V-AKT murine thymoma viral oncogene homolog/mammalian target of rapamycin activation induces a module of metabolic changes contributing to growth in insulin-induced hepatocarcinogenesis

Authors


  • Potential conflict of interest: Nothing to report.

  • M.E. and F.D. were supported by the Deutsche Forschungsgemeinschaft DFG (grant nos.: Ev168/2-1 and Do622/2-1). V.D.M., G.G., S.M., and G.D. were supported, in part, by a fellowship from the Master and Back Program, Sardegna Ricerche, Regione Autonoma della Sardegna.

Abstract

Mounting epidemiological evidence supports a role for insulin-signaling deregulation and diabetes mellitus in human hepatocarcinogenesis. However, the underlying molecular mechanisms remain unknown. To study the oncogenic effect of chronically elevated insulin on hepatocytes in the presence of mild hyperglycemia, we developed a model of pancreatic islet transplantation into the liver. In this model, islets of a donor rat are transplanted into the liver of a recipient diabetic rat, with resulting local hyperinsulinism that leads to the development of preneoplastic lesions and hepatocellular carcinoma (HCC). Here, we investigated the metabolic and growth properties of the v-akt murine thymoma viral oncogene homolog/mammalian target of rapamycin (AKT/mTOR) pathway, a major downstream effector of insulin signaling, in this model of insulin-induced hepatocarcinogenesis. We found that activation of insulin signaling triggers a strong induction of the AKT/mTOR cascade that is paralleled by increased synthesis of fatty acids, cholesterol, and triglycerides, induction of glycolysis, and decrease of fatty acid oxidation and gluconeogenesis in rat preneoplastic and neoplastic liver lesions, when compared with the healthy liver. AKT/mTOR metabolic effects on hepatocytes, after insulin stimulation, were found to be mTORC1 dependent and independent in human HCC cell lines. In these cells, suppression of lipogenesis, glycolysis, and the pentose phosphate pathway triggered a strong growth restraint, despite insulin administration. Noticeably, metabolic abnormalities and proliferation driven by insulin were effectively reverted using the dual PI3K/mTOR inhibitor, NVP-BEZ235, both in vitro and in vivo. Conclusions: The present results indicate that activation of the AKT/mTOR cascade by unconstrained insulin signaling induces a defined module of metabolic alterations in hepatocytes contributing to aberrant cell growth. Thus, inhibition of AKT/mTOR and related metabolic changes might represent a novel preventive and therapeutic approach to effectively inhibit insulin-induced hepatocarcinogenesis. (Hepatology 2012;)

Human hepatocellular carcinoma (HCC) incidence is rapidly rising in Western Europe and the United States.1 However, the well-established causal relationship between chronic hepatitis B and C infection and hepatocarcinogenesis does not fully explain this escalation in HCC occurrence, with one quarter of the cases remaining idiopathic.1 Recently, nonalcoholic steatohepatitis (NASH) has received attention for its potential causal role in hepatocarcinogenesis.2-5 Indeed, several case-control studies indicate that HCC patients with cryptogenic cirrhosis show clinical and demographic features suggestive of NASH, as compared with age- and sex-matched HCC patients of viral or alcoholic etiology.1, 5 The most compelling evidence for an association between NASH and HCC comes from studies examining the correlation of HCC with two conditions strongly related to NASH: obesity and diabetes.6, 7 Significantly associated with obesity, insulin resistance and related hyperinsulinemia are known to contribute significantly to hepatic steatosis.8, 9 Similarly, type II diabetes mellitus (T2DM) has been proposed as a risk factor for HCC through the development of NASH.1 Accordingly, recent investigations indicate that T2DM people are at a significantly higher risk to develop liver cancer than normoglycemic people.1, 10-13 Furthermore, although highly debated, recent studies suggest that treatment with insulin analogs, such as insulin glargine, increases the incidence of various tumor types, including HCC.14, 15 Previous reports indicate that insulin regulates many metabolic pathways in hepatocytes and promotes hepatocellular growth.16-18 However, the molecular mechanisms linking deregulation of insulin to human hepatocarcinogenesis remain obscure.

To study the effect of chronically elevated secretion of insulin on hepatocytes, we developed a model of pancreatic islet transplantation into the liver.19-22 In this model, pancreatic islets of donor rats are transplanted into the liver of recipient diabetic rats. Because of the low number of transplanted islets, a mild systemic hyperglycemia persists and constitutes a constant stimulus for the islet grafts to synthesize and secret insulin into the blood. The resulting local hyperinsulinism in the liver acini, located downstream of the islet grafts, leads to morphological changes in the affected hepatocytes (e.g., an excessive storage of glycogen and lipids combined with a high cell turnover), resulting in a sharp demarcation of these altered acini from the surrounding liver tissue.19-22 These changes are accompanied by alterations of carbohydrate and lipid metabolism (i.e., up-regulation of glycolysis, pentose-phosphate pathway, fatty acid synthase expression, and down-regulation of gluconeogenesis).19-22 At the molecular level, cytoplasmic translocation and up-regulation of the insulin receptor (IR) and induction of IRS-1, Raf-1, and Mek-1 occurred.23 These adaptive alterations are morphologically and biochemically equivalent to preneoplastic foci of altered hepatocytes, known as clear cell foci, characteristic of chemically induced hepatocarcinogenesis models.24 In the first 3 months after transplantation, these adaptive/preneoplastic lesions are limited to the downstream liver acini, but then they gradually expand into the neighboring liver, finally progressing to HCC between 6 and 24 months.19-22 Thus, this model can be regarded as a prototypical example of carcinogenesis originating from cells with a deregulated metabolism.

The v-akt murine thymoma viral oncogene homolog (AKT)/mammalian target of rapamycin (mTOR) pathway is a major effector of the insulin cascade,25 but a direct link between AKT/mTOR activation, deregulation of cell metabolism, and insulin-driven carcinogenesis has not been demonstrated to date. Thus, we investigated the functional role of the AKT/mTOR pathway in the rat model of insulin-driven hepatocarcinogenesis. Our data underline a pivotal function of the AKT/mTOR cascade in the growth of liver lesions induced by local, chronic insulin deregulation through the induction of specific metabolic changes and promotion of hepatocellular proliferation. Furthermore, the present findings open the possibility of inhibiting AKT/mTOR signaling as a novel therapeutic approach for the treatment of human HCC, especially when associated with deregulation of insulin signaling.

Abbreviations:

ACAC, acetyl-coenzyme A carboxylase; ACADM, acyl-CoA dehydrogenase; ACLY, ATP citrate lyase; AICAR, 5-aminoimidazole-4-carboxamide 1-beta-D-ribofuranoside; AKR1B10, Aldo-keto reductase family 1, member B10; AKT, v-akt murine thymoma viral oncogene homolog; ALDOA, aldolase A; AMPK, AMP-activated kinase; 6-AN, 6-aminonicotinamide; 3-BrPA, 3-bromopyruvate; BW, body weight; chREBP, carbohydrate-responsive element-binding protein; CoA, coenzyme A; COP1, constitutive photomorphogenic protein 1; 2-DG, 2-deoxy-D-glucose; DMSO, dimethyl sulfoxide; 4EB-P1, eukaryotic translation initiation factor 4E binding protein 1; ECHS1, enoyl-CoA hydratase 1; FASN, fatty acid synthase; FOXO1, forkhead box O1; G6PD, glucose-6-phosphate dehydrogenase; G6Pase, glucose-6-phosphatase; HCC, hepatocellular carcinoma; HIF-1α; hypoxia-inducible factor 1-α; HK2, hexokinase 2; HMGCR, 3-hydroxy-3-methylglutaryl-CoA-reductase; IHC, immunohistochemistry; INSIG2, insulin-induced gene 2; IR, insulin receptor; LDH, lactate dehydrogenase; LDHA, lactate dehydrogenase A; MKP3, mitogen-activated protein kinase phosphatase 3; MAPK, mitogen-activated protein kinase; mTOR, mammalian target of rapamycin; mTORC, mTOR complex; NASH, nonalcoholic steatohepatitis; p70S6K, p70 S6 kinase; PCK1, phosphoenolpyruvate carboxykinase 1; PGC-1α, PPARγ, coactivator 1 alpha; PI3K, phosphoinositide 3-kinase; PPARγ, peroxisome proliferator-activated receptor gamma; PHLPP, pleckstrin homology domain leucine-rich repeat protein phosphatase; PRKCλ/ι, protein kinase C lambda/iota; Raptor, regulatory-associated protein of mTOR; RHEB, Ras homolog enriched in brain; RPIA, ribose 5-phosphate isomerase A; RPS6, ribosomal protein S6; SCD, stearoyl-CoA desaturase; SGK1, serum/glucocorticoid-regulated kinase1; SD, standard deviation; siRNAs, short interfering RNAs; SREBP, sterol regulatory element binding protein; T2DM, type II diabetes mellitus; TRB3, tribbles homolog 3; USP2, ubiquitin-specific protease 2.

Materials and Methods

Rat Tissue Specimens, Transplantation Procedure, and Treatment.

Rat liver specimens were obtained from previous experiments. Healthy livers, pools of liver preneoplastic lesions isolated with the use of a stereomicroscope, and HCC were used. The transplantation procedure was performed as previously described.19, 20 Briefly, diabetes was induced in adult inbred male Lewis rats (250-300 g) by treatment with a single subcutaneous dose of streptozotocin (80 mg/kg body weight [BW]) and was defined by a nonfasting blood-glucose level higher than 400 mg/dL, manifesting between 1 and 3 days after the administration of streptozotocin. Islets of Langerhans were isolated from nondiabetic littermates and transplanted into the liver of recipient rats through the portal vein. A low number of islets (250-450 islet grafts per animal) was transplanted so that mild hyperglycemia (250-300 mg/dL) persisted for at least 10 months after transplantation. During infusion, the branch supplying the left part of the liver was clamped, thus ensuring that the transplants were embolized only into the right part of the liver and the left part served as an intraindividual control. As an additional control, the livers of nondiabetic rats not undergoing transplantation were used. However, because there were no significant differences in the parameters examined between the intraindividual controls and the nondiabetic rats not undergoing transplantation, the data of these two groups were combined and referred to as the control liver. Animals were sacrificed under anesthesia between 2 days and 24 months after transplantation. Groups of rats were subjected to daily administration of the phosphoinositide 3-kinase (PI3K)/mTOR dual inhibitor, NVP-BEZ235 (kindly provided by Novartis, Basel, Switzerland), dissolved in 1% methylcellulose at a concentration of 10 mg/kg BW for 4 weeks. Rats were housed, fed, and treated according to the German Animal Protection Law and approved by the Local Government of Mecklenburg-Vorpommern.

Cell Lines and Treatments.

Transfection of Hep3B and HLE cell lines with short interfering RNAs (siRNAs) and treatment with specific inhibitors were performed as described in the Supporting Materials and Methods.

Immunoblotting.

Hepatic tissue samples were homogenized and processed as previously reported.26 Nitrocellulose membranes were probed with specific primary antibodies (Supporting Table 1).

Statistical Analysis.

Tukey-Kramer's test was used to evaluate statistical significance. P < 0.05 was considered significant. Data are expressed as means ± standard deviation (SD).

See the Supporting Materials and Methods for detailed descriptions of materials and methods.

Results

Activation of the AKT/mTOR Pathway in Rat Liver Lesions.

First, we investigated by immunoblotting the levels of AKT, mTOR, and their effectors, including Ras homolog enriched in brain (RHEB), p70 S6 kinase (p70S6K), ribosomal protein S6 (RPS6), eukaryotic translation initiation factor 4E binding protein 1 (4E-BP1), and hypoxia-inducible factor 1-alpha (HIF-1α), acting downstream of the mTOR complex 1 (mTORC1), and serum/glucocorticoid-regulated kinase 1 (SGK1), which acts downstream of mTORC2 (Fig. 1A). A progressive up-regulation of total and activated AKT and mTOR, and of p70S6K, RHEB, RPS6, HIF-1α, inactivated/phosphorylated 4E-BP1, and activated/phosphorylated SGK1 occurred in preneoplastic and neoplastic rat lesions, when compared with control liver (Fig. 1A; Supporting Fig. 1). Phosphorylated/activated AKT, mTOR, and inactivated/phosphorylated 4E-BP1 levels were further confirmed by immunohistochemistry (IHC) (Fig. 1B). Also, levels of AMP-activated kinase (AMPK) alpha proteins were assessed, because AMPKs are known to negatively modulate de novo lipogenesis induced by mTOR27 (Fig. 1A; Supporting Fig. 1). AMPKα1 levels were equivalent in healthy livers, preneoplastic foci and HCCs, whereas those of AMPKα2 and activated/phosphorylated AMPKα levels were down-regulated in preneoplastic foci and HCCs (Fig. 1A; Supporting Fig. 1). In accord, the levels of markers of AMPK activation, including phosphorylated/inactivated 3-hydroxy-3-methylglutaryl-coenzyme A (CoA) reductase (HMGCR), phosphorylated/inactivated acetyl-CoA carboxylase (ACAC), and phosphorylated/inactivated regulatory-associated protein of mTOR (Raptor), were lowest in preneoplastic foci and HCCs, when compared with healthy livers (Fig. 1A; Supporting Fig. 1).

Figure 1.

Activation of AKT/mTOR pathway and loss of AMPKα activation in preneoplastic foci and HCC induced by pancreatic islet transplantation into the liver of diabetic rats, when compared with normal livers. (A) Representative immunoblotting of the major proteins downstream of AKT belonging to the mTOR pathway. Whole cell lysates were prepared from normal livers (NL), preneoplastic foci (FOCI), and HCC and immunoblotted with indicated antibodies. Of note, as a consequence of AMPKα inactivation, inactivated/phosphorylated levels of HMGCR, ACAC, and Raptor decreased in preneoplastic foci and HCC, whereas total levels of HMGCR and ACAC were induced and Raptor levels were equivalent, respectively, when compared with control livers. (B) Representative IHC of the activation of the AKT/mTOR pathway, showing increased expression of activated/phosphorylated AKT, activated/phosphorylated mTOR, and the mTOR target inactivated/phosphorylated 4EBP1, each in a preneoplastic focus (left column) and a HCC (right column). A transplanted islet graft, surrounded by the glycogenotic and PAS-positive preneoplastic focus (i.e., hyperinsulinemic liver acinus), is marked (I) in the PAS reaction and in the serial section of p-AKT IHC. In contrast to the other proteins, p-4EBP1 staining is both nuclear and cytoplasmic. The lower edge of the panel represents 3.3 mm in the PAS stain and p-AKT of the focus, 0.7 mm in p-mTOR and p-4EBP1 staining of the focus, and 0.35 mm in all panels of the HCC. PAS, Periodic Acid-Schiff.

Activation of Lipogenesis and Inhibition of Fatty Acids Oxidation in Rat Lesions.

Because we recently demonstrated that activation of the AKT/mTOR pathway induces aberrant lipogenesis,26 we assessed by immunoblotting whether the same occurred in the rat model. Levels of proteins involved in fatty acid biosynthesis, including fatty acid synthase (FASN), ACAC, and stearoyl-CoA desaturase 1 (SCD1), were progressively increased in preneoplastic liver foci and HCC, when compared with unaltered liver tissues (Fig. 2A; Supporting Fig. 2). In contrast, ATP citrate lyase (ACLY) overexpression peaked in foci, but was still higher in HCC, when compared with control liver. Aldo-keto reductase family 1, member B10 (AKR1B10), which binds and prevents ACAC degradation by the proteasome, thus increasing fatty acid synthesis,28 was instead up-regulated only in HCC. Similarly, AKR1B10-ACAC complexes (sign of AKR1B10 prolipogenic activity) were equivalent in control liver and preneoplastic foci, but significantly increased in HCC. Upstream inducers of lipogenesis, including carbohydrate-responsive element-binding protein (chREBP), protein kinase C lambda/iota (PRKCλ/ι), and peroxisome proliferator-activated receptor gamma (PPARγ), were progressively increased from preneoplastic foci to HCC. Furthermore, ubiquitin-specific protease 2a (USP2a), which sustains FASN activity by impeding its ubiquitin-dependent degradation in the liver,26 was induced in preneoplastic and neoplastic lesions. In contrast, the negative regulator of sterol regulatory element-binding protein (SREBP)-1-mediated lipogenesis, insulin-induced gene 2 (INSIG2),29 was strongly down-regulated in preneoplastic foci and HCC (Fig. 2A; Supporting Fig. 2).

Figure 2.

Activation of de novo lipogenesis and suppression of fatty acid β-oxidation in preneoplastic foci and HCC induced by pancreatic islet transplantation into the liver of diabetic rats, when compared with normal livers. (A) Representative immunoblotting of the major proteins involved in lipid biosynthesis, fatty acid β-oxidation, and transcription factors promoting lipogenesis. Whole cell lysates were prepared from normal livers (NL), preneoplastic foci (FOCI), and HCC and immunoblotted with indicated antibodies. (B) Representative IHC of preneoplastic lesions and HCC (lower right panel) for components (SCD1, ACAC, ACLY, and HMGCR) and transcription factors (SREBP1 and chREBP) of lipid and cholesterol metabolism. Upper and middle row as well as left and center panel of the bottom row show serial sections of the same preneoplastic lesions, respectively. Periodic Acid-Schiff reaction is shown to outline the preneoplastic foci. The lower edge of the panel represents 1.9 mm in the upper row and the left and center panel of the bottom row and 0.4 mm (center and right panel) in the middle row and in the HCC.

As concerns the cholesterol cascade, SREBP-2 and HMGCR were progressively induced from preneoplastic lesions to HCCs (Fig. 2A; Supporting Fig. 2).

Furthermore, IHC showed strong immunoreactivity for ACAC, ACLY, SCD1, SREBP-1, chREBP, and HMGCR in preneoplastic foci and HCCs, but not in unaltered surrounding liver tissues and control livers (Fig. 2B). Also, triglyceride and cholesterol levels as well as fatty acids biosynthesis were all significantly increased in rat preneoplastic foci and HCCs, when compared to control livers (Supporting Fig. 3).

In contrast, levels of fatty oxidation and proteins involved in this process, including mitochondrial acyl-CoA dehydrogenase (ACADM) and enoyl-CoA hydratase 1 (ECHS1), were progressively reduced in rat preneoplastic foci and HCCs (Fig. 2A; Supporting Figs. 2 and 3).

Induction of Glycolysis and Pentose Phosphate Pathways and Suppression of Gluconeogenesis in Rat Lesions.

Because of the role of the AKT pathway in glucose metabolism30 and previous results in the rat preneoplastic foci,20 we determined the levels of proteins involved in glycolysis, pentose phosphate cascade, and gluconeogenesis (Fig. 3; Supporting Fig. 4). As concerns glycolysis, we found concomitant up-regulation of hexokinase 2 (HK2), aldolase A (ALDOA), and lactate dehydrogenase A (LDHA) in preneoplastic and neoplastic rat lesions (Fig. 3A). An equivalent pattern was detected when assessing the levels of lactate dehydrogenase (LDH) activity in the sample collection (Fig. 3B). Similarly, proteins involved in the pentose phosphate pathway, including glucose-6-phosphate dehydrogenase (G6PD) and ribose 5-phosphate isomerase A (RPIA), were up-regulated in rat preneoplastic foci and HCCs. Also, G6PD activity was more elevated in preneoplastic foci, when compared with healthy livers, and was highest in HCCs (Fig. 3C). On the other hand, the enzymes involved in gluconeogenesis, including phosphoenolpyruvate carboxykinase 1 (PCK1) and glucose-6-phosphatase (G6Pase), and the key gluconeogenic transcription coactivator, PPARγ, coactivator 1 alpha (PGC-1α), were down-regulated in the same lesions (Fig. 3A). Because mitogen-activated protein kinase (MAPK) phosphatase 3 (MKP-3) promotes hepatic gluconeogenesis by dephosphorylating forkhead box O1 (FOXO1) at serine 256,31 we determined the levels of MKP-3 and phosphorylated/inactivated FOXO1 in the rat samples. MKP-3 expression was decreased, and phosphorylated/inactivated levels of FOXO1 were augmented in rat liver lesions (Fig. 3A; Supporting Fig. 4), further confirming the reduction of gluconeogenesis induced by insulin.

Figure 3.

Activation of glycolysis and the pentose phosphate pathway as well as suppression of gluconeogenesis in preneoplastic foci and HCC induced by pancreatic islet transplantation into the liver of diabetic rats, when compared with healthy livers, as detected by immunoblotting (A), LDH assay activity (B), and G6PD activity (C). Whole cell lysates were prepared from normal livers (NL), preneoplastic foci (FOCI), and HCC and immunoblotted with indicated antibodies. LDH and G6PD activity were assessed using the LDH and G6PD activity assays (BioVision, Inc., Milpitas, CA), according to the manufacturer's instructions.

Metabolic Alterations Induced by AKT/mTOR Are Necessary for Insulin-Driven Growth of HCC Cells.

To determine whether the metabolic changes induced by AKT/mTOR cascade play a role in insulin-driven growth, the Hep3B and HLE cell lines were subjected to insulin administration concomitant with inhibition of de novo lipogenesis, fatty acid oxidation, glycolysis, or pentose phosphate pathways. Suppression of lipogenesis by siRNA-mediated silencing of SREBP-1 and SREBP-2, led to reduction of Hep3B (Fig. 4A) and HLE (not shown) cell proliferation and induction of apoptosis (Supporting Fig. 5). A significant reduction in cell proliferation and induction of apoptosis was also detected after treatment with fatty oxidation inducers (e.g., 5-aminoimidazole-4-carboxamide 1-beta-D-ribofuranoside [AICAR] and metformin), glycolysis inhibitors (e.g., 2-deoxy-D-glucose [2-DG] and 3-bromopyruvate [3-BrPA]), and with the G6PD inhibitor, 6-aminonicotinamide (6-AN), in Hep3B (Fig. 4B; Supporting Fig. 5) and HLE (not shown) cell lines. Noticeably, combined treatment with SREBP-1/2 siRNA, metformin, 2-DG, and 6-AN resulted in a much more pronounced growth restraint of Hep3B (Fig. 4C; Supporting Fig. 5) and HLE (not shown) cells, when compared with treatment using SREBP-1/2 siRNA, 2-DG, or 6-AN alone, implying a synergistic, antineoplastic function of the four treatments when used combinatorially.

Figure 4.

Insulin-induced growth depends on metabolic alterations induced by AKT/mTOR in human HCC cell lines. (A) Effect of suppression of lipogenesis or induction of fatty acid oxidation on the proliferation rate (as assessed by bromodeoxyuridine incorporation) of Hep3B cells treated with insulin. Silencing of SREBP1 and SREBP2 genes by siRNA (SREBP1/2 siRNA), and the fatty acid oxidation inducers, AICAR and metformin, were used to assess the effect of inhibiting lipogenesis and inducing fatty acid oxidation, respectively. (B) Effect of suppression of glycolysis via 2-DG and 3-BrPA administration or the pentose phosphate pathway with the G6PD inhibitor, 6-AN, on Hep3B cell proliferation after insulin administration. (C) Combinatorial treatment with SREBP1 siRNA, metformin, 2-DG, and 6-AN strongly restrains the proliferation of Hep3B cells treated with insulin, when compared with single treatments. Equivalent results were obtained in the HLE cell line (not shown). Each bar represents mean ± SD of three independent experiments conducted in triplicate. Tukey-Kramer's test: P < 0.001 a, versus control (untreated cells); b, versus DMSO; c, versus insulin-treated cells; d, versus insulin-treated cells and drug solvent (DMSO); e, versus insulin-treated cells and SREBP1/2 siRNA or 2-DG; f, versus cells treated with insulin and AICAR, 3-BrpA, or metformin (in C); g, versus cells treated with insulin and 2-DG; h, versus cells treated with insulin and 6-AN. (D) Representative immunoblotting showing effective silencing of SREBP1 and SREBP2 genes after concomitant transfection of 500 ng of specific siRNAs against the same genes in the Hep3B cell line.

Metabolic Alterations Induced by Insulin Through AKT/mTOR Are mTORC1 Dependent and Independent.

Because mTORC1 is a major effector of AKT metabolic properties,25 we determined whether mTORC1 is responsible for the observed effect on metabolism induced by insulin. For this purpose, Hep3B and HLE cell lines were subjected to insulin treatment concomitant with inhibition of either mTORC1 or AKT. In the Hep3B cell line, a rise in the AKT pathway was detectable as early as 10 minutes after insulin administration (data not shown). Levels of the AKT cascade remained elevated 24 and 36 hours after insulin supplementation (Fig. 5A,B) and were associated with a significant increase in HCC cell proliferation and survival (Fig. 5C,D). Hep3B cell growth was considerably inhibited by a decrease in cell proliferation and induction of apoptosis when insulin administration was associated with rapamycin (an mTORC1 inhibitor) treatment (Fig. 5C,D). Of note, treatment of Hep3B cells with the AKT1/2 inhibitor or the PI3K/mTOR dual inhibitor, NVP-BEZ235, led to a much more pronounced growth inhibition (Fig. 5C,D). At the molecular level, rapamycin treatment induced a down-regulation of the proteins involved in de novo lipogenesis, glycolysis, and the pentose 6-phosphate pathway and an up-regulation of ACADM and ECHS1 (Fig. 5A,B). However, the expression of AKR1B10, USP2a, chREBP, and PRKCλ/ι remained unaffected after rapamycin administration (Fig. 5A). Also, levels of the negative regulators of lipogenesis, INSIG2 and AMPKα2, were not rescued by rapamycin (Fig. 5A). Furthermore, levels of proteins involved in gluconeogenesis, including G6Pase, PGC-1α, MKP-3, and phosphorylated/inactivated FOXO1, were unmodified in rapamycin-treated cells (Fig. 5B). In contrast, the use of either AKT1/2 inhibitor or NVP-BEZ235 had a remarkable effect on the levels of all the proteins involved in lipogenesis, glycolysis, pentose phosphate, and gluconeogenesis in Hep3B cells (Fig. 5A,B). Treatment with either the AKT1/2 inhibitor or NVP-BEZ235 strongly down-regulated AKR1B10, USP2a, chREBP, and PRKCλ/ι proteins, whereas it up-regulated INSIG2 and AMPKα2 (Fig. 5A). The latter resulted in a significantly more pronounced inhibition of lipid biosynthesis and induction of fatty acid oxidation, when compared with rapamycin treatment (Supporting Fig. 6). LDH and G6PD activity was significantly reduced after the treatment with AKT1/2, NVP-BEZ235, and rapamycin, when compared with untreated cells, with the three drugs showing an equivalent lowering effect (Supporting Fig. 7). Equivalent results were obtained in HLE cells (data not shown).

Figure 5.

Insulin metabolic effects by the AKT pathway rely on mTORC1-dependent and -independent mechanisms. Insulin administration to Hep3B cells induces activation of de novo lipogenesis, suppression of fatty acid β-oxidation (A), up-regulation of glycolysis and the pentose phosphate pathway, down-regulation of gluconeogenesis (B), increase in cell proliferation (C), and decrease in apoptosis (D). Of note, inactivation of mTORC1 by rapamycin influences the levels of many, but not all, proteins downstream of AKT responsible for the observed metabolic effects. In contrast, treatment of Hep3B cells with the AKT1/2 inhibitor and the PI3K/mTOR inhibitor, NVP/BEZ235, strongly modulate the levels of all AKT effectors (A and B) and remarkably inhibit HCC cell growth in vitro (C and D). The latter implies the existence of an mTOR-dependent and -independent mechanism responsible for the biologic effect of insulin on HCC cells. (A and B) Representative immunoblotting on levels of AKT downstream effectors. (C) Effect of mTORC1 suppression by rapamycin and of AKT inhibition by AKT1/2 inhibitor or NVP/BEZ235 on proliferation of Hep3B cells treated with insulin. (D) Effect of mTORC1 suppression by rapamycin and of AKT inhibition by AKT1/2 inhibitor or NVP/BEZ235 on apoptosis of Hep3B cells treated with insulin. Each bar represents mean ± SD of three independent experiments conducted in triplicate. Tukey-Kramer's test: P < 0.002 a, versus control (untreated cells); b, versus DMSO; c, versus insulin-treated cells; d, versus cells treated with insulin and rapamycin; e, versus cells treated with insulin and AKT1/2 inhibitor.

The PI3K/mTOR Inhibitor, NVP-BEZ235, Reverses the Metabolic Features and Decreases Proliferation of Rat Preneoplastic Foci.

The pathogenetic link between deregulated insulin signaling and activation of the AKT pathway was further investigated in vivo. For this purpose, a group of transplanted rats was subjected to treatment with the dual PI3K/mTOR inhibitor, NVP-BEZ235. This drug was chosen because of its striking effect on Hep3B cell growth (Fig. 5C,D) to overcome the possible resistance to rapamycin that was previously described32 and to inhibit the molecular changes induced by AKT in an mTORC1-independent manner. Macroscopically, the livers treated with NVP-BEZ235 were characterized by the presence of paler spotty areas (the preneoplastic foci), when compared with control rats (Fig. 6A). This was at least partly the result of the depletion of the fat content in the lesions, as assessed by Oil-Red-O lipid staining (Supporting Fig. 8). At the cellular level, NVP-BEZ235 administration resulted in an ∼78% reduction of hepatocellular proliferation in foci subjected to NVP-BEZ235 treatment, when compared with corresponding lesions from rats treated with solvent alone (Fig. 6; Supporting Fig. 9A,B). Also, a decrease in apoptosis in rat preneoplastic foci treated with NVP-BEZ235 was detected (Supporting Fig. 10). At the molecular level, NVP-BEZ235 drastically decreased the levels of all the members of the AKT/mTOR cascade investigated and the expression of enzymes involved in lipogenesis, glycolysis, and the pentose phosphate pathway while triggering the up-regulation of AMPKα2 and proteins involved in fatty acid β-oxidation and gluconeogenesis (Fig. 6B-D). Also, NVP-BEZ235 induced a significant reduction in cholesterol levels, triglyceride levels, fatty acid biosynthesis, and LDH and G6PD activity as well as induction of fatty acid oxidation, when compared with preneoplastic foci from untreated rats (Supporting Fig. 11).

Figure 6.

Effect of treatment with the PI3K/mTOR dual inhibitor, NVP-BEZ235, on the growth of preneoplastic lesions induced by pancreatic islet transplantation in the liver. (A) In control rats (solvent only), the foci appear macroscopically as yellowish areas and become paler after NVP-BEZ235 administration. Treatment with the latter drug also results in marked reduction in proliferation, as assessed by bromodeoxyuridine labeling (B), as well as in decreased levels of proteins involved in glycolysis (C), lipogenesis (B, last four panels; D), and up-regulation of proteins involved in gluconeogenesis (C). Whole cell lysates were prepared from normal livers (NL), preneoplastic foci either treated with solvent alone (n = 8) or with NVP-BEZ235 (n = 8), and immunoblotted with indicated antibodies.

Suppression of TRB3, PHLPPs, and Up-regulation of the Phosphoinositide 3′-Kinases, PIK3CA and PIK3CB, in Rat HCC.

Because the metabolic effects on hepatocytes induced by insulin in the rat model are presumably mainly paracrine, the molecular alterations that we detected can be regarded as a direct consequence of insulin deregulation, mostly in preneoplastic foci, but not in HCC. Thus, we investigated whether alterations occurring exclusively in tumors might explain the unrestrained activation of the AKT/mTOR pathway in rat HCC (Supporting Fig. 12). Phosphatase and tensin homolog,33 the main AKT inhibitor, was progressively up-regulated from preneoplastic foci to HCC, excluding its involvement in the perpetuation of AKT/mTOR signaling in rat liver tumors. On the other hand, tribbles homolog 3 (TRB3), which inhibits AKT activation induced by insulin in the liver,34 was strongly up-regulated in preneoplastic foci, but suppressed in HCC. Accordingly, TRB3-AKT complexes (a marker of AKT inhibition) were elevated in preneoplastic foci and were extremely low in HCC. Because TRB3 counteracts de novo lipogenesis by stimulating degradation of ACAC through the constitutive photomorphogenic protein 1 (COP1) E3 ubiquitin ligase,35 we determined the levels of TRB3-ACAC and ACAC-COP1 complexes and the amount of ubiquitinated ACAC. TRB3-ACAC, ACAC-COP1, and ACAC ubiquitinated levels were highest in preneoplastic foci and lowest in HCC (Supporting Fig. 12). These data imply the presence of a mechanism at least partly counteracting ACAC activity (and AKT signaling) in the preneoplastic foci that is selectively lost at the tumor stage. Furthermore, pleckstrin homology domain leucine-rich repeat protein phosphatases 1 and 2 (PHLPP1 and PHLPP2), which control the amplitude and duration of AKT signaling by catalyzing its dephosphorylation,36 were selectively down-regulated in HCC.

Furthermore, we assessed the levels of the upstream inducers of AKT, namely, the phosphoinositide 3′-kinase catalytic subunit isoforms. Noticeably, PIK3CA and PIK3CB were up-regulated only in HCC, whereas no significant differences in the levels of PIK3CD and PIK3CG were detected in rat healthy livers, preneoplastic foci, and HCC.

Discussion

Mounting evidence suggests a role for insulin deregulation in human hepatocarcinogenesis.1, 9, 10 In our rat model, intrahepatic transplantation of a low number of pancreatic islets is able to ameliorate, but not to normalize, hyperglycemia, thus perpetuating local hyperinsulinemia and inducing a well-defined sequence of morphological events in the downstream hepatocytes, leading to the development of preneoplastic foci and HCC.19-22 In the present study, we investigated the function of the AKT/mTOR pathway as a downstream effector of the insulin-signaling cascade in this rat model. Overall, our results imply a central role of AKT signaling in mediating multiple biochemical and molecular events (e.g., up-regulation of lipogenesis, glycolysis, and the pentose phosphate pathway as well as down-regulation of fatty acid β-oxidation and gluconeogenesis) induced by insulin deregulation. Of note, subsequent experiments in human HCC cell lines supplemented with insulin showed that suppression of lipogenesis, glycolysis, and/or the pentose phosphate pathway results in growth restraint of these cells. This finding clearly indicates that the metabolic alterations induced by the AKT/mTOR cascade mediate at least some of the growth properties of insulin, thus linking metabolic alterations and aberrant liver growth in this model. Importantly, with the exception of ACLY levels, the metabolic alterations identified in preneoplastic foci were even more pronounced in frankly malignant HCC, suggesting a prominent role of AKT signaling and related metabolic changes both in early and late stages of hepatocarcinogenesis in this model. Nevertheless, it is important to notice that the aforementioned metabolic alterations presumably depend on, at least partly, different molecular mechanisms in preneoplastic and neoplastic rat liver lesions. Indeed, these metabolic changes can be easily explained for the preneoplastic foci, which are confined to the anatomic borders of the liver acinus and drain hyperinsulinemic blood from islet grafts. In HCC, however, the often scattered islet graft remnants can only be partly responsible for these metabolic alterations, although they can be regularly demonstrated within tumors.21 Although the intralesional insulin concentration cannot be measured, it can be assumed that the former hyperinsulinemia, induced by the islet grafts, is significantly diminished within HCC. Thus, the metabolic alterations detected in the tumors cannot exclusively be explained as a consequence of increased insulin signaling. Previous findings indicate that the IR is overexpressed in rat HCC, but not in preneoplastic foci.23 The latter finding might suggest that elevated levels of IR might provide a higher sensitivity for insulin signaling in HCC, despite the absence of elevated insulin levels. In the present study, we show that suppression of the AKT inhibitors, TRB3, PHLPP1, and PHLPP2, and up-regulation of AKT and its upstream inducers, PIK3CA and PIK3CB, occur exclusively in rat HCC. These alterations, together with the peculiar up-regulation of the ACAC stabilizer, AKR1B10, in HCC, indicate the existence in rat liver tumors of a complex genetic program leading to the perpetuation of the molecular mechanism that is solely dependent on insulin signaling in the preneoplastic foci. Additional molecular mechanisms might contribute to metabolic alterations in rat HCC and are currently under investigation.

At the molecular level, in accord with recent studies,29, 37, 38 we show that AKT signaling exerts its effects on metabolism through mTORC1-dependent and -independent mechanisms (Fig. 7). Under insulin growth-promoting stimuli, selective inhibition of mTORC1 by rapamycin triggered a significant decrease in glycolysis, a less pronounced reduction of lipogenesis, and no effect on both gluconeogenesis and some lipogenesis-related proteins (e.g., AKR1B10, USP2a, PRKCλ/ι, chREBP, AMPKα2, and INSIG2) in HCC cell lines. On the other hand, use of either the AKT1/2 inhibitor or concomitant suppression of PI3K and mTOR promoted a much stronger growth restraint, a more pronounced fall in lipid biosynthesis, and reactivation of gluconeogenesis in HCC cells supplemented with insulin. Besides their pathogenetic significance, the present results support the use of PI3K/mTOR and mTORC1/2 dual inhibitors, rather than mTORC1 single inhibitors, in the treatment of HCC with activated AKT. In accord with this hypothesis, it has been recently shown that NVP-BEZ235 exerted a remarkable antitumor activity in all HCC cell lines tested.39

Figure 7.

Schematic representation of the molecular mechanisms whereby AKT promotes metabolic changes and cell growth in hepatocytes. Activation/phosphorylation of AKT triggers lipid biosynthesis through mTORC1-dependent (induction of SREBP1, SREBP2, and PPARγ, and suppression of ACADM and ECHS1) and -independent mechanisms (up-regulation of USP2a, chREBP, and PRKCλ/ι and inhibition of AMPKα2 and INSIG2). AKT induces also the activation of glycolysis and the pentose phosphate pathway through mTORC1, whereas the latter is not involved in the suppression of gluconeogenesis (inhibition of PCK1, PGC1-α, G6Pase, and MKP-3). Aberrant activation of de novo lipogenesis, glycolysis, and the pentose phosphate pathway by AKT after insulin stimulation ultimately leads to hepatocyte proliferation (i.e., cell growth). Symbols: arrows, activation; blunted arrows, inhibition.

Among the targets strongly reactivated by the PI3K/mTOR dual inhibitor, NVP-BEZ235, in our rat model of insulin-driven hepatocarcinogenesis was AMPKα2. Because the AMPK system stimulates fatty acid oxidation (thus counteracting lipid biosynthesis) and alleviates hyperglycemia and hyperlipidemia,27 it might represent a pivotal metabolic tumor suppressor and a target for liver cancer prevention and therapy. In accord with this hypothesis, we found that the AMPK inducers, AICAR and metformin, were able to significantly restrain the growth of human HCC cell lines supplemented with insulin. Also, recent evidence indicates that metformin reduces liver-related death and the risk of HCC development in diabetic patients affected by T2DM and significantly prolongs the overall survival of diabetic patients with early-stage liver cancer.13, 40-42 Thus, these data together envisage the possibility of using PI3K/mTOR inhibitors and/or AMPK inducers both in the prevention of HCC development in patients affected by diabetes and metabolic syndrome and in the treatment of human HCC associated with the activation of the insulin-signaling cascade.

In summary, we showed that insulin deregulation triggers a number of metabolic alterations in the rat liver through the AKT/mTOR cascade that are associated with the appearance of preneoplastic foci. The metabolic changes induced by AKT after insulin chronic secretion occur through both mTORC1-dependent and -independent mechanisms. The activation of the AKT/mTOR cascade and the related metabolic alterations are maintained in HCC, although hyperinsulinemia is only one of the mechanisms among others responsible for the aforementioned changes. Thus, AKT has a central role in mediating the biologic and metabolic effects of insulin on hepatocytes and represents a promising target for the treatment of liver cancer.

Ancillary