Dysfunctional AMPK Activity, Signalling Through Mtor Andsurvival in Response to Energetic Stress in LKB1-Deﬁcient Lung Cancer

Home , MARK1, The PJs

Oncogene (2007) 26, 1616–1625 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Dysfunctional AMPK activity, signalling through mTOR andsurvival in response to energetic stress in LKB1-deﬁcient lung cancer

J Carretero1, PP Medina1, R Blanco1, L Smit2, M Tang1, G Roncador3, L Maestre3, E Conde1,4, F Lopez-Rios4, HC Clevers2 and M Sanchez-Cespedes1

1Lung Cancer Group, Spanish National Cancer Centre (CNIO), Madrid, Spain; 2Hubrecht Laboratory, Center for Biomedical Genetics, Utrecht, The Netherlands; 3Monoclonal Antibodies Unit, (CNIO), Hospital Universitario 12 de Octubre, Madrid, Spain and 4Pathology Department, Hospital Universitario 12 de Octubre, Madrid, Spain

LKB1, mutatedin Peutz–Jeghers andin sporadiclung Giardiello et al., 1987). We recently discovered that tumours, phosphorylates a group of protein kinases named LKB1 is also somatically inactivated in sporadic lung AMP-activatedprotein kinase (AMPK)-relatedkinases. adenocarcinomas (Sanchez-Cespedes et al., 2002), a Among them is included the AMPK, a sensor of cellular clear evidence that LKB1 mutations are not circum- energy status. To investigate the relevance of LKB1 in scribed to the PJS. lung carcinogenesis, we study several lung cancer cells LKB1 is a serine–threonine kinase that functions in a with andwithout LKB1-inactivating mutations. We report heterotrimeric complex with the inactive pseudokinase, that LKB1-mutant cells are deficient for AMPK activity STE20-related adaptor (STRAD) and the armadillo andrefractory to mTOR inhibition upon glucose depletion repeat scaffolding-like protein, MO25. The latter but not growth-factor deprivation. The requirement for stabilizes the binding of STRAD to LKB1 and wild-type LKB1 to properly activate AMPK is further relocalizes LKB1 from the nucleus to the cytoplasm demonstrated in genetically modified cancer cells. In (Baas et al., 2003; Boudeau et al., 2003). addition, LKB1-deficient lung primary tumours had Several studies have demonstrated that ectopic LKB1 diminishedAMPK activity, assessedby complete absence can suppress the growth of LKB1-deficient tumour cells or low level of phosphorylation of its critical substrate, (Tiainen et al., 1999; Karuman et al., 2001; Jimenez acetyl-CoA carboxylase. We also demonstrate that LKB1 et al., 2003). LKB1 also controls the regulation of wild-type cells are more resistant to cell death upon VEGF and Cox-2 expression (Ylikorkala et al., 2001; glucose withdrawal than their mutant counterparts. Rossi et al., 2002). As happens in the PJS, Lkb1 Finally, modulation of AMPK activity did not affect heterozygous mice show gastrointestinal hamartomas PI3K/AKT signalling, an advantage for the potential use and hepatocellular carcinomas (Nakau et al., 2002). of AMPK as a target for cancer therapy in LKB1 wild- Vascular abnormalities have also been reported type tumours. Thus, sustainedabrogation of cell energetic (Ylikorkala et al., 2001). checkpoint control, through alterations at key genes, LKB1 functions as a master upstream kinase of a appear to be an obligatory step in the development of some group of protein kinases named AMPK-related kinases, lung tumours. homologous to the AMP-activated protein kinase. The Oncogene (2007) 26, 1616–1625. doi:10.1038/sj.onc.1209951; most studied substrate of LKB1 is AMPK itself (Hawley published online 4 September 2006 et al., 2003; Woods et al., 2003), a sensor of cellular energy status that is activated by phosphorylation at Keywords: LKB1; AMPK; lung cancer; energy stress Thr 172 in the presence of high levels of AMP (Hardie, 2003). Several physiological and pathological stresses lead to AMPK activation, including exercise, hypoxia, ischaemia, heat shock and low glucose. Upon activation, AMPK phosphorylates multiple downstream targets to Introduction normalize adenosine triphosphate (ATP) levels, including acetyl-CoA carboxylase (ACC). Interestingly, Germ line mutations of LKB1, also known as STK11, AMPK can also phosphorylate and activate the tuberin give rise to Peutz–Jeghers syndrome (PJS) (Hemminki protein (or TSC2), the product of the tuberous sclerosis et al., 1998; Jenne et al., 1998), characterized by complex 2 gene (TSC2) (Inoki et al., 2003), and LKB1 is pigmentation anomalies, development of hamartomas required for the repression of mTOR under low ATP and increased risk of cancer (Jeghers et al., 1949; conditions in cultured cells in an AMPK- and TSC2- dependent manner (Brugarolas et al., 2004; Corradetti Correspondence: Dr M Sanchez-Cespedes, Lung Cancer Group, et al., 2004). Mutations of TSC1 and TSC2 genes are Spanish National Cancer Centre (CNIO), 28029 Madrid, Spain. E-mail: [email protected] linked to the tuberous sclerosis syndrome, which shares Received 19 June 2006; revised 21 July 2006; accepted 21 July 2006; similarities with the PJS, including the presence of published online 4 September 2006 hamartomas (Consortium TECS, 1993). TSC2 binds to Deficient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1617 the TSC1 protein, forming a functional complex that in AMPK activation. First, we generated a monoclonal inhibits the phosphorylation of S6K and 4EBP, two key antibody against LKB1. This specific antibody recog- regulators of translation (reviewed by Manning and nizes a unique band in the immunoblot at 52 kDa. Cantley, 2003). Figure 1a depicts the absence or presence of LKB1 In addition to AMPK, other LKB1 substrates include protein in LKB1-mutant and wild-type cells, respec- the MARK1-4 kinases implicated in the regulation of tively. The mutational status of other common cancer cell polarity (Lizcano et al., 2004). This is consistent genes is also indicated. The antibody probably targets with the role of LKB1 in inducing complete polarity in an epitope at the C-terminal region of the protein intestinal epithelial cells (Baas et al., 2004). because no band could be detected when ectopic SL-8 Here, we report our investigation of the biological protein (an LKB1 mutant from a Peutz–Jeghers characteristics of lung cancer cells carrying LKB1- individual lacking the C-terminal region) was intro- inactivating mutations. Our results confirm that duced in LKB1-deficient lung cancer cells (data not LKB1 regulates AMPK activity and demonstrate shown). Consistent with this, neither was the LKB1 that LKB1 deficiency impairs inhibition of mTOR band observed in the H23 cells with a nonsense activity in response to agents that increase intracellular mutation at exon 8, which is predicted to encode for a levels of AMP but not to the deprivation of growth shorter LKB1 protein that lacks the C-terminal region. factors. STRADa is required to enhance LKB1 activity and cytosolic localization, so we first tested the levels of STRADa protein in our panel of cancer cell lines. As Results depicted in Figure 1a, LKB1 parallels STRADa protein expression. Low levels of STRADa protein were LKB1-mutant lung cancer cells are deficient in AMPK detected in LKB1-deficient cells, in contrast to their activity LKB1 wild-type counterparts (similar results were The LKB1-mutant and wild-type lung cancer cells obtained by Baas and Clevers in A549, H23 and H522 (Carretero et al., 2004) were assayed for their proficiency cells; personal communication). Thus, the presence of

a KRAS ++ + + + TP53 +++ + + + P16 + + + + + + + LKB1 + + + + ERBB2 + -H23 -A427 -H441 -Calu-3 -H522 -A549 -H1299 -H2126 LKB1-

STRADα -

GAPDH-

b LKB1 mutant LKB1 wild type

2 ** 1.5 -H23 -A427 -A549 -H2126 -H441 -H522 -H1299 -Calu 3 pAMPK- 1 0.5 AMPK-

Densitometric units of 0

GAPDH- to GAPDH pAMPK relative H23- A427- A549- H522- H441- Calu3- H2126- H1299-

c (+) Glucose (-) Glucose -H23 -A427 -A549 -H522 -H1299 -Calu 3 -H23 -A427 -A549 -H522 -H1299 -Calu 3 pACC-

GAPDH-

Figure 1 Lack of LKB1, decreased STRADa and defective AMPK activation in LKB1-mutant cells. (a) Western-blot (WB) of LKB1 and STRADa according to the presence ( þ ) or absence (À)ofLKB1 mutations. (b) WB of pAMPK following treatment with AICAR (1 mM, 4 h). Densitometric analysis of the results of three independent experiments is also shown. The error bars denote s.d. (*Po0.001; t-test) (c) WB of pACC after glucose depletion for 4 h. Total AMPK and GAPDH are shown for protein loading comparison. Representative blots are shown.

Oncogene Deficient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1618 endogenous STRADa ensures the functionality of and H23 cells. Cells were transfected with a tetracycline the complex in LKB1 wild-type cells. LKB1-mutant (tet) repression expression construct and a tet-repressor- cells have decreased AMPK phosphorylation at controlled expression vector containing the entire LKB1 T172 (pAMPK) or AMPK activity (Figure 1b and c), coding sequence. Individual colonies were picked and even upon treatment with 5-aminoimidizole-4-carbo- analysed for expression of LKB1 by Western blotting. xamide riboside (AICAR), an AMPK activator As shown in Figure 2a, several clones expressed that is converted to the AMP-mimetic ZMP, or upon inducible LKB1 in both types of cell. LKB1 protein glucose starvation. AMPK activity was indicated was detectable after 12 h following induction with by the level of phosphorylation at Ser-79 of its critical doxycyclin. For subsequent studies, we selected the substrate ACC, the rate-limiting enzyme of the H42-5/H42-19 and the A91-5/A91-7 clones from the long-chain fatty acid synthesis. In contrast, LKB1 H23 and A549 cells, respectively. Immunofluorescence wild-type cells had high levels of pAMPK and (Figure 2b) confirmed that these clones had ectopic pACC upon AICAR treatment or glucose depletion. LKB1 expression in more than 90% of the cells (data AMPK and ACC phosphorylation were apparent in not shown). LKB1 wild-type cells, even before glucose depletion. A STRADa protein is almost undetectable in the faint band of pAMPK or pACC was detected in parental H23 and A549 cells, so we first tested the LKB1-mutant cells upon AICAR treatment or upon levels of STRADa upon reintroduction of wild-type glucose depletion, suggesting the existence of other LKB1. Interestingly, ectopic expression of LKB1 kinases that phosphorylate AMPK to some degree, as increased endogenous STRADa (Figure 2c), indicating has already been proposed (Woods et al., 2003; Shaw that the presence of LKB1 either prevents degradation et al., 2004a). of STRADa or promotes its gene transcription. We next evaluated whether ectopic LKB1 restored Ectopic wild-type LKB1 restores AMPK activity and AMPK activity in the presence of the appropriate increases STRADa protein levels stimuli. As shown in Figure 2c, glucose starvation lead To further evaluate the impact of LKB1 on AMPK to a considerable increase of pACC upon induction of activation, we introduced wild-type LKB1 into the A549 LKB1 with doxycyclin in the clones.

a H23 A549 b H42-5 H42-18 H42-19 A91-3 A91-5 A91-7 DOX -+-+-+ -+-+ - + LKB1-

c A91-7 6 A549-TR * DOX -+ A549-TR A91-7 A549-TR A91-7 A91-7 AICAR -+-+ -+ -+ 4 DOX -+-+ α STRAD - -pACC- 2 *

0 LKB1- -GAPDH- 1

Densitometric units of (+)Gluc (-)Gluc (+) Gluc (-) Gluc to GAPDH relative pACC GAPDH-

d siLKB1 C siLKB1 C -+ -+AICAR

LKB1- pACC- H441 H1299 GAPDH-

pACC- H1299 GAPDH- Figure 2 Modulation of AMPK activity in LKB1 genetically modiﬁed cells. (a) WB analysis of clones (H23 and A549 parental) expressing the tet-repressor-controlled LKB1 in the presence or absence of doxycyclin. (b) Immunoﬂuorescence with the anti-LKB1 antibody in the A91-7 clone. (c) On the left, WB of ectopic LKB1 and endogenous STRADa in the A91-7 clone after 24 h doxycyclin induction with or without AICAR (1 mM for 4 h). On the right, WB of pACC in LKB1-expressing clones grown with or without glucose. Cells expressing the tet-repressor alone are shown for control. Densitometric analysis of the results of three independent experiments after doxycyclin induction is also shown. The error bars denote s.d. (*Po0.001; t-test). (d) On the left, the pTER-LKB1 knocks down LKB1 expression as analysed by WB. On the right, WB of pACC in indicated cells transfected with pTER-LKB1. Representative blots are shown.

Oncogene Deficient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1619 LKB1 depletion impairs AMPK activation upon glucose Given the dual role of mTOR in integrating signalling depletion or AICAR treatment from nutrient availability and response to growth We evaluated the modifications in AMPK activation in factors, we also studied the ability of LKB1-mutant lung cancer cells after depleting wild-type LKB1 by and wild-type cells to affect mTOR activity on growth- siRNA expression. To this end, we used a construct in factor starvation. We observed that the presence of the plasmid pTER carrying a sequence that had LKB1 mutations did not affect the efficient inhibition of previously been demonstrated to downregulate LKB1 mTOR activity in the absence of growth factors, as efficiently (Baas et al., 2004). We performed a transient indicated by the decrease in pAKT and pS6 levels when co-transfection assay of pTER-LKB1 and pGFP, growth factors were removed from the culture followed by selection in a flow cytometric cell sorting. (Figure 3c). Conversely, LKB1 wild-type cells showed a As determined by Western blotting, siRNA was able to variable ability to modulate mTOR activity under such downregulate LKB1 expression efficiently (Figure 2d). circumstances. Constitutive activation of AKT is evident Compared with the control, cells transfected with the in the Calu-3 cells and was probably due to ErbB2 gene pTER-LKB1 had decreased levels of pACC following amplification. It is particularly worth noting that the treatment with AICAR. Some examples of these LKB1 wild-type H1299 cells inhibit mTOR activity on observations are also depicted in Figure 2d. growth factors but not glucose depletion.

LKB1-deﬁcient cells are refractoryto mTOR inactivation LKB1-mutant lung primaryadenocarcinomas have in response to glucose but not upon deprivation of growth reduced levels of AMPK activity factors We also investigated whether this observations were We tested the ability of LKB1-mutant and wild-type reproduced in primary tumours. We performed immu- cells to inactivate mTOR signalling in response to nostaining of LKB1 and pACC proteins in 37 lung energetic collapse. As a subrogate marker for mTOR primary adenocarcinomas, including seven tumours activity, we used phospho-235/236 S6 ribosomal protein carrying LKB1 mutations leading to truncated protein (pS6), a substrate of the translational regulator p70 (Fernandez et al., 2004). Overall, 13 (35%) of the lung ribosomal S6 kinase 1 (S6K), itself a downstream adenocarcinomas, including all LKB1 mutants, effector of mTOR. We found that mutant cells were were negative for LKB1 immunostaining. Some exam- highly resistant to mTOR inactivation, as measured by ples are depicted in Figure 4a. Levels of pACC protein the levels of pS6, following glucose depletion or AICAR were categorized as absent/low or moderate/high treatment (Figure 3a). Figure 3b shows the gradual according to the intensity of staining. According to this increase and decrease in the levels of pACC and pS6, classiﬁcation, 10 (77%) of the tumours negative and respectively, in the LKB1 wild-types Calu-3 and H441 seven (29%) positive for LKB1 immunostaining, had but not in the LKB1 mutants A549 and H23 cells, absent/low pACC levels (P ¼ 0.007; Fisher exact test) following progressive decrease in glucose concentrations. (Figure 4b).

a b Glucose Glucose Glucose Glucose -AICAR -Gfree -Control pACC- H23

pS6- ant A427

mut GAPDH- A549 Calu-3 A549 H441 H23

LKB1

H2126 c H23 A427 A549 H2126 H441 H522 H1299 Calu 3 H1299 -+ - + -+ - + -+ -+ -+ -+ pAKT- H522 pS6-

wild type H441 tAKT- 4

LKB1 Calu-3 2

pS6 0 Figure 3 LKB1-deﬁcient cells are more resistant to mTOR inactivation, as indicated by the maintenance of pS6 levels, following glucose starvation but not growth factor deprivation. (a) WB of pS6 in the presence of AICAR (1 mM) or glucose starved for 4 h. (b) WB of pACC and pS6.GAPDH for protein loading comparison. Cells were incubated in 25, 5 and 0 mM of glucose, 4 h. (c) Cells placed in serum-free medium (À) to eliminate growth factors or with 10% FCS ( þ ) for 6 h. Densitometric analysis of the results of pAKT relative to total AKT. Representative blots are shown.

Oncogene Deﬁcient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1620 a LKB1 mutant LKB1 wild type LKB1 (+) LKB1 (–)

pACC (–) pACC (+)

25 20 17 15 3 10 10 7

No of tumours 5 0 LKB1-mutants LKB1-wild types Figure 4 Immunostaining of LKB1 and pACC in primary lung adenocarcinomas. (a) Example of an LKB1-mutant and wild-type tumour, negative and positive for LKB1 and pACC immunostaining, respectively. (b) White and grey bars indicate low and high levels of pACC, respectively. Most LKB1-negative tumours had very low levels of pACC protein (Po0.05; Fisher’s exact test).

Complete activation of AMPK following glucose the A549 and the H23 cells that were reconstituted with depletion rescue cell viabilityin LKB1wild-type cells wild-type LKB1 and assayed the ability of AICAR or To test whether the responses to conditions or drugs glucose depletion to affect cell proliferation. As indi- that interfere with the mTOR signalling were dependent cated in Figure 5c, ectopic and stable wild-type LKB1 on the status of the LKB1 gene, we evaluated the effect increased the number of cells. Examples of cell viability on cell proliferation and viability of lung cancer cells following AICAR treatment in glucose-depleted cells following treatment with AICAR, rapamycin and LY- are depicted in Figure 5d. Thus, LKB1-mutant cells are 294002. We observed that cell growth inhibition induced sensitive to conditions that deplete available intracel- by each compound varied in the distinct cells but was lular energy owing to their deficient AMPK activity. independent of LKB1 gene status (examples for AICAR and rapamycin are shown in Figure 5a). We also Analysis of the functional interaction between LKB1/ compared how AICAR treatment affected cell viability AMPK and PI3K/AKT signalling in cancer cells following glucose withdrawal in the distinct tumour The presence of a negative feedback loop triggered by an cells. In contrast to their mutant counterparts, LKB1 overactivation of mTOR leads to the inhibition of wild-type cells showed significant reduced cell death insulin-responsive PI3K/AKT signalling. This inhibitory after AICAR treatment (Figure 5b). The inset shows feedback occurs as a result of an mTOR-dependent how AICAR treatment under conditions of glucose phosphorylation of IRS-1 that uncouples its interaction depletion yielded maximum pAMPK levels in the with the insulin receptor (Manning, 2004). Here, we Calu-3 cells. The effect was also evident after restoring tested whether modulation of mTOR through LKB1/ wild-type LKB1 in deficient cells. To this end, we used AMPK affects PI3K/AKT signalling, as this may have

Oncogene Deﬁcient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1621

a 100 100 H23 80 A427 80 A549 H2126 60 H441 60 H522 40 H1299 40 Growth inhibition (%) Growth

Growth inhibition (%) Growth CALU3

20 20

0 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Log [AICAR] (mM) Log [RAP] (nM) b c 14 * - + - + AICAR + + - - Glucose 12 100 a,b A91-7 H42-19 a 10 80 Calu-3 8 a 60 6 to control % cell death 4 * 40 * 2 20 % cell proliferation relative % cell proliferation 0 H23 A549 H2126 H522 H1299 CALU3 0 - DOX + DOX + DOX + AICAR

- GLUCOSE - GLUCOSE + AICAR

d Control -Glucose -Glucose + AICAR Control -Glucose -Glucose+ AICAR

H23 H522

H1299 A549

Calu3 H2126

LKB1 mutant LKB1 wild type Figure 5 Treatment with AICAR protects against glucose-depletion-triggered cell death in LKB1 wild-type but not in mutant cells. (a) Cells were counted following crystal violet staining. Each concentration was assayed in triplicate. (b) Percentage of cell death after glucose depletion (white bars) or after glucose depletion and 1 mM AICAR for 4 h (grey bars). Cell death was determined as the percentage of propidium iodide-positive cells. The inset shows a WB of pAMPK and total AMPK of the Calu-3 treated with AICAR and/or glucose starved. Data are presented as the mean7s.d. of triplicates (*Po0.05; t-test).(c) Percentage of cell proliferation relative to control in the indicated clones after glucose withdrawal before (ÀDOX) or after ( þ DOX) doxycyclin induction and after doxycyclin induction and AICAR treatment ( þ DOX þ AICAR) for 24 h (a, po0.01 versus -DOX; b, Po0.01 versus þ DOX, one- way ANOVA test). Cell proliferation was determined by crystal violet staining (d) Examples of cell death in cells treated as in B. implications for the targeting of AMPK in cancer same observations were obtained for the H42-19 clone therapy. We first examined how ectopic expression of (Supplementary Figure 1). Similarly, a progressive LKB1 in LKB1-deficient cancer cells exposed to distinct decrease in glucose concentration in the Calu-3 cells treatments influenced the levels of several components clearly activated AMPK and reduced levels of pS6, of the pathways. As expected, doxycyclin induction in while pAKT remained constant (Figure 6c). In the clones increased pAMPK activity and concomitantly Figure 6a, it is also observed that the LY-294002 reduced levels of pS6 upon AICAR activation but compound decreased pAMPK and pACC levels. In fact, did not significantly increase AKT phosphorylation increasing LY-294002 concentration lead to a clear (Figure 6a and b). Inhibition of mTOR activity by reduction of pACC and pAMPK levels in cancer cell rapamycin notably activated AKT (Figure 6a). The lines (Supplementary Figure 2). However, a nonspecific

Oncogene Deﬁcient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1622 a DOX – + b - DOX – + 2h 4h 30‘ 1h 2h 4h 30‘ 1h 2h 4h C A A LY+A R+A LY R AICAR + – + pAKT- pACC- pAMPK- pAKT- pACC- tAKT- pS6- GAPDH- GAPDH- A91-5 A91-7

c 25 10 5 2 10 Gluc. (mM) d pACC- 0 1 5 20 CC (µM) pACC- pAKT- pAKT- tAKT- pS6- pS6-

GAPDH- GAPDH- Figure 6 Modulation of AMPK activity does not signiﬁcantly change pAKT levels. (a) WB of the A91-5 clone following doxycyclin induction in untreated cells (c), with 1 mM AICAR for 4 h (a) and 50 mM LY294002 (A þ LY) or 20 nM rapamycin (A þ R) for 4 h; 50 mM LY294002 (LY); 20 nM rapamycin (R). (b) WB of the A91-7 clone following doxycyclin induction upon 1 mM of AICAR treatment at different times. (c) WB of the Calu-3 cells at increasing amounts of glucose (Gluc) for 4 h. (d) WB of the Calu-3 cells at increasing amounts of compound C (CC) for 10 h and 1 mM AICAR for 4 h. Representative blots are shown.

effect of the compound on the activity of LKB1 cannot (CaMKK) (Hawley et al., 2005), our observations be discarded. On the other hand, increasing the unequivocally show that these cells are incapable of concentration of the selective AMPK antagonist Com- fully activating AMPK when energetically stressed, pound C (CC), in the constitutive-PI3K-AKT active which is evidence in favour of a role for AMPK Calu-3 cells, markedly decreased AMPK activity while inactivation in lung carcinogenesis. AMPK becomes increasing pS6, although the levels of pAKT remained active in situations that increase the ratio of AMP/ATP, unchanged (Figure 6d). The same results were observed such as low glucose availability and hypoxia (Hardie, after inhibiting AMPK activity with CC in the clones 2003). This rarely occurs in normal cells under expressing ectopic LKB1 (data not shown). It is of physiological conditions, with the exception of muscle particular note that at 20 mM, CC substantially reduced cells. Our finding that some LKB1 wild-type lung pS6 levels, probably as a result of some unspecific effects primary tumours and cell lines have high endogenous of this compound at high concentrations. Thus, AMPK activity implies that cancer cells are exposed to modulation of AMPK activity apparently did not affect a considerably high AMP/ATP ratio, probably as a PI3K signalling, which may be a finding of relevance to consequence of their continuous cell growth. the eventual use of AMPK as a possible cancer therapy The LKB1-specific adaptor protein, STRADa, which in LKB1 wild-type tumours. activates LKB1 and translocates it to the nucleus, is essential for LKB1 catalytic activity (Baas et al., 2003). Our observations reveal that STRADa protein is present Discussion in LKB1 wild-type but is almost undetectable in LKB1- mutant cells. The fact that STRADa levels can be The presence of LKB1-inactivating mutations in cancer restored following ectopic expression of wild-type LKB1 cells attests to their relevance in cancer development. A suggests that LKB1 promotes either STRADa tran- breakthrough in the understanding of LKB1 biological scription or STRADa protein stabilization. Likewise, function was the identification of the energy sensor, STRADa protein is detectable in HeLa cells stably AMPK, as an in vitro and in vivo substrate of LKB1 expressing wild-type LKB1 but not in control cells (Hawley et al., 2003; Woods et al., 2003). In response to (Hawley et al., 2003). Moreover, ectopic expression of specific stresses, LKB1 phosphorylates and activates STRADa induced LKB1 protein stabilization in a colon AMPK, which in turn activates TSC2-inhibiting signal- epithelial cell line stably expressing LKB1 (Baas et al., ling through mTOR (Corradetti et al., 2004; Shaw et al., 2004). Thus, it appears that STRADa and LKB1 2004b). The LKB1 role in AMPK activation has been regulate each other’s protein levels, contributing to the demonstrated in cell-free systems, in some LKB1- overall activity of LKB1 in the complex and driving the deficient cancer cells such as HeLa and in mouse localization of LKB1 within the cell. It is interesting to embryonic fibroblasts (MEFs) from Lkb1-knockout note that, whereas LKB1 localization is predominantly mice. Although LKB1-mutant lung cancer cells are able cytoplasmic in the primary tumours, LKB1 is detected to phosphorylate AMPK to some extent, probably in both cytoplasm and nucleus in the LKB1-stable through the Ca2 þ /CaM-dependent protein kinase kinase transfectant cells. This implies that, in spite of stronger

Oncogene Deficient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1623 STRADa expression, the availability of STRADa and/ energetically costly processes such as DNA replication or MO25a in the transfectants still limits LKB1 and cell division. Finally, the distinct ability to activate positioning in the cytoplasm. AMPK yields differences in cell survival in response to Lack of inhibition of mTOR activity by AICAR or energy deprivation among LKB1-mutant and LKB1 glucose depletion in LKB1-deficient cells was described wild-type cells that could be exploited in cancer therapy. previously (Corradetti et al., 2004). Here, we demonstrated that deficient LKB1 activity precludes the negative regulation of signalling through mTOR in Material andmethods response to the low availability of glucose in a variety of lung cancer cells, whereas does not affect the ability to Reagents, cell lines and lung primary tumours respond to growth factors. Paradoxically, some tumour AICAR was obtained from Toronto Research Chemicals cells can tolerate glucose depletion to some extent if they (North York, Canada), Rapamycin and LY-294002 from Sigma (St Louis, MO, USA) and compound C from are able to metabolize nonglycolytic bioenergetic sub- Calbiochem (San Diego, CA, USA). The cell lines were strates through AMPK (Elstrom et al., 2004; Buzzai obtained from the American Type Culture Collection (ATCC, et al., 2005). In fact, complete activation of AMPK using Rockville, MD, USA), tested negative for mycoplasma AICAR in cells with constitutively active AKT prevented infection and grown under recommended conditions. Thirty- glucose depletion-mediated cell death (Elstrom et al., seven lung primary tumours from formalin-fixed, paraffin- 2004; Plas and Thompson, 2005). Our observations embedded tissue blocks were used for the tissue microarray demonstrate that this effect is dependent on the presence (TMA), as described previously (Conde et al., 2006). of LKB1 activity because AICAR treatment prevented cell death upon glucose depletion only in wild-type LKB1 Generation of LKB1 monoclonal antibody cells. It is not clear, how LKB1-mutant cells are more Two BALB/c mice were injected intraperitoneally (three times resistant to mTOR inhibition, in response to glucose at 15-day intervals) with 100 mg MBP–LKB1 fusion protein depletion, whereas they are more sensitive to cell death. plus Freund’s adjuvant. See technical details as Supplementary The observation that AMPK activity facilitates tumour methods (Mason et al., 1983). The LKB1 antibody is now cell survival in LKB1 wild-type tumours and that some commercially available at Abcam, Cambridge, UK (code ab15095) and Santa CruzBiotechnology (Santa Cruz,CA, lung tumours carry increased AMPK activity indicate USA) (code sc-32245). that inhibition of AMPK may be used as antineoplastic therapy in a subset of lung tumours. Supporting this Antibodies, Western blots and immunohistochemistry hypothesis, reduction of AMPK levels was shown to Anti-phospho-ACC (S79), anti-phospho-AKT (S473), anti- decrease cell viability after glucose deprivation in human AKT, anti-phospho AMPK (T172), anti-AMPK and anti- tumour cell lines (Kato et al., 2002). phospho-S6 (S235/236) were obtained from Cell Signalling There is an inhibitory feedback loop of AKT activity Technology (Beverly, MA, USA). Technical details of the due to sustained mTOR activation that triggers down- Western blots are included as Supplementary methods. regulation of the insulin receptor substrate-1 and -2 Immunohistochemical staining of LKB1 and phospho-acetyl (IRS-1 and IRS-2) mediated by p70S6K (Manning, CoA carboxylase (ACC) proteins was performed on 3-mm-thick 2004). Derived from this effect it has been shown that sections from the TMA as described previously (Conde et al., inhibition of mTOR in cancer cells and in patient 2006). Immunostaining was evaluated by a pathologist (EC). tumours causes activation of AKT kinase and induction of IRS-1 (Sun et al., 2005; O’Reilly et al., 2006). Cell culture, transfection and proliferations assays Moreover, TSC2-deficient cells are unable to activate Cells were plated in culture flasks in DMEM (Invitrogen/ AKT in response to growth factors, which may account GIBCO, Grand Island, NY, USA) medium containing 10% (v/v) foetal bovine serum (FBS), 2 mML-glutamine, 50 mg/ml for the low incidence of tumours induced by TSC2 penicillin/streptomycin and 2.5 mg/ml fungizone. Details of the deficiency in mice (Zhang et al., 2003; Manning et al., cultures and transfection assays are included as Supplementary 2005). Our data show that LKB1-deficient tumour cells methods. For the cell proliferation assays, cells were seeded on retained the ability to activate AKT and mTOR in 96-well plates at a density of 5000 cells/well and were allowed response to growth factors and that mTOR inhibition or to grow for 24 h before adding the drugs. Cell death was activation driven by the modulation of AMPK activity determined as the percentage of propidium iodide-positive does not lead to substantial changes in pAKT levels. cells. For details see the Supplementary methods section. Each These observations agree with the lack of AKT acti- experimental condition was assayed in triplicate. vation following ectopic expression of LKB1 (Corradetti et al., 2004) or with no changes in AMPK activity upon Immunocytochemical identification of LKB1 protein transformation with AKT (Buzzai et al., 2005). In For subcellular localization of LKB1, we performed an contrast, others detected AKT activation following immunocytochemical assay. A549 and H23 clones stably AICAR treatment in both LKB1-mutant and wild-type transfected with wild-type LKB1 were grown on coverslips MEFs (Shaw et al., 2004b) or other crosstalk between After 24 h, the coverslips were fixed in cold methanol (2 min at À201C). The preparation was washed in PBS and blocked with AMPK and PI3K/AKT signalling (Hahn-Windgassen 3% BSA/0.1% Triton X-100 in PBS. Monoclonal LKB1 et al., 2005; Tzatsos and Kandror, 2006). antibody was diluted 1:50 and incubated for 1 h. Excess In conclusion, our present observations confirm the antibody was removed by washing three times in blocking role of LKB1 in AMPK activation and the abrogation medium. Labelling was revealed with anti-mouse IgG-Alexa488 of energetic checkpoints in lung cancer cells to maintain (Molecular Probes, Eugene, OR, USA) at 1:200 and incubated

Oncogene Deﬁcient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1624 for 1 h. All preparations were mounted with DePeX mounting Microscopy and Flow Cytometry, Immunohistochemistry medium (BDH Laboratory Supplies, Poole, UK). and Protein Technology Units of the CNIO. The work was supported by the Spanish Ministerio de Educacio´ n Acknowledgements (SAF2005-00626). M Sanchez-Cespedes is supported by the Ramon y Cajal Programme and PP Medina by the Comunidad We thank our collaborators in the Tumour Bank Network. Autonoma de Madrid; R Blanco is supported by the Fondo de We also acknowledge the technical help of the Confocal Investigaciones Sanitarias (FIS).

References

Baas AF, Boudeau J, Sapkota GP, Smit L, Medema R, Hardie DG. (2003). The AMP-activated protein kinase Morrice NA et al. (2003). Activation of the tumour cascade: the key sensor of cellular energy status. Endocrinol- suppressor kinase LKB1 by the STE20-like pseudokinase ogy 144: 5179–5183. STRAD. EMBO J 22: 3062–3072. Hawley SA, Boudeau J, Reid JL, Mustard KJ, Udd L, Makela Baas AF, Kuipers J, van der Wel NN, Batlle E, Koerten HK, TP et al. (2003). Complexes between the LKB1 tumor Peters PJ et al. (2004). Complete polarization of single suppressor, STRADalpha/beta and MO25alpha/beta are intestinal epithelial cells upon activation of LKB1 by upstream kinases in the AMP-activated protein kinase STRAD. Cell 116: 457–466. cascade. J Biol 2: 28. Boudeau J, Baas AF, Deak M, Morrice NA, Kieloch A, Hawley SA, Pan DA, Mustard KJ, Ross L, Bain J, Schutkowski M et al. (2003). MO25alpha/beta interact Edelman AM et al. (2005). Calmodulin-dependent protein with STRADalpha/beta enhancing their ability to bind, kinase kinase-beta is an alternative upstream kinase for activate and localize LKB1 in the cytoplasm. EMBO J 22: AMP-activated protein kinase. Cell Metab 2: 9–19. 5102–5114. Hemminki A, Markie D, Tomlinson I, Avizienyte E, Roth S, Brugarolas J, Lei K, Hurley RL, Manning BD, Loukola A et al. (1998). A serine/threonine kinase gene Reiling JH, Hafen E et al. (2004). Regulation of defective in Peutz–Jeghers syndrome. Nature 18: 184–187. mTOR function in response to hypoxia by REDD1 and Inoki K, Zhu T, Guan KL. (2003). TSC2 mediates cellular the TSC1/TSC2 tumor suppressor complex. Genes Dev 18: energy response to control cell growth and survival. Cell 115: 2804–2893. 577–590. Buzzai M, Bauer DE, Jones RG, Deberardinis RJ, Jeghers H, McKusic VA, KatzKH. (1949). Generalized Hatzivassiliou G, Elstrom RL et al. (2005). The glucose intestinal polyposis and melanin spots of the oral mucosa, dependence of Akt-transformed cells can be reversed by lip, and digits: a syndrome of diagnostic significance. N Engl pharmacologic activation of fatty acid beta-oxidation. J Med 241: 1031–1036. Oncogene 24: 4165–4173. Jenne DE, Reimann H, Nezu J, Friedel W, Loff S, Carretero J, Medina PP, Pio R, Montuenga LM, Sanchez- Jeschke R et al. (1998). Peutz–Jeghers syndrome is caused Cespedes M. (2004). Novel and natural knockout lung by mutations in a novel serine threonine kinase. Nat Genet cancer cell lines for the LKB1/STK11 tumor suppressor 18: 38–44. gene. Oncogene 23: 4037–4040. Jimenez AI, Fernandez P, Dominguez O, Dopazo A, Sanchez- Conde E, Angulo B, Tang M, Morente M, Torres-Lanzas J, Cespedes M. (2003). Growth and molecular profile of lung Lopez-Encuentra A et al. (2006). Molecular context of the cancer cells expressing ectopic LKB1: down-regulation of epidermal growth factor receptor mutations: evidence for the phosphatidylinositol 30-phosphate kinase/PTEN path- the activation of mTOR/S6K signaling. Clin Cancer Res 12: way. Cancer Res 63: 1382–1388. 710–717. Karuman P, Gozani O, Odze RD, Zhou XC, Zhu H, Shaw R Consortium TECS. (1993). Identification and characterization et al. (2001). The Peutz–Jeghers gene product LKB1 is of the tuberous sclerosis gene 2 on chromosome 16. Cell 75: a mediator of p53-dependent cell death. Mol Cell 7: 1305–1315. 1307–1319. Corradetti MN, Inoki K, Bardeesy N, DePinho RA, Kato K, Ogura T, Kishimoto A, Minegishi Y, Nakajima N, Guan KL. (2004). Regulation of the TSC pathway by Miyazaki M et al. (2002). Critical roles of AMP-activated LKB1: evidence of a molecular link between tuberous protein kinase in constitutive tolerance of cancer cells to sclerosis complex and Peutz–Jeghers syndrome. Genes Dev nutrient deprivation and tumor formation. Oncogene 21: 18: 1533–1538. 6082–6090. Elstrom RL, Bauer DE, Buzzai M, Karnauskas R, Harris MH, Lizcano JM, Goransson O, Toth R, Deak M, Morrice NA, Plas DR et al. (2004). Akt stimulates aerobic glycolysis in Boudeau J et al. (2004). LKB1 is a master kinase that cancer cells. Cancer Res 64: 3892–3899. activates 13 kinases of the AMPK subfamily, including FernandezP, Carretero J, Medina PP, JimenezAI, Rodriguez- MARK/PAR-1. EMBO J 23: 833–843. Perales S, PazMF et al. (2004). Distinctive gene expression Manning BD. (2004). Balancing AKT with S6K: implication of human lung adenocarcinomas carrying LKB1 mutations. for both metabolic diseases and tumorigenesis. J Cell Biol Oncogene 23: 5084–5091. 167: 399–403. Giardiello FM, Welsh SB, Hamilton SR, Offerhaus GJ, Manning BD, Cantley LC. (2003). United at last: the Gittelsohn AM, Booker SV et al. (1987). Increased risk of tuberous sclerosis complex gene products connect the cancer in the Peutz–Jeghers syndrome. N Engl J Med 316: phosphoinositide 3-kinase/Akt pathway to mammalian 1511–1514. target of rapamycin (mTOR) signalling. Biochem Soc Trans Hahn-Windgassen A, Nogueira V, Chen CC, 31: 573–578. Skeen JE, Sonenberg N, Hay N. (2005). Akt activates Manning BD, Logsdon MN, Lipovsky AI, Abbott D, the mammalian target of rapamycin by regulating cellu- Kwiatkowski DJ, Cantley LW. (2005). Feedback inhibition lar ATP level and AMPK activity. J Biol Chem 280: of Akt signaling limits the growth of tumors lacking Tsc2. 32081–33209. Genes & Dev 19: 1773–1778.

Oncogene Deﬁcient AMPK and energy stress in LKB1-mutant cells J Carretero et al 1625 Mason DY, Cordell JL, Pulford KAF. (1983). Techniques kinase directly activates AMP-activated kinase and regulates in Immunocytochemistry. London Academic Press: London, apoptosis in response to energy stress. Proc Natl Acad Sci p 17. USA 101: 3329–3335a. Nakau M, Miyosh H, Seldin MF, Imamura M, Oshima M, Sun SY, Rosenberg LM, Wang X, Zhou Z, Yue P, Fu H et al. Taketo MM. (2002). Hepatocellular carcinoma caused by (2005). Activation of Akt and eIF4E survival pathways loss of heterozygosity in lkb1 gene knockout mice. Cancer by rapamycin-mediated mammalian target of rapamycin Res 62: 4549–4553. inhibition. Cancer Res 65: 7052–7058. O’Reilly KE, Rojo F, She QB, Solit D, Mills GB, Smith D Tiainen M, Ylikorkala A, Makela TP. (1999). Growth et al. (2006). mTOR inhibition induces upstream receptor suppression by Lkb1 is mediated by a G(1) cell cycle arrest. tyrosine kinase signaling and activates Akt. Cancer Res 66: Proc Natl Acad Sci USA 96: 9248–9251. 1500–1508. Tzatsos A, Kandror KV. (2006). Nutrients suppress phospha- Plas DR, Thompson CB. (2005). Akt-dependent transforma- tidylinositol 3-kinase/Akt signaling via raptor-dependent tion: there is more to growth than just surviving. Oncogene mTOR-mediated insulin receptor substrate 1 phosphoryla- 24: 7435–7442. tion. Mol Cell Biol 26: 63–76. Rossi DJ, Ylikorkala A, Korsisaari N, Salovaara R, Luukko K, Woods A, Johnstone SR, Dickerson K, Leiper FC, Fryer LG, Launonen V et al. (2002). Induction of cyclooxygenase-2 in a Neumann D et al. (2003). LKB1 is the upstream kinase in mouse model of Peutz–Jeghers polyposis. Proc Natl Acad Sci the AMP-activated protein kinase cascade. Curr Biol 13: USA 99: 12327–12332. 2004–2008. Sanchez-Cespedes M, Parrella P, Esteller M, Nomoto S, Ylikorkala A, Rossi DJ, Korsisaari N, Luukko K, Alitalo K, Trink B, Engles JM et al. (2002). Inactivation of LKB1/ Henkemeyer M et al. (2001). Vascular abnormalities and STK11 is a common event in adenocarcinomas of the lung. deregulation of VEGF in Lkb1-deﬁcient mice. Science 293: Cancer Res 62: 3659–3662. 1323–1326. Shaw RJ, Bardeesy N, Manning BD, LopezL, Kosmatka M, Zhang H, Cicchetti G, Onda H, Koon HB, DePinho RA et al. (2004a). The LKB1 tumor suppressor Asrican K, Bajraszewski N et al. (2003). Loss of Tsc1/Tsc2 negatively regulates mTOR signaling. Cancer Cell 6: 91–99b. activates mTOR and disrupts PI3K-Akt signaling Shaw RJ, Kosmatka M, Bardeesy N, Hurley RL, Witters LA, through downregulation of PDGFR. Clin Invest 112: DePinho RA et al. (2004b). The tumor suppressor LKB1 1223–1233.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene