Akt-Dependent Proapoptotic Effects of Dietary Restriction on Late-Stage Management of a Phosphatase and Tensin Homologue/Tuberous Sclerosis Complex 2–Deficient Mouse Astrocytoma

Malignant astrocytomas are the most common primary brain tumors and represent a leading cause of cancer-related deaths in children and the elderly (1–4). The inability to effectively manage astrocytomas has been due, in part, to the unique anatomic and metabolic environment of the brain that prevents the complete resection of tumor tissue and impedes the delivery of therapeutic agents. Although glucose is the preferred metabolic fuel for healthy neurons and glia, these cells can transition to noncarbohydrate fuels (ketone bodies) for energy under low glucose conditions (5, 6). In contrast, astrocytomas lack metabolic flexibility and are more susceptible than normal brain tissues to cell death mediated by oxidative stress or ATP depletion in response to reduced glucose availability (7–10). Hence, therapies that can exploit the differences in energy metabolism between normal brain cells and brain tumor cells should be effective for tumor management (5).

Defects in components of the phosphatase and tensin homologue (PTEN)/Akt signaling pathways are associated with the antiapoptotic and glycolytic features of many tumors, including astrocytomas (11–13). Akt is a major effector of receptor tyrosine kinases (RTK), which modulate cell growth, apoptosis, protein synthesis, and energy metabolism (13–15). Most astrocytomas exhibit constitutive Akt phosphorylation in association with reduced PTEN tumor suppressor expression or with increased activation of growth factor receptors (14, 16, 17). Akt inhibits apoptosis through inactivation of the proapoptotic factors BAD and procaspase-9 (12, 13, 18, 19). Akt also phosphorylates and inhibits the tuberous sclerosis (TSC1/TSC2) protein complex, leading to activation of the mammalian target of rapamycin (mTOR) and the up-regulation of cell growth and protein synthesis (13, 14, 20). Hence, targeting Akt-associated signaling pathways could be effective in managing astrocytomas.

Besides Akt activation, astrocytomas can also express deficiencies in the TSC2 protein (21, 22). Tumors lacking TSC2 expression are highly angiogenic, exhibit constitutive mTOR activity, and are acutely prone to apoptosis under energy restriction (13, 15, 20, 21, 23). The hypoxia-inducible transcription factor-1α (HIF-1α) is a major downstream effector of Akt that influences glucose flux and metabolism in gliomas and other tumors exhibiting the Warburg effect (13, 15). Astrocytomas frequently overexpress HIF-1α and associated target genes to include the type 1 glucose transporter protein (GLUT1) and those encoding glycolytic enzymes (13). It is noteworthy that gliomas and other tumors with defects in PTEN/Akt/TSC2 signaling are dependent on glycolysis for survival (15, 18, 24). Consequently, astrocytomas may be manageable with therapies that lower glucose availability, such as dietary restriction.

Dietary restriction (DR) has long been recognized to improve health, promote longevity, and reduce both the incidence and growth of many experimental tumors (5, 8, 25–30). The reduction in tumor growth is due in large part to dietary caloric restriction, although reductions in some minerals (e.g., copper) may contribute to therapeutic efficacy (31, 32). Our previous findings in experimental astrocytomas showed that DR is proapoptotic, antiangiogenic, and anti-inflammatory (5, 7, 8, 10, 31, 33). These effects arise in part from a systemic down-regulation of glucose and insulin-like growth factor-I (IGF-I) levels. The transition from glucose to ketone bodies as the primary energy source for the brain under DR exploits the genetic defects in brain tumor cells while enhancing the health and vitality of normal neurons and glia according to principles of evolutionary biology and metabolic control theory (5, 7, 10). In this regard, DR stands apart from all other therapies for malignant astrocytoma.

Although prior studies showed that DR could manage mouse and human astrocytomas when initiated early in tumor development (before or shortly after tumor implantation, early-onset DR), no prior studies have identified the mechanisms involved or determined if DR is also effective in managing astrocytoma growth when the diet is initiated at late stages of tumor development (late-onset DR). The objective of this study was to examine the efficacy of DR as a therapy for the late-stage management of the CT-2A malignant mouse astrocytoma and to identify potential mechanisms of action. CT-2A is a rapidly growing, poorly differentiated, and highly angiogenic malignant astrocytoma that overexpresses the NG2 angiogenic proteoglycan (8, 34, 35). Immunohistochemical analyses of CT-2A showed staining for glial fibrillary acidic protein and the Sox9 and Sox10 oligodendrocyte cell precursor markers compared with normal neural cells in adjacent brain tissue (36). Moreover, ¹H nuclear magnetic resonance spectroscopy revealed differences between the tissue metabolic profile in CT-2A and contralateral normal brain tissue that are consistent with those found in malignant human gliomas to include increased levels of lactate, alanine, glycine, and glutamate (36). Increased lactate levels in CT-2A probably imply greater utilization of glucose and glycolysis in the tumor relative to normal brain tissue (36). Warburg proposed that defective mitochondrial respiration underlies the glycolytic phenotype in tumor cells (37, 38). In support of this view, recent experiments completed by Kiebish et al. showed that CT-2A cells contain numerous defects in the content and distribution of inner mitochondrial membrane lipids compared with either normal mouse astrocytes or mouse brain tissue (39). These lipid defects are also associated with reduced activities of several enzymes of the mitochondrial electron transport chain, making it improbable that CT-2A cells can generate an adequate supply of ATP through respiration for survival. Thus, CT-2A shares histologic, immunohistochemical, proliferative, angiogenic, and metabolic characteristics with high-grade human astrocytomas.

In this study, we found that the CT-2A astrocytoma is similar to human astrocytomas in exhibiting diminished PTEN and TSC2 protein expression and increased expression of the glycolytic enzyme pyruvate kinase type-M2 (PKM2). We also found that late-onset DR reduced CT-2A tumor growth, delayed malignant progression, and significantly extended survival. Furthermore, our findings emphasize an important role for autocrine/paracrine activation of the IGF-I/Akt signaling pathway in potentiating the antiapoptotic phenotype of astrocytomas and suggest that DR targets this signaling pathway to reduce CT-2A astrocytoma growth.

Mice and experimental astrocytoma. Mice of the C57BL/6J strain were obtained from The Jackson Laboratory and were propagated in the Boston College Animal Care Facility as previously described (40). Adult male mice (∼14 wk of age) were used in this study and were housed individually in plastic cages with filter tops containing Sani-Chip bedding. Cotton nesting pads were provided for all mice for warmth and the mouse room was maintained at 22 ± 1°C on a 12 h light/12 h dark cycle. The syngeneic malignant mouse astrocytoma, CT-2A, was originally produced by implantation of a chemical carcinogen, 20-methylcholanthrene, into the cerebrum of C57BL/6J mice and was classified as malignant astrocytoma (34). The procedures for animal use were in strict adherence to the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care Committee at Boston College.

Antibodies, primers, and reagents. Antibodies were obtained from Cell Signaling against phospho-PTEN (S-380, R-382/383), total PTEN, PKM2, phospho-Akt (S-473), total Akt, phospho-mTOR (S-2448), phospho-BAD (S-136), caspase-9, caspase-3, phospho–IGF-IRβ (Y-1135/1136)/IRβ (Y-1150/1151), total IGF-IRβ, and TSC1 and for the lysis buffer. An anti–HIF-1α antibody was purchased from Abcam. Anti-GLUT1 antiserum was kindly provided as a gift from I.A. Simpson (Department of Neural and Behavioral Sciences, Hershey Medical Center, Hershey, PA). Anti-TSC2 (tuberin), anti–β-actin, goat anti-rabbit IgG-HRP, and goat anti-mouse IgG-HRP antibodies were obtained from Santa Cruz Biotechnology, Inc. The TRIzol reagent was obtained from Invitrogen. Oligo(dT) primers, Moloney murine leukemia virus reverse transcriptase, and Taq polymerase were purchased from Promega. Information for IGF-IR (forward 5′-TGACTTCTGCTCAAATGCTCC-3′ and reverse 5′-TGTTCCTGGTGTTTATGTCCC-3′), IGF-I (forward 5′-GCTCTTCAGTTCGTGTGTGG-3′ and reverse 5′-TTGGGCATGTCAGTGTGG-3′), HIF-1α (forward 5′-GGTGCTAACAGATGACGGCGAC-3′ and reverse 5′-TCAGCACCAAGCACGTCATGGG-3′), and β-actin (forward 5′-TGTGATGGTGGGAATGGGTCAG-3′ and reverse 5′-TTTGATGTCACGCACGATTTCC-3′) genes were obtained from the National Center for Biotechnology Information under Genbank accession numbers NM_010513, AF_440694, NM_010431, and NM_007393, respectively. 2,2,2-Tribromoethanol and tert-amyl alcohol were obtained from Sigma. The Stanbio Enzymatic Glucose Assay kit was obtained from StanBio Laboratories. The Lactate Assay kit was obtained from the Biomedical Research Service Center. The DC Protein Assay kit was purchased from Bio-Rad.

Intracerebral tumor implantation. Small tumor fragments were implanted into the right cerebral hemisphere of C57BL/6J mice using a trocar as we described previously (41). Briefly, mice were anesthetized with 2,2,2-tribromoethanol, i.p., and their heads were shaved and swabbed with 70% ethanol under sterile conditions. Small tumor fragments (∼1 mm³, estimated using a 1 mm × 1 mm grid) from a donor mouse were implanted into the right cerebral hemisphere of anesthetized recipient mice. All of the mice recovered from the surgical procedure and were returned to their cages when fully active. Initiation of tumors from intact tumor fragments is preferable to initiation from cultured cells because the fragments contain an already established microenvironment that facilitates rapid tumor growth (41).

Dietary regimens, body weight, and food intake measurements. All mice received PROLAB RMH 3000 chow. This contained a balance of mouse nutritional ingredients and delivers 4.1 kcal/g gross energy, where fat, carbohydrate, protein, and fiber comprised 55, 520, 225, and 45 g/kg of the diet, respectively. Mice were separated into individual cages 2 wk before tumor implantation, during which time body weight and food intake measurements were recorded every 3 d to gather baseline information. For the intracerebral tumor study, all tumor-bearing mice were fed ad libitum for the first 10 d after tumor implantation and were then randomly assigned to one of two diet groups that received rodent chow in either unrestricted (UR; n = 9) or restricted (DR; n = 9) amounts. The two groups were matched for body weight (∼26.9 g) before the initiation of DR. The feeding regimen for the DR mice was designed to reduce body weights by ∼30% relative to values recorded 9 d after tumor implantation. More specifically, the DR feeding regimen involved alternating days of fasting and calorically restricted feeding. On the feeding days, DR mice were provided with 75% of the average amount of energy consumed by the UR group on the previous day. Body weight and food intake of all mice was recorded daily (1:00 p.m. to 3:00 p.m.). The food intake for the UR mice was determined daily by subtracting the weight of the food pellets remaining in the food hopper from the amount that was provided 24 h earlier. New food was provided every 4 d. For mice in the DR group, food pellets were dropped directly into each cage for easy access. Water was provided ad libitum for all mice.

Analysis of DR on malignant progression and survival. To examine the effect of DR on malignant progression, we conducted a survival study in which subcutaneous tumor volume served as an end point as we previously described (10). CT-2A tumor fragments, suspended in 200 μL of PBS, were injected s.c. into the left lateral flanks of adult male C57BL/6J mice. DR was initiated 14 d after tumor inoculation, when subcutaneous tumors were ∼1,000 mm³ in volume. On day 14, the tumor-bearing mice were randomly separated into either unrestricted (UR, n = 7) or DR (n = 8) groups that were matched for body weight and tumor volume. The diets for the UR and DR groups were identical to those described above. DR mice that achieved the 30% body weight reduction were then fed a limited amount of rodent chow on a daily basis to maintain this level of weight reduction. Tumor volumes were determined every 2 to 5 d by measuring the two perpendicular diameters of the tumor using electronic calipers. Volumes were computed from the equation of an ellipsoid (length² × width/2) as we previously described (10). Tumor growth rates (% × day⁻¹) were determined from the equation ln(2)/DT, where DT represents the time necessary for a tumor to double in volume from the value measured 14 d after implantation (42). Survival was defined as the time necessary for subcutaneous tumor volume to exceed 2,500 mm³. Mice were euthanized when tumors ulcerated or when mice showed morbidity as defined by our Institutional Animal Care and Use Committee guidelines.

Plasma glucose and lactate measurements. Mice were anesthetized with isoflurane and euthanized by exsanguination, involving collection of blood from the heart (i.e., ventricles) into heparinized Eppendorf tubes using a tuberculin syringe outfitted with a 25G needle. The blood was centrifuged at 3,000 × g for 10 min, the plasma supernatant was collected, and aliquots (two fractions per aliquot of plasma supernatant) were stored at −80°C before analysis. Plasma glucose and lactate levels were measured in a spectrophotometer using the appropriate enzymatic assay.

Tissue processing. Following euthanasia by cardiac puncture, tumors were dissected from normal-appearing brain tissue and immediately placed on dry ice to reduce tissue degradation. Residual normal-appearing contralateral (i.e., left) brain tissue was also removed and frozen on dry ice. Frozen specimens were weighed and stored at −80°C before RNA and protein isolation. CT-2A tumor and contralateral normal brain tissues were dissected into two sections, with one half being used for RNA isolation and the other half for protein isolation.

Western blot analysis. Frozen CT-2A tumor and contralateral normal brain tissues were homogenized in ice-cold lysis buffer containing 20 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1 mmol/L Na₂EDTA, 1 mmol/L EGTA, 1% Triton, 2.5 mmol/L NaPP_i, 1 mmol/L α-glycerophosphate, 1 mmol/L Na₃PO₄, 1 μg/mL leupeptin, and 1 mmol/L phenylmethylsufonyl fluoride. The same protocol was used to prepare protein homogenates of CT-2A cells and mouse astrocytes. Lysates were transferred to 1.7 mL Eppendorf tubes, mixed on a rocker for 1 h at 4°C, and then centrifuged at 8,100 × g for 20 min. Supernatants were collected and protein concentrations were estimated using the Bio-Rad detergent-compatible protein assay. Approximately 10 to 40 μg of total protein from each tissue sample and 50 to 60 μg of protein from each in vitro sample were denatured with SDS-PAGE sample buffer [63 mmol/L Tris-HCl (pH 6.8), 10% glycerol, 2% SDS, 0.0025% bromphenol blue, and 5% 2-mercaptoethanol] and were resolved by SDS-PAGE on 4% to 12% Bis-Tris gels (Invitrogen). Proteins were transferred to a polyvinylidene difluoride immobilon TM-P membrane (Millipore) overnight at 4°C and blocked in either 5% nonfat powdered milk or 5% bovine serum albumin in TBS with Tween 20 (pH 7.6) for 1 to 3 h at room temperature. Membranes were probed with primary antibodies overnight at 4°C with gentle shaking. The blots were then incubated with the appropriate secondary antibody for 1 h at room temperature (∼22°C) and bands were visualized with enhanced chemiluminescence. Each membrane was stripped and reprobed for β-actin as an internal loading control and the ratio of the indicated protein to β-actin was analyzed by scanning densitometry (FluorChem 8900 Software).

Semiquantitative reverse transcription-PCR. Total RNA was isolated from tumor and normal brain tissues according to the manufacturer's protocol. The same protocol was used to prepare RNA homogenates of CT-2A cells and mouse astrocytes. RNA concentration and purity were determined by spectrophotometric measurements at 260 and 280 nm, respectively. Single-stranded cDNA was synthesized from total RNA (3 μg) in a 20-μL reaction with Moloney murine leukemia virus reverse transcriptase according to the manufacturer's protocol. cDNA was used for PCR amplification of the 212-, 365-, 318-, and 363-bp regions of the CDS for IGF-IR (nt 1034-1447), IGF-I (nt 103-314), HIF-1α (nt 303-667), and β-actin (nt 207-719) mRNA transcripts, respectively. Gradient PCR was done to obtain optimal primer annealing temperatures. To determine the optimal linear range for semiquantitative reverse transcription PCR (RT-PCR), PCR was done at increasing cycle numbers. PCR amplification was done with Taq DNA polymerase (Promega) using similar protocols for each gene in a 25-μL reaction volume for each sample. PCR products (12 μL) were separated on 1.0% agarose gels containing ethidium bromide, visualized with UV light, and analyzed using the FluorChem 8900 software.

Cell culture conditions. The CT-2A mouse astrocytoma cell line was established as previously described (34). The C8-D1A astrocyte cell line was purchased from the American Type Culture Collection. Both the CT-2A astrocytoma and the C8-D1A astrocyte cell line originated from C57BL/6J mice. Cell lines were maintained in DMEM supplemented with 10% fetal bovine serum and 0.5% penicillin/streptomycin. The cells were cultured in a CO₂ incubator with a humidified atmosphere containing 95% air and 5% CO₂ at 37°C. To determine the influence of glucose on the expression of IGF-I and IGF-IR, CT-2A cells and mouse astrocytes were incubated with 0 to 25 mmol/L glucose and with 0 to 4 mmol/L glutamine in serum-free DMEM for a total of 12 to 18 h. To harvest cells, the flasks were washed once with PBS and treated with trypsin-EDTA for 2 min. Cells were removed from the flask with gentle pipetting in PBS, transferred to 15 mL conical centrifuge tubes, centrifuged for 3 min at 1,000 rpm, and the pellets were washed again with PBS. Finally, the pellets were lysed with lysis buffer and were centrifuged at 8,100 × g for 20 min at 4°C. The lysates were collected and stored at −80°C for protein analysis. Both floating and adhered cells were collected and lysed when preparing homogenates for Western blot analysis and semiquantitative RT-PCR.

Statistical analysis. Body weight, food intake, tumor growth, and plasma metabolite levels were analyzed by the Student's t test to perform a two-sided pairwise comparison among the UR and DR groups (SPSS 14.0). In each figure, error bars are mean ± SE. Western blot and RT-PCR analyses represent semiquantitative estimates of the amount of the indicated protein or mRNA that is present in a tissue extract. The former was taken into account in the statistical evaluation of the data.

The CT-2A astrocytoma is deficient in the PTEN and the TSC2 tumor suppressor proteins and overexpresses PKM2. Western blot analysis showed that PTEN and TSC2 protein expression was significantly lower in CT-2A tumor tissue than in contralateral normal brain tissue (Fig. 1A and B). The level of PTEN (S-380,R-382/383) phosphorylation was also less in the CT-2A tumor tissue than in the normal brain tissue (Fig. 1A). The PTEN and TSC2 protein deficiencies in the CT-2A cells were confirmed in vitro when compared with the level of expression in the transformed nontumorigenic syngeneic mouse astrocyte cell line (Fig. 1D and E). In addition, the expression of PKM2 was significantly greater in CT-2A tumor than in the contralateral normal brain, indicating that CT-2A is more glycolytic than normal brain tissue (Fig. 1C). Although CT-2A was also found to be TSC1 protein deficient in vivo, the level of TSC1 expression in CT-2A cells in vitro was not significantly different from the level detected in control mouse astrocytes (data not shown). Viewed together, these data show that the expression of the PTEN and the TSC2 tumor suppressor protein is less in CT-2A tumor tissue than in normal brain tissue and that these deficiencies correspond with increased expression of the PKM2 marker for glycolysis (43, 44). It is thus far unknown whether the PTEN and TSC2 tumor suppressor protein deficiencies in CT-2A are caused by genetic, epigenetic, or regulatory abnormalities. However, the deficiencies in both of the TSC proteins in vivo, taken together with the finding that PTEN and TSC2 are deficient in CT-2A cells in vitro compared with control mouse astrocytes, indicate that the latter two phenomena are likely responsible for these observations.

DR reduces body weight and plasma glucose/lactate levels and improves the health of tumor-bearing mice. The intermittent feeding paradigm significantly reduced total caloric (energy) intake and produced a stair-step-like reduction in the average body weight for the DR group between 10 and 18 days after tumor implantation (Fig. 2A-C). The average body weight for the DR group was ∼25% less than that for the UR group by cf18 days after tumor implantation (P < 0.001). This DR feeding regimen resulted in a 63% dietary restriction for the DR group relative to the UR group (Fig. 2C). The reduced energy intake for the DR group was also associated with significantly lower plasma glucose and lactate levels (Fig. 2D and E). It is necessary to mention that the plasma lactate levels of mice in the DR group were similar to those of healthy non–tumor-bearing C57BL/6J mice (8.7 ± 0.4 mmol/L) under identical conditions. Despite a reduction in total body weight, the DR mice seemed healthy and were more active than the UR mice as assessed by ambulatory and grooming behavior. No signs of vitamin or mineral deficiency were observed in the DR mice according to standard criteria for mice (10). These findings are consistent with the well-recognized health benefits of transient diet restriction in rodents (7, 9, 30).

DR reduces intracerebral tumor growth, delays malignant progression, and increases survival in CT-2A–bearing mice. CT-2A tumor weight was ∼70% less in the DR group than in the UR group despite the late onset of DR (Fig. 3A). The rate of subcutaneous tumor growth was ∼63% lower in the DR group than in the UR group (7.2 ± 1.0% × day⁻¹ for the DR group versus 19.2 ± 3.6% × day⁻¹ for the UR group; Fig. 3B). To examine the effect of DR on malignant progression, we conducted a survival study in which tumor volume served as an end point for mice injected s.c. with CT-2A tissue. DR significantly increased the survival of mice bearing subcutaneous tumors (Fig. 3C). Collectively, these results show that DR significantly reduced late-stage tumor growth, delayed malignant progression, and improved the survival of mice bearing the CT-2A astrocytoma.

DR suppresses constitutive Akt phosphorylation but not mTOR or p70S6K phosphorylation in CT-2A. Western blot analysis was done to examine the effects of DR on Akt signaling in CT-2A. The expression of phosphorylated Akt (S-473) was significantly greater in the CT-2A tumor tissue than in normal brain tissue under unrestricted feeding conditions (Fig. 4A). Phosphorylated Akt (S-473) expression was significantly lower in tumors from the DR group than in tumors from the UR group. Expression of phosphorylated mTOR (S-2448) was greater in the tumor tissue than in the normal brain tissue. DR had no significant effect on mTOR (S-2448) phosphorylation (Fig. 4B) or on the phosphorylation state of the mTOR substrate p70S6K (T-389; Fig. 4C). Viewed together, these results indicate that phosphorylated Akt (S-473) and phosphorylated mTOR (S-2448) are expressed constitutively in the CT-2A astrocytoma, and also show that DR reduced expression of phosphorylated Akt (S-473) but did not alter phosphorylation of either mTOR (S-2448) or of p70S6K (T-389) in this model of malignant astrocytoma.

DR reduces HIF-1α and GLUT1 expression in the CT-2A astrocytoma. Western blot analysis and RT-PCR were done to examine the effects of DR on HIF-1α and GLUT1 expression in contralateral normal brain and in CT-2A. The expression of HIF-1α mRNA and protein were up-regulated in CT-2A compared with contralateral normal brain (Fig. 5A and B). A higher molecular weight isoform of the GLUT1 protein, specific to endothelial cells (45), was also overexpressed in CT-2A compared with normal brain (Fig. 5B). DR significantly decreased the expression of HIF-1α and GLUT1 in CT-2A but increased the expression of a lower molecular eight glial cell-specific GLUT1 isoform in contralateral normal brain (45). The expression of the lower molecular weight GLUT1 isoform was not apparent in CT-2A, although a doublet was visible in the size range of the higher molecular weight isoform (Fig. 5B). On the whole, these data show that expression of HIF-1α and GLUT1 were up-regulated in CT-2A tumor tissue compared with contralateral normal brain tissue. Furthermore, DR reduced HIF-1α and GLUT1 expression in CT-2A tumor tissue but increased GLUT1 expression in contralateral normal brain tissue.

Expression of IGF-I and IGF-IR in the CT-2A astrocytoma, influence of glucose on IGF-I, and IGF-IR expression. Western blot analysis and RT-PCR were done to examine the effects of DR (in vivo) or glucose (in vitro) on the expression of IGF-I and IGF-IR in contralateral normal brain, in mouse astrocytes, and in the CT-2A astrocytoma. The results show that mRNA expression of both IGF-I and IGF-IR as well as protein expression of IGF-IR was significantly greater in the CT-2A tumor tissue than in normal brain tissue under UR feeding conditions (Fig. 6A and B). DR reduced IGF-I mRNA expression and IGF-IR protein expression in CT-2A (Fig. 6A and B). In addition, the level of IGF-I receptor/insulin receptor (IGF-IR/IR) tyrosine phosphorylation was lower in tumors from the DR group than in tumors from the UR group (Fig. 6C). Viewed together, these data indicate that the IGF-IR pathway is active in the CT-2A astrocytoma and that DR could target IGF-IR signaling to inhibit tumor growth.

Relative to nontumorigenic mouse astrocytes, CT-2A cells exhibited constitutive tyrosine phosphorylation of IGF-IR/IR and overexpressed IGF-IR (Fig. 6D). To determine the influence of glucose on the IGF-IR pathway in mouse astrocytes and in CT-2A cells, we incubated cells in serum-free medium containing 0 to 25 mmol/L glucose for a total of 12 to 18 hours. Cells were then harvested and both RNA and protein lysates were prepared. The results of Western blot analysis showed that the level of IGF-IR/IR tyrosine phosphorylation in CT-2A cells increased as the concentration of glucose in the culture medium increased (Fig. 6E). In contrast, glucose did not seem to affect IGF-IR protein expression in either CT-2A cells or in mouse astrocytes. The dose-dependent increase in the level of IGF-IR/IR tyrosine phosphorylation in CT-2A cells was associated with a dose-dependent increase in IGF-I mRNA expression (Fig. 6F). These findings indicate that DR reduces constitutive expression of IGF-I and IGF-IR and also that glucose stimulates IGF-I transcription and increases tyrosine phosphorylation of IGF-IR/IR in CT-2A tumor cells. Thus, IGF-I could potentially act as an autocrine/paracrine factor in this mouse astrocytoma.

DR inhibits BAD phosphorylation and reduces procaspase-9/procaspase-3 expression in CT-2A. We next determined whether the DR-induced reduction of Akt phosphorylation in CT-2A was associated with changes in BAD (S-136) phosphorylation and procaspase-9/procaspase-3 expression. BAD (S-136) phosphorylation was significantly lower in tumors from the DR group than in tumors from the UR group (Fig. 7). This decrease in BAD phosphorylation corresponded with a significant reduction in procaspase-9 and procaspase-3 expression in tumors from the DR group relative to the UR group (Fig. 7), suggesting that DR increased procaspase cleavage in the CT-2A astrocytoma. These findings correspond with prior evidence that DR is proapoptotic in CT-2A (31). Taken together, these results indicate that the DR-induced reduction of Akt (S-473) phosphorylation in CT-2A was associated with the dephosphorylation of BAD (S-136) as well as with enhanced cleavage of procaspase-9/procaspase-3.

Most astrocytomas have constitutive phosphorylation of Akt in association with reduced PTEN expression or with increased receptor tyrosine kinase activation (11, 12, 13, 14, 17). Many astrocytomas are also TSC2 protein deficient (21, 23). Astrocytomas and other tumors harboring PTEN/Akt/TSC2 pathway abnormalities are heavily dependent on glycolysis to satisfy their bioenergetic requirements for survival (15, 18, 24). Thus, astrocytomas are potentially manageable with therapies that can target glycolysis without causing toxicity. Although DR, which reduces glucose/growth factor metabolism, is effective in managing experimental mouse and human astrocytomas when initiated early in tumor development (within a few days of tumor implantation; refs. 8, 31, 33), no prior studies have identified the mechanisms involved or have determined if DR is also effective in managing astrocytoma growth when initiated at late stages of tumor development (weeks after tumor implantation). We now show for the first time that late-onset DR could reduce tumor growth, delay malignant progression, and significantly extend survival in mice bearing the syngeneic PTEN/TSC2–deficient CT-2A astrocytoma. We also confirm the association of PTEN/Akt signaling and IGF-IR pathway activity with the antiapoptotic and glycolytic phenotype of an experimental astrocytoma. Our findings suggest that the suppression of Akt-dependent antiapoptotic pathways, in concert with diminished TSC2 expression and persistently elevated mTOR activation, may contribute to the antitumor effects of DR. The potential mechanisms by which DR manage the CT-2A astrocytoma is presented in Fig. 8.

We found that PTEN and TSC2 protein expression was significantly lower in the CT-2A astrocytoma and cultured CT-2A cells than in contralateral normal brain or in control C57BL/6J mouse astrocytes. These deficiencies were associated with constitutive mRNA expression of IGF-I, elevated mRNA and protein expression of IGF-IR, as well as with increased phosphorylation of Akt and mTOR. Because Akt is a major effector of the IGF-IR pathway in human astrocytomas (46), the up-regulation of IGF-I, phosphorylated IGF-IR, total IGF-IR, and phosphorylated Akt could confer a growth and survival advantage to the CT-2A tumor cells. We found that DR decreased expression of IGF-I mRNA, IGF-IR tyrosine phosphorylation, total IGF-IR protein, and phosphorylated Akt in CT-2A; however, DR did not affect mTOR or p70S6K phosphorylation. Our current findings agree with in vitro evidence that glucose availability can influence the expression of IGF-I mRNA in an experimental glioma (47, 48) and support studies showing that reduced expression of components of the IGF-IR pathway and phosphorylated Akt contribute to the growth-inhibitory effects of DR in experimental rodent tumors (25, 31, 49). It seems worth noting that DR does not influence PTEN or TSC2 protein expression in either CT-2A tumor tissue or in contralateral normal brain tissue (data not shown), suggesting that the tumor growth–inhibitory effects of DR on the CT-2A astrocytoma do not involve altered PTEN or TSC2 expression.

Circulating IGF-I is not readily transported across the blood-brain barrier, and local production is believed to be the primary source of brain IGF-I (46). Although IGF-I is overproduced in many astrocytomas, increased protein levels are not detected in the cerebrospinal fluid, suggesting that IGF-I acts locally in an autocrine/paracrine manner to influence tumor malignancy (46). Consequently, our findings suggest that a local down-regulation of IGF-I and IGF-IR expression and reduced IGF-IR/IR tyrosine phosphorylation could contribute in part to the antitumor effects of DR. The observation that glucose stimulates IGF-I transcription and IGF-IR/IR tyrosine phosphorylation in a dose-dependent manner in CT-2A cells in vitro lends additional support to the contention that IGF-I may act as an autocrine/paracrine growth factor in this experimental astrocytoma model. Although the IR is expressed in several regions of the adult brain and can be expressed in human gliomas (50, 51), the insulin molecule itself is not produced in the central nervous system or in the central nervous system tumors and must cross the blood-brain barrier to reach the receptor. Furthermore, a recent study showed that less than 1% of peripherally given insulin enters the brain from the circulation in rodents (51). Because DR is known to reduce circulating insulin levels (51), it seems unlikely that the IR plays an appreciable role in mediating the growth-inhibitory effects of DR in the CT-2A astrocytoma. In addition, our data are consistent with reports that tumors deficient in TSC2 maintain a high level of mTOR activity, even when the phosphorylation of Akt is inhibited, and are also more susceptible to the proapoptotic and growth-inhibitory effects of energy restriction as would occur under DR (15, 20, 23, 52, 53).

The phosphorylation and inactivation of BAD (on S-136) and procaspase-9 mediate, in part, the antiapoptotic actions of Akt activation (18, 19). BAD transmits proapoptotic signals generated during glucose/growth factor deprivation. We found that BAD was constitutively phosphorylated on S-136 in CT-2A compared with contralateral normal brain and showed that DR suppressed BAD phosphorylation and increased procaspase-9/procaspase-3 cleavage. BAD stimulates apoptosis by heterodimerizing with and inactivating the antiapoptotic proteins Bcl-2 and Bcl-x (18, 19). DR is known to reduce Bcl-2 and Bcl-xL expression and to increase the expression of Bax, Apaf-1, caspase-9, and caspase-3 in experimental carcinomas, suggesting that DR could inhibit tumor growth in part by inducing mitochondrial-dependent apoptosis mediated by the dephosphorylation of BAD (54). Our findings agree with biochemical and morphologic evidence that DR is proapoptotic in malignant astrocytomas and support evidence that BAD coordinates glucose/IGF-I homeostasis and the induction of apoptosis (7, 18, 19, 31). Overall, our findings show that reduced glucose availability and IGF-I expression play a key role in suppressing Akt and in mediating the proapoptotic effects of DR in the CT-2A astrocytoma.

We also found that the expression of HIF-1α, GLUT1 (larger molecular weight isoform specific to BBB endothelial cells), and PKM2 were higher in CT-2A than in contralateral normal brain and that DR decreased HIF-1α at transcriptional and posttranscriptional levels while also reducing GLUT1. These observations agree with evidence that (a) HIF-1α, GLUT1, and PKM2 are constitutively overexpressed in malignant astrocytomas (55, 56); (b) reduced PTEN expression and constitutive activation of Akt could potentiate HIF-1α protein stabilization and increase HIF-1α transcription (16, 52); and (c) increased HIF-1α expression is associated with the up-regulation of molecules that promote cell growth, glucose uptake, and glycolysis (52, 57). Our findings suggest an important role for glucose/IGF-I metabolism in sustaining an elevated level of HIF-1α activity in CT-2A. The observation that the proapoptotic effects of DR in CT-2A were associated with reduced expression of HIF-1α and GLUT1 agrees with evidence that elevated HIF-1α and GLUT1 expression may confer protection against apoptosis in malignant astrocytomas (18). Furthermore, the up-regulation of a GLUT1 isoform specific to brain endothelial cells and the down-regulation of this glycoprotein by DR may also reflect the antiangiogenic effects of DR in this malignant astrocytoma (8, 31). In contrast to the CT-2A astrocytoma, DR up-regulated a lower molecular weight GLUT1 isoform specific to glial cells in normal brain tissue. These findings could represent the necessity of intact mTOR signaling in regulating GLUT1 protein expression in normal brain but not in CT-2A under low glucose conditions.

The redistribution of ATP between essential and nonessential processes is imperative for assuring cell survival when glycolytic energy becomes limiting as would occur under DR (5, 15). Most of the ATP generated through glucose metabolism in normal cells are used to maintain the activity of ion-motive ATPase pumps and protein synthesis (5, 58). Under energy restriction, cells reduce nucleotide and protein synthesis to preserve enough ATP to maintain pump activity and ionic homeostasis. Cells with impaired mitochondria, which are unable to budget the ATP necessary to maintain ionic and osmotic equilibria, can die from metabolic catastrophe involving ionic imbalances, oxidative stress, and failure of ATPase pumps (5, 15, 58). To this end, the DR-induced down-regulation of glycolysis mediated by the suppression of IGF-I/Akt signaling will impair viability in tumor cells, which cannot readily transition to fuels other than glucose (5, 59). Overall, our findings exemplify the potential efficacy and versatility of DR as a broad-spectrum inhibitor of astrocytoma growth and suggest that DR may extend survival in patients with advanced PTEN/TSC2–deficient malignant astrocytomas.

No potential conflicts of interest were disclosed.

We thank Alexis M. Jones, Katherine E. Kurgansky, Linh Ta, and John G. Mantis for technical assistance and Michael A. Kiebish, Anthony C. Faber, and Thomas C. Chiles for helpful comments while drafting the manuscript. We thank I.A. Simpson for kindly providing anti-GLUT1 antiserum as a gift.