The LKB1 tumour suppressor phosphorylates and activates AMPK (AMP-activated protein kinase) when cellular energy levels are low, thereby suppressing growth through multiple pathways, including inhibiting the mTORC1 (mammalian target of rapamycin complex 1) kinase that is activated in the majority of human cancers. Blood glucose-lowering Type 2 diabetes drugs also induce LKB1 to activate AMPK, indicating that these compounds could be used to suppress growth of tumour cells. In the present study, we investigated the importance of the LKB1–AMPK pathway in regulating tumorigenesis in mice resulting from deficiency of the PTEN (phosphatase and tensin homologue deleted on chromosome 10) tumour suppressor, which drives cell growth through overactivation of the Akt and mTOR (mammalian target of rapamycin) kinases. We demonstrate that inhibition of AMPK resulting from a hypomorphic mutation that decreases LKB1 expression does not lead to tumorigenesis on its own, but markedly accelerates tumour development in PTEN+/− mice. In contrast, activating the AMPK pathway by administration of metformin, phenformin or A-769662 to PTEN+/− mice significantly delayed tumour onset. We demonstrate that LKB1 is required for activators of AMPK to inhibit mTORC1 signalling as well as cell growth in PTEN-deficient cells. Our findings highlight, using an animal model relevant to understanding human cancer, the vital role that the LKB1–AMPK pathway plays in suppressing tumorigenesis resulting from loss of the PTEN tumour suppressor. They also suggest that pharmacological inhibition of LKB1 and/or AMPK would be undesirable, at least for the treatment of cancers in which the mTORC1 pathway is activated. Most importantly, our results demonstrate the potential of AMPK activators, such as clinically approved metformin, as anticancer agents, which will suppress tumour development by triggering a physiological signalling pathway that potently inhibits cell growth.
- AMP-activated protein kinase (AMPK)
- mammalian target of rapamycin (mTOR)
- phosphatase and tensin homologue deleted on chromosome 10 (PTEN)
The majority of cancers possess mutations that stimulate proliferation, growth and survival by activating the PI3K (phosphoinositide 3-kinase)–Akt–mTOR (mammalian target of rapamycin) signalling pathways [1,2]. A frequent mutation leading to the activation of this pathway results in loss of the PTEN (phosphatase and tensin homologue deleted on chromosome 10) lipid phosphatase, which breaks down the PtdIns(3,4,5)P3 second messenger formed following activation of PI3K [3,4]. Heterozygous PTEN+/− mice spontaneously develop a variety of tumours [5–11] and tumorigenesis in these mice is suppressed by reducing expression of PDK1 (phosphoinositide-dependent kinase 1), an upstream activator of Akt  or by deletion of the Akt1 isoform . Akt drives cell proliferation, at least in part, through its ability to activate mTORC1 (mTOR complex 1), via phosphorylation of TSC (tuberous sclerosis complex) 2 and PRAS40 (proline-rich Akt substrate of 40 kDa) (reviewed in [14,15]). Proliferation and growth of PTEN-deficient tumours is inhibited by treatment with the mTORC1 inhibitor rapamycin or by knockdown of Raptor (regulatory associated protein of mTOR) [16,17]. Furthermore, overexpression of the Rheb GTPase, an activator of mTORC1, stimulates growth and proliferation of Akt1/Akt2-deficient cells .
There has been much interest in dissecting the role of other pathways that modulate mTOR signalling. One of these comprises AMPK (AMP-activated protein kinase), which is phosphorylated and activated by the LKB1 tumour suppressor protein kinase when cellular energy levels are low [18,19]. Under these conditions, 5′-AMP levels are elevated, and 5′-AMP binds to AMPK, inducing a conformational change. This protects AMPK from dephosphorylation by an as yet unidentified phosphatase, thereby promoting its phosphorylation and activation by LKB1 [20,21]. Once activated, AMPK functions to restore energy levels by phosphorylating key cellular substrates, leading to the stimulation of ATP-generating pathways. Activation of AMPK also inhibits energy-consuming processes such as protein synthesis and growth. These effects are mediated in part through inhibition of mTORC1 resulting from phosphorylation of TSC2 by AMPK .
Pathways controlled by mTORC1 are activated in cells and tumours possessing reduced AMPK activity as a result of loss of LKB1 [2,23]. Moreover, low-energy stress or oral hypoglycaemic anti-Type 2 diabetes drugs, such as metformin or phenformin that activate AMPK, fail to inactivate mTORC1 or inhibit cell growth and proliferation in LKB1-deficient cells [2,23,24]. In addition, humans lacking one allele of the LKB1 gene develop Peutz–Jeghers cancer syndrome, which bears a phenotypic resemblance to other conditions resulting from inappropriate activation of mTORC1, including Cowden's syndrome (loss of PTEN) or tuberous sclerosis (loss of TSC1 or TSC2) [25,26]. LKB1 is also mutated in certain sporadic human cancers and with particularly high frequency in lung cancer [27,28].
These findings suggest that inhibition of the LKB1–AMPK signalling axis in cancer is undesirable, as it would be expected to promote tumorigenesis. However, it has also been reported that LKB1-deficient fibroblasts undergo a greater degree of cell death under energy-deprived conditions . Thus cancer cells lacking the ability to control their energy levels through the AMPK pathway may be more susceptible to stresses that lower energy levels. Furthermore, inhibition of AMPK has recently been shown to markedly sensitize a variety of cancer cells to cisplatin, a chemotherapeutic agent . Thus inhibition of the LKB1–AMPK pathway might actually be beneficial for the treatment of cancer, promoting apoptosis of cancer cells by interfering with their ability to manage energy stress and by increasing the effectiveness of cytotoxic compounds.
In the present study, we explored the effects of lowering the expression of LKB1 and stimulating AMPK on tumour development in PTEN+/− mice. Our findings demonstrate the crucial role that the LKB1–AMPK signalling network plays in controlling the tumorigenesis that results from the loss of PTEN. The data suggest that the use of clinically approved AMPK activators, such as metformin, may be useful for the treatment or prevention of cancer in humans. Our observations also demonstrate that inhibiting LKB1-dependent pathways for the treatment of tumours in which the mTOR pathway is activated is highly undesirable.
Complete™ protease inhibitor cocktail tablets were from Roche. Taq DNA polymerase and pre-cast SDS/polyacrylamide Bis-Tris gels were from Invitrogen. Protein G–Sepharose and [γ-32P]ATP were purchased from Amersham Bioscience. Phosphocellulose P81 paper was from Whatman. Triton X-100 and dimethyl pimelimidate were from Sigma. AICAR (5-aminoimidazole-4-carboxamide 1-β-D-ribofuranoside) was from Toronto Research Chemicals. Phenformin hydrochloride and 1,1-dimethylbiguanide hydrochloride (metformin) were from Sigma–Aldrich. A-769662 was synthesized as described previously .
Anti-AMPKα1 was raised in sheep against the peptide CTSPPDSFLDDHHLTR (residues 344–358 of rat AMPKα1, with an additional cysteine residue for coupling), anti-LKB1 for immunoprecipitation and immunoblotting was raised in sheep against the peptide TFIHRIDSTEVIYQPR (residues 24–39 of mouse LKB1). Antibodies against LKB1 (#3050) for immunohistochemistry, phospho-AMPKα (Thr172) (#2535), phospho-ACC (acetyl-CoA carboxylase) (Ser79) (#3662), pan-ACC (#3661), S6 ribosomal protein (#2212), phospho-S6 ribosomal protein (Ser235) (#2211), S6K1 (S6 kinase 1) (#9202), phospho-S6K1 (Thr389) (#9205), phospho-Akt (Thr308) (#9275), phospho-Akt (Ser473) (#9271), FOXO1 (Forkhead box O1) (#9462) and PTEN (#9559) were purchased from Cell Signaling Technology. The anti-Akt1 antibody (#05-591) was from Upstate Biotechnology. The anti-Ki67 antibody (#VP-K452) was from Vector Laboratories. Secondary antibodies coupled to horseradish peroxidase were from Pierce.
Mice breeding, genotype analysis and tumour analysis
All animal studies were approved by the University of Dundee Ethics Committee and performed under a U.K. Home Office project licence. The generation and genotyping of the PTEN+/− mice  and the LKB1 hypomorphic mice  have been described previously. The parental LKB1fl/+ and PTEN+/− used for these experiments had been backcrossed with C57BL/6J for over seven generations before initiating the crosses for the present study. Littermates with different genotypes of LKB1 and PTEN were derived as described in Figure 1(A) and maintained under standard husbandry conditions for a period of up to 14 months of age. During this period, mice were monitored weekly for tumour development. According to our U.K. Home Office licence, any animal that displayed an obvious external tumour of over 1.44 cm2 or showed signs of sickness was killed and subjected to necropsy and pathological analysis after tissue fixation in 10% formalin as described previously .
Tumour slices were generated as described previously . Immunohistochemical staining was performed with automated procedures at the University of Dundee Tissue Bank at Ninewells Hospital. The antibodies used in the staining were as described above. Antibody staining was detected using the rabbit Vectastain ABC kit (Vector Laboratories). Sections were viewed on a Nikon Eclipse 600 microscope, and images were captured on a Nikon DXM1200 digital camera with EclipseNet software.
Assay of LKB1 and AMPK
LKB1 and AMPK activities were assayed as described previously . In brief, 500 μg of different mouse tissues and ES cell (embryonic stem cell) lysates were used to immunoprecipitate LKB1, and 50 μg of lysates were used for AMPKα1 immunoprecipitation. Kinase activities were measured employing the LKBtide peptide  for LKB1 or AMARA peptide for AMPKα1 . One m-unit of activity was defined as that which catalysed the incorporation of 1 pmol of 32P into the substrate per min.
Generation of LKB1/PTEN-knockout ES cells
Female LKB1fl/+PTEN+/− mice were induced to superovulate by the injection of PMSG (pregnant mare serum gonadotropin). This was followed 48 h later by the injection of HCG (human chorionic gonadotropin). These mice were then mated with male LKB1fl/+PTEN+/− mice. Blastocysts were removed at 2.5 days post-coitus and cultured on 24-well plates on a feeder layer of mitotically inactivated primary mouse embryonic fibroblasts for 1–2 weeks to allow the ES cells to grow. Wells were trypsinized, and 80% of the aliquot was frozen in two batches, while the remaining 20% was used to grow cells for DNA preparation. Cells were genotyped by PCR for PTEN  or LKB1 . In order to generate LKB1−/− ES cell lines, LKB1fl/flPTEN−/−, LKB1fl/flPTEN+/+ ES cells obtained as above were subjected to Cre-mediated deletion to remove the floxed LKB1 allele as described previously . After transfection by electroporation with 30 μg of pMC-CrePuro, the cells were seeded on to feeder layers under puromycin (1 μg/ml) selection, and colonies were picked in duplicate into 96-well tissue culture plates. One of each selected colonies was subjected to G418 selection (0.2 mg/ml). The sensitive colonies were expanded and analysed by PCR and immunoblotting to confirm the deletion of LKB1 alleles. The LKB1+/+PTEN−/− ES cell line was generously provided to us by Dr Pier P. Pandolfi (Harvard Medical School, Boston, MA, U.S.A.) .
Cell culture, stimulation and cell lysis
LKB1+/+PTEN+/+, LKB1+/+PTEN−/−, LKB1−/−PTEN−/− and LKB1−/−PTEN+/+ ES cells were grown on gelatinized tissue culture plastic in DMEM (Dulbecco's modified Eagle's medium) (Sigma) containing 10% (v/v) FBS (fetal bovine serum) (Hyclone) supplemented with 1000 mg/l glucose, 0.1 mM non-essential amino acids, antibiotics (100 units of penicillin G and 100 μg/ml streptomycin), antimycotic (1 μg/ml ciproxin infusion), 2 mM L-glutamine, 0.1 mM 2-mercaptoethanol and 25 ng/ml murine leukaemia inhibitory factor. The ES cells were cultured to 80% confluence on 10 cm diameter dishes and then stimulated with the indicated agonists as described in the Figure legends. The cells were lysed in 0.5 ml of ice-cold lysis buffer [50 mM Tris/HCl, pH 7.5, 1 mM EGTA, 1 mM EDTA, 1% (w/v) Triton X-100, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose, 0.1% (v/v) 2-mercaptoethanol and Complete™ protease inhibitor cocktail (one tablet per 50 ml)] and centrifuged at 12000 g for 15 min at 4 °C. The supernatants were divided into aliquots, frozen in liquid nitrogen, and stored at −80 °C until use. Protein concentrations were determined by the Bradford method using BSA as a standard.
Cell growth assay
ES cell lines were seeded in 12-well plates (30000 cells per well) in DMEM containing 10% FBS in the presence or absence metformin or phenformin. The cell medium was changed every 24 h with freshly dissolved metformin or phenformin added. After 72 h, the cells were washed once with 1× PBS, and fixed in 4% (v/v) paraformaldehyde in PBS for 15 min. After washing once with water, the cells were stained with 0.1% Crystal Violet in 10% ethanol for 20 min and washed three times with water. Crystal Violet was extracted from cells with 0.5 ml of 10% (v/v) ethanoic (acetic) acid for 20 min. The eluate was then diluted 1:10 in water, and absorbance at 590 nm was quantified.
Administration of mice with AMPK activators
The indicated strains of mice were administered with different AMPK activators in their drinking water (metformin, 300 mg/kg of body weight per day; phenformin, 300 mg/kg of body weight per day; A-769662, 30 mg/kg of body weight per day). Drugs were dissolved from powder freshly every day into the drinking water, taking into account the mouse body weight and the volume of water the animal consumed each day. No other water was available for the mice to drink, and the volume of water consumed by each mouse per day was noted and used to calculate the dose of drugs required. A-769662 was dissolved in DMSO to 1 M, before diluting into drinking water. The LKB1fl/+PTEN+/− mice were maintained under the standard husbandry conditions described above and monitored weekly for tumour development as well as body weight and food intake. Blood glucose levels were measured using the Ascensia Breeze blood glucose meter (Bayer), blood lactic acid was measured using the Accutrend Lactate Meter (Roche) and serum insulin was measured using an ELISA Kit (Crystal Chem).
Generation of LKB1 hypomorphic PTEN+/− mice
To investigate how suppression of the LKB1 pathway affected tumorigenesis, we crossed PTEN+/− mice which spontaneously develop tumours in multiple tissues after ∼8 months of age [5–11], with LKB1fl/fl hypomorphic mice in which the expression of LKB1 is reduced 5–10-fold in most tissues analysed . Employing crosses described in Figure 1(A), we generated littermate strains of mice expressing normal levels of LKB1 and PTEN (LKB1+/+PTEN+/+), reduced LKB1–normal PTEN (LKB1fl/fl-PTEN+/+), normal LKB1–deficient PTEN (LKB1+/+PTEN+/−), moderately reduced LKB1–deficient PTEN (LKB1fl/+PTEN+/−) and reduced LKB1–deficient PTEN (LKB1fl/flPTEN+/−). All mouse strains were of normal size and, apart from the LKB1fl/flPTEN+/− mice that were born at ∼50% of the expected number, the other mouse strains were recovered in the expected Mendelian frequency (Figure 1B, cross 2). In the PTEN+/− mice, the level of the PTEN protein in liver (Figure 1C), heart, spleen and intestine (Supplementary Figure S1A at http://www.BiochemJ.org/bj/412/bj4120211add.htm) was about half that observed in tissues derived from PTEN+/+ mice. The LKB1fl/fl mice had markedly reduced levels of LKB1 protein and activity in all tissues examined (Figure 1C and Supplementary Figure S1A). In tissues derived from LKB1fl/+ mice, intermediate levels of LKB1 protein and activity were observed. In the LKB1fl/fl hypomorphic mice, AMPK activity and its phosphorylation at the residue phosphorylated by LKB1 (Thr172), as well as phosphorylation of the AMPK substrate ACC, were markedly reduced in liver (Figure 1D), spleen and intestine (Supplementary Figure S1B), establishing that reduction of LKB1 expression resulted in inhibition of AMPK activity in vivo. Consistent with the decreased AMPK activity, LKB1fl/fl hypomorphic mice displayed increased mTORC1 activity, assessed by enhanced phosphorylation of S6K1 at Thr389 and the ribosomal S6 protein at Ser235 (Figure 1D and Supplementary Figure S1B). The LKB1fl/+ mice with only moderately reduced LKB1 protein possessed levels of AMPK activity and phosphorylation of ACC similar to those of wild-type mice (Figure 1D and Supplementary Figure S1B).
Analysis of tumour formation
Groups of ∼30 mice of each genotype were monitored for tumours and were killed when they exhibited large externally palpable tumours, reduced body weight, became unwell or reached 14 months of age. A necroscopy was performed, and the tissues were fixed in 10% formalin and subjected to detailed histopathological examination (Table 1). By 14 months of age, the LKB1 hypomorphic mice not deficient in PTEN (LKB1fl/flPTEN+/+) displayed no significant tumours (Figure 2A and Table 1). Therefore reduction in LKB1 expression alone is not sufficient to induce tumorigenesis. Consistent with previous work [5–11], tumour development in LKB1+/+PTEN+/− mice was observed at ∼7 months of age, and, by 14 months, ∼70% had one or more tumours. Strikingly, PTEN+/− mice with moderately reduced LKB1 expression (LKB1fl/+PTEN+/−), developed tumours earlier, commencing at 4 months, and, by 9 months, ∼70% of mice developed tumours (Figure 2A). Consistent with the reduction of LKB1 levels accelerating tumorigenesis in PTEN+/− mice, animals with the most pronounced reduction of LKB1 expression (LKB1fl/flPTEN+/−), developed tumours at 3 months, and, by 6 months, ∼70% of animals displayed tumours (Figure 2A).
Pathological analysis indicated that the types and morphology of tumours arising in PTEN+/− mice with normal and reduced levels of LKB1 were broadly similar (Table 1). Representative histopathological sections of these tumours are shown in Figure 2(B). The tumours comprised lymphoma, intestinal polyps, prostate carcinoma, breast adenocarcinoma, phaeochromocytoma and prostatic intraepithelial neoplasia (Figure 2B). The proportion of mice displaying lymphoma, intestinal polyps, phaeochromocytoma and prostate carcinoma were moderately (<2-fold) higher in the LKB1 hypomorphic PTEN+/− mice compared with the PTEN+/− mice expressing normal levels of LKB1 (Table 1). The intestinal polyps were found in both the small and large intestine, but were more numerous in the small intestine. In some animals, single polyps were observed, but, in others, multiple polyps, up to ten, were found. They ranged in size from small pedunculated polyps projecting into, but not obstructing, the lumen and measuring less than 1 mm in diameter to very large extensive sessile growths occupying more than 10 mm of bowel wall and obstructing the lumen (Figure 2B). The larger and more numerous polyps were seen in animals with the LKB1fl/flPTEN+/− genotype. Architecturally, the polyps were biphasic complex growths with distorted glands and a cellular stroma (Figure 2B and Supplementary Figure S2 at http://www.BiochemJ.org/bj/412/bj4120211add.htm). The gland epithelium was composed of crypts, absorptive epithelium mostly on the surface, numerous mucin-secreting cells and Paneth cells. There was no significant atypia, and the mitotic figures were mostly within the crypt like population. These surface epithelial cells, but not the stromal component, stained with a broad-spectrum anti-cytokeratin antibody. The stromal component had the morphology of smooth muscle, with spindle-shaped cells with abundant eosinophilic cytoplasm and cigar-shaped nuclei lying within a loose extracellular matrix. This stromal component stained in immunocytochemistry experiments with an antibody to the α-smooth muscle isoform of actin and the intermediate filament desmin confirming the smooth muscle phenotype. As with the epithelium, there was no atypia, and mitoses were infrequent. We found that 15–30% of LKB1+/+PTEN+/− and LKB1fl/flPTEN+/− animals also developed a number of minor pre-neoplastic lesions, with the most common being prostatic intraepithelial neoplasia and intratubular germ cell neoplasia (Table 1).
We examined intestinal polyps from the LKB1+/+PTEN+/− and LKB1fl/flPTEN+/− animals and found that, in polyps of both genotypes, FOXO1 was predominantly in the cytosol (Figure 2C), indicating that Akt was activated. This was confirmed by the finding of high levels of phosphorylation of Akt at Thr308 and phosphorylation of the S6 protein at Ser235 predominantly in the epithelial cells of the polyps from both genotypes (Figure 2C). Comparable levels of Ki67 staining were also seen in the epithelium and smooth muscle cells of the polyps from LKB1+/+PTEN+/− and LKB1fl/flPTEN+/−animals. The staining with Ki67 showed regional variation in the epithelium, being most marked in the crypts and sparse on the surface epithelium. The staining of the smooth muscle cells was distributed more randomly. Analogous analysis was undertaken for the lymphoma and phaeochromocytoma, and no significant differences in the staining patterns were detected between the tumours arising in the control and experimental mice (results not shown).
Role of the LKB1–AMPK pathway in modulating mTOR signalling and growth of ES cells
We next explored the role that the LKB1–AMPK pathway plays in regulating Akt and mTOR utilizing single LKB1+/+PTEN−/− or LKB1−/−PTEN+/+, as well as double LKB1−/−PTEN−/− ES cells, generated as described in the Experimental section. The PTEN-deficient (LKB1+/+PTEN−/− and LKB1−/−PTEN−/−) ES cells displayed elevated levels of phosphorylation of Akt and mTORC1 activity, as judged by increased phosphorylation of S6K1 and S6 protein (Figure 3A). LKB1−/−PTEN+/+ cells possessed low levels of Akt phosphorylation, but elevated levels of phosphorylation of S6K1 at Thr389, similar to those observed in LKB1+/+PTEN−/− cells (Figure 3A). We also investigated how treatment of these ES cells with activators of AMPK affected Akt and mTORC1. Treatment of wild-type LKB1+/+PTEN+/+ or LKB1+/+PTEN−/− cells with the AMP mimetic AICAR or either of the anti-Type 2 diabetes drugs metformin or phenformin stimulated phosphorylation of AMPK as well as ACC. There was also decreased phosphorylation of S6K1 as well as its substrate S6 protein, suggestive of inhibition of mTORC1 (Figure 3B). In contrast, treatment of LKB1-deficient (LKB1−/−PTEN+/+ or LKB1−/−PTEN−/−) cells with AMPK activators failed to induce phosphorylation of AMPK or ACC or inhibit phosphorylation of S6K1 or S6 protein, indicating that, in LKB1-deficient cells, AMPK was not activated, and neither was mTORC1 inhibited (Figure 3B). Treatment of ES cells with AMPK activators did not inhibit phosphorylation of Akt at Thr308 (Figure 3B).
We also analysed the effect of metformin and phenformin on the growth of ES cell lines. Consistent with studies performed in other cell lines , the growth of wild-type LKB1+/+PTEN+/+ as well as LKB1+/+PTEN−/− ES cells was significantly inhibited in a dose-dependent manner by metformin (Figure 3C) or phenformin (Figure 3D). In contrast, the growth of LKB1-deficient LKB1−/−PTEN+/+ or LKB1−/−PTEN−/− ES cells, was inhibited to a markedly lower extent by metformin or phenformin (Figures 3C and 3D). These observations support the notion that LKB1 is required for the growth-suppressing effects of metformin or phenformin.
Effect of AMPK-activating drugs on PTEN tumorigenesis
The finding that a reduction in LKB1 expression accelerated tumour development in PTEN+/− animals and that AMPK is required to inhibit mTORC1 and cell growth in PTEN-deficient cells led us to hypothesize that activation of AMPK would suppress tumorigenesis in mice. To test this, we studied how administration of PTEN+/− mice with AMPK-activating agents affected tumour development. We used metformin or phenformin that activates AMPK indirectly, most likely by decreasing ATP regeneration through the inhibition of mitochondrial respiration (reviewed in [18,19]). We also used a recently formulated compound A-769662, which activates AMPK both allosterically and by inhibiting dephosphorylation of AMPK at Thr172, the residue phosphorylated by LKB1 [20,35,36]. Control experiments indicated that mice administered with any of these drugs over a 2-week period possessed significantly elevated levels of AMPK activity/phosphorylation and ACC phosphorylation in liver (Figure 4A), spleen (Figure 4B) and intestine (Figure 4C). A-769662 stimulated AMPK activity to a similar extent in all tissues as observed following an intraperitoneal injection of the AMP mimetic drug AICAR. Phenformin and metformin also enhanced AMPK activity in the liver, but were less potent than A-769662 or AICAR at stimulating AMPK in the spleen or intestine (Figure 4).
To investigate the effects of AMPK activators on tumour development, we utilized LKB1fl/+PTEN+/− animals, as these develop tumours within 4 months. Control studies revealed that administering LKB1fl/+ mice with phenformin, metformin or A-769662 induced a similar level of AMPK activation as observed in LKB1+/+ mouse tissues (Figure 4). We also found that AMPK activators induced similar activation of AMPK in LKB1fl/+ and LKB1+/+ ES cells (Supplementary Figure S3 at http://www.BiochemJ.org/bj/412/bj4120211add.htm). Thus the slight reduction in LKB1 levels in LKB1fl/+ animals does not significantly interfere with the ability of phenformin, metformin or A-769662 to stimulate AMPK. Groups of ten LKB1fl/+PTEN+/− mice from 4 weeks of age were either given no drug or prescribed one of the AMPK-activating drugs. The mice tolerated long-term administration of AMPK-activating compounds, and their body weight, levels of blood glucose, insulin and lactate, and food intake were not affected significantly (Supplementary Figure S4 at http://www.BiochemJ.org/bj/412/bj4120211add.htm). In the control group given no drug, tumour formation commenced as expected by 4 months of age, and 100% of mice developed tumours by 8 months (Figure 5A). Strikingly, tumour formation was markedly delayed in all mice administered one of the AMPK activators. By 6 months of age, none of the animals on phenformin or A-769662 had developed tumours at a stage when 60% of the control animals had tumours. Development of tumours in mice treated with phenformin and A-769662 started at 7 months of age. By 10 months of age, 60% of mice on phenformin and 80% on A-769662 possessed tumours. After 12 months, 20% of animals treated with phenformin or 10% with A-769662 displayed no tumours (Figure 5A). Consistent with metformin being a weaker activator of AMPK than phenformin or A-769662, we observed that metformin was less effective at inhibiting tumorigenesis. For mice prescribed metformin, tumour development commenced at 5 months, with all animals developing tumours by 12 months. We have also repeated this study with another set of ten LKB1fl/+PTEN+/− mice for each experimental condition. These mice are currently 10 months old, and the results obtained so far are similar, with all the control mice having developed tumours and tumour development being markedly delayed in mice administered with AMPK activators (X. Huang and D. R. Alessi, unpublished work). Pathological investigation of the tumours indicated that the types and morphology of tumours arising in animals administered with AMPK activators were similar to those observed in control animals not given a drug (Table 2). Representative histopathological sections of intestinal polyps and lymphoma and prostatic intraepithelial neoplasia are shown in Figure 5(B).
Previous work has shown that heterozygous LKB1+/− mice develop intestinal polyps resembling tumours seen in humans with Peutz–Jeghers syndrome. However, it has been controversial whether the tumours resulted from LKB1 haploinsufficiency or loss of heterozygosity (reviewed in ). Interestingly, hypomorphic LKB1fl/flPTEN+/+ mice do not develop more tumours, despite expressing lower levels of LKB1 in all tissues, than the tumour-developing LKB1+/− animals. This suggests that loss of heterozygosity rather than haploinsufficiency accounts for the tumour formation observed in LKB1+/− mice. It is possible that complete loss of cellular LKB1 and hence AMPK activity is required to induce tumorigenesis through the mTORC1 pathway. In contrast, the partial AMPK activity observed in LKB1fl/fl mice may be sufficient to inhibit mTORC1-driven tumorigenesis. This would explain why the activity of the mTORC1 pathway is not markedly elevated in the liver, spleen and intestine tissues of LKB1fl/fl mice (Figure 1D and Supplementary Figure S1B), compared with previous work undertaken in LKB1−/− cell lines and tissues [2,37]. It is also possible that loss of LKB1 alone without mutations in other oncogenes or tumour suppressors is insufficient to drive tumour formation. This might explain why in human cancer, loss of LKB1 is rarely found alone and frequently occurs in conjunction with mutations in other oncogenes such as Ras [28,38]. Thus the LKB1–AMPK pathway might operate to modulate and suppress inappropriately high mTORC1 activity, rather than acting as a direct stimulator of mTORC1 in the absence of any other activating stimulus.
The acceleration of tumorigenesis in LKB1 hypomorphic PTEN+/− mice provides a genetic demonstration of the co-operation between the LKB1 and PTEN pathways in a mouse model of cancer. Our data indicate that the rate and extent of tumour development in PTEN+/− mice correlates with the level of LKB1 expression. Even LKB1fl/+PTEN+/− mice that only have a moderate (∼40%) decrease in LKB1 expression, develop tumours significantly earlier than the LKB1+/+PTEN+/− animals (Figure 2A). As mentioned above, others have postulated that inhibition of LKB1 or AMPK might actually be therapeutically advantageous for the treatment of cancer in humans, as suppression of this pathway might reduce the ability of tumour cells to tolerate and survive low energy and/or chemotherapy [2,29]. However, our findings indicate that long-term inhibition of the LKB1–AMPK pathway would not be desirable at least for the treatment of tumours that possess elevated PI3K–mTOR pathway signalling. Moreover, we would recommend that protein kinase inhibitors being developed for cancer treatment are counterscreened to ensure that they do not inhibit the LKB1–AMPK signalling pathway. As LKB1 phosphorylates and activates other protein kinases related to AMPK [32,39], we cannot rule out the possibility that reduction in the activity of one or more of these enzymes contributes to activation of mTORC1 and/or acceleration of tumorigenesis in LKB1 hypomorphic PTEN+/− mice. The role of AMPK-related kinases in controlling mTOR has yet to be investigated, although metformin and phenformin do not stimulate the activity of AMPK-related kinases [32,40]. A recent report demonstrates that heterozygous LKB1+/− mice are markedly sensitized to the chemical carcinogen 7,12-dimethylbenz(α)anthracene, although this effect was attributed to an AMPK–mTOR-independent pathway that has yet to be characterized . It would be interesting to investigate whether this was mediated through one or more of the AMPK-related kinases.
It is also likely that AMPK suppresses cell growth by phosphorylating substrates other than mTORC1 regulators. One study suggests that AMPK inhibits cell-cycle progression by controlling the phosphorylation and stability of p27kip, a cell-cycle inhibitor . However, whether AMPK directly phosphorylates p27kip has yet to be established. The stabilization of p27kip was postulated to enable cancer cells to better survive conditions of nutrient and energy stress . In a similar vein, AMPK was claimed to enhance survival of hypoxic cancer cells by stabilizing HIF (hypoxia-inducible factor)-1α . Taken together, these findings indicate that, in addition to suppressing proliferation and growth, activation of the AMPK pathway may also facilitate the survival of some tumour cells. If this were the case, AMPK activators when used in cancer therapy might have more profound tumour-regression effects if utilized in combination with apoptosis-inducing therapies. Other potential AMPK anticancer targets include p53, which supposedly, following phosphorylation by AMPK, induces apoptotic cell death rather than enhancing survival [44,45]. Recent work also demonstrates that AMPK regulates epithelial cell polarity in part by controlling the assembly and disassembly of tight junctions [46,47]. The ability of AMPK to maintain cell polarity under situations of energy stress is likely to also contribute to its anti-tumour effects. Loss of LKB1 in the pancreas has also been demonstrated to lead to marked disorganization of acinar polarity, as well as cytoskeletal and tight junction defects . Whether this is mediated through AMPK and/or AMPK-related kinases is unknown. Much more work is required to delineate the diverse effects of activating AMPK activity in cancer cells and to define the substrates that AMPK phosphorylates to regulate growth, proliferation, survival and polarity.
Loss-of-function mutations in LKB1 are particularly common in human lung cancers [27,28]. Consistent with observations of previous studies [5–11], we found that PTEN+/− mice, whether hypomorphic for LKB1 or not, failed to develop lung cancer. In human lung cancer, loss of LKB1 is frequently associated with gain-of-function mutations in Ras [27,28]. This indicates that lung tumour formation may be driven by co-operation between the LKB1–mTOR and Ras pathways. It has also been proposed that exposure of lung cells to inhaled carcinogens plays a role in promoting tumour development in LKB1-deficient lung tissue . Further work is required to understand why mutations of LKB1 are so prevalent in lung cancer and the role that Ras and/or carcinogens play in promoting transformation of LKB1-deficient lung cells. It would also be interesting to investigate how modulating the levels of LKB1 expression or activating AMPK affects lung cancer as well as other tumours harbouring activating mutations of Ras. Mutation of LKB1 in lung cancers was also reported to enhance metastasis ; however, we found no increased metastasis of tumours between the PTEN+/− and normal or hypomorphic LKB1 mice.
Before the discovery that the LKB1 tumour suppressor activated AMPK, there was little interest in the role of AMPK in cancer. However, the ability of AMPK to gauge and control cellular energy places it in an ideal position to ensure that cell growth and proliferation is coupled to the availability of a sufficient supply of nutrients and energy. The finding that three distinct drugs that activate AMPK delayed tumorigenesis in PTEN+/− mice suggests that activators of AMPK could have therapeutic benefit for the treatment of cancer in humans. Metformin has been used as a frontline therapy for Type 2 diabetes for over 50 years and is inexpensive and well-tolerated in humans. Phenformin is a more potent activator of AMPK than metformin and is also more effective at reducing blood glucose levels in diabetic patients, but was withdrawn from clinical use in the U.S.A. in 1977, as it caused lactic acidosis in some patients . For the treatment of some cancers, side effects such as lactic acidosis may be more acceptable than for a chronic disease like diabetes, and the reintroduction of phenformin for cancer therapy should be explored. It may also be possible to initiate research to define the patients who are at risk of developing phenformin-induced lactic acidosis. Metformin and phenformin both activate AMPK indirectly, probably by decreasing cellular energy levels by inhibiting mitochondrial respiration (reviewed in [18,19]). Since the discovery that metformin exerts its blood-glucose-lowering effects by activating AMPK, pharmaceutical companies have devoted significant resources to develop drugs that activate AMPK directly for the treatment of Type 2 diabetes. The discovery of A-769662 was therefore a landmark finding , as it demonstrated that the development of such activators was feasible. The recent elucidation of the three-dimensional structure of AMPK subunits [21,50], coupled with further structural analysis to dissect the mechanism by which A-769662 interacts with AMPK, should accelerate the discovery of novel allosteric activators of AMPK. The present study suggests that these compounds should not only be assessed for their anti-Type 2 diabetes properties, but also for their ability to suppress and treat human cancer. A small-scale epidemiology population-based case-control study was carried out to address whether diabetics prescribed metformin were protected from developing cancer suggested that those taking metformin were 23% less likely to get cancer with the reduction in risk rising to 40% for people who had been administered the drug for a longer period of time . In future work, it would also be of considerable interest to investigate whether AMPK-activating drugs act synergistically with other anticancer agents, including those that inhibit the mTOR signalling pathway. It is possible that some of the inhibitors developed against the components of the mTOR pathway would be rendered more effective at suppressing cancer and prolonging lifespan, if prescribed in combination with metformin or phenformin. AMPK is also activated in humans during exercise, and several studies have reported that exercise is beneficial in the prevention of certain cancers . The effect of exercise in reducing cancer frequency is likely to be complex, but the depletion of whole-body ATP levels during exercise would raise AMPK activity and have the potential to inhibit cell growth and prevent cancer.
We thank Alan Ashworth (The Breakthrough Breast Cancer Research Centre, Institute of Cancer Research, London, U.K.) for the LKB1fl/fl hypomorphic mice, Ramon Parsons (Institute for Cancer Genetics, Department of Pathology and Cell Biology, Columbia University, New York, NY, U.S.A.) for the PTEN+/− mice, Pier Paolo Pandolfi (Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, U.S.A.) for the LKB1+/+PTEN−/− ES cell line, the Tissue Bank of the University of Dundee for histological preparation, George Thompson (University of Dundee, Ninewells Hospital) for help with immunohistochemistry analysis, Gail Fraser (MRC Protein Phosphorylation Unit, University of Dundee) for assistance with genotyping of mice and the antibody purification team (Division of Signal Transduction Therapy, University of Dundee) co-ordinated by Hilary McLauchlan and James Hastie for generation and purification of antibodies. We also thank the Association for International Cancer Research (D. R. A.), Diabetes UK (D. R. A.) the Medical Research Council (D. R. A.), the Moffat Charitable Trust (D. R. A.), Tenovus Tayside (S. F.) and the pharmaceutical companies supporting the Division of Signal Transduction Therapy (AstraZeneca, Boehringer-Ingelheim, GlaxoSmithKline, Merck & Co., Merck KGaA and Pfizer) for financial support.
Abbreviations: ACC, acetyl-CoA carboxylase; AICAR, 5-aminoimidazole-4-carboxamide 1-β-D-ribofuranoside; AMPK, AMP-activated protein kinase; DMEM, Dulbecco's modified Eagle's medium; ES cell, embryonic stem cell; FBS, fetal bovine serum; FOXO1, Forkhead box O1; mTOR, mammalian target of rapamycin; mTORC1, mTOR complex 1; PI3K, phosphoinositide 3-kinase; PTEN, phosphatase and tensin homologue deleted on chromosome 10; S6K1, S6 kinase 1; TSC, tuberous sclerosis complex
- © The Authors Journal compilation © 2008 Biochemical Society