Hexokinase-2-mediated aerobic glycolysis is integral to cerebellar neurogenesis and pathogenesis of medulloblastoma
© Gershon et al.; licensee BioMed Central Ltd. 2012
Received: 16 March 2012
Accepted: 9 August 2012
Published: 23 January 2013
While aerobic glycolysis is linked to unconstrained proliferation in cancer, less is known about its physiological role. Why this metabolic program that promotes tumor growth is preserved in the genome has thus been unresolved. We tested the hypothesis that aerobic glycolysis derives from developmental processes that regulate rapid proliferation.
We performed an integrated analysis of metabolism and gene expression in cerebellar granule neuron progenitors (CGNPs) with and without Sonic Hedgehog (Shh), their endogenous mitogen. Because our analysis highlighted Hexokinase-2 (Hk2) as a key metabolic regulator induced by Shh, we studied the effect of conditional genetic Hk2 deletion in CGNP development. We then crossed Hk2 conditional knockout mice with transgenic SmoM2 mice that develop spontaneous medulloblastoma and determined changes in SmoM2-driven tumorigenesis.
We show that Shh and phosphoinositide 3-kinase (PI3K) signaling combine to induce an Hk2-dependent glycolytic phenotype in CGNPs. This phenotype is recapitulated in medulloblastoma, a malignant tumor of CGNP origin. Importantly, cre-mediated ablation of Hk2 abrogated aerobic glycolysis, disrupting CGNP development and Smoothened-induced tumorigenesis. Comparing tumorigenesis in medulloblastoma-prone SmoM2 mice with and without functional Hk2, we demonstrate that loss of aerobic glycolysis reduces the aggressiveness of medulloblastoma, causing tumors to grow as indolent lesions and allowing long-term survival of tumor bearing mice.
Our investigations demonstrate that aerobic glycolysis in cancer derives from developmental mechanisms that persist in tumorigenesis. Moreover, we demonstrate in a primary tumor model the anti-cancer potential of blocking aerobic glycolysis by targeting Hk2.
See commentary article:http://www.biomedcentral.com/1741-7007/11/3
KeywordsWarburg effect Aerobic glycolysis Medulloblastoma Smoothened Brain tumor Cerebellum
Aerobic glycolysis, the metabolism of glucose to lactate despite the availability of oxygen, is observed in diverse cancers, a phenomenon known as the Warburg effect[1, 2]. Indeed, many cancers, including brain tumors, demonstrate increased glucose utilization, suggesting that glycolytic metabolism may confer a selective advantage[3, 4]. Less is known about metabolic adaptations during development. Examining these adaptations is important because metabolic patterns that support cancerous growth may derive from genetic programs that evolved to support developmental growth.
Neurogenesis, like tumorigenesis, requires rapid cellular proliferation, but under precise control. In human brain development, over 80 billion cerebellar granule neurons (CGNs) are generated in the first 6 months of life. Many of the developmental milestones observed in the first year of life are directly attributed to proper formation of cerebellar neural circuits involving the granule neurons. Excessive proliferation and retarded maturation of CGNPs, often driven by mutations in neurodevelopmental genes, give rise to medulloblastoma, the most common malignant brain tumor in children[5, 6]. We hypothesized that aerobic glycolysis is integral to the regulated proliferation of neural progenitors, and that aerobic glycolysis in cancer may result from the abnormal persistence of metabolic programs that are typically restricted to development. We therefore investigated the relationship between glucose metabolism and neural progenitor function during cerebellar development and medulloblastoma pathogenesis.
Postnatal neurogenesis in the cerebellum presents an ideal opportunity to study metabolic dynamics of neurogenesis under aerobic conditions. CGNs are the most numerous cells in the brain, and arise from CGNPs that proliferate in the external granule cell layer (EGL) in a wave of neurogenesis that occurs postnatally and lasts until postnatal day (P) 15 in mice. CGNPs thus proliferate under normoxic conditions, and mouse cerebellum may be sampled at defined time points to include proliferating neural progenitors or exclusively post-mitotic neurons. As CGNPs terminally differentiate, they migrate from the EGL to the internal granule cell layer (IGL) such that position in the cerebellum corresponds with differentiation state. CGNPs are readily cultured and maintain their proliferative behavior in vitro in serum-free media supplemented with Shh and insulin[8, 9]. If Shh is withdrawn, CGNPs exit the cell cycle and differentiate – such that after 24 hours in culture without Shh, proliferation is minimal. Importantly, activating mutations in the Shh pathway have been found in human medulloblastoma and can recapitulate tumorigenesis in transgenic mice, including the ND2:SmoA1 and SmoM2 lines that express constitutively active alleles of Smoothened[5, 10–13]. These animal models consistently implicate CGNPs as proximal cells of origin for Shh-driven medulloblastoma. Here, we examine glucose metabolism in CGNPs, CGNs and Smoothened-induced, murine medulloblastomas in order to determine whether aerobic glycolysis originates in neural development and whether this metabolic pattern is essential to the pathogenesis of embryonal cancers of the nervous system.
Mice were handled in compliance with the guidelines of the University of North Carolina Animal Care and Use Committee. NeuroD2:SmoA1 mice were provided by Dr James Olson (Fred Hutchinson Cancer Research Center, Seattle, WA, USA) and SmoM2 mice (Bl6 background) were purchased from Jackson Laboratories (Bar Harbor, ME, USA). hGFAP-cre mice were generously provided by Dr Eva Anton (University of North Carolina, Chapel Hill, NC, USA); these mice were initially obtained in the FVB/N background, and were crossed into the Bl6 background at least 10 times. Hk2fl/fl mice were obtained from the European Mouse Mutant Archive and are documented on the archive’s website. In brief, these mice (deposited by Dr Eija Pirinen) harbor LoxP sites at intron 3 and intron 10 of the Hk2 gene, such that exons 4 to 10 are deleted in the presence of cre recombinase. Hk2fl/fl mice were crossed at least 5 times with Bl6 mice prior to the experimental breeding. Medulloblastomas were detected by daily observation for abnormalities of head shape and movement, and animals were sacrificed at the onset of tumor symptoms, specifically ataxia, weight loss or movement disorder. For EdU experiments, mouse pups at P10 were injected intraperitoneally (IP) with 50 μl HBSS containing EdU (250 μM; catalogue number A10044; Life Technologies, Grand Island, NY, USA) and sacrificed after 24 hours. All animal handling and protocols were carried out in accordance with established practices as described in the National Institutes of Health Guide for Care and Use of Laboratory Animals and as approved by the Animal Care and Use Committee of the University of North Carolina (IACUC #10-126.0).
Cell culture techniques
In vitro metabolism studies
For enzymatic measurement of lactate, media were sampled after 48 hours in culture and lactate was quantified by the l-Lactate Assay Kit (catalogue number 1200011002; Eton Bioscience Durham, NC, USA) using the manufacturer’s protocol. For 18-fluorodeoxiglucose (18FDG) studies, CGNPs were cultured for 48 hours, incubated for 40 minutes in 2 μCi 18FDG in glucose-containing DMEM/F12 supplemented as indicated, washed twice, and collected. The concentration of FDG was less than 0.1 nM, and control experiments in which Shh-treated CGNPs were treated with either normal media or with media containing 1 nM 2-deoxyglucose demonstrated no change in lactate production or CGNP proliferation, measured by incorporation of EdU (data not shown). Radioactivity was measured by gamma counter (2470 Wizard2; PerkinElmer Waltham, MA, USA) and normalized to the activity measured in the initial media. For oxygen consumption rate (OCR) measurements, after 48 hours in culture with either vehicle or Shh, CGNPs were changed to fresh media and the OCR was measured using a Seahorse XF24 (Seahorse Bioscience, North Billerica, MA, USA) following the manufacturer’s protocol. The electron transport uncoupling agent trifluorocarbonylcyanide phenylhydrazone (FCCP; 300 nM) was added, and OCR measurements were then repeated immediately. For proliferation assays, EdU was added to the cell culture and visualized using the manufacturer’s protocol (catalogue number C10337; Life Sciences), and positive cells were counted using Leica-Metamorph software (Molecular Devices). For Hk activity assays, cells from 3 replicate wells per condition, or snap-frozen cerebella from 3 replicate mice per genotype, were lysed and processed for colorimetric assay per manufacturer’s protocol (Hexokinase Assay Kit, catalogue number E-111; Biomedical Research Service Center, SUNY, Buffalo, NY, USA).
For NMR-based metabolomic analysis, cells were plated in 12-well plates in 650 μl media and then 50 μl media samples were harvested at the indicated time points. Cell counts on the day of media harvest demonstrated that all wells contained 95 to 105% of the mean number of cells, and there was no statistically significant variation in cell number in wells treated with Shh or vehicle (data not shown). Media samples were processed as previously described. Briefly, proton (1H) spectra were acquired at 25°C on a 14.1 T Varian INOVA spectrometer (600 MHz 1H frequency) equipped with a CapNMR™ microcoil (Magnetic Resonance Microsensors Corp, Savoy, MN, USA). The 1H spectra were acquired using a one-pulse sequence with presaturation of the water resonance using a 90° flip angle, and a total repetition time of 12.65 seconds. The peak areas in the 1H spectra were determined using Chenomx NMR processing software version 7.1 (Edmonton, Alberta, Canada). First, spectra were zero-filled to 32,000 points, and were line broadened using a 0.5 Hz exponential Gaussian function. Chemical shifts presented were obtained from the Human Metabolome Database. Concentrations were calculated from the 1H spectra by comparing peak areas with the peak for 2,2′,3,3′-duetero-trimethyl propionate. Concentration values were then normalized for the cell number in each well, and the results were analyzed by two-way analysis of variance with Bonferroni correction. For statistical comparison of multiple spectra, we performed orthogonal partial least squares discriminant analysis using ACD Labs 12.0 1D NMR Processor (ACD Labs Toronto, Ontario, Canada) to zero-fill to 32,000 points, with a 0.5 Hz exponential Gaussian function applied, then spectra were binned into 0.005 ppm segments and values were exported to SIMCA-P + 11 (Umetrics Umeå, Sweden). Loading coefficients and P- scaled correlation coefficients were exported to MatLab (Mathworks, Natick, MA, USA) and plotted as the y value (loading coefficient) and color coded (correlation coefficient).
In vivo metabolism studies
To measure cerebellar glucose uptake, mouse pups at P5 or P20 were injected IP with 0.2 mCi 18FDG; after 40 minutes, pups were rapidly decapitated and the cerebella and forebrain were harvested by dissection. Tissue samples were washed and weighed, and incorporated radioactivity was quantified by gamma counter (2470 Wizard2; PerkinElmer). Incorporated counts from the cerebellum were normalized for tissue weight and for dose to the brain, as measured by incorporated radioactivity in the frontal lobe sample from the same animal.
Magnetic resonance spectroscopy (MRS) acquisitions were performed at 9.4 T on a Bruker BioSpec 94/30 MRI system (Bruker BioSpin, Bilerica, MA, USA). A volume of interest was placed on the pup cerebellum region based on T2-weighted images with a size of 11.5 mm3. A point-resolved spectroscopy sequence was used for single-voxel signal acquisition (Echo Time = 1.4 ms; total repetition time = 20,000 ms; 64 × 64 matrix size). The spectrum was adjusted with the water signal at 4.7 ppm as a reference.
18FDG positron emission tomography/computed tomography (PET/CT) imaging was performed on a PET/CT scanner (GE eXplore Vista PET/CT; GE Helathcare Worldwide, Waukesha, WI, USA). Under isoflurane anesthesia, mice underwent intravenous administration of 500 μCi 18FDG and computed tomography scan. Thirty minutes after 18FDG injection, PET/CT imaging was acquired over 10 minutes. Images were reconstructed using ordered subset expectation maximization algorithms, and were normalized to dose and animal weight to generate standardized uptake values of the final images.
Histology and immunohistochemistry
Mouse brain and tumor tissue were embedded in paraffin and sectioned to 5 μm thickness. H & E-stained sections were prepared using standard techniques. EdU was detected using the Click-iT® EdU Alexa Fluor 488 Imaging Kit (catalogue number C10337; Life Sciences), as per the manufacturer’s protocol. Immunohistochemistry (IHC) was performed on paraffin-embedded sections after deparaffinization in Histoclear, rehydration in a graded ethanol series, and antigen retrieval by heating to boiling in 10 mM citrate buffer pH 6.0 in a pressure cooker for 15 minutes and then transferring to PBS. For Hk2 detection, tissue was not embedded in paraffin but rather was sectioned by Vibratome to 100 μm thickness and stained by IHC without antigen retrieval. IHC was performed as previously described using primary antibodies: Hk1 (catalogue number 2024; Cell Signaling Technologies, Danvers, MA, USA), GFP (catalogue number 600-101-215; Rockland Immunochemicals, Gilbertsville, PA, USA), Hk2 (catalogue number 2867; Cell Signaling), Calbindin (catalogue number 2173; Cell Signaling), CD31 (catalogue number 3528 Cell Signaling), NeuN (catalogue number MAB377; Millipore, Billerica, MA, USA), proliferating cell nuclear antigen (PCNA, catalogue number 2586; Cell Signaling), and p27 (catalogue number 3686; Cell Signaling). After EdU and IHC staining, nuclei were counterstained with 4′6-diamino-2-phenylindole (DAPI; catalogue number D1306; Life Sciences), diluted 200 ng/ml in PBS for 5 minutes, and immunoreactivity was evaluated with a Leica epifluorescence DM5000B microscope (Leica Microsystems, Wetzlar, Germany). Stained slides were then scanned using an Aperio ScanScope XT (Vista, CA, USA).
Western blot analysis
Cultured cells, whole cerebella, and tumors were lysed by homogenization in lysis buffer (catalogue number 9803; Cell Signaling). Protein concentrations were quantified using the Bicinchoninic acid method (catalogue number 23227; Thermo Scientific Asheville, NC, USA) and equal concentrations of protein were resolved on SDS-polyacrylamide gels then transferred to polyvinylidene fluoride membranes. Immunologic analysis was performed on a SNAP ID device (Millipore) using the manufacturer’s protocol with primary antibodies to β-actin (catalogue number 4970; Cell Signaling), Hk1 (catalogue number 2024; Cell Signaling), Hk2 (catalogue number 2867; Cell Signaling), Cyclin D2 (catalogue number 3741; Cell Signaling), insulin-like growth factor (IGF) receptor (catalogue number 9750; Cell Signaling), phospho-IGF receptor (catalogue number 6113; Cell Signaling), Akt (catalogue number 4685; Cell Signaling), pAkt (catalogue number 4060; Cell Signaling), HP-Hif1a (catalogue number 3434; Cell Signaling), phospho-AMP-activated kinase (catalogue number 2535; Cell Signaling), phospho-Acyl-CoA carboxylase (catalogue number 3661; Cell Signaling), caspase-3 (cC3, catalogue number 9664; Cell Signaling), GFP (catalogue number 600-101-215; Rockland), Smo (catalogue number AB72130; Abcam, Cambridge, MA, USA), and Cip2A (catalogue number SC-80660; Santa Cruz Biotechnology Santa Cruz, CA, USA). Secondary antibodies were anti-rabbit IgG horseradish peroxidase (catalogue number 7074; Cell Signaling), and anti-mouse IgG horseradish peroxidase (catalogue number 7076; Cell Signaling). Antibody conjugates were visualized by chemiluminescence (catalogue number RPN2106; GE Healthcare).
Total RNA was prepared from CGNPs using the RNeasy Mini Kit (catalogue number 74104; Qiagen, Valencia, CA) as per protocol. First-strand cDNA was synthesized using the Invitrogen SuperScript III Kit (catalogue number 18080-051, Life Sciences). To prevent amplification from genomic DNA, PCR primers were designed to span at least one intron, and PCR products were cloned and sequenced to verify identity. The PCR primers were: Hk2, ATTGTCCAGTGCATCGCGGA and AGGTCAAACTCCTCTCGCCG; Cyclin D2, GCGTGCAGAAGGACATCCA and CACTTTTGTTCCTCACAGACCTCTAG; and β-actin, ATGCTCTCCCTCACGCCATC and CAGGATTCCATACCCAAGA. PCR reactions were run on an ABI 7500Fast instrument, using ABI Fast Sybr Green master mix (catalogue number 4385612; Applied Biosystems Carlsbad, CA, USA), cycling between 95 and 60°C, as per the manufacturer’s protocol, for 50 cycles. The threshold cycle (CT) was determined by ABI proprietary software. PCR efficiency for each primer pair was measured by amplifying a series of copy number standards from cloned, sequenced PCR products and used to calculate the fold-change, using β-actin as the reference standard.
Shh signaling induces aerobic glycolysis in CGNPs
To determine whether mitogenic signaling alters the glucose metabolism of neural progenitors, we compared lactate generation, glucose uptake and oxygen consumption of CGNPs cultured in the presence or absence of Shh. We isolated CGNPs from P5 mouse pups and cultured them in serum-free, N2-supplemented media, with Shh or vehicle as indicated. After 48 hours in culture, only Shh-treated CGNPs continued to proliferate (Figure1A). Starting from fresh media at 24 hours, from 24 to 48 hours in culture, Shh-treated CGNPs accumulated 180% more lactate than Shh-deprived CGNPs that exited the cell cycle (Figure1B). Shh-induced lactate production did not depend on the high glucose and K+ concentrations of typical CGNP media, as Shh induced comparable lactate production in CGNPs maintained in CGNP media (18 mM glucose, 25 mM KCl), DMEM/F12 (4 mM KCl) or low-glucose DMEM (5.6 mM glucose; see Additional file1: Figure S1). Shh-treated CGNPs also demonstrated differential uptake of 18FDG when exposed briefly to the tracer in freshened 18 mM glucose culture media (Figure1B). Despite increased glucose uptake and lactate production, Shh-treated CGNPs did not increase the OCR, measured as picomoles per minute in real-time by an XF Extracellular flux Analyzer (Seahorse Bioscience) and normalized for the number of cells per well. Importantly, both vehicle-treated and Shh-treated CGNPs increased the OCR briskly and equally when exposed to the respiratory chain uncoupling agent FCCP (data not shown), indicating that CGNPs were not constrained by the availability of oxygen. Taken together, these results demonstrate that Shh induced CGNPs to increase metabolism of glucose to lactate under conditions in which oxygen was not limiting.
Concentrations of selected metabolites in Shh- or vehicle-containing CGNP media at the indicated times
ΔShh vs. ΔV
0 hours (μM)
6 hours (μM)
Δ6 hours (μM)
0 hours (μM)
6 hours (μM)
Δ6 hours (μM)
109 ± 5
105 ± 2
−4 ± 5
111 ± 3
127 ± 7
15 ± 8
160 ± 4
394 ± 4
234 ± 6
165 ± 6
348 ± 13
183 ± 15
997 ± 49
1,002 ± 22
5 ± 53
990 ± 16
989 ± 26
−2 ± 30
196,834 ± 934
17,312 ± 328
−2,372 ± 990
19,449 ± 749
19,399 ± 405
−49 ± 852
2,443 ± 99
2,037 ± 31
−407 ± 103
2,509 ± 92
2,515 ± 65
6 ± 113
223 ± 21
221 ± 6
−2 ± 22
215 ± 10
254 ± 8
39 ± 13
334 ± 27
4,952 ± 114
4,618 ± 117
299 ± 15
2,012 ± 49
1,713 ± 51
578 ± 29
483 ± 17
−95 ± 33
555 ± 23
560 ± 4
5 ± 23
461 ± 30
484 ± 37
23 ± 47
413 ± 14
487 ± 14
74 ± 20
558 ± 14
499 ± 21
−59 ± 24
540 ± 22
552 ± 17
12 ± 28
Concentrations of each metabolite were calculated from NMR spectra, with 6 replicate wells per condition. Data presented as mean ± SEM. For statistical analysis, two-way analysis of variance with Bonferroni correction was applied, and P values were calculated for the contrast of change in Shh (ΔShh) versus change in vehicle (ΔV).
Glycolytic phenotype persists in medulloblastoma
Medulloblastoma cells, like CGNPs, are highly proliferative. To determine whether the high glucose flux observed in mitotic CGNPs persists in medulloblastoma, we used 18FDG PET/CT to compare glucose uptake in tumor-bearing and wild-type mice. We consistently detected strong glucose uptake within ND2:SmoA1-induced medulloblastomas (Figure2C). Elevated glucose uptake in murine medulloblastoma is consistent with reported PET scan results in human medulloblastoma and confirms that medulloblastomas share the glycolytic phenotype of CGNPs.
Hk2 is induced by Shh-pathway activation and persists in medulloblastoma
Hk enzymes catalyze the first step in glucose metabolism. While there are four homologous Hk genes, Hk1 and Hk2 have been frequently associated with aerobic glycolysis[4, 19]. To identify proteins that mediate the glycolytic phenotype of CGNPs and medulloblastoma, we examined the expression of Hk1 and Hk2 in CGNPs, CGNs, and ND2:SmoA1-induced medulloblastoma.
Previous investigations have validated Western blot for Cyclin D2 as a marker of Shh-induced proliferation, and we therefore compared Cyclin D2 and Hk2 in both isolated CGNPs and in whole cerebellar lysates at progressive points in postnatal development. Importantly, Hk2 expression corresponded closely with the expression of Cyclin D2 (Figure3A,C) both with exposure to Shh in vitro, and in vivo throughout the period of postnatal neurogenesis. Hk2 and Cyclin D2 were expressed at P6 and P8, and both proteins were down-regulated by P14, as neurogenesis wanes. Hk2 and Cyclin D2 were strongly up-regulated in SmoA1-induced medulloblastoma. As with CGNPs in vitro, expression of Hk1 varied inversely with expression of Hk2 (Figure3C).
To determine whether induction of Hk2 by Shh operates through transcription regulation, we compared the abundance of Hk2 mRNA in CGNPs maintained in the presence or absence of Shh. We prepared cDNA from four sets of isolated CGNPs maintained with or without Shh and measured abundance of transcripts encoding Hk2, Cyclin D2 and β-actin by quantitative real-time RT-PCR. We included Cyclin D2 as a known target of Shh-signaling and β-actin as a loading control. Shh increased expression of Hk2, driving a fold-change of 10.7 ± 2.6 (mean ± SEM), comparable with the fold increase for Cyclin D2 of 7.5 ± 1.7 (Figure3D).
To identify the specific cells expressing Hk1 and Hk2, we examined cerebellar sections using IHC. In the P7 cerebellum, differentiated CGNs residing in the IGL uniformly expressed Hk1, while the Hk1 protein was undetectable in the entire population of CGNPs throughout the EGL (Figure3E, P7 Hk1 panel). In contrast, Hk2 was expressed evenly throughout the EGL during neurogenesis (Figure3E, P7 Hk2 panel). Hk2 was absent from the region of the EGL after the CGNP population had completely migrated to the IGL (Figure3E, P16 panel), indicating Hk2-expressing cells were specifically the CGNPs. Importantly, with the development of medulloblastoma, Hk2 expression resumed (Figure3E, medulloblastoma panel). Thus, while undifferentiated cells including CGNPs and medulloblastoma expressed Hk2, Hk1 was expressed by their differentiated progeny.
Previous investigation demonstrated that Shh and insulin/IGF/PI3K signaling pathways converge in CGNPs to activate N-myc. We therefore tested whether N-myc activity mediated developmental induction of Hk2. To disrupt N-myc activity and block potential complementation through C-myc, we treated CGNPs cultured with Shh and N2 with the Myc inhibitor 10058-F4, which blocks association with Max, an interaction required for Myc transcriptional regulation. We then compared expression of Hk2 with that of Cyclin D2, a marker of Shh-induced proliferation, and Cip2A, a known myc target previously demonstrated to be inhibited by 10058-F4[28, 29]. We found 10058-F4 reduced Shh-mediated induction of Hk2, Cip2A and Cyclin D2 in a dose-dependent manner (Figure4C). These data implicate myc–max transcriptional activation in the induction of Hk2 by the combination of Shh and PI3K signaling.
Hk2 is required for Shh-induced aerobic glycolysis
Importantly, increased glycolysis in response to Shh was significantly reduced in CGNPs from hGFAP-cre;Hk2fl/fl mice. We compared lactate production in Shh-treated CGNPs isolated either from hGFAP-cre;Hk2fl/fl mice or from littermate Hk2fl/fl controls that lacked the hGFAP-cre transgene. We found that while Hk2-deficient CGNPs proliferated in response to Shh, they generated 40% less lactate than CGNPs from littermates with intact Hk2 (Figure5C). NMR analysis of media samples also demonstrated decreased lactate production (Figure5D). Orthogonal partial least squares discriminant analysis identified lactate as the only metabolite to change significantly with deletion of Hk2; a trend toward decreased glucose utilization with Hk2 deletion was also detected, but this change was not statistically significant relative to the overall glucose concentration (data not shown). Genetic deletion of Hk2 thus reduced glycolysis without causing CGNPs to catabolize alternative energy substrates, consistent with increasing efficiency of glucose utilization through oxidative phosphorylation.
Deletion of Hk2 disrupts CGNP development
Deletion of Hk2 disrupts medulloblastoma growth
In contrast to the rapidly progressive Hk2 wildtype tumors, medulloblastomas with Hk2 deletion were markedly less malignant (Figure7C). While 100% of P12 hGFAP-cre;SmoM2;Hk2fl/fl mice developed the characteristic occipital expansion that is the first sign of tumor, median survival was 31 days (P < 0.00002) and 30% survived event-free to the end of the experiment at 100 days. Importantly, these long-term survivors were able to breed with wildtype mice to give rise to hGFAP-cre;SmoM2;Hk2fl/+ progeny that developed malignant tumors, demonstrating the efficacy of the inherited SmoM2 and hGFAP-cre alleles. All hGFAP-cre;SmoM2;Hk2fl/+ mice, like hGFAP-cre;SmoM2;Hk2+/+ mice, died by P20 (data not shown). Western blot analysis demonstrated expression of the SmoM2-YFP fusion protein in both hGFAP;SmoM2;Hk2+/+ and hGFAP;SmoM2;Hk2fl/fl mice (Figure7D), while IHC for GFP revealed SmoM2-YFP expression throughout the cerebella of hGFAP-cre; SmoM2 mice, both in tumor and in adjacent, differentiated regions, in both the Hk2+/+ and Hk2f/fl genotypes (Figure7E). Mosaicism for SmoM2 was not observed in any GFP-stained sections. Mice with the hGFAP-cre;SmoM2;Hk2fl/fl genotype thus expressed SmoM2 and developed tumors, but these tumors progressed less rapidly than tumors with at least one functional allele of Hk2.
Hk2 links energy metabolism and maintenance of an undifferentiated state
Along with increased differentiation, Hk2-deficient tumors demonstrated increased micro-vascularization (Figure8B) and increased localization of proliferative cells along blood vessels (Figure8C). Between regions of perivascular proliferation, tumor cells exited the cell cycle and up-regulated the early differentiation marker p27 (Figure8C). The close correlation between proliferation and distance from capillaries suggested that sustained proliferation in Hk2-deficient tumors might depend on availability of oxygen. Accordingly, reduced tumor growth might result from inability to meet energy needs outside the perivascular region.
To probe for a link between impaired energy production and reduced tumor growth in Hk2-deficient medulloblastoma, we compared phosphorylation of AMPk in cerebella from hGFAP-cre;SmoM2;Hk2+/+ mice and hGFAP-cre;SmoM2;Hk2fl/fl mice. At P15, cerebella from both genotypes expressed SmoM2, but hGFAP-cre;SmoM2;Hk2+/+ cerebella were almost entirely replaced by tumor while hGFAP-cre;SmoM2;Hk2fl/fl cerebella contained both tumor and differentiated tissue. We integrated the AMPk results with a comparison of proliferation and apoptosis by Western blot analysis for specific markers (Figure8F). AMPk functions as an intracellular energy sensor, becoming phosphorylated in response to cellular energy scarcity. In Hk2-deficient medulloblastomas, phosphorylation of AMPk was greatly increased, consistent with energy scarcity. Phosphorylation of Acc1, a known target of AMPk, was also increased, indicating that the detected increase in AMPk activity was functionally relevant. We examined cC3 because we have previously found that medulloblastomas demonstrate continuous baseline apoptosis demonstrable with cC3 staining. Interestingly, both Cyclin D2 and cC3 were markedly less abundant in Hk2fl/fl medulloblastomas, demonstrating that while energy scarcity correlated with reduced proliferation, it did not provoke cell death. Deletion of Hk2 thus reduced the ability of progenitors to remain undifferentiated without altering survival. Taken together, these findings support a model in which: 1) aerobic glycolysis supports the undifferentiated progenitor phenotype of medulloblastoma by preserving energy homeostasis; and 2) impaired energy homeostasis caused by Hk2 deletion promotes differentiation, both disrupting patterning in development and reducing tumor growth.
Since its initial observation by Otto Warburg, aerobic glycolysis has been documented in a variety of cancer cells and in non-neoplastic thymocytes. The selective advantage of aerobic glycolysis in cancer has been directly tested by comparing growth rates of xenograft tumors with and without PkM2 or Hk2. Why cells should harbor a genetic program that promotes cancer growth, however, has been unclear. Our data demonstrate that aerobic glycolysis is a developmental program that is co-opted in the course of in vivo tumorigenic transformation. Importantly, we identified Hk2 as a key mechanism through which developmental signaling molecules induce aerobic glycolysis. Moreover, the Hk2-dependent, glycolytic metabolism of mitotic CGNPs was maintained in medulloblastoma and was essential to cancer pathogenesis: disrupting glycolysis through conditional deletion of Hk2 markedly reduced tumor growth while increasing differentiation. These findings reveal cancer cells exploiting the specialized energy metabolism of developmental progenitors to maintain an undifferentiated state and malignant potential.
We noted that Hk2 up-regulation exerted profound effects on glucose metabolism in CGNPs that exceeded the change in Hk activity measured in lysates. Our data indicate that Hk2 is required for maximal glycolysis, and functions optimally to shunt glucose toward glycolysis in the intact cell. Hk2 is known to localize to the outer mitochondrial membrane, and this subcellular localization is critical to its pro-glycolytic effect. Accordingly, it is not surprising that the measured effect of increased Hk2 in lysates underestimates the influence of Hk2 on glucose metabolism in live cells.
The nature of the benefit that glycolysis confers on dividing cells is controversial. Since oxidative phosphorylation generates more ATP per glucose molecule, an open question is why cells metabolize any glucose to lactate when oxygen is not limiting. Potential cellular benefits of aerobic glycolysis for mitotic cells may include supplying intermediaries for lipid and nucleic acid synthesis, or enabling high glucose flux. High glucose flux has been observed to exert an anti-apoptotic effect through diverse mechanisms including Bax inactivation, prevention of cytochrome c release, or redox inactivation of cytochrome c. Our findings that genetic deletion of Hk2 disrupted both energy homeostasis and the balance between proliferation and differentiation provide new insight into the question of how aerobic glycolysis can support progenitor function.
A key aspect of the progenitor state is the maintenance of self-renewal capacity that allows daughter cells to remain proliferative after cell division. In the course of developmental CGNP proliferation, the capacity for self-renewal is gradually lost and the CGNP population declines until proliferation ceases. In contrast, under the influence of SmoM2-driven tumorigenesis, self-renewal is maintained and the EGL grows unconstrained. Hk2 deletion disrupted the maintenance of self-renewal capacity, causing premature differentiation of CGNPs in the EGL, manifested as focal disorganization. In medulloblastoma, however, where self-renewal capacity does not typically wane, Hk2 deletion caused a much greater effect, disrupting unconstrained growth, promoting increased differentiation and increasing the probability of animal survival. The abnormal migration patterns of Hk2-deleted CGNPs and the reduced growth of Hk2-deleted tumors thus consistently highlight a role for Hk2-driven glycolysis in maintaining the undifferentiated progenitor state.
The increased vascularity of Hk2-deleted EGL and medulloblastoma implicate oxygen homeostasis in progenitor function. By up-regulating Hk2-dependent glycolysis, wildtype CGNPs reduce oxygen dependence at the cost of increased utilization of glucose (Figure1). By blocking Shh-driven glycolysis, Hk2 deletion increased the dependence of CGNPs and medulloblastoma on vascular support. The ectopic capillaries formed in the Hk2-deficient EGL(Figure6) and the tumor cells in Hk2fl/fl medulloblastoma that proliferated along capillaries (Figure7), both demonstrated increased vascular dependence caused by loss of Hk2. In Hk2-deficient cerebella, increased vascularity effectively compensated for the loss of aerobic glycolysis and cerebellar development was largely preserved despite local failures of migration and differentiation. In Hk2-deficient medulloblastomas, however, increased vascularization was insufficient to compensate, perhaps due to the increased tumor mass or alternatively due to increased metabolic demand. Importantly, vascular compensation did not prevent cellular energy scarcity, as demonstrated by activation of AMPk. Therefore, with increased vessel support, oxidative phosphorylation could meet the energy needs of developmentally regulated neurogenesis, but not of unconstrained tumor growth. Importantly, the consequence of energy failure in medulloblastoma was a loss of progenitor state and progression through developmentally appropriate differentiation. Similarly, recent investigation has demonstrated that AMPk activation blocks the reprogramming of mouse embryonic fibroblasts into induced stem cells; in each case, activation of AMPk blocks the maintenance of progenitor self-renewal and promotes terminal differentiation.
An important benefit of aerobic glycolysis for CGNPs may be to prevent energy failure that could limit proliferative potential. Neural stem cells of the forebrain and hippocampus are known to require the support of a perivascular niche[44–47]. In contrast to these discrete sites of postnatal neurogenesis, the EGL, where CGNPs proliferate, extends over a broad area; to achieve even growth throughout the EGL, CGNPs must proliferate at a constant rate regardless of distance from the supportive niche. Aerobic glycolysis, through reduced oxygen dependence at the cost of increased glucose dependence, may reduce the need for perivascular support. While oxygen must be obtained from capillaries, glucose can be mobilized from intracellular stores or provided by neighboring cells through active transport. Aerobic glycolysis may thus release progenitors from the perivascular niche, and disrupting Hk2 may inhibit tumorigenesis by constraining progenitors to regions of niche support.
Consistent with an integral role in neurogenesis, we found aerobic glycolysis to be regulated by developmental signaling. In particular, we found that induction of glycolysis requires co-incident activation of Shh and insulin/IGF/PI3K pathways. Importantly, we identified Myc–Max-mediated transcriptional regulation as a specific downstream effector necessary for this regulation. Such a role, would be consistent with previous investigations that have linked c-myc to the Warburg effect in other cell types[48, 49]. Modulation of intracellular calcium has also been identified as a point of convergence for Shh and PI3K signaling and could play a key role in mediating Shh effects on CGNP glycolysis. Alongside our finding that Shh regulates carbohydrate metabolism, recently published work has demonstrated that Shh down-regulates fatty acid oxidation in favor of lipid biosynthesis. Importantly, we found that blockade of Shh pathway-induced glycolysis in medulloblastoma caused both reduced proliferation and inactivation of Acc1, a critical regulator of lipid metabolism. These findings together define a mitogen-induced metabolic configuration in which carbohydrate and lipid metabolism are integrated to optimally support progenitor proliferation.
Our results demonstrate that aerobic glycolysis is primarily a neurodevelopmental program that is co-opted in medulloblastoma tumorigenesis to promote neoplastic growth. We found that medulloblastoma and neural progenitors share specific metabolic requirements that distinguish them from post-mitotic brain cells. Our results showing increased survival in medulloblastoma-bearing mice with Hk2 deletion have revealed an unexpected link between cellular metabolism and differentiation state. In Hk2-deleted tumors where Shh-driven glycolysis was prevented, the consequence was increased differentiation and reduced tumor growth. Together, these findings connect developmental signaling pathways with patterns of metabolism in cancer while also demonstrating the potential efficacy of metabolic therapy for medulloblastoma through targeting of Hk2.
Cerebellar granule neuron
Cerebellar granule neuron progenitor
Dulbecco’s modified Eagle’s medium
External granule cell layer
Fetal calf serum
Green fluorescent protein
- H & E:
Hematoxylin and eosin
Insulin-like growth factor
Internal granule cell layer
Magnetic resonance spectroscopy
Nuclear magnetic resonance
Oxygen consumption rate
Proliferating cell nuclear antigen
Polymerase chain reaction
Positron emission tomography
TRG is supported by grants from the National Institutes of Health (NIH; 1K08NS077978-01), the St. Baldrick’s Foundation, the Morgan Adams Foundation and the Matthew Larson Brain Tumor Foundation. CRM is supported in part by the Damon Runyon Cancer Research Foundation (CI-45-09). This project is supported in part by NIH grants NS042197 and GM078366 to MD and P30DK056350 to the University of North Carolina at Chapel Hill Nutrition Obesity Research Center.
- Vander Heiden MG, Cantley LC, Thompson CB: Understanding the warburg effect: the metabolic requirements of cell proliferation. Science. 2009, 324: 1029-1033. 10.1126/science.1160809.PubMed CentralView ArticlePubMedGoogle Scholar
- Dang CV: Links between metabolism and cancer. Genes Dev. 2012, 26: 877-890. 10.1101/gad.189365.112.PubMed CentralView ArticlePubMedGoogle Scholar
- Jones RG, Thompson CB: Tumor suppressors and cell metabolism: a recipe for cancer growth. Genes Dev. 2009, 23: 537-548. 10.1101/gad.1756509.PubMed CentralView ArticlePubMedGoogle Scholar
- Wolf A, Agnihotri S, Micallef J, Mukherjee J, Sabha N, Cairns R, Hawkins C, Guha A: Hexokinase 2 is a key mediator of aerobic glycolysis and promotes tumor growth in human glioblastoma multiforme. J Exp Med. 2011, 208 (2): 313-26. 10.1084/jem.20101470.PubMed CentralView ArticlePubMedGoogle Scholar
- Grimmer MR, Weiss WA: Childhood tumors of the nervous system as disorders of normal development. Curr Opin Pediatr. 2006, 18: 634-638. 10.1097/MOP.0b013e32801080fe.View ArticlePubMedGoogle Scholar
- Marino S: Medulloblastoma: developmental mechanisms out of control. Trends Mol Med. 2005, 11: 17-22. 10.1016/j.molmed.2004.11.008.View ArticlePubMedGoogle Scholar
- Hatten ME, Heintz N: Mechanisms of neural patterning and specification in the development cerebellum. Ann Rev Neurosci. 1995, 18: 385-408. 10.1146/annurev.ne.18.030195.002125.View ArticlePubMedGoogle Scholar
- Wechsler-Reya RJ, Scott MP: Control of neuronal precursor proliferation in the cerebellum by Sonic Hedgehog. Neuron. 1999, 22: 103-114. 10.1016/S0896-6273(00)80682-0.View ArticlePubMedGoogle Scholar
- Dudek H, Datta SR, Franke TF, Birnbaum MJ, Yao R, Cooper GM, Segal RA, Kaplan DR, Greenberg ME: Regulation of neuronal survival by the serine-threonine protein kinase Akt. Science. 1997, 275: 661-665. 10.1126/science.275.5300.661.View ArticlePubMedGoogle Scholar
- Zurawel RH, Allen C, Chiappa S, Cato W, Biegel J, Cogen P, de Sauvage F, Raffel C: Analysis of PTCH/SMO/SHH pathway genes in medulloblastoma. Genes Chromosomes Cancer. 2000, 27: 44-51. 10.1002/(SICI)1098-2264(200001)27:1<44::AID-GCC6>3.0.CO;2-V.View ArticlePubMedGoogle Scholar
- Zurawel RH, Allen C, Wechsler-Reya R, Scott MP, Raffel C: Evidence that haploinsufficiency of Ptch leads to medulloblastoma in mice. Genes Chromosomes Cancer. 2000, 28: 77-81. 10.1002/(SICI)1098-2264(200005)28:1<77::AID-GCC9>3.0.CO;2-Y.View ArticlePubMedGoogle Scholar
- Hallahan AR, Pritchard JI, Hansen S, Benson M, Stoeck J, Hatton BA, Russell TL, Ellenbogen RG, Bernstein ID, Beachy PA, Olson JM: The SmoA1 mouse model reveals that notch signaling is critical for the growth and survival of sonic hedgehog-induced medulloblastomas. Cancer Res. 2004, 64: 7794-7800. 10.1158/0008-5472.CAN-04-1813.View ArticlePubMedGoogle Scholar
- Mao J, Ligon KL, Rakhlin EY, Thayer SP, Bronson RT, Rowitch D, McMahon AP: A novel somatic mouse model to survey tumorigenic potential applied to the hedgehog pathway. Cancer Res. 2006, 66: 10171-10178. 10.1158/0008-5472.CAN-06-0657.PubMed CentralView ArticlePubMedGoogle Scholar
- Kenney AM, Cole MD, Rowitch DH: Nmyc up-regulation by sonic hedgehog signaling promotes proliferation in developing cerebellar granule neuron precursors. Development. 2003, 130: 15-28. 10.1242/dev.00182.View ArticlePubMedGoogle Scholar
- Tikunov AP, Johnson CB, Lee H, Stoskopf MK, Macdonald JM: Metabolomic investigations of American oysters using H-NMR spectroscopy. Mar Drugs. 2010, 8: 2578-2596. 10.3390/md8102578.PubMed CentralView ArticlePubMedGoogle Scholar
- Human Metabolome Database.http://www.hmdb.ca,
- Pfaffl MW, Horgan GW, Dempfle L: Relative expression software tool (RESTÂ©) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic Acids Res. 2002, 30: e36-10.1093/nar/30.9.e36.PubMed CentralView ArticlePubMedGoogle Scholar
- Gururangan SMRCP, Hwang EMD, Herndon JEIIPD, Fuchs HMDPD, George TMD, Coleman REMD: [18F]Fluorodeoxyglucose-positron emission tomography in patients with medulloblastoma. Neurosurgery. 2004, 55: 1280-1289.View ArticlePubMedGoogle Scholar
- Mathupala SP, Ko YH, Pedersen PL: Hexokinase II: cancer’s double-edged sword acting as both facilitator and gatekeeper of malignancy when bound to mitochondria. Oncogene. 2006, 25: 4777-4786. 10.1038/sj.onc.1209603.PubMed CentralView ArticlePubMedGoogle Scholar
- Kenney AM, Rowitch DH: Sonic hedgehog promotes G1 Cyclin expression and sustained cell cycle progression in mammalian neuronal precursors. Mol Cell Biol. 2000, 20: 9055-9067. 10.1128/MCB.20.23.9055-9067.2000.PubMed CentralView ArticlePubMedGoogle Scholar
- Riddle SR, Ahmad A, Ahmad S, Deeb SS, Malkki M, Schneider BK, Allen CB, White CW: Hypoxia induces hexokinase II gene expression in human lung cell line A549. Am J Physiol Lung Cell Mol Physiol. 2000, 278: L407-L416.PubMedGoogle Scholar
- Gwak G-Y, Yoon J-H, Kim KM, Lee H-S, Chung JW, Gores GJ: Hypoxia stimulates proliferation of human hepatoma cells through the induction of hexokinase II expression. J Hepatol. 2005, 42: 358-364. 10.1016/j.jhep.2004.11.020.View ArticlePubMedGoogle Scholar
- Mathupala SP, Rempel A, Pedersen PL: Glucose catabolism in cancer cells. J Biol Chem. 2001, 276: 43407-43412. 10.1074/jbc.M108181200.View ArticlePubMedGoogle Scholar
- Hall RK, Granner DK: Insulin regulates expression of metabolic genes through divergent signaling pathways. J Basic Clin Physiol Pharmacol. 1999, 10: 119-133.View ArticlePubMedGoogle Scholar
- Kenney AM, Widlund HR, Rowitch DH: Hedgehog and PI-3 kinase signaling converge on Nmyc1 to promote cell cycle progression in cerebellar neuronal precursors. Development. 2004, 131: 217-228. 10.1242/dev.00891.View ArticlePubMedGoogle Scholar
- Yin X, Giap C, Lazo JS, Prochownik EV: Low molecular weight inhibitors of Myc-Max interaction and function. Oncogene. 2003, 22: 6151-6159. 10.1038/sj.onc.1206641.View ArticlePubMedGoogle Scholar
- Wenzel ASM: The mycN/max protein complex in neuroblastoma. Short review. Eur J Cancer. 1995, 31A: 516-519.View ArticlePubMedGoogle Scholar
- Kerosuo L, Fox H, Perälä N, Ahlqvist K, Suomalainen A, Westermarck J, Sariola H, Wartiovaara K: CIP2A increases self-renewal and is linked to Myc in neural progenitor cells. Differentiation. 2010, 80: 68-77. 10.1016/j.diff.2010.04.003.View ArticlePubMedGoogle Scholar
- Khanna A, Backelman C, Hemmes A, Junttila MR, Wiksten J-P, Lundin M, Junnila S, Murphy DJ, Evan GI, Haglund C: MYC-dependent regulation and prognostic role of CIP2A in gastric cancer. J Natl Cancer Inst. 2009, 101: 793-805. 10.1093/jnci/djp103.View ArticlePubMedGoogle Scholar
- Zhuo L, Theis M, Alvarez-Maya I, Brenner M, Willecke K, Messing A: hGFAP-cre transgenic mice for manipulation of glial and neuronal function in vivo. Genesis. 2001, 31: 85-94. 10.1002/gene.10008.View ArticlePubMedGoogle Scholar
- Schüller U, Heine VM, Mao J, Kho AT, Dillon AK, Han Y-G, Huillard E, Sun T, Ligon AH, Qian Y: Acquisition of granule neuron precursor identity is a critical determinant of progenitor cell competence to form Shh-induced medulloblastoma. Cancer Cell. 2008, 14: 123-134. 10.1016/j.ccr.2008.07.005.PubMed CentralView ArticlePubMedGoogle Scholar
- Miyazawa K, Himi T, Garcia V, Yamagishi H, Sato S, Ishizaki Y: A role for p27/Kip1 in the control of cerebellar granule cell precursor proliferation. J Neurosci. 2000, 20: 5756-5763.PubMedGoogle Scholar
- David C: The AMP-activated protein kinase cascade – a unifying system for energy control. Trends Biochem Sci. 2004, 29: 18-24. 10.1016/j.tibs.2003.11.005.View ArticleGoogle Scholar
- Ha J, Daniel S, Broyles SS, Kim KH: Critical phosphorylation sites for acetyl-CoA carboxylase activity. J Biol Chem. 1994, 269: 22162-22168.PubMedGoogle Scholar
- Garcia I, Crowther AJ, Gama V, Ryan Miller C, Deshmukh M, Gershon TR: Bax deficiency prolongs cerebellar neurogenesis, accelerates medulloblastoma formation and paradoxically increases both malignancy and differentiation. Oncogene. 2012, Epub ahead of printGoogle Scholar
- Brand K, Aichinger S, Forster S, Kupper S, Neumann B, NÜRnberg W, Ohrisch G: Cell-cycle-related metabolic and enzymatic events in proliferating rat thymocytes. Eur J Biochem. 1988, 172: 695-702. 10.1111/j.1432-1033.1988.tb13944.x.View ArticlePubMedGoogle Scholar
- Christofk HR, Vander Heiden MG, Harris MH, Ramanathan A, Gerszten RE, Wei R, Fleming MD, Schreiber SL, Cantley LC: The M2 splice isoform of pyruvate kinase is important for cancer metabolism and tumour growth. Nature. 2008, 452: 230-233. 10.1038/nature06734.View ArticlePubMedGoogle Scholar
- Mathupala SP, Ko YH, Pedersen PL: Hexokinase-2 bound to mitochondria: cancer’s stygian link to the ‘Warburg effect’ and a pivotal target for effective therapy. Semin Cancer Biol. 2009, 19: 17-24. 10.1016/j.semcancer.2008.11.006.PubMed CentralView ArticlePubMedGoogle Scholar
- Vazquez A, Liu J, Zhou Y, Oltvai Z: Catabolic efficiency of aerobic glycolysis: the Warburg effect revisited. BMC Sys Biol. 2010, 4: 58-10.1186/1752-0509-4-58.View ArticleGoogle Scholar
- Rathmell JC, Fox CJ, Plas DR, Hammerman PS, Cinalli RM, Thompson CB: Akt-directed glucose metabolism can prevent bax conformation change and promote growth factor-independent survival. Mol Cell Biol. 2003, 23: 7315-7328. 10.1128/MCB.23.20.7315-7328.2003.PubMed CentralView ArticlePubMedGoogle Scholar
- Gottlob K, Majewski N, Kennedy S, Kandel E, Robey RB, Hay N: Inhibition of early apoptotic events by Akt/PKB is dependent on the first committed step of glycolysis and mitochondrial hexokinase. Genes Dev. 2001, 15: 1406-1418. 10.1101/gad.889901.PubMed CentralView ArticlePubMedGoogle Scholar
- Vaughn AE, Deshmukh M: Glucose metabolism inhibits apoptosis in neurons and cancer cells by redox inactivation of cytochrome c. Nat Cell Biol. 2008, 10: 1477-1483. 10.1038/ncb1807.PubMed CentralView ArticlePubMedGoogle Scholar
- Vazquez-Martin A, Vellon L, Quirós PM, Cufí S, de Galarreta Ruiz E, Oliveras-Ferraros C, Martin AG, Martin-Castillo B, López-Otín C, Menendez JA: Activation of AMP-activated protein kinase (AMPK) provides a metabolic barrier to reprogramming somatic cells into stem cells. Cell Cycle. 2012, 11: 974-989. 10.4161/cc.11.5.19450.View ArticlePubMedGoogle Scholar
- Shen Q, Goderie SK, Jin L, Karanth N, Sun Y, Abramova N, Vincent P, Pumiglia K, Temple S: Endothelial cells stimulate self-renewal and expand neurogenesis of neural stem cells. Science. 2004, 304: 1338-1340. 10.1126/science.1095505.View ArticlePubMedGoogle Scholar
- Palmer TD, Willhoite AR, Gage FH: Vascular niche for adult hippocampal neurogenesis. J Comp Neurol. 2000, 425: 479-494. 10.1002/1096-9861(20001002)425:4<479::AID-CNE2>3.0.CO;2-3.View ArticlePubMedGoogle Scholar
- Wurmser AE, Palmer TD, Gage FH: Cellular interactions in the stem cell niche. Science. 2004, 304: 1253-1255. 10.1126/science.1099344.View ArticlePubMedGoogle Scholar
- Louissaint A, Rao S, Leventhal C, Goldman SA: Coordinated interaction of neurogenesis and angiogenesis in the adult songbird brain. Neuron. 2002, 34: 945-960. 10.1016/S0896-6273(02)00722-5.View ArticlePubMedGoogle Scholar
- Robey IF, Stephen RM, Brown KS, Baggett BK, Gatenby RA, Gillies RJ: Regulation of the Warburg effect in early-passage breast cancer cells. Neoplasia. 2008, 10: 745-756.PubMed CentralView ArticlePubMedGoogle Scholar
- Dang CV, Le A, Gao P: MYC-induced cancer cell energy metabolism and therapeutic opportunities. Clin Cancer Res. 2009, 15: 6479-6483. 10.1158/1078-0432.CCR-09-0889.PubMed CentralView ArticlePubMedGoogle Scholar
- Belgacem YH, Borodinsky LN: Sonic hedgehog signaling is decoded by calcium spike activity in the developing spinal cord. Proc Natl Acad Sci U S A. 2011, 108: 4482-4487. 10.1073/pnas.1018217108.PubMed CentralView ArticlePubMedGoogle Scholar
- Bhatia B, Hsieh M, Kenney AM, Nahle Z: Mitogenic Sonic hedgehog signaling drives E2F1-dependent lipogenesis in progenitor cells and medulloblastoma. Oncogene. 2011, 30: 410-422. 10.1038/onc.2010.454.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.