

Figure 6. AMPK Promotes p53 Phosphorylation at Ser-15

(A) Wild-type MEFs infected with control vector (Vec) or vector expressing active AMPK (CA-AMPK) were analyzed by Western blot for p53protein expression. Expression of myc-tagged CA-AMPK was detected using anti-myc antibody.(B) AMPK activates p53-responsive promoters. Wild-type (p53+/+) or p53 null (p53−/−) MEFs were transfected with empty vector or activeAMPK (CA-AMPK) along with a luciferase reporter containing tandem p53 consensus binding sites (pG13) or the human p21 reporter (pWWP).Promoter activity is expressed as the ratio of luciferase to renilla activity (RLU, relative light units) 48 hr posttransfection. The data representthe mean ± SD for triplicate samples.(C) The AMPK phosphorylation site in rat ACCα and potential sites found in human and mouse p53.(D) AICAR induces p53 phosphorylation. Human 293T cells were treated with various concentrations of AICAR for 24 hr, and the level ofphospho-Ser-15 p53 was examined by Western blot.(E) AMPK activity induces Ser-18 phosphorylation of p53. H1299 cells were cotransfected with mouse p53 and active AMPK (CA-AMPK) orcontrol vector. Levels of phospho-Ser-18 p53, p53, and p21 were determined 48 hr posttransfection by Western blot.(F) AMPK kinase activity promotes phosphorylation of p53 at Ser-18. Wild-type (myc-AMPK) or dominant-negative (myc-DN-AMPK) AMPK

AMPK Regulates p53-Dependent Cell-Cycle Arrest289

and growth curves for each genotype were establishedp53, and then the cells were subjected to low-glucose

was transfected into 293T cells, and cells were subjected to low-glucose conditions 24 hr posttransfection. AMPK was immunoprecipitatedfrom lysates using anti-myc antibodies and used to phosphorylate recombinant substrates in vitro. Recombinant GST fusion proteins express-ing amino acids 1–98 of mouse p53 (GST-p53(1-98)) or containing a serine to alanine mutation at amino acid 18 (GST-p53 S18A(1-98)) wereused to assess phosphorylation of p53 by AMPK. GST-fusion proteins expressing SAMS peptide (GST-SAMS) or containing a mutation atthe phosphate acceptor site (GST-SAMSM) were used to control for AMPK activity. Recombinant proteins incorporating 32P were detectedby autoradiography.(G) Glucose limitation induces p53 phosphorylation. Full-length mouse p53 was cotransfected with vector, wild-type AMPKα2 (myc-AMPK),or dominant-negative AMPKα2 (myc-DN-AMPK) into H1299 cells. Cells were incubated for 16 hr in the presence of full (25 mM) or low (0.5mM) glucose. Lysates were analyzed for Ser-18 p53 phosphorylation.(H) Ser-18 of p53 is required to mediate cell-cycle arrest in response to glucose limitation. MEFs from p53-deficient (p53−/−), wild-type (p53+/+),and p53S18A/S18A animals were cultured in the presence of 0.5 mM glucose, and proliferation was measured 24 hr later by BrdU incorporation.Data are presented as the percentage of BrdU+ cells relative to control cells grown in 25 mM glucose (mean ± SD).(I) AMPK requires phosphorylation of p53 at Ser-18 to induce arrest in MEFs. Early-passage MEFs from wild-type (p53+/+) and p53S18A/S18A

animals were infected with vector alone (Vec) or active AMPK (CA-AMPK), and cell counts after three days of culture were measured. Finalcell numbers were normalized relative to the initial cell input and presented as mean ± SD. Statistically significant differences as calculatedby paired Student’s t test are indicated (n.s., no significant difference).

p53 activity was investigated. Levels of endogenousp53 protein were elevated in cells expressing activeAMPK (Figure 6A). In addition, p53-dependent tran-scription was strongly induced by active AMPK in wild-type MEFs (p53+/+) (Figure 6B). p53 is activated inresponse to cellular stress in part through posttransla-tional modifications that promote its stabilization (Vo-gelstein et al., 2000). The motif surrounding serine 15of p53 (serine 18 in mouse), a site known to undergophosphorylation during p53 activation, is similar to themotif on ACC phosphorylated by AMPK (Figure 6C),raising the possibility that this site is also targeted byAMPK kinase activity. Treatment of 293T cells withAICAR promoted a dose-dependent increase in p53phosphorylation at Ser-15 (Figure 6D), consistent withprevious observations (Imamura et al., 2001). To testwhether phosphorylation of p53 at this residue is medi-ated by the activity of AMPK, p53 null H1299 cells werecotransfected with CA-AMPK and mouse p53 (Figure6E). Coexpression of CA-AMPK with p53 promotedphosphorylation of p53 at Ser-18, confirming thatAMPK activity was sufficient to induce this phosphory-lation event. Coexpression of CA-AMPK and p53 alsoled to a synergistic increase in endogenous p21 pro-tein levels.

To assess whether AMPK immunoprecipitated fromglucose-deprived cells can phosphorylate p53, we per-formed an in vitro AMPK kinase assay using a GST-p53fusion protein containing the N terminus of mouse p53(amino acids 1–98) as a substrate (Canman et al., 1998).Wild-type AMPK but not the dominant-negative mutantcould induce the phosphorylation of the p53 fragment(Figure 6F). This AMPK-induced phosphorylation wasonly observed using wild-type p53 sequence as sub-strate. A p53 substrate containing a serine to alaninesubstitution at position 18 failed to be phosphorylated.While these data cannot rule out the possibility that p53is directly phosphorylated by an AMPK-associated ki-nase that coimmunoprecipitates with AMPK, theydemonstrate that AMPK kinase activity triggered byglucose deprivation promotes p53 phosphorylationspecifically at serine 15 (serine 18 of mouse p53).

To confirm that AMPK activity is required for p53phosphorylation under conditions of glucose limitation,wild-type or dominant-negative AMPK was cotrans-fected into H1299 cells along with full-length mouse

stress (0.5 mM). Cells expressing wild-type AMPK dis-played increased Ser-18 phosphorylation of mousep53 under low-glucose conditions. In contrast, Ser-18phosphorylation in response to low glucose wasblocked by dominant-negative AMPK (Figure 6G). Thebiological importance of Ser-18 phosphorylation in thep53-dependent response to glucose limitation was de-termined using MEFs harboring a germline knock-inmutation that converts Ser-18 of mouse p53 to alanine(denoted p53S18A/S18A) (Chao et al., 2000). While MEFswild-type for p53 were unable to proliferate under low-glucose conditions, p53S18A/S18A MEFs displayed a pro-liferative rate nearly identical to that of p53 null cells(Figure 6H).

Finally, to assess whether AMPK-dependent p53phosphorylation is required for AMPK-induced cell-cycle arrest, the ability of AMPK to induce cell-cyclearrest in p53S18A/S18A MEFs was examined. In contrastto wild-type MEFs in which CA-AMPK induces a prolif-erative arrest, p53S18A/S18A mutant MEFs expressingCA-AMPK failed to undergo cell-cycle arrest and prolif-erated at a similar rate to controls (Figure 6I).

p53 Controls AMPK-Induced Cellular SenescenceThe induction of a persistent G1 arrest in the face ofthe continued ability to undertake translation and cellgrowth is a hallmark of cellular senescence. Recent evi-dence has associated a progressive bioenergetic com-promise with this process (Zwerschke et al., 2003). Toassess the influence of AMPK activity on replicativesenescence, constitutively active (CA-AMPK) and do-minant-negative (DN-AMPK) mutants of AMPK were in-troduced into low-passage primary MEFs by retroviral-mediated gene transfer, and the infected cells weresubjected to serial passage following a modified 3T3protocol. Growth curves of the infected MEFs demon-strated that elevated AMPK activity induced a prema-ture proliferative arrest in primary MEFs, while suppres-sion of AMPK activity extended replicative life span(Figure 7A).

p53 has been implicated as a critical component ofthe senescence program (Itahana et al., 2001). Giventhe link established between AMPK and p53, the in-volvement of AMPK in the replicative senescence ofprimary MEFs lacking p53 was examined. Constitutivelyactive AMPK was introduced into p53+/− or p53−/− MEFs,

Molecular Cell290

Figure 7. AMPK Induces p53-Dependent Premature Senescence in MEFs

(A) Serial passage of primary MEFs expressing AMPK mutants. Early passage MEFs from p53+/+ embryos infected with vector expressingdominant-negative AMPK (DN-AMPK), constitutively active AMPK (CA-AMPK), or control vector (Vec) were subjected to serial passage by amodified 3T3 protocol. Population doublings (PDL) at each time point are expressed as the mean ± SD for triplicate cultures.(B) AMPK cannot induce senescence in p53-deficient MEFs. Early-passage MEFs from p53+/− or p53−/− embryos infected with control vector(Vec) or vector expressing active AMPK (CA-AMPK) were subjected to serial passage following selection. Population doublings (PDL) areplotted as a function of time.(C) Senescence-associated β-galactosidase activity in MEF cultures. p53+/− or p53−/− MEFs infected with vector control (Vec) or active AMPK(CA-AMPK) were replated after passage 4 and then stained for SA-β-gal activity. Representative images are shown.(D) Quantitation of SA-β-gal-positive cells from (C). The data represent the mean ± SD for triplicate samples.

following selection (Figure 7B). p53+/− MEFs expressing etCA-AMPK underwent premature growth arrest relativepto those infected with control retrovirus. The AMPK-texpressing cultures also displayed enhanced senes-acence-associated β-galactosidase (SA-β-gal) activity,land loss of p53 negated both these effects (Figures 7Cland 7D).dc

Discussion gw

Initiation of cell division commits cells to a metaboli- acally demanding process. Recently, it has been sug- Tgested that mammalian cells require specific extracel- clular signals to license their ability to take up nutrients. aIf true, this suggests that there exist mechanisms by bwhich cells coordinate proliferation relative to their abil- pity to take up specific extracellular metabolites. To suc- ecessfully undergo cell division, cells must have a suffi- pcient supply of essential nutrients to support both ATP pproduction and macromolecular synthesis. We report a

lhere that the G /S transition in mammalian cells is teth-

1

red to glucose availability by AMPK. Glucose restric-ion can induce a reversible cell-cycle arrest in theresence of sufficient growth factors and amino acidso support TOR-dependent growth. Induction of AMPKctivity results in cell-cycle arrest at levels of extracellu-

ar glucose that are still capable of supporting the pro-iferative expansion of cells. The ability of AMPK to in-uce cell-cycle arrest is dependent on p53. Theoupling of glucose availability to cell-cycle pro-ression has the hallmarks of a nutrient-sensing path-ay that turns off cell proliferation before glucose avail-bility drops to levels that cannot sustain cell viability.he existence of such a glucose-sensitive cell-cycleheckpoint may act as a buffer to prevent cells fromccumulating to levels that are not ultimately sustaina-le. Disruption of this checkpoint through loss of p53ermits unchecked proliferation despite limiting nutri-nts but ultimately leads to decreased cell viability. Byromoting senescence, AMPK-dependent activation of53 may promote the conservation of the remainingvailable glucose to support the survival and physio-

ogic function of the existing cell.

AMPK Regulates p53-Dependent Cell-Cycle Arrest291

In addition to the cell-cycle effects demonstratedhere, recent work indicates that AMPK activation canalso regulate protein translation. AMPK activity sup-presses translation by inhibiting the activities of elon-gation factor-2 (eEF2) or ribosomal S6-Kinase (S6K). Inaddition, AMPK can phosphorylate TSC2 in responseto acute energy starvation, leading to stabilization ofthe TSC1-TSC2 complex and an inhibition of mTOR,S6K, and eIF4E signaling (Inoki et al., 2003). The tumorsuppressor LKB1 acts upstream of AMPK to mediatethese effects (Corradetti et al., 2004; Shaw et al., 2004).These results have raised the possibility that an AMPK-dependent reduction in translation could prolong theG1 phase of the cell cycle and thus slow proliferation.However, our observation that glucose levels of 0.5–1mM promote AMPK-dependent cell-cycle arrest asso-ciated with TOR-dependent increases in cell size sug-gest that AMPK can mediate its cell-cycle effects with-out significantly affecting translation. Together theseobservations suggest that cell cycle and cell size aredifferentially controlled by nutrient availability. Underthe conditions reported here, proliferation is more sen-sitive to glucose limitation than is growth control. Whilecomplete depletion of glucose can inhibit translation byeither AMPK-dependent inhibition of mTOR (Inoki et al.,2003) or by inducing an ER stress response (Kaufmanet al., 2002), the cell-cycle arrest observed here occursat glucose levels that are able to sustain proliferativeexpansion of MEFs either deficient for p53 or in whichAMPK activity is suppressed.

The tumor suppressor p53 is one of the primarygenes that regulates cellular commitment to DNA repli-cation and cell division (Vogelstein et al., 2000). Thepresent data indicate that p53 plays an essential role inthe cell-cycle arrest triggered by glucose limitation.Like cells with impaired AMPK activity, p53 null MEFsfail to arrest in low-glucose conditions. The depen-dence of AMPK on p53 for its antiproliferative effectsis supported by the resistance of p53-deficient cells toproliferative arrest induced by AMPK activation, eitherby low glucose, AICAR, or overexpression of activeAMPK. Thus, AMPK appears to activate a nutrient-sen-sitive signaling pathway that initiates p53-dependentcell-cycle arrest during times of energy deficiency. Acti-vation of this stress pathway appears to be specific tonutrient limitation; inhibition of AMPK activity does notantagonize other p53-dependent stress responsessuch as cell-cycle arrest induced by DNA damage.Thus, p53 can serve as a metabolite sensor and coordi-nate reversible cell-cycle arrest in response to glucosedeprivation. Surprisingly, although the activation of p53induced by DNA damage is associated with reducedsurvival, the metabolic activation of p53 is associatedwith the enhanced ability of cells to survive progressiveglucose depletion. Based on these results, we proposethat the ability of AMPK to induce p53 activity repre-sents a key integration point between the cellular meta-bolic state and the cell cycle.

p53 activity is maintained at low levels in normal cellsthrough the activity of Mdm2, which decreases thefunction of tetrameric p53 and targets it for degradationby the proteasome (Haupt et al., 1997; Kubbutat et al.,1997). Posttranslational modifications triggered follow-ing cellular insult, including phosphorylation at serine

15 (serine 18 in mouse), allow p53 to escape negativeregulation. Phosphorylation of p53 at Ser-15 in the Nterminus promotes p53 stabilization in part by disrupt-ing the p53-Mdm2 interaction (Shieh et al., 1997). More-over, phosphorylation at this site promotes the recruit-ment of cofactors such as p300 (Lambert et al., 1998),which acetylate p53 and further augment its ability toinduce cell-cycle arrest (Barlev et al., 2001). Phosphor-ylation of mouse p53 at Ser-18 is a critical event in thep53 response to DNA damage (Chao et al., 2000). Thisprocess is mediated by the ataxia telangiectasia-mutated (ATM) kinase, which targets p53 phosphoryla-tion at this residue in response to DNA damage (Baninet al., 1998; Canman et al., 1998). AMPK activation re-sults in phosphorylation of this same serine residue inresponse to energetic stress. In the absence of bioen-ergetic stress, treatment of cells with AICAR can lead top53 phosphorylation (Imamura et al., 2001). Consistentwith a role for AMPK in p53 phosphorylation, we findthat this AICAR-dependent p53 phosphorylation is notblocked by caffeine, an inhibitor of ATM kinase activity,suggesting that the effect of AICAR on Ser-15 phos-phorylation is ATM independent (data not shown). Theinability of AMPK to induce cell-cycle arrest when Ser-18 of mouse p53 is mutated confirms that this phos-phorylation event is critical for control of proliferationwhen glucose levels are low. This response is likelymodified by further changes in the metabolic state:SIRT1 has been shown to regulate p53 action throughdeacetylation in an NAD-regulated fashion (Luo et al.,2001; Vaziri et al., 2001).

Replicative senescence is proposed to be a tumorsuppressor mechanism in response to bioenergeticstress (Ben-Porath and Weinberg, 2004). Environmentalfactors that place oxidative stress on cells promote theearly onset of senescence (Parrinello et al., 2003). Suchstresses cause cells to shuttle available glucose fromthe glycolytic pathway to the pentose phosphate shuntto increase the production of reducing equivalents.Such a change would reduce the available intracellularglucose required to support ATP production. Consis-tent with this, the AMP:ATP ratio increases by approxi-mately 30-fold in senescent cells (Zwerschke et al.,2003). This increased level of AMP could lead to AMPK-dependent activation of p53 as documented here. Thetumor suppressor LKB1 is likely involved in this path-way, as Lkb1−/− MEFs, which are impaired in their abilityto activate AMPK (Hawley et al., 2003), do not senesce(Bardeesy et al., 2002). Activation of a p53-dependentmetabolic cell-cycle checkpoint by AMPK enables cellsto survive unfavorable growth conditions. To this end,the involvement of AMPK in the regulation of senes-cence may at least partially explain its role as a tumorsuppressor (Kato et al., 2002).

Experimental Procedures

Materials and DNA ConstructsMyc-tagged AMPKα2 cDNA constructs (wild-type and dominant-negative K45R mutant) have been described previously (Mu et al.,2001). Active AMPK (CA-AMPK) was generated by introducing aT172D mutation into AMPKα2 truncated at residue 312. All AMPKcDNAs were subcloned into pcDNA3 or the MSCV-based murineretroviral vector MigCD8t. A S18A mutation was introduced into the

Molecular Cell292

mouse p53 sequence by PCR mutagenesis, and both wild-type and A2mutant p53 were subcloned into pEF6-HisC (Invitrogen). Expres-

sion constructs encoding GST fusion proteins GST-SAMS (SAMS Aasequence: HMRSAMSGLHLVKRR), GST-SAMS mutant (SAMSM:

HMRSAMAGLHLVKRR), GST-p53(1-98), and GST-p53 S18A(1-98) Tpwere generated by PCR. 5-aminoimidazole-4-carboxamide ribo-

side (AICAR) was obtained from Toronto Research Chemicals. itb

Cell Lines, Transfection, and Retroviral Transduction (Primary MEFs (p53+/+, p53+/−, and p53−/−) were prepared from day pE13 embryos and passaged as previously described (Zong et al., T2004). p53S18A/S18A MEFs were generated as previously described B(Chao et al., 2003). Transfections were conducted using Lipofec- [tamine 2000 (Invitrogen) or Fugene6 (Roche). High-titre retrovirus (was generated by transfecting retroviral vector (20 �g) and packag- Ging virus DNA (10 �g) into 293T cells using standard calcium phos- mphate transfection methods. Virus was collected 48 and 72 hr post- utransfection and cleared prior to use. MEFs (7.5 × 105/10 cm dish) swere infected four times over 2 days using 4 ml of retroviral super-natant and 8 �g/ml polybrene. CD8+ cells were sorted 4 days post- Sinfection and grown for 2 days following sorting prior to use in Eproliferation assays. The level of infection was measured by flow ccytometry using anti-human CD8 antibodies (eBioscience). r

s(

Measurement of Cell Proliferation, Cell-Cycle�

Analysis, and Cell Viabilityt

Cell proliferation curves in all MEF lines were determined by celld

counting using trypan blue exclusion. For low-glucose culture con-i

ditions, cells were washed twice with PBS, then cultured in glu-cose-free DMEM (Invitrogen) supplemented with 10% dialyzed FBS(dFBS) (Invitrogen) and concentrations of glucose as indicated. Ra-

Adiation-induced cell-cycle arrest was measured 16 hr following ex-posure of cells to 5.5 Gy of γ radiation. Actively proliferating cells

Wwere assayed by incorporation of BrdU. For cell-cycle analysis,

HMEFs (1 × 105/30 mm dish) were pulsed with BrdU, then trypsin-

Kized, fixed in 70% ethanol, labeled with anti-BrdU antibodies and

Wpropidium iodide (Sigma), and analyzed by flow cytometry. Cell via-

Dbility was determined by flow cytometry using PI as a viability dye.

PPTDetermination of Cell Size

Cell size of viable MEFs was measured by flow cytometry. MEFswere trypsinized and resuspended in FACS buffer (phosphate-buf- Rfered saline containing 1.5% BSA and 0.02% sodium azide) con- Rtaining 1 �g/ml Hoechst 33342, and the mean FSC intensity was Ameasured for G1 phase cells using an LSR flow cytometer (Becton PDickenson).

R

Western Blots BCells were lysed in modified RIPA buffer (150 mM NaCl, 1% NP-40, L50 mM Tris-Cl [pH 8.0], 0.1% SDS) supplemented with the following (protease and phosphatase inhibitors: Na3VO4 (100 �M), PMSF (1 DmM), leupeptin (1 �g/ml), benzamidine (1 �g/ml). Cleared lysates

Bwere resolved by SDS-PAGE and transferred to nitrocellulose. BlotsSwere probed as previously described (Rathmell et al., 2003) using(the following antibodies: anti-p53 (Ab7) (Oncogene Research Prod-yucts), anti-p53 (FL-393), anti-p21 (C-19), anti-actin (I-19) (SantaBCruz Biotechnology), anti-myc tag, anti-p53 (total and phospho-zSer-15), anti-p21, and anti-His tag (Cell Signaling Technology). Rab-tbit polyclonal antibodies against total AMPK and phospho-Thr-172tAMPK have been described (Mu et al., 2001).Ba

Promoter AssaysCAMPK-dependent activation of p53-responsive promoters was as-gsessed by luciferase assay using a Dual Luciferase Assay kit (Pro-vmega). 3T3 MEFs (wild-type p53 or p53−/−) were plated at 2 × 105

ocells/35 mm dish and transfected 12–16 hr later with 1 �g CA-CAMPK plasmid or vector control, 0.75 �g promoter construct (p53-uresponsive promoter pG13 or human p21 promoter pWWP [el-Deiry

et al., 1993]), and 0.25 �g Renilla control. Luciferase activity was Cymeasured 48 hr posttransfection on a TR717 Luminometer (Tropix).

MPK Kinase Assay93T cells (5 × 106/10 cm dish) were transfected with 20 �g ofMPK plasmid (wild-type AMPK or DN-AMPK) or control plasmid,nd cells were harvested 24 hr later. Cells were lysed in 50 mMris/HCl (pH 7.4), 250 mM mannitol, 50 mM NaF, 1 mM sodiumyrophosphate, 1 mM benzamidine, 1 mM PMSF, 5 �g/ml trypsin

nhibitor, and 0.5% Triton X-100, and cleared by centrifugation, andhe supernatant fraction was precleared with protein-G agaroseeads for 1 hr. The lysate was incubated with anti-myc antibody

clone 9E10, 2 �g/2 mg protein) for 1 hr, followed by immunopreci-itation with 30 �l protein-G agarose beads (50:50 slurry) for 2 hr.he beads were washed and used in a kinase reaction in HEPES-rij buffer containing 5 mM MgCl2, 0.2 mM ATP with 0.5 �Ci/�l

γ-32P]ATP, 0.3 mM AMP, and 1 �g of recombinant protein substrateproduced as described by Canman et al. [1998]). RecombinantST-p53 fusion proteins containing the first 98 amino acids ofouse p53 (GST-p53(1-98) or mutant GST-p53 S18A(1-98)) were

sed as substrates. Recombinant GST-SAMS and GST-SAMSM fu-ion proteins were used to control for AMPK kinase activity.

erial Passage and Detection of Senescent Cellsarly-passage MEFs (<passage 2) were infected with supernatantsontaining MigCD8t, MigCD8t-DN-AMPK, or MigCD8t-CA-AMPKetrovirus, sorted 3 days postinfection, and subjected to serial pas-age according to a modified 3T3 protocol. Population doublingsPDL) per passage were calculated using the following formula:PDL = log2(CF/CI), where CI is the initial number of cells and CF is

he number of cells after 3 days of culture. Senescent cells wereetected by measuring the percent β-galactosidase-positive cells

n three separate fields.

cknowledgments

e would like to thank Wei-Xing Zong, Chi Li, Tullia Lindsten, Peterammerman, Ryan Cinalli, Celeste Simon, Brian Keith, and Connierawczyk for technical help and critical reading of this manuscript.e also thank Richard Feldman (University of Maryland), Wafik el-eiry, Donna George, Julie Leu, and Steve Reiner (University ofennsylvania) for the gift of reagents. R.G.J. is supported by aostdoctoral Fellowship from the Cancer Research Institute (USA).his work is funded in part by grants from the NIH and NCI.

eceived: September 10, 2004evised: January 27, 2005ccepted: March 25, 2005ublished: April 28, 2005

eferences

anin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C.W., Chessa,., Smorodinsky, N.I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y.

1998). Enhanced phosphorylation of p53 by ATM in response toNA damage. Science 281, 1674–1677.

ardeesy, N., Sinha, M., Hezel, A.F., Signoretti, S., Hathaway, N.A.,harpless, N.E., Loda, M., Carrasco, D.R., and DePinho, R.A.

2002). Loss of the Lkb1 tumour suppressor provokes intestinal pol-posis but resistance to transformation. Nature 419, 162–167.

arlev, N.A., Liu, L., Chehab, N.H., Mansfield, K., Harris, K.G., Hala-onetis, T.D., and Berger, S.L. (2001). Acetylation of p53 activatesranscription through recruitment of coactivators/histone acetyl-ransferases. Mol. Cell 8, 1243–1254.

en-Porath, I., and Weinberg, R.A. (2004). When cells get stressed:n integrative view of cellular senescence. J. Clin. Invest. 113, 8–13.

anman, C.E., Lim, D.S., Cimprich, K.A., Taya, Y., Tamai, K., Saka-uchi, K., Appella, E., Kastan, M.B., and Siliciano, J.D. (1998). Acti-ation of the ATM kinase by ionizing radiation and phosphorylationf p53. Science 281, 1677–1679.

arling, D. (2004). The AMP-activated protein kinase cascade—anifying system for energy control. Trends Biochem. Sci. 29, 18–24.

arlson, M., Osmond, B.C., and Botstein, D. (1981). Mutants ofeast defective in sucrose utilization. Genetics 98, 25–40.

AMPK Regulates p53-Dependent Cell-Cycle Arrest293

Chao, C., Saito, S., Anderson, C.W., Appella, E., and Xu, Y. (2000).Phosphorylation of murine p53 at ser-18 regulates the p53 re-sponses to DNA damage. Proc. Natl. Acad. Sci. USA 97, 11936–11941.

Chao, C., Hergenhahn, M., Kaeser, M.D., Wu, Z., Saito, S., Iggo, R.,Hollstein, M., Appella, E., and Xu, Y. (2003). Cell type- and pro-moter-specific roles of Ser18 phosphorylation in regulating p53 re-sponses. J. Biol. Chem. 278, 41028–41033.

Corradetti, M.N., Inoki, K., Bardeesy, N., DePinho, R.A., and Guan,K.L. (2004). Regulation of the TSC pathway by LKB1: evidence ofa molecular link between tuberous sclerosis complex and Peutz-Jeghers syndrome. Genes Dev. 18, 1533–1538.

el-Deiry, W.S., Tokino, T., Velculescu, V.E., Levy, D.B., Parsons, R.,Trent, J.M., Lin, D., Mercer, W.E., Kinzler, K.W., and Vogelstein, B.(1993). WAF1, a potential mediator of p53 tumor suppression. Cell75, 817–825.

Hardie, D.G., Carling, D., and Carlson, M. (1998). The AMP-acti-vated/SNF1 protein kinase subfamily: metabolic sensors of theeukaryotic cell? Annu. Rev. Biochem. 67, 821–855.

Haupt, Y., Maya, R., Kazaz, A., and Oren, M. (1997). Mdm2 pro-motes the rapid degradation of p53. Nature 387, 296–299.

Hawley, S.A., Boudeau, J., Reid, J.L., Mustard, K.J., Udd, L., Ma-kela, T.P., Alessi, D.R., and Hardie, D.G. (2003). Complexes betweenthe LKB1 tumor suppressor, STRADalpha/beta and MO25alpha/beta are upstream kinases in the AMP-activated protein kinasecascade. J. Biol. 2, 28.

Holley, R.W., and Kiernan, J.A. (1974). Control of the initiation ofDNA synthesis in 3T3 cells: low-molecular weight nutrients. Proc.Natl. Acad. Sci. USA 71, 2942–2945.

Imamura, K., Ogura, T., Kishimoto, A., Kaminishi, M., and Esumi, H.(2001). Cell cycle regulation via p53 phosphorylation by a 5#-AMPactivated protein kinase activator, 5-aminoimidazole- 4-carboxa-mide-1-beta-D-ribofuranoside, in a human hepatocellular carci-noma cell line. Biochem. Biophys. Res. Commun. 287, 562–567.

Inoki, K., Zhu, T., and Guan, K.L. (2003). TSC2 mediates cellularenergy response to control cell growth and survival. Cell 115,577–590.

Itahana, K., Dimri, G., and Campisi, J. (2001). Regulation of cellularsenescence by p53. Eur. J. Biochem. 268, 2784–2791.

Kato, K., Ogura, T., Kishimoto, A., Minegishi, Y., Nakajima, N., Miya-zaki, M., and Esumi, H. (2002). Critical roles of AMP-activated pro-tein kinase in constitutive tolerance of cancer cells to nutrient de-privation and tumor formation. Oncogene 21, 6082–6090.

Kaufman, R.J., Scheuner, D., Schroder, M., Shen, X., Lee, K., Liu,C.Y., and Arnold, S.M. (2002). The unfolded protein response in nu-trient sensing and differentiation. Nat. Rev. Mol. Cell Biol. 3, 411–421.

Kubbutat, M.H., Jones, S.N., and Vousden, K.H. (1997). Regulationof p53 stability by Mdm2. Nature 387, 299–303.

Lambert, P.F., Kashanchi, F., Radonovich, M.F., Shiekhattar, R., andBrady, J.N. (1998). Phosphorylation of p53 serine 15 increases in-teraction with CBP. J. Biol. Chem. 273, 33048–33053.

Luo, J., Nikolaev, A.Y., Imai, S., Chen, D., Su, F., Shiloh, A., Gua-rente, L., and Gu, W. (2001). Negative control of p53 by Sir2alphapromotes cell survival under stress. Cell 107, 137–148.

Mu, J., Brozinick, J.T., Jr., Valladares, O., Bucan, M., and Birnbaum,M.J. (2001). A role for AMP-activated protein kinase in contraction-and hypoxia-regulated glucose transport in skeletal muscle. Mol.Cell 7, 1085–1094.

Pardee, A.B. (1974). A restriction point for control of normal animalcell proliferation. Proc. Natl. Acad. Sci. USA 71, 1286–1290.

Parrinello, S., Samper, E., Krtolica, A., Goldstein, J., Melov, S., andCampisi, J. (2003). Oxygen sensitivity severely limits the replicativelifespan of murine fibroblasts. Nat. Cell Biol. 5, 741–747.

Plas, D.R., Rathmell, J.C., and Thompson, C.B. (2002). Homeostaticcontrol of lymphocyte survival: potential origins and implications.Nat. Immunol. 3, 515–521.

Rathmell, J.C., Fox, C.J., Plas, D.R., Hammerman, P.S., Cinalli,R.M., and Thompson, C.B. (2003). Akt-directed glucose metabolism

can prevent Bax conformation change and promote growth factor-independent survival. Mol. Cell. Biol. 23, 7315–7328.

Salt, I.P., Johnson, G., Ashcroft, S.J., and Hardie, D.G. (1998). AMP-activated protein kinase is activated by low glucose in cell linesderived from pancreatic beta cells, and may regulate insulin re-lease. Biochem. J. 335, 533–539.

Shamji, A.F., Nghiem, P., and Schreiber, S.L. (2003). Integration ofgrowth factor and nutrient signaling: implications for cancer biol-ogy. Mol. Cell 12, 271–280.

Shaw, R.J., Bardeesy, N., Manning, B.D., Lopez, L., Kosmatka, M.,DePinho, R.A., and Cantley, L.C. (2004). The LKB1 tumor suppres-sor negatively regulates mTOR signaling. Cancer Cell 6, 91–99.

Shieh, S.Y., Ikeda, M., Taya, Y., and Prives, C. (1997). DNA damage-induced phosphorylation of p53 alleviates inhibition by MDM2. Cell91, 325–334.

Vaziri, H., Dessain, S.K., Ng Eaton, E., Imai, S.I., Frye, R.A., Pandita,T.K., Guarente, L., and Weinberg, R.A. (2001). hSIR2(SIRT1) func-tions as an NAD-dependent p53 deacetylase. Cell 107, 149–159.

Vogelstein, B., Lane, D., and Levine, A.J. (2000). Surfing the p53network. Nature 408, 307–310.

Wilson, W.A., and Roach, P.J. (2002). Nutrient-regulated protein ki-nases in budding yeast. Cell 111, 155–158.

Woods, A., Johnstone, S.R., Dickerson, K., Leiper, F.C., Fryer, L.G.,Neumann, D., Schlattner, U., Wallimann, T., Carlson, M., and Car-ling, D. (2003). LKB1 is the upstream kinase in the AMP-activatedprotein kinase cascade. Curr. Biol. 13, 2004–2008.

Zong, W.X., Ditsworth, D., Bauer, D.E., Wang, Z.Q., and Thompson,C.B. (2004). Alkylating DNA damage stimulates a regulated form ofnecrotic cell death. Genes Dev. 18, 1272–1282.

Zwerschke, W., Mazurek, S., Stockl, P., Hutter, E., Eigenbrodt, E.,and Jansen-Durr, P. (2003). Metabolic analysis of senescent humanfibroblasts reveals a role for AMP in cellular senescence. Biochem.J. 376, 403–411.