Leptin activates Akt in oesophageal cancer cells via multiple atorvastatin-sensitive small GTPases

Obesity is a risk factor for Barrett’s oesophagus and oesophageal adenocarcinoma. Adipose tissue secretes the hormone leptin. Leptin is a growth factor for several cell types, including Barrett’s cells and oesophageal adenocarcinoma cells. Statins are associated with reduced rates of Barrett’s oesophagus and oesophageal cancer and exhibit anti-cancer effects in vitro. The mechanisms of these effects are not fully established. We have examined the effects of leptin and the lipid-soluble statin, atorvastatin, on signalling via monomeric GTP-binding proteins and Akt. Proliferation and apoptosis were assessed in OE33 cells. Akt activity was quantified by cell-based ELISA and in vitro kinase assay. Specific small-molecule inhibitors and a dominant-negative construct were used to reduce Akt activity. Small GTPases were inhibited using transfection of dominant-negative plasmids, prenylation inhibitors and pretreatment with atorvastatin. Leptin stimulated Akt activity and cell proliferation and inhibited camptothecin-induced apoptosis in an Akt-sensitive manner. Leptin induced phosphorylation of Bad and FOXO1 in an Akt-sensitive manner. Leptin activated Ras, Rac, RhoA and cdc42. Transfection of dominant-negative plasmids confirmed that leptin-induced Akt activation required Ras, RhoA cdc42 but not Rac. Atorvastatin inhibited leptin-induced activation of Ras, RhoA, cdc42 and Akt. Co-treatment with mevalonate prevented these effects of atorvastatin. The protein kinase Akt is essential to the growth-promoting and anti-apoptotic effects of leptin in oesophageal adenocarcinoma cells. Akt is activated via Ras-, Rho- and cdc42-dependant pathways. Atorvastatin reduces leptin-induced Akt activation by inhibiting prenylation of small GTPases. This may explain the reduced incidence of oesophageal adenocarcinoma in statin-users.


Introduction
OAC is an important cause of mortality and morbidity. In the USA and Europe, incidence rates have increased by approximately 7-fold in the last 40 years [1]. Survival rates for established OAC are poor (5-year survival less than 20%) [2,3]. In most cases, OAC is thought to develop from the precursor lesion, BO, which is a metaplastic phenotypic change in the oesophageal mucosa from squamous to glandular-type. Subsequently, cancer develops along a dysplasiacarcinoma sequence. The pathogenesis of BO and OAC is complex but the two most established modifiable risk factors for both are acid-reflux and obesity (particularly central obesity) [4]. The interaction of these is complex. Several lines of evidence of have suggested that actions and interactions between acid and peptides secreted by adipose tissue, at the level of the oesophageal epithelium, may promote the development of OAC in addition to the combination of effects of lifestyle and effects on the gastrooeosphageal junction [5][6][7].

3
We have demonstrated previously that the hormone leptin, produced by adipose tissue, has important effects in stimulating proliferation and inhibiting apoptosis in Barrett's epithelial cells [8,9]. Levels of leptin rise in proportion to adipose tissue mass [5]. Increased leptin levels are an independent risk factor for the development of both BO and progression along the dysplasia-carcinoma sequence [10,11]. Functional leptin receptors are expressed on the mucosa of BO and cultured Barrett's cell lines [5,8,9] Leptin is also secreted into the gastric lumen by gastric chief cells and theoretically Barrett's mucosa is exposed to leptin from both the circulation and gastric refluxate [12,13].
In experimental in vitro models, acid-reflux also stimulates proliferation and inhibits apoptosis in malignant and non-malignant Barrett's cell lines [14,15]; the combined effects of acid and leptin are synergistic [16]. Leptin has also been shown to increase the resistance of oesophageal cancer to chemotherapy in vivo and in vitro [17]. The mechanisms of these effects are incompletely defined.
The serine-threonine protein kinase Akt (also known as protein kinase B) is important in many systems in controlling proliferation and apoptosis. We have previously reported that Akt activation increases in Barrett's mucosa along the nondysplastic to low-grade dysplasia to high-grade dysplasia sequence. Akt is activated in oesophageal Barrett's cells by acid-reflux [18] and by hormones such as gastrin and glycine-extended gastrin as well as leptin [7,8,19,20]. In colon cancer cells, leptin-induced Akt activation is essential to the anti-apoptotic and proliferative actions [21]. Previously studies have outlined some of the pathways downstream of Akt that mediate the anti-apoptotic effects in Barrett's epithelial cells but there are much less data on the mechanisms that are involved in the upstream pathways leading to activation of Akt in this system [18,22]. Membrane-bound small GTPbinding proteins (GTPases) are important regulators of the signalling between cell-membrane receptors and intracellular signals, including protein kinases such as Akt. These GTP-binding proteins of the Ras, Rho, Rac and cdc42 subfamilies have been implicated in a variety of cellular processes including proliferation, cell-survival, invasion and inflammation [23]. Activation of the small GTP-binding proteins by both typical tyrosine kinase-linked growth factor receptors, including leptin [24,25], and 7-transmembrane G-protein coupled receptors has been described [26][27][28]. Previously, Ras has been shown to mediate anti-apoptotic signalling in Barrett's cells [29,30], but the role of these GTP-binding proteins in OAC and Barrett's mucosa have not been fully explored. We have also previously shown that statins (HMG-CoA reductase inhibitors) have anti-cancer effects in OAC cells in vitro [29] and are associated with a reduced incidence of OAC in epidemiological studies [31], and the mechanisms of these effects have not been fully described but may involve small GTPases.
Therefore, in this study, we have examined the hypothesis that GTP-binding proteins are involved in the regulation of Akt activation by leptin and that atorvastatin, a potent lipidsoluble statin, interferes with pro-oncogenic leptin signalling at the level of membrane-bound GTPases.

Cell culture, proliferation and apoptosis
OE33 OAC cells were cultured as described previously [8,32]. For proliferation experiments, cells were serumstarved for 24 h before stimulation with recombinant human leptin (1nM) (Bachem, UK), and then proliferation was assessed using a BrdU incorporation assay as previously described [20]. Apoptosis was induced by exposing serum-free leptin-treated cells with camptothecin (Merck, UK), and apoptosis quantified with the ApoPercentage assay as described previously [19]. We have previously confirmed that these assays provide comparable data to that obtained by cell counting and the MTT assay [8,9] and caspase-3 activity and quantification of intracellular nucleosomes [16,19]. For inhibitor studies, the Akt inhibitors, API-2 and GSK690693 (both 1 μm) [33,34] (Tocris, Abingdon, UK), were added 60 min prior to leptin. Cells were treated with adenoviral vector containing DN-Akt (Vector Biolabs, Pennsylvania, USA) 24 h prior to stimulation with leptin as described [26].

Akt activation and downstream activity
Serum-starved OE33 cells were stimulated with leptin, subsequently phosphorylated and non-phosphorylated Akt, Bad and FOXO1 (Forkhead box O1) were quantified in formalinfixed cells using commercially-available cell-based ELISAs (Active Motif, Waterloo, Belgium) as described previously [18]. Akt activity was additionally measured using an in vitro kinase activity as described previously [20].

Inhibition of Akt reduces leptin-stimulated proliferation and evasion of apoptosis
Leptin stimulated proliferation of OE33 cells, and this was significantly blocked (>80%) by 2 separate Akt-specific inhibitors and transfection with a DN-Akt construct (Fig. 1a). Leptin also demonstrated a significant anti-apoptotic effect. Leptin reduced camptothecin-induced apoptosis by 40%. Pretreatment with both Akt inhibitors and the DN-Akt plasmid abolished this anti-apoptotic effect (Fig. 1b). Effect on Akt activity. After 5 min, cells were lysed and Akt kinase activity quantified in Akt immunoprecipitates using a specific kinase (K-LISA) kit. Results are expressed as phosphorylating activity compared to untreated controls. Results are expressed as mean ± SEM, N = 5, *P < 0.01 vs control untreated, **P < 0.01 vs leptin-treated

Leptin-stimulated Akt is enzymatically active and leads to downstream signalling
The activation of Akt by leptin was confirmed using 2 different assays. Leptin-stimulated Akt activation was assessed by assay measuring Akt phosphorylation (increased by 187%) as well as by a specific kinase assay (increased by 215%). Pretreatment with Akt inhibitors or the DN-Akt effectively blocked Akt activation, with similar effects in both assays (Fig. 2). Total Akt levels were unaltered by leptin or smallmolecule inhibitors (data not shown), consistent with previously published data [7,8,16].
To confirm that leptin-stimulated Akt was functionally active, we further assessed downstream markers of Akt activity as read-outs. Leptin stimulation led to an increase in the phosphorylation of both Bad (increased by 150%) (Fig. 3a) and FOXO1 (increased by 87%) (Fig. 3a, b) and in both cases this was significantly reduced by the Akt inhibitors and the DN-Akt construct.

Akt activation by leptin requires small GTP-binding protein activity
The activation of protein kinases is controlled in many cases by small GTP-binding protein. (a) Effect on Akt phosphorylation. After 5 min, cells were formalinfixed and subsequently phosphorylated Akt and total Akt levels quantified in fixed cells using a specific cell-based ELISA. Results for each well were normalised for cell content using subsequent crystal violet staining. Results are expressed as the ratio of phosphorylated/ total kinase relative to control-treated cells. (b) Effect on Akt activity. After 5 min, cells were lysed and Akt kinase activity quantified in Akt immunoprecipitates using a specific kinase (K-LISA) kit. Results are expressed as phosphorylating activity compared to untreated controls. Results are expressed as mean ± SEM, N = 5, *P < 0.01 vs control untreated, **P < 0.05 vs leptin-treated, ***P < 0.01 vs leptintreated activation by >80%, DN-RhoA by 50-57% and DN-cdc42 by about 25%. DN-Rac1 had no effect on Akt activation. A similar pattern of results was seen whether Akt was assessed with either a cell-based ELISA for phosphorylated protein (Fig. 4a) or in a kinase assay (Fig. 4b).
Small GTP-binding protein is active when localised to the cell membrane; this localisation requires post-translational modification by prenylation with the addition of a farnesyl group (to Ras) and geranylgeranyl groups to Rac, Rho and cdc42). Therefore, we performed further studies using specific inhibitors of farnesylation and geranylgeranylation.
The farnesyl transferase inhibitor FTI276 reduced leptinstimulated Akt activation by 85% and the geranylgeranyl transferase inhibitor GGTI298 reduced Akt activation by approximately 75%. Statins reduce the ability to prenylate GTP-binding protein by reducing the availability of farnesyl and geranylgeranyl intermediates subsequent to the inhibition of the mevalonate synthetic pathway [35]. Pretreatment with the lipid-soluble statin, atorvastatin, also inhibited Akt activation by 50%. Co-treatment with mevalonate prevented this inhibitory effect (Fig. 5).

Leptin stimulates small GTP-binding protein activity
To confirm the involvement of small GTPase in leptin signalling, the activity of all 4 classes of GTPase was assessed using specific assays. Leptin stimulation potently increased the activity of Ras, Rac, Rho and cdc42 (Fig. 6). The farnesyltransfer inhibitor inhibited the activity of Ras, whilst the geranylgeranylation inhibitor reduced the activation of Rac, Rho and cdc42. Pretreatment with atorvastatin reduced leptin-stimulated activation of all classes of GTP-binding protein (by 54-76%) and co-treatment with mevalonate significantly ameliorated this effect in all cases.

Conclusions
In this study, we have demonstrated for the first time the essential requirement for the small GTPases Ras, Rho and cdc42 in leptin-induced Akt activation in oesophageal cancer cells. We have also shown that the lipid-soluble statin, atorvastatin, inhibits leptin-induced pro-oncogenic signalling by reducing signalling via these membranelinked GTPases.
We have confirmed that leptin exerts important proproliferative and anti-apoptotic effects on OAC cells. We have demonstrated the essential role of the protein kinase Akt in this pathway. We confirmed increased leptin-stimulated Akt activity using two separate methods. Similar effects were seen using 3 different methods of inhibiting Akt. Although the two different small-molecule Akt inhibitors used have been reported to be Akt selective, most kinase inhibitors are not completely specific; hence, we performed confirmatory experiments using transfection with a dominant-negative adenoviral construct. All these results were consistent and clearly implicate Akt as a downstream mediator in the effects of leptin in oesophageal cancer cells.
The specific downstream effectors of the Akt activity in this model remain to be clarified. Important actions of Akt include phosphorylation of Forkhead (FOXO) transcription factors, which leads to subsequent sequestration in the cytosol and inhibition of transcriptional activity and phosphorylation and reduced activity of the proapoptotic protein Bad. These have been implicated in the anti-apoptotic effect mechanisms of other signalling systems, such as that induced by gastrin [36]. We used these two markers primarily as read-outs to confirm the functional activity of Akt in OE33 cells. Our results confirm the functional enzymatic activity of Akt in the system, although, at present, we cannot definitely implicate either Bad or FOXO1 as downstream mediators of Akt. FOXO1 is known to be an essential downstream mediator of the central anorectic effects of activation of the leptin receptor [37]. Other downstream targets of Akt, such as the NF-κB pathway and stabilisation of COX-2 mRNA, are likely to be important in the anti-apoptotic effects of Akt in Barrett's oesophageal cells [19,20]. Recent studies have also further highlighted the potential central role of Akt in Barrett's oesophageal cancer progression: signalling via the insulin/insulin-like growth factor-1 axis and HER2 receptor also activates Akt along the Barrett's dysplasiacancer sequence [38]. This latter study did not examine the upstream activators of Akt or the downstream effectors.
Our results show that activation of the leptin receptor activates small GTP-binding proteins of all 4 subfamilies studied (Ras, Rho, Rac and cdc42). The data from the dominant-negative transfection studies clearly demonstrated that although leptin-induced Akt activation was independent of Rac activation, there was a requirement for all three other GTP-binding protein families. Inhibition of Ras had by far the greatest inhibitory effect on leptininduced Akt activity, and there were smaller effects from inhibiting RhoA and cdc42. These data show that leptininduced Akt activation involves Ras, Rho and cdc42. There did not seem to be an absolute requirement for any specific GTP-binding protein, although Ras seemed to have the dominant role in signalling. Previously, leptin has been shown to activate all of Ras, Rho, Rac and cdc42 in a variety cell models, although the exact significance of these changes and the downstream effectors have not been fully described. In colon cancer cells, leptin activates Rho, Rac and cdc42 which are involved in enhancing migration and invasion [24,39]. There are a wealth of data examining downstream signalling from the leptin receptor, this is concentrated on protein kinases and transcription factors and there are relatively few studies examining the specific role of the small GTPase families [40].
We further explored different pharmacological methods to manipulate GTP-binding protein signalling and the effects on Akt activation. The results with the prenylation inhibitors were consistent with the dominant-negative plasmids. Ras is predominantly farnesylated, whilst Rho, Rac and cdc42 are geranylgeranylated [35]. The addition of the lipid tails allows localisation of the GTP-binding protein to the inner side of the cell membrane, where they are functionally involved in signalling from many receptors. The farnesyl transferase inhibitor reduced Akt activation more than the geranylgeranyl transferase inhibitor, in keeping both with the dominant role of Ras in leptin-induced Akt activation, and the subsidiary but important roles of Rho and cdc42.
Previous studies, both from our own group and others, have shown that statins have anti-proliferative and proapoptotic actions in oesophageal adenocarcinoma cells [29,41,42]. Clinical epidemiological studies have shown associations between statin use and reduced incidences of both BO and OAC [43][44][45][46] and improved overall and cancer-specific survival after surgery for OAC [47], suggesting that these effects of statins do have clinical correlations. However, the exact mechanism(s) underlying these apparent protective associations are not fully defined [5], but the clinical and experimental data do support further the further exploration of statins as chemopreventative agents in BO.
Statins inhibit the essential early step in the pathway for the formation of mevalonate. Mevalonate is the precursor to cholesterol but also the precursor to farnesyl pyrophosphate and geranylgeranyl pyrophosphate, which are the required substrates for the prenylation and membrane localisation of small GTP-binding proteins. We pretreated OE33 cells with the lipid-soluble statin atorvastatin at approximately the same concentration as can be found with high-dose clinical use (40-80 mg daily) [48,49]. Atorvastatin significantly reduced by Akt activation and GTP-binding protein activity. These effects of atorvastatin were abolished by adding back mevalonate. Mevalonate is downstream of HMG-CoA reductase in the biosynthetic pathways and is the precursor farnesyl and geranylgeranyl pyrophosphate, suggesting that the effects of atorvastatin were mediated by the reduction in the availability of intermediates for prenylation and resultant reduction in GTP-binding protein signalling.
Although potentially useful anti-cancer effects of statins in vitro against oesophageal cancer cells and non-malignant Barrett's oesophageal cells have been documented, consistent with the clinical observations, some of these studies have been criticised for using what were clearly supraphysiological statin concentrations [50]. In this current study, we specifically used atorvastatin concentrations that can be achieved with standard dosing regimens [48,49]. The inhibitory effects of atorvastatin demonstrated in this study were clearly less than those obtained with the more specific experimental tools to inhibit Akt and small GTP-binding proteins (such as the dominant-negative constructs and prenylation inhibitors). This presumably reflects the less than total inhibition of the mevalonate pathway possible with pharmacologically feasible statin levels: even then this atorvastatin concentration is well over the IC 50 for HMG-CoA-reductase (15.2 nM [51]. Higher degrees of apoptosis and inhibition of proliferation have been achieved using considerably higher concentrations of statins (up to 50 μm) [41,42]. The specificity of the pharmacological effects and clinical relevance of these effects are open to question.
Recently, alternative methods of inhibition of small GTPase signalling by statins have been described: atorvastatin and pitavastatin inhibited Rac signalling by enhancing degradation of Rac in a prenylation-independent manner [52]. In our current study, the effects of atorvastatin were reversed by mevalonate favouring a prenylation-dependant pathway.
The effects of statins in potentially preventing the development or progression of OAC in vivo are undoubtedly complex, but our current data suggest that direct effects at the level of the oesophageal epithelial cells may be involved. Although the experimental cell-line studies implicating leptin signalling, GTPases and the ameliorating effect of statins [29,53] are all consistent with the epidemiological and clinical studies [43,[54][55][56], these particular pathways have not been examined as yet in the animal models of Barrett's carcinogenesis [57].
Similarly, the mechanisms linking obesity and the development of OAC are complex. Direct effects of adipose tissue mediators such as leptin are but one of the likely factors including lifestyle, environment and disruption of the antireflux barrier [5,6]. However, our studies further confirm that leptin can have important pro-neoplastic effects directly on the oesophageal glandular epithelium. The protein kinase Akt has an essential role in these leptin-mediated effects and the small GTP-binding proteins Ras, Rho and cdc42 are involved in regulation Akt activity. Further studies examining the pharmacological modulation of Akt in oesophageal neoplasia and pre-neoplasia are warranted. Statins, in this case atorvastatin, by affecting GTP-binding protein prenylation and activation can reduce Akt activation and the potential role of statins to prevent OAC should be explored further.

3
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.