Aldehyde dehydrogenase activity is necessary for beta cell development and functionality in mice
- 1.8k Downloads
Pancreatic beta cells maintain glucose homeostasis and beta cell dysfunction is a major risk factor in developing diabetes. Therefore, understanding the developmental regulatory networks that define a fully functional beta cell is important for elucidating the genetic origins of the disease. Aldehyde dehydrogenase activity has been associated with stem/progenitor cells and we have previously shown that Aldh1b1 is specifically expressed in pancreas progenitor pools. Here we address the hypothesis that Aldh1b1 may regulate the timing of the appearance and eventual functionality of beta cells.
We generated an Aldh1b1-knockout mouse line (Aldh1b1 tm1lacZ) and used this to study pancreatic development, beta cell functionality and glucose homeostasis in the absence of Aldh1b1 function.
Differentiation in the developing pancreas of Aldh1b1 tm1lacZ null mice was accelerated. Transcriptome analyses of newborn and adult islets showed misregulation of key beta cell transcription factors and genes crucial for beta cell function. Functional analyses showed that glucose-stimulated insulin secretion was severely compromised in islets isolated from null mice. Several key features of beta cell functionality were affected, including control of oxidative stress, glucose sensing, stimulus-coupling secretion and secretory granule biogenesis. As a result of beta cell dysfunction, homozygous mice developed glucose intolerance and age-dependent hyperglycaemia.
These findings show that Aldh1b1 influences the timing of the transition from the pancreas endocrine progenitor to the committed beta cell and demonstrate that changes in the timing of this transition lead to beta cell dysfunction and thus constitute a diabetes risk factor later in life.
Gene Expression Omnibus (GEO) accession: GSE58025
KeywordsAldehyde dehydrogenase Beta cell development Beta cell dysfunction Beta cell transcriptome Diabetes risk factor Hyperglycaemia Insulin secretion
AMP-activated protein kinase
Dolichos biflorus agglutinin
Days post coitum
Glucagon-like peptide-1 receptor
Glucose-stimulated insulin secretion
Intraperitoneal glucose tolerance test
Intraperitoneal insulin tolerance test
Intraperitoneal pyruvate tolerance test
NK6 homeobox 1
Postnatal day 1
Pancreatic and duodenal homeobox 1
Reactive oxygen species
Extensive genome-wide association studies have identified over 80 genomic susceptibility loci for type 2 diabetes. The majority of the discovered risk variants have been linked to beta cell dysfunction rather than to insulin resistance . Thus, elucidating the developmental regulatory networks defining a fully functional beta cell is important for understanding the genetic origins of the disease and deriving mature beta cells from stem cells.
All pancreatic cell types are derived from progenitors that emerge at the posterior foregut region of the definitive endoderm and expand to form a branched epithelium surrounded by mesenchyme . Subsequently, acinar progenitors are confined at the tips, while trunk cells become endocrine/duct bipotent progenitors. Neurogenin 3-positive (NGN3+) endocrine progenitors arise in the trunk and migrate into the mesenchyme leaving behind ductal progenitors . The emergence of progenitor pools and differentiated cell types is temporally regulated but the mechanisms implicated are unknown [3, 4]. After birth, normal beta cell development continues with the maturation of the stimulus–secretion coupling machinery, the enhancement of glucose sensing, the increase in the number of insulin-containing secretory granules (SGs) and the establishment of the appropriate beta cell mass through extensive proliferation. Several transcription factors implicated in progenitor and endocrine specification also play key overlapping roles in the postnatal expansion, maturation and maintenance of adult beta cells. Their expression levels are also important for beta cell functionality [5, 6, 7, 8, 9, 10, 11, 12]. Disturbances of the intrauterine milieu can lead to islet defects and diabetes later in life but little is known about the molecular mediators of this effect [13, 14, 15].
Aldehyde dehydrogenase (ALDH) activity is increasingly associated with stem/progenitor cells and it is hypothesised that it contributes to the maintenance of the progenitor status [16, 17, 18, 19]. Mitochondrial Aldh1b1 is specifically expressed in all pancreas progenitors in the mouse embryo but not in mature endocrine cells. Its expression in the adult is confined to rare centroacinar-like cells with pancreas stem/progenitor characteristics [18, 20]. Here we addressed the hypothesis that Aldh1b1 may regulate the timing of the appearance and eventual functionality of beta cells.
Aldh1b1 tm1lacz-knock-in mouse strain
The Aldh1b1 tm1(KOMP)Vlcg ES line (KOMP; UC Davis, CA, USA) was used to generate the Aldh1b1 tm1lacZ allele (electronic supplementary material [ESM] Fig. 1). Animal maintenance and experimentation were in accordance with international guidelines and subjected to ethics approval from the competent veterinary committees of Athens and TU Dresden. Animals were assigned to experiments randomly on condition of fulfilling the conditions of the experiment (genotype, sex, age) as described, and the scoring/outcome assessment was blind to group assignment.
Fasting glucose measurements and glucose, insulin and pyruvate tolerance tests
For the intraperitoneal glucose tolerance test (IPGTT), male mice were fasted for 16 h overnight and injected intraperitoneally with d-glucose (Sigma-Aldrich, St Louis, MO, USA) at 2 g/kg. Blood from the tail vein was used to determined glucose level using the Contour XT monitoring system (Bayer, Leverkusen, Germany) and insulin level was determined in the supernatant fraction following clotting using an ultrasensitive mouse insulin ELISA kit (Mercodia, Uppsala, Sweden). Details of the intraperitoneal insulin tolerance test (IPITT) and intraperitoneal pyruvate tolerance test (IPPTT) procedures are provided in ESM Methods.
Immunostaining and morphometric analysis
Dissected pancreases were fixed in 4% wt/vol. paraformaldehyde and processed for immunofluorescence using standard procedures [18, 21]. Details of antibodies used and morphometric analyses are provided in ESM Methods.
Islet isolation and functional analyses
Islets were isolated from male mice and maintained as described . Standard procedures were used for insulin secretion assays, measurements of intracellular calcium ([Ca2+]i) changes , detection of reactive oxygen species (ROS) , ATP measurements  and transmission electron microscopy . Details are provided in ESM Methods.
RNA isolation, real-time PCR, RNA Seq and accession numbers
Total RNA was prepared using the RNeasy kit (Qiagen, Hilden, Germany), first-strand cDNA preparation, real-time PCR and library preparations for RNA sequencing and bioinformatics analyses were done according to standard procedures. Details are provided in ESM Methods. Raw and normalised data were deposited in GEO (www.ncbi.nlm.nih.gov/geo) under accession number GSE58025.
Hormone measurements and liver glycogen content assay
Pancreatic insulin, proinsulin and glucagon content, as well as liver glycogen content, were assayed as described in ESM Methods.
Western blotting and X-gal staining
Details on western blotting and X-gal staining procedures are provided in ESM Methods.
Statistical significance was determined by Student’s t test for two-tailed distributions of unpaired groups. For analyses of wild-type and null mice at different time points (with the exception of gestation time points) two-way ANOVA with Bonferroni post hoc test for differences between means were conducted. The SEM is provided unless otherwise stated and p < 0.05 was considered significant. For differences in islet size distributions and means, statistical significance was determined using the Wilcoxon rank-sum test and p < 0.001 was considered significant.
Premature lineage commitment during development in the absence of Aldh1b1 expression
We generated the Aldh1b1 tm1lacZ mouse line (ESM Fig. 1 a, b) and analysed pancreas development in the Aldh1b1 tm1lacZ null embryos in which Lacz expression recapitulated expression in the pancreas (ESM Fig. 1 c–h).
The number of epithelial cells in mitosis (Phosphohistone-H3+/E-cadherin+ cells) transiently increased in the Aldh1b1 tm1lacZ null pancreases at 14.5 dpc, thus compensating for the accelerated differentiation and maintaining the size of the organ (Fig. 1k and data not shown). Apoptosis levels, detected by immunofluorescence for activated caspase-3, were very low and similar to controls excluding implication of ALDH1B1 in progenitor survival (data not shown).
Expression of the sex determining region Y-box 9 (SOX9), pancreatic and duodenal homeobox 1 (PDX1) and pancreas specific transcription factor 1a (PTF1a) progenitor markers at 14.5 dpc and RNA Seq analysis of 13.5 and 15.5 dpc total pancreases did not detect other significant quantifiable differences between wild-type and Aldh1b1 tm1lacZ null pancreases, suggesting that only a subset of progenitor cells are affected at successive time points. Careful analysis of ALDH1B1 expression showed that the protein persisted in nascent C-PEP+ cells until 15.5 dpc but not later (Fig. 1i, j and data not shown).
Therefore, loss of Aldh1b1 function resulted in premature differentiation, balanced by increased mitosis, suggesting a dual role for Aldh1b1 in regulating both timing of commitment and proliferation of progenitors. We then asked whether loss of ALDH1B1 in progenitors and nascent beta cells affected islet patterning and beta cell function.
Islet patterning is defective in newborn Aldh1b1 tm1lacZ null mice
Cellular analysis revealed a striking heterogeneity of Aldh1b1 tm1lacZ null islets at postnatal day 1 (P1). Expression of the transcription factors PDX1 and NK6 homeobox 1 (NKX6.1) [12, 27, 28, 29, 30] was homogeneous in control islets whereas a large number of Aldh1b1 tm1lacZ null islets showed absent, substantially weaker NKX6.1 expression (47%) or strongly increased expression (18%). Some (35%) showed substantially weaker PDX1 expression compared with controls and co-expression of NKX6.1 and PDX1 within cells of the same islet was not uniform (ESM Fig. 2 a, b, b’ and data not shown) (n = 76). This was accompanied by variable levels of C-PEP immunofluorescence in the Aldh1b1 tm1lacZ null P1 pancreases, where 50% of the islets had either reduced or absent staining compared with controls (ESM Fig. 2c, d, d’). In addition, the number of phosphohistone-H3 (PH3)+ cells was increased by nearly 40% in the null islets (ESM Fig. 2f and data not shown). However, Aldh1b1 tm1lacZ P1 null pancreases had similar weight, beta cell mass, islet size distribution and insulin, proinsulin and glucagon content compared with controls. The relative numbers of alpha, beta and delta cells remained similar (Fig. 4d, ESM Figs 2 e, 3 g–j, l and data not shown).
Taken together these data suggest that Aldh1b1 is necessary for proper fetal development of beta cells and the upregulation of key transcription factors.
Islet patterning defects are maintained and expanded in adulthood
These findings showed that defects and heterogeneity in islet patterning persisted into adulthood and to determine their extent we compared the transcriptome of Aldh1b1 tm1lacZ null and control islets isolated at week 8. The number of misregulated genes had now expanded more than sixfold (Fig. 2b, c) and gene ontology analyses suggested that key components of beta cell function were affected (see below). Strikingly, expression of most transcription factors necessary to maintain islet functionality, with the notable exception of the downregulation of Nfe2l2 (Nrf2), a gene protecting beta cells from oxidative stress , was upregulated, suggesting an effort towards functional compensation (Fig. 2d and ESM Table 1). Consistent with the increased mitosis rates, expression of several cell cycle and mitosis-associated genes was altered in the Aldh1b1 tm1lacZ null islets (ESM Fig. 5a and ESM Table 2). Continued mitosis in the islets resulted in a 3.6-fold expansion of the beta cell mass and a twofold increase in the median islet size in Aldh1b1 tm1lacZ nulls by week 20 (Fig. 4a–d, ESM Fig. 2e and ESM Fig. 3a).
These findings showed that early defects persisted and expanded in the islets of Aldh1b1 tm1lacZ null mice, suggesting impaired functionality.
Null mice are glucose intolerant and develop age-dependent hyperglycaemia
To establish whether these molecular defects compromised glucose homeostasis we first determined blood glucose levels following overnight fasting of 6- to 52-week-old mice. From 12 weeks onwards Aldh1b1 tm1lacZ null mice developed hyperglycaemia, which accelerated with age (Fig. 4e) . IPGTTs showed that the onset of hyperglycaemia was preceded by reduced glucose tolerance and significant delay in acute insulin secretion in young adult Aldh1b1 tm1lacZ null mice. The first-phase response in the Aldh1b1 tm1lacZ null mice took nearly three times as long to peak without reaching control levels and a second-phase response was absent (Fig. 4f, g). Reduced glucose tolerance was further exacerbated at week 20 and acute insulin secretion was equally weak at week 20 (ESM Fig. 3b, c).
Hormone levels were not reduced at week 8 or week 20 in Aldh1b1 tm1lacZ null pancreases and pancreas weight was similar to that in wild-type controls (see above and data not shown). In fact, there was a transient increase in insulin and proinsulin content as well as a 2.6-fold increase in glucagon content in the null pancreases at week 8 (ESM Fig. 3j). The insulin-to-proinsulin ratio remained remarkably similar between Aldh1b1 tm1lacZ null and controls until week 20 (ESM Fig. 3g–i). Total pancreas hormone levels returned to normal at week 20 and this, coupled with the 3.6-fold expansion in beta cell mass at that stage (ESM Figs 3g, h, j and Fig. 4d), implied lower insulin concentration in cells.
ALDH1B1 is not expressed in the fat or skeletal muscle of adult mice, as demonstrated by the lack of detectable β-galactosidase activity in Aldh1b1 tm1lacZ heterozygotes (data not shown). To exclude the possibility that the defects in glucose homeostasis observed in Aldh1b1 tm1lacZ null mice are due to secondary defects arising from the liver or other peripheral tissues, we conducted a series of experiments that included IPITTs (ESM Fig. 3d), measurement of glucagon levels in the blood (ESM Fig. 3l), IPPTTs and measurement of glucose-6-phosphatase gene transcript levels as indicators of gluconeogenesis (ESM Fig. 3m, n), and measurement of hepatic glycogen stores (ESM Fig. 3m). In all assays, Aldh1b1 tm1lacZ null mice behaved similarly to wild-type controls.
Taken together, these findings suggested that the Aldh1b1 tm1lacZ null mice are glucose intolerant due to islet defects in glucose-stimulated insulin secretion (GSIS).
Aldh1b1 tm1lacZ null islets are defective in GSIS
These findings showed that Aldh1b1 tm1lacz null islets had severe defects in glucose sensing, stimulus-coupled insulin secretion or both.
Glucose sensing is impaired in the Aldh1b1 tm1lacZ null islets
We then investigated glucose uptake, glycolytic flux and ATP generation in the Aldh1b1 tm1lacZ null mutants. Immunostaining for Glut2 (Slc2a2) did not reveal changes in expression at either week 6 or week 20 (ESM Fig. 4 i, j). However, GLP1R immunostaining was severely decreased in all Aldh1b1 tm1lacZ islets, indicating the loss of its potentiating effect in glucose uptake (Fig. 5f, g). RNA Seq data showed that expression of c-myc, a central regulator of glycolytic gene expression , was strongly repressed and expressions of several genes involved in the glycolytic pathway and genes encoding enzymes diverting glutamine in the tricarboxylic acid cycle were deregulated (Fig. 5e and ESM Table 1). In mammals, low energy levels are reflected by increased AMP:ATP or ADP:ATP ratios and this leads to the activation of AMP-activated protein kinase (AMPK) by phosphorylation . Levels of P-AMPKA in the Aldh1b1 tm1lacZ null islets were substantially higher than in controls suggesting that null beta cells were energy depleted (Fig. 5i, j). ATP levels, determined by chemiluminescence, were indeed reduced by nearly twofold in isolated Aldh1b1 tm1lacZ null islets (Fig. 5b). Additionally, a very large number of ribosomal protein genes were downregulated (ESM Fig. 5e and ESM Table 2). This also explains the weak expression of PDX1 and NKX6.1 at the protein level (Fig. 3a–c) despite the upregulation of the corresponding genes (Fig. 2d and ESM Table 1).
Beta cells are vulnerable to sustained oxidative stress due to the low expression of antioxidant enzymes . Control mice maintained or induced expression of several genes involved in protection from and repair of oxidative damage in the transition from P1 to the adult stage. This induction did not take place in the Aldh1b1 tm1lacZ null islets (Fig. 5k and ESM Table 1), consistent with the repression of Nfe2l2 (Fig. 2d) . Staining of islets isolated at week 20 with the live ROS indicator CM-H2DCFDA indicated a much higher ROS content in the null islets (Fig. 5l, m). Additionally, immunofluorescence for 4-hydroxynonenal, an indicator of excessive oxidation of unsaturated fatty acids, showed that Aldh1b1 tm1lacZ null islets were more strongly stained (ESM Fig. 4k, l). Thus, these findings suggested that Aldh1b1 tm1lacZ null islets are subject to oxidative stress.
Taken together these data showed that Aldh1b1 tm1lacZ null islets are energy depleted and exposed to high oxidative load.
Stimulus-coupling secretion is impaired in the Aldh1b1 tm1lacZ null islets
We then investigated the expression and function of the constituent components of stimulus-coupling secretion machinery. The RNA Seq data showed that expression of several K+ and Ca2+ transporters was misregulated in the Aldh1b tm1lacZ islets (ESM Fig. 5b and ESM Table 2). Additionally, there was a strong upregulation of several K+ channel genes in Aldh1b1 tm1lacZ null islets, including Kcnj11, a type 2 diabetes risk factor  (Fig. 5n and ESM Table 1), and expression of adenylate cyclases Adcy4 and Adcy5 was repressed (Fig. 5h and ESM Table 1). To determine whether defects from membrane depolarisation and downstream of it contributed to the phenotype, we first stimulated insulin secretion in isolated islets with 100 μmol/l tolbutamide, a K+ channel antagonist, in the presence of basal levels of glucose. Following stimulation, the insulin concentration went from 79.2 ± 8.6 pmol/l to 13.9 ± 0.15 nmol/l in the wild-type islet medium and reached just 3.27 ± 0.15 nmol/l in the Aldh1b1 tm1lacZ null islet medium, suggesting that the efficiency of events starting with K+-mediated membrane depolarisation and leading to insulin secretion was compromised (Fig. 5c).
The L-type voltage-gated Ca2+ channel genes Cacna1c and Cacna1d are essential for rodent pancreatic beta cell function [39, 40] and polymorphisms in Cacna1d have been associated with type 2 diabetes . RNA Seq data showed that Ca2+ channel genes, including Cacna1c and Cacna1d, were upregulated in the adult Aldh1b1 tm1lacZ islets (Fig. 5n and ESM Table 1). We isolated potential K+ channel defects by directly stimulating Ca2+ influx in isolated islets with 2 μmol/l of the specific L-type voltage-sensitive Ca2+ channel agonist BayK8644 in the presence of basal levels of glucose. Following stimulation of wild-type islets, insulin concentration in the medium reached 0.58 ± 0.04 nmol/l but remained virtually unchanged in the Aldh1b1 tm1lacZ null islets, suggesting that increased expression of voltage-gated Ca2+ channels was not reflected at the protein level and/or that formation of SGs and exocytosis were affected (Fig. 5d). We then followed, by fluorescence, changes in the intracellular Ca2+ concentration in response to glucose stimulation in isolated islets. The average response was reduced in the week 20 Aldh1b1 tm1lacZ null islets (six mice, n = 59) compared with the wild-type controls (four mice, n = 34) and this was attributed to a group of islets (14 out of 59) with very low or negligible response (Fig. 5o, p). This is consistent with the islet heterogeneity in the nulls demonstrated by variable levels of NKX6.1, PDX1 and C-PEP immunostaining (ESM Fig. 4 e–h). Thus, reduced Ca2+ mobilisation contributes to, but is not solely responsible for, the observed phenotype.
Taken together, these data show that the functionality of the stimulus-coupling insulin secretion machinery of Aldh1b1 tm1lacZ null beta cells is impaired at multiple levels.
Deterioration of glucose homeostasis with age
To understand the substantial hyperglycaemia that developed with age, we analysed mice at week 52. Glucose tolerance had further deteriorated and acute insulin secretion was weak and monophasic (ESM Fig. 6 a, b). Energy depletion and oxidative stress were again detected in Aldh1b1 tm1lacZ null islets but not in controls. Additionally, increased cell death was now detected in all null islets but not in controls (ESM Fig. 6 c–h). Also, proinsulin processing was now less efficient in the Aldh1b1 tm1lacZ nulls and pancreas glucagon levels were increased (ESM Fig. 3 i, j). As a result of chronic hyperglycaemia the nulls developed insulin resistance by week 52 (ESM Fig. 3f). Thus, a combination of persisting defects, detected earlier, as well as cell death, defective proinsulin processing and insulin resistance, cause the exacerbation of the phenotype in old age.
Beta cell dysfunction is a clinical hallmark of the progression to type 2 diabetes . Studies in humans and rodents have associated nutrient depletion or environmental insults during fetal growth with increased susceptibility to adult onset of metabolic disease and beta cell dysfunction but little is known about the implicated molecular players [13, 14, 15]. The Aldh1b1 tm1lacZ null mice provide a model whereby early defects in fetal and neonatal stages manifest later in islet functional defects and deterioration of glucose homeostasis, capturing aspects of type 2 diabetes patients. Early defects in null young adult mice (weeks 6–8) are manifested both molecularly and functionally. Similar islet heterogeneity is encountered in all mice examined and therefore the phenotype among mice is relatively uniform. The presence of normal islets mitigates the phenotype.
Since Aldh1b1 is expressed only in pancreatic progenitors and briefly in newly formed fetal beta cells, the origin of the documented islet defects is embryonic. The relatively mild phenotype could be due to the robust expression of other ALDH genes including mitochondrial ones in the developing pancreas (ESM Table 3). Islet heterogeneity may result because either compensating ALDH genes are not homogeneously expressed in the progenitors or shifting the differentiation window of endocrine cells at an earlier time deprives a subset of differentiating cells from essential signals.
Reduced expression of several transcription factors that play overlapping roles in establishing and maintaining beta cell functionality [5, 6, 8, 9, 10, 11, 12, 27, 44] was significant at P1 in the islets of Aldh1b1 tm1lacZ null mutants whereas misregulation of other beta cell functional components was manifested mostly later. Expression of these transcription factors was subsequently upregulated in the islets of Aldh1b1 tm1lacZ null adults suggesting an adaptive response. The increase in beta cell mass is also such a response but does not address the functional defects arising from the misregulation of essential transcription factors. Thus, misregulation of the beta cell transcription factor network is the primary result of Aldh1b1 functional inactivation and underlies the multiplicity of defects in the islets of Aldh1b1 tm1lacZ null mice.
Availability of low ROS concentrations appears to be critical for self-renewal of some tissue stem cells [45, 46]. ALDH1B1 activity may participate in ROS level regulation in the pancreatic progenitors and, therefore, the regulation of the timing of lineage commitment. Premature commitment may limit the exposure of progenitors to necessary patterning signals. Alternatively, ALDH1B1 may modulate metabolic processes in the mitochondrion and, therefore, affect the levels of key metabolites necessary for the enzymes employed in epigenetic regulation of gene expression . It has indeed been hypothesised that nutrient availability during fetal growth may lead to epigenetic changes predisposing to diabetes later in life . Metabolomics of pancreatic progenitors and analysis of epigenetic changes in newly formed beta cells may clarify the underlying molecular mechanism.
In summary, the findings suggest that deregulation οf the transcription factors that control beta cell specification, maturation and maintenance is the main early effect of Aldh1b1 inactivation and this occurs at the level of progenitor/committed endocrine cells during development. This has implications for the conversion of pluripotent stem cells into functional mature beta cells and the restoration of dysfunctional beta cells in type 2 diabetes. In addition, the findings indicate that genetic predisposition to type 2 diabetes may arise from mutations in developmental genes, which may not necessarily result in an early postnatal phenotype but manifest later in life.
We thank M. Gouti (Francis Crick Institute, London), Z. Kanaki and A. Klinakis (BRFAA, Athens) for mouse ES cell culture and blastocyst injections, I. Kasampalidis (BRFAA, Athens) for statistical analysis of the islet size distributions, T. Kurth and S. Kretzschmar (TU Dresden) for EM embedding and sectioning and the personnel of the animal houses at BRFAA and MTZ/TU Dresden for expert animal husbandry. The Aldh1b1 mouse strain used for this research project was created from ES cell clone Aldh1b1_Ae11, generated by the Mouse Biology Program (www.mousebiology.org), and made into ES cells by the KOMP Repository (WWW.KOMP.org) at the University of California Davis.
This work was supported by grants from the European Foundation for the Study of Diabetes (EFSD/LILLY) (AG), the General Secretariat for Research and Technology (ESPA 2011, SYN-1-1496) (Greece) (AG), the Juvenile Diabetes Foundation (Hellas) (AG), the German Centre for Diabetes Research (DZD) (AG, SS), the Emmy Noether Program of the German Research Foundation (DFG) (SS), the CRTD – DFG Research Center for Regenerative Therapies Dresden (SS), the Cluster of Excellence and the DFG - SFB/Transregio 127 (SS), the DFG (AD) and the SFB655 of the DFG (AD).
IS and AG conceived and designed the study. All authors were involved with the acquisition, analysis and interpretation of data. AG wrote the manuscript, with critical revision from IS, VA, EN, DA, JS, AM, AD, MS and SS. All authors approved the final version of the paper. IS and AG are the guarantors of this work.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
- 35.Dang CV, Le A, Gao P (2009) MYC-induced cancer cell energy metabolism and therapeutic opportunities. Clin Cancer Res 15:6479–6483Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.