Glucose metabolism in the liver activates the transcription of various genes encoding enzymes of glycolysis and lipogenesis and also G6pc (glucose-6-phosphatase). Allosteric mechanisms involving glucose 6-phosphate or xylulose 5-phosphate and covalent modification of ChREBP (carbohydrate-response element-binding protein) have been implicated in this mechanism. However, evidence supporting an essential role for a specific metabolite or pathway in hepatocytes remains equivocal. By using diverse substrates and inhibitors and a kinase-deficient bisphosphatase-active variant of the bifunctional enzyme PFK2/FBP2 (6-phosphofructo-2-kinase–fructose-2,6-bisphosphatase), we demonstrate an essential role for fructose 2,6-bisphosphate in the induction of G6pc and other ChREBP target genes by glucose. Selective depletion of fructose 2,6-bisphosphate inhibits glucose-induced recruitment of ChREBP to the G6pc promoter and also induction of G6pc by xylitol and gluconeogenic precursors. The requirement for fructose 2,6-bisphosphate for ChREBP recruitment to the promoter does not exclude the involvement of additional metabolites acting either co-ordinately or at downstream sites. Glucose raises fructose 2,6-bisphosphate levels in hepatocytes by reversing the phosphorylation of PFK2/FBP2 at Ser32, but also independently of Ser32 dephosphorylation. This supports a role for the bifunctional enzyme as the phosphometabolite sensor and for its product, fructose 2,6-bisphosphate, as the metabolic signal for substrate-regulated ChREBP-mediated expression of G6pc and other ChREBP target genes.
- carbohydrate-response element-binding protein (ChREBP)
- fructose 2,6-bisphosphate
- glucose 6-phosphate
The liver regulates blood glucose homoeostasis by production of glucose by glycogenolysis and gluconeogenesis in the post-absorptive state and by conversion of glucose into glycogen and triacylglycerol in the fed state. Insulin and glucose act co-ordinately in promoting glycogen storage and also stimulation of glycolysis and lipogenesis through elevation of F2,6P2 (fructose 2,6-bisphosphate), an allosteric activator of phosphofructokinase-1  and through transcriptional regulation of glycolytic and lipogenic genes [2,3]. The main transcription factor involved in regulation of gene expression by glucose in liver is ChREBP (carbohydrate-response element-binding protein) , which belongs to the Mondo family of transcription factors . Its paralogue MondoA is expressed ubiquitously and is involved in glucose regulation in peripheral tissues . ChREBP forms heterodimers with its binding partner Mlx (Max-like protein X) , which is expressed ubiquitously and forms heterodimers with other transcription factors [8,9]. ChREBP expression in liver is induced by dietary carbohydrate  and it is also elevated in models of hyperphagia such as the ob/ob mouse [10,11].
ChREBP mediates transcriptional regulation of glycolytic and lipogenic enzymes that are elevated in fatty liver [2,3], as shown by the attenuated expression of these enzymes and improvement of fatty liver by ChREBP knockdown in the ob/ob mouse [10,11]. This supports a role for ChREBP in the conversion of carbohydrate into fat [2,3]. However, gene microarray studies in hepatocytes have identified several other potential ChREBP target genes that are induced by glucose through Mlx-dependent mechanisms . One notable case is G6pc, encoding the catalytic unit of glucose-6-phosphatase which catalyses the final reaction in glucose production. The glucose-induction of G6pc, which has been replicated in several studies in vitro [12–14] and in vivo in models of diabetes or hyperglycaemia [15,16], is regarded as a paradoxical response to glucose  because induction of G6pc by glucose aggravates the hyperglycaemia further. Expression of G6pc, like the expression of ChREBP and glycolytic enzymes, is elevated in the ob/ob mouse and repressed by ChREBP knockdown [10,11]. The high G6pc activity in the ob/ob model is attributed to the failure of insulin signalling  and the reversal by ChREBP knockdown to restoration of insulin signalling . However, binding of ChREBP to the G6pc promoter was shown in islet β-cells which express G6pc at lower levels than hepatocytes . The glucose-induction of G6pc may be a mechanism for intracellular homoeostasis of phosphorylated intermediates . Understanding the metabolic links between glucose metabolism and ChREBP activation may help to clarify the additional functions of ChREBP.
A high glucose concentration stimulates expression of ChREBP, its translocation to the nucleus and binding to the promoter elements . Changes in covalent modification and allosteric mechanisms are implicated in ChREBP activation [21–23]. However, the identity of the signalling metabolite(s) that link glucose metabolism with ChREBP activation remains unsettled. Several studies have argued in support of either X5P (xylulose 5-phosphate), as an activator of a phosphatase that promotes dephosphorylation of ChREBP , or for G6P (glucose 6-phosphate) [25,26]. The latter is supported by association studies showing increased gene expression at elevated G6P levels or by the use of 2-DG (2-deoxyglucose) that is phosphorylated on the 6-position. A limitation of association studies is that G6P is connected to several metabolic pathways such as the glycogenic, glycolytic and glucosamine pathways, and intermediates of these pathways would therefore be expected to change during experimentally induced perturbation of G6P. Although most studies have focused on the question of whether G6P or X5P is the candidate signalling molecule for ChREBP activation [24–26], there have been few systematic studies addressing the potential roles of multiple metabolites. In order to advance further the hypothesis that ChREBP activation is a mechanism for homoeostasis of phosphorylated intermediates , the aim of the present study was to identify candidate metabolites involved in glucose-regulated G6pc gene expression. We have identified F2,6P2 as an essential metabolite in glucose-induced expression of G6pc and other ChREBP target genes. Our results do not exclude the involvement of G6P and X5P or other mechanisms.
S4048  was a gift from Dr D. Schmoll (Aventis Pharma, Frankfurt, Germany). 5-Chloro-2-[N-(2,5-dichlorobenzenesulfonamido)]-benzoxazole, an FBPI (fructose-1,6-bisphosphatase inhibitor), was from Calbiochem. [1,2-3H]2-DG was from PerkinElmer.
Hepatocyte culture and treatment with adenoviral vectors
Hepatocytes were isolated from male Wistar rats fed ad libitum . Procedures conformed to Home Office Regulations and were approved by Newcastle University Ethical Committee. Hepatocytes were suspended in MEM (minimal essential medium) supplemented with 5% (v/v) neonatal calf serum, seeded on gelatin-coated (1 mg/ml) multiwell plates or coverslips for immunostaining and cultured for 2–4 h to allow cell attachment . Treatment with adenoviral vectors was in serum-free MEM between 2 and 4 h after cell attachment. Adenoviral vectors for expression of Mlx, MondoA, ChREBP, PFK-WT [wild-type PFK2/FBP2 (6-phosphofructo-2-kinase–fructose-2,6-bisphosphatase)] and a kinase-deficient variant, PFK-KD (S32D, T55V) were as described previously [12,29]. After cell attachment, the medium was replaced by MEM containing 5 mM glucose and 10 nM dexamethasone and the hepatocytes were cultured overnight (16–18 h).
Incubations with substrates
Incubations for substrate-regulated gene expression were performed (after pre-culture of the hepatocytes for 16–18 h) in fresh MEM containing 10 nM dexamethasone, 10 nM insulin and 5 mM glucose (unless stated otherwise) and with the substrates/inhibitors indicated. Parallel incubations were performed for RNA extraction (4 h of incubation) and metabolite determination (1 or 4 h of incubation). Metabolite data reported are for 1 h of incubation unless otherwise stated. Glycolysis was determined from the metabolism of [3-3H]glucose as in  and from pyruvate and lactate formation determined enzymically . Metabolism of 2-DG was determined using [1,2-3H]2-DG (6 μCi/ml). Incubations were for 4 h and were terminated by collection of the medium for determination of 3H2O  followed by two rapid washes of the hepatocyte monolayer with 150 mM saline and extraction of the cells in 0.1 M NaOH for determination of labelled cellular metabolites and glycogen .
F2,6P2 was determined after cell extraction in 0.15 M NaOH . For other cell metabolites, incubations were stopped by draining the medium and snap-freezing in liquid nitrogen. The cells were extracted in either 5% (w/v) [2-DG6P (2-deoxyglucose 6-phosphate)] or 10% (w/v) [ATP, G6P, F6P (fructose 6-phosphate), X5P, NAG (N-acetylglucosamine) metabolites] HClO4, and the deproteinized supernatant (following centrifugation at 11000 g for 15 min) was neutralized with 3 M K2CO3. G6P and F6P were determined fluorimetrically by coupling NADPH formation to reduction of resazurin with diaphorase (λex 530 nm, λem 590 nm, Spectramax, Me5) as in  except that phosphoglucoisomerase (1 μg/ml) was included for determination of F6P. Determination of 2-DG6P was carried out using the method described in , except that G6P dehydrogenase was 6 units/ml and two sets of standards (G6P and 2-DG6P) were used, which reached end point within 1 min and 6 h respectively. The 2-DG6P standards were used to determine the kinetic end point, whereas the G6P standards were used for estimation of concentration. Results were not corrected for the labile nature of 2-DG6P under acidic conditions and are an underestimate by approximately 30–60%. X5P was determined as described in , but by coupling to resazurin as described in . NAG metabolites and the activity of glutamine:fructose-6-phosphate amidotransferase were determined as in . ATP was determined using a luciferase assay kit (Sigma–Aldrich) to establish substrate, inhibitor and protein overexpression conditions that do not affect cell ATP. 2-DG lowered ATP at 20 mM and overexpression of MondoA lowered ATP unless Mlx was co-expressed. No ATP depletion was observed with the substrate incubation conditions reported in the present study. Metabolites are expressed as pmol/mg of protein (F2,6P2) or nmol/mg of protein (other metabolites) or, where indicated, as the fold change relative to 5 mM glucose control (represented as unity).
Real-time RT (reverse transcription)–PCR
RNA was extracted in TRIzol® (Invitrogen)  and cDNA was synthesized from 1 μg of RNA with random hexamers and Superscript reverse transcriptase (Invitrogen). Real-time RT–PCR was performed in a total volume of 10 μl containing 50 ng of reverse-transcribed RNA and 5 ng of forward and reverse primers in a Roche Capillary Light Cycler, with initial denaturation at 95°C for 10 min followed by 40–50 cycles of 95°C for 15 s, 58°C for 7 s and 72°C for 15 s. The primers used were: G6pc, 5′-CTACCTTGCGGCTCACTTTC-3′ (forward) and 5′-ATCCAAGTGCGAAACCAAAC-3′ (reverse); Pklr (pyruvate kinase liver/red blood cells), 5′-CTGGAACACCTCTGCCTTCTG-3′ (forward) and 5′-CACAATTTCCACCTCCGACTC-3′ (reverse); Gck (glucokinase), 5′-TTGCTCTAAGGGGACCAGAA-3′ (forward) and 5′-GGAACGAGGGAGAGAAGGAC-3′ (reverse); TXNIP (thioredoxin-interacting protein), 5′-ACCAGTGTCTGCCAAAAAGG-3′ (forward) and 5′-GCCATTGGCAAGGTAAGTGT-3′ (reverse); Mlx, 5′-TCTGTCCCCAACACAGATGA-3′ (forward) and 5′-ACGATGGCTTTGCTGAGTTT-3′ (reverse); ChREBP, 5′-GGGACATGTTTGATGACTATGTC-3′ (forward) and 5′-AATAAAGGTCGGATGAGGATGCT-3′ (reverse); and MondoA, 5′-ATCCACAGCGGCCACTTCATG-3′ (forward) and 5′-TCATGCACTCGAAGAGCTTGG-3′. All mRNA values are expressed as the fold change relative to the 5 mM glucose control (represented as unity).
ChIP (chromatin immunoprecipitation) assays
Binding of transcription factors to the gene promoters were determined using the Upstate Biotechnology ChIP assay kit (Millipore) essentially as described in . Hepatocytes were cultured in 150 cm2 dishes and were treated with adenoviral vector (PFK-KD) and incubated under otherwise identical conditions as for the mRNA studies. After pre-clearing  cell supernatants were incubated overnight (4°C) with 6 μg of IgG against ChREBP (NB400-135, Novus Biologicals), Mlx (sc-14705, Santa Cruz Biotechnology) or control IgG (sc-2027, Santa Cruz Biotechnology). DNA was recovered by phenol/chloroform extraction and amplified by Touchdown real-time PCR with primer sequences amplifying promoter regions of the Pklr [5′-GGATGCCCAATATAGCCTCA-3′ (forward) and 5′-CCATGCTGCTACGTTGCTTA-3′ (reverse)], and G6pc [5′-GCATCAGCCCTGTGTGAATA-3′ (forward) and 5′-GAGTTGAGGGCAAACAGAGC-3′ (reverse)] genes.
Proteins were fractionated by SDS/PAGE and transferred on to nitrocellulose membranes. Membranes were incubated overnight with antibodies against PFK2/FBP2-pSer32 or PFK2/FBP2-bisphosphatase domain . After washing, membranes were incubated for 1 h with horseradish-peroxidase-conjugated anti-rabbit or anti-chicken IgG (Dako), and immunoreactive bands were visualized by enhanced chemiluminescence (GE Healthcare).
Hepatocyte monolayers on gelatin-coated coverslips were fixed at the end of the incubation with 4% (w/v) paraformaldehyde  and immunostained with anti-ChREBP antibody (NB400-135, Novus Biologicals) and Alexa Fluor® 488-conjugated anti-rabbit IgG (Molecular Probes). Nuclei were counterstained using Hoechst 33342. Cells were imaged using either a Nikon E400 or a Leica TCS-CP2-UV microscope with a ×63 NA (numerical aperture) 1.3 oil-immersion objective. For quantification >100 hepatocytes were imaged, and ChREBP subcellular localization was scored as to whether: (i) nuclear>cytoplasm; (ii), nuclear=cytoplasm; (iii) nuclear<cytoplasm. Nuclei were located using Hoechst 33342 staining. The number of cells showing predominantly nuclear localization was expressed as the percentage of cells imaged.
Incubations were in duplicate or triplicate in each experiment and, unless indicated otherwise, results are means±S.E.M. for the number of hepatocyte preparations. Statistical analysis was by Student's paired t test or two-way ANOVA with Bonferroni correction for ChIP assays.
Expression of G6pc mRNA correlates with G6P levels
Studies in vivo have shown an association between high G6P levels and increased expression of ChREBP target genes using the chlorogenic derivative S4048, a potent inhibitor of the G6P transporter that markedly elevates hepatic G6P levels [35,36]. S4048 is a powerful experimental tool in vitro because it raises G6P at a high glucose concentration (>5 mM) or high gluconeogenic flux, but has a negligible effect at 5 mM glucose in the absence of gluconeogenic precursors . To test whether G6pc mRNA levels correlate with G6P levels, hepatocytes were incubated for 4 h with 25 mM glucose or with 5 mM DHA (dihydroxyacetone) which is converted into G6P by gluconeogenesis, and with either S4048 or with an FBPI. G6P levels were elevated (P<0.01) with 25 mM glucose or DHA and were markedly augmented by S4048. However, the elevation by DHA was attenuated (P<0.02) by the FBPI (Figure 1A). G6pc mRNA was elevated by 25 mM glucose and by DHA and augmented by S4048 and attenuated by the FBPI in parallel with the changes in G6P (Figure 1B). Thus G6pc mRNA correlated significantly (r=0.83, P<0.01) with G6P (Figure 1C). To confirm that the induction of G6pc by 25 mM glucose is dependent on glucose metabolism to G6P, we tested the effect of 5-thioglucose, a hexokinase inhibitor that abolishes the increase in G6P . In the combined presence of 3 mM 5-thioglucose and 25 mM glucose, there was no elevation of G6pc mRNA relative to 5 mM glucose (1.3±0.2, n=3, P>0.05 compared with 1.0 for 5 mM glucose). This rules out a direct effect of glucose on G6pc expression. To test for possible effects of S4048 that are independent of the elevation of G6P levels (Figure 1A), we determined the effect of S4048 at 5 mM glucose, because S4048 does not raise G6P levels at 5 mM glucose . We found no effect of S4048 on G6pc mRNA at 5 mM glucose (1.1±0.7, n=10, P>0.05 compared with 1.0 for 5 mM glucose). Cumulatively, these results support the validity of S4048 as a tool to study metabolite control of gene expression, they exclude a direct effect of glucose on G6pc mRNA expression and they demonstrate a correlation between G6pc mRNA and G6P.
2-DG does not induce G6pc mRNA
To test whether elevation of G6P can mimic the action of glucose on G6pc mRNA, we used 2-DG, which is phosphorylated on the 6-position, but not metabolized further by phosphoglucoisomerase. This hexose is commonly used to distinguish between effects of H6P (hexose 6-phosphate) as opposed to downstream metabolites of glycolysis. Low rates of metabolism of 2-DG by the glycogenic and pentose phosphate pathways have been reported in some experimental systems . We confirmed cellular accumulation of 2-DG6P during incubation with 2-DG and this was enhanced by S4048 (Figure 2A), indicating that 2-DG6P is a substrate for the G6P transporter and for G6pc. 2-DG did not affect cellular ATP at concentrations up to 10 mM (results not shown). Using 3H-labelled 2-DG, we confirmed cellular accumulation of 3H-labelled metabolites that was enhanced by S4048 (Figure 2B). Incorporation of label into glycogen accounted for 3–10% of total cellular 3H-labelled metabolites, but label incorporation into 3H2O was not detectable. This indicates metabolism of 2-DG by the glycogenic pathway, but not by the pentose phosphate pathway. Incubation with 2-DG did not induce G6pc mRNA (Figure 2C), despite levels of 2-DG6P (Figure 2A) exceeding the G6P concentration at 25 mM glucose (Figure 1). In additional experiments in the presence of 25 mM glucose or 2-DG (5 and 10 mM) also did not augment G6pc mRNA (results not shown). The expression of the ChREBP target Pklr was also not affected by 2-DG (results not shown), as reported previously . 2-DG has been shown to induce TXNIP in islet β-cells [40,41] and other non-hepatic cells , and both ChREBP and its paralogue MondoA have been implicated in the transcriptional control of TXNIP in non-hepatic cells (for a comprehensive review, see ). We found that TXNIP mRNA was induced by 25 mM glucose and by 2-DG (Figure 2D). S4048 enhanced the effect of 25 mM glucose, but not the effect of 2-DG (Figure 2D), despite the elevation of 2-DG6P by S4048 (Figure 2A). This implies an effect of 2-DG that is independent of its phosphorylation as reported previously for TXNIP induction in islet β-cells . Cumulatively, these results show that, whereas treatment with 2-DG causes marked accumulation of the corresponding H6P (Figure 2A) to levels exceeding those of G6P at 25 mM glucose and induces TXNIP mRNA, it does not elevate either G6pc mRNA or Pklr mRNA. This suggests that one or more metabolites of glucose that are not generated from 2-DG are essential for glucose-regulated G6pc gene expression.
Effects of overexpression of ChREBP or MondoA on gene expression
We next tested the effects of overexpression of ChREBP or MondoA on the glucose-induction of G6pc. Preliminary experiments titrating the adenoviral vector concentration showed that overexpression of ChREBP caused a small but significant increase in cell ATP, whereas overexpression of MondoA decreased ATP (control, 8.9±0.5; ChREBP, 10.3±0.7*; MondoA, 6.5±0.8** nmol/mg; *P<0.02; **P<0.002). Lowering of ATP by MondoA was not observed if the binding partner Mlx was co-expressed (control, 8.9±0.5; Mlx 9.3±0.8, Mlx+MondoA, 10.3±0.7*; Mlx+ChREBP 10.4±0.7*, nmol/mg; *P<0.02). MondoA was therefore co-expressed with Mlx and compared with overexpression of Mlx alone. Treatment with the adenoviral vectors for ChREBP, Mlx and MondoA was associated with an increase in mRNA for the respective genes (Figures 2E–2G). Expression of G6pc (Figure 2H) and Pklr (Figure 2I) were significantly increased by expression of ChREBP at 5 mM glucose, and G6pc was also increased at 25 mM glucose. However, expression of Mlx alone or in combination with MondoA had no effect on expression of either G6pc or Pklr (Figures 2H and 2I), confirming that ChREBP, but not Mlx, limits the expression of these genes at 5 mM glucose. Expression of TXNIP (Figure 2J) at 25 mM glucose was increased by expression of MondoA plus Mlx, but not by Mlx alone or by expression of ChREBP. Cumulatively, these results indicate a role for ChREBP in the glucose-induction of G6pc.
Elevated F2,6P2 levels or glycolysis are essential for glucose-induction of G6pc
Since 2-DG is not a substrate for phosphoglucoisomerase and therefore cannot be converted into F6P or metabolized by downstream pathways including formation of F2,6P2, glycolysis and the hexosamine pathway, we next asked whether the glucose-induction of G6pc requires glucose metabolism by glycolysis. Flux through glycolysis in hepatocytes is markedly dependent on the concentration of F2,6P2, which is synthesized from F6P by the kinase activity of the bifunctional enzyme PFK2/FBP2 . F2,6P2 is degraded by the bisphosphatase activity of PFK2/FBP2 and, accordingly, the cellular F2,6P2 is determined by the concentration of the substrate, F6P (which is in equilibrium with G6P) and by the kinase/bisphosphatase ratio of PFK2/FBP2. The latter is dependent on phosphorylation of the N-terminus (Ser32) which favours an increase in the bisphosphatase/kinase ratio and on allosteric effectors [1,42]. To test for a role for either F2,6P2 or for downstream flux through glycolysis, we expressed a kinase-deficient bisphosphatase-active variant of PFK2/FBP2 (PFK-KD) to unbalance the kinase/bisphosphatase ratio and thereby degrade the F2,6P2 generated at elevated glucose. Incubation with 25 mM glucose caused a 5-fold elevation of F2,6P2 levels (Figure 3A). This was markedly attenuated (>70%) by expression of PFK-KD (Figure 3A). However, the hepatic G6P content was unchanged by PFK-KD at 25 mM glucose (Figure 3B). This establishes the validity of PFK-KD as a tool for lowering F2,6P2 levels at 25 mM glucose independently of the H6P substrate. Flux through glycolysis at 25 mM glucose determined from [3-3H]glucose metabolism (Figure 3C) or formation of pyruvate plus lactate (Figure 3D) was decreased by PFK-KD, as expected from the lowering of F2,6P2 levels . The lowering of F2,6P2 levels and glycolysis with PFK-KD was associated with marked attenuation of G6pc mRNA (Figure 3E) and Pklr mRNA (Figure 3F), However, expression of GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was unaffected by PFK-KD (results not shown) and likewise expression of TXNIP mRNA was also not affected by PFK-KD at 25 mM glucose (2.65±0.38 compared with 2.77±0.29, means±S.E.M., n=8). We next asked whether the repression of G6pc and Pklr mRNA caused by lowering of F2,6P2 levels with PFK-KD is confined to glucose-inducible and/or ChREBP target genes? We therefore determined Gck mRNA because Gck is repressed by glucose through an Mlx-dependent but ChREBP-independent mechanism . In contrast with the suppression of G6pc and Pklr mRNA (Figures 3E and 3F), Gck mRNA was significantly elevated by PFK-KD (Figure 3G). To determine whether the effects of PFK-KD (Figures 3E–3G) are due to depletion of F2,6P2 as distinct from an effect of PFK-KD protein independent of its catalytic activity, we overexpressed PFK-WT (Figures 3H–3J). This raised F2,6P2 levels at 25 mM glucose (Figure 3A), but had a negligible effect on glycolysis (Figure 3B), presumably because of a saturating effect of endogenous levels . Expression of PFK-WT had converse effects compared with PFK-KD on G6pc mRNA (Figure 3H) and Pklr mRNA (Figure 3I) and lowered Gck mRNA at 5 mM glucose (Figure 3J). This confirms that the changes in mRNA expression caused by PFK-KD (Figures 3E–3G) are caused by depletion of F2,6P2 or suppression of glycolysis, but not by an effect of PFK2/FBP2 protein independently of its catalytic activity.
DHA-induced G6pc mRNA correlates with F2,6P2
We next asked whether F2,6P2 can account for the correlation between G6pc mRNA and G6P levels in the experiment shown in Figure 1. Cellular F2,6P2 levels under the incubation conditions shown in Figure 1 were elevated by DHA and attenuated by FBPI (Figure 4A) in parallel with the changes in G6P (cf. Figure 1A). S4048 augmented the elevation of F2,6P2 levels by both 25 mM glucose and DHA. However, the fold increase in concentration relative to 5 mM glucose was greater for G6P than for F2,6P2 (Figure 4B, x-axis compared with y-axis; 40-fold compared with 10-fold) and G6pc mRNA levels correlated more strongly with F2,6P2 levels than with G6P levels (r=0.94 compared with r=0.84; Figure 4C), implicating a potential role for F2,6P2 in the control of G6pc mRNA.
We next asked whether the suppression of G6pc mRNA by PFK-KD (Figure 3E) is due to suppression of glycolysis (Figures 3C and 3D) or to an effect of F2,6P2 independent of glycolysis. To address this point, we tested the effect of PFK-KD on G6pc induction with DHA as the substrate. DHA is phosphorylated by triokinase to dihydroxyacetone phosphate and this is metabolized by both gluconeogenesis and glycolysis. During incubation with 5 mM DHA, expression of PFK-KD markedly lowered F2,6P2 levels (Figure 4D) and significantly raised G6P levels (Figure 4E). The latter is consistent with an inhibitory effect of F2,6P2 on fructose-1,6-bisphosphatase 1 . Formation of pyruvate and lactate, a measure of flux through glycolysis, was raised during incubation with 5 mM DHA in both untreated cells and in cells treated with PFK-KD (Figure 4F). In experiments with titrated PFK-KD expression, F2,6P2 levels (Figure 4G) and G6pc mRNA (Figure 4H) were attenuated in parallel, and Gck mRNA (Figure 4I) was raised at 5 mM glucose with increasing PFK-KD titre. The lowering of G6pc mRNA with increasing PFK-KD titre correlated strongly (r=0.99, P<0.0001) with F2,6P2 (Figure 4J). This supports a role for F2,6P2 in the regulation of G6pc mRNA that is not explained by changes in glycolysis.
Elevated F2,6P2 levels are essential for xylitol-induced G6pc mRNA
Xylitol mimics the glucose-induction of G6pc  and other ChREBP targets . The proposed mechanism is that X5P, generated from xylitol or from glucose by the pentose phosphate pathway, activates PP2A (type 2A protein phosphatase) that dephosphorylates ChREBP . We therefore asked whether the attenuation by PFK-KD of G6pc induction with glucose or DHA (Figures 3 and 4) is associated with lowering of X5P levels and whether elevation of F2,6P2 levels is essential for the induction of G6pc by xylitol. Xylitol, like glucose and DHA, raised F2,6P2 levels (Figure 5A) and G6pc mRNA (Figure 5B), and both of these effects were attenuated by PFK-KD. X5P levels were significantly raised by 5 mM DHA (34%), 25 mM glucose (2-fold) and xylitol (Figure 5C). PFK-KD did not lower X5P levels with DHA, glucose or xylitol (Figure 5C) and it also did not lower H6P levels (F6P and G6P) with any of the substrates (Figure 5D). This shows that the suppression of G6pc mRNA by PFK-KD is not explained by suppression of X5P or H6P and that elevation of F2,6P2 levels is essential for induction of G6pc by xylitol. To test whether xylitol affects G6pc by mechanisms additional to elevation of F2,6P2, we determined the effect of xylitol in the presence of 25 mM glucose (Figures 5E and 5F). Xylitol did not increase F2,6P2 further when combined with 25 mM glucose (Figure 5E), but it markedly enhanced G6pc mRNA (Figure 5F). Thus, whereas F2,6P2 is essential for induction of G6pc by xylitol at 5 mM glucose (Figure 5B), a metabolite derived from xylitol markedly enhances G6pc mRNA at 25 mM glucose.
Suppression of F2,6P2 does not affect hexosamine pathway flux
A role for the hexosamine pathway in regulating G6pc expression  or ChREBP function  was suggested previously from incubations with glucosamine. The hexosamine pathway involves amidation of F6P with glutamine to glucosamine 6-phosphate catalysed by GFAT (glutamine:fructose-6-phosphate amidotransferase) and its further metabolism to UDP-GlcNAc (UDP-N-acetylglucosamine), the substrate for modification of serine/threonine residues by O-linked N-acetylglucosamine transferase (Figure 6A). We therefore asked whether the attenuation of G6pc mRNA levels by lowering F2,6P2 with PFK-KD could be explained by inhibition of flux through the hexosamine pathway. The cell content of NAG metabolites was elevated by 25 mM glucose and reversed by the GFAT inhibitor 6-diazo-5-oxo-L-norleucine (Figure 6B), in parallel with inhibition of GFAT activity (Figure 6C) and consistent with increased hexosamine pathway flux at 25 mM glucose. Treatment with PFK-KD caused a small but significant increase in NAG metabolites at 5 mM glucose, but had no effect at 25 mM glucose (Figure 6D), indicating that the suppression of G6pc by PFK-KD at 25 mM glucose (Figure 3) is not associated with decreased flux through the hexosamine pathway. Glucosamine, which enters the hexosamine pathway after GFAT activity, increased NAG metabolite levels (Figure 6E), but, unlike 25 mM glucose, it did not affect G6pc mRNA levels (Figure 6F). This shows that changes in flux through the hexosamine pathway cannot account for the repression of G6pc mRNA by PFK-KD at 25 mM glucose.
F2,6P2 is essential for ChREBP recruitment to the G6pc and Pklr promoters
The attenuation of G6pc mRNA (Figure 3) by suppression of F2,6P2 with PFK-KD could be due to inhibition of transcription or to post-transcriptional mechanisms such as increased mRNA decay. To test for the former possibility, we determined the recruitment of Mlx and ChREBP to the G6pc and Pklr promoters by ChIP assays under similar conditions as for the mRNA analysis (Figure 3). Incubation of hepatocytes with 25 mM glucose plus S4048 for 4 h increased binding of Mlx and ChREBP to the Pklr and G6pc promoters (Figure 7). This was attenuated by suppression of F2,6P2 with PFK-KD, indicating a role for F2,6P2 in ChREBP recruitment to these promoters.
Elevated F2,6P2 levels are essential for glucose-induced ChREBP translocation
Glucose-regulated gene expression is dependent on translocation of ChREBP from the cytoplasm to the nucleus [21,22]. We tested whether F2,6P2 has a role in ChREBP translocation. For these experiments, wild-type ChREBP was overexpressed, and the hepatocytes were pre-cultured with glucagon to sequester ChREBP in the cytoplasm. Subcellular location of ChREBP and F2,6P2 content were determined in timed incubations at 5 mM or 25 mM glucose in medium without glucagon (Figures 8A and 8B). Translocation of ChREBP to the nucleus were more rapid at 25 mM glucose (Figure 8A). Depletion of F2,6P2 with PFK-KD counteracted the effect of 25 mM glucose on F2,6P2 and ChREBP translocation, whereas PFK-WT had no effect on translocation (Figures 8C–8E). This suggests a role for F2,6P2 in glucose-induced ChREBP translocation.
Role of dephosphorylation of PFK2/FBP2-Ser32 during substrate regulation
The above studies show that elevation of F2,6P2 levels at 25 mM glucose is essential for ChREBP translocation and recruitment to the G6pc promoter (Figures 7 and 8). Phosphorylation of the liver isoform of PFK2/FBP2 on Ser32 by PKA (protein kinase A) increases the bisphosphatase/kinase ratio, whereas dephosphorylation by a X5P-sensitive PP2A has the converse effect . We asked whether the elevation of F2,6P2 levels by 25 mM glucose is associated with dephosphorylation of PFK2/FBP2 on Ser32. The antiserum against pSer32  detects multiple bands (Figure 9A, upper blot). However, its selectivity for PFK2/FBP2-pSer32 was confirmed from lack of immunoreactivity at 50 kDa to S32A and S32D variants of PFK2/FBP2 (Figure 9A). Incubation of hepatocytes with glucagon (2 or 10 nM) markedly increased pSer32 immunoreactivity (Figure 9B), and the effect of 2 nM glucagon (but not 10 nM glucagon) was attenuated by incubation with xylitol or high glucose without or with S4048 (Figure 9B, middle panel). In the absence of glucagon, there was no immunoreactivity to pSer32. To confirm that this lack of immunoreactivity to pSer32 in the absence of glucagon was not due to low affinity of the antibody, we expressed PFK2-WT and tested for changes in pSer32 under conditions of elevated protein expression (Figure 9C). No decrease in immunoreactivity to Ser32 in response to glucose was detected at elevated expression of PFK2/FBP2 when the hepatocytes were not exposed to glucagon (Figure 9C). Two conclusions can be drawn: first, that high glucose or xylitol concentrations can reverse the phosphorylation of Ser32 by 2 nM glucagon; and, second, that the elevation of F2,6P2 by high glucose concentrations (without or with S4048) in the absence of glucagon is not caused by dephosphorylation of pSer32. This supports a major role for either a substrate effect (H6P concentration) or for allosteric control of PFK2/FBP2 activity by H6P or other metabolites . It implicates PFK2/FBP2 as a ‘metabolite sensor’ that responds to elevated phosphorylated intermediates through changes in F2,6P2 which induces ChREBP recruitment to the G6pc promoter.
G6pc catalyses the final reaction in hepatic glucose production, and its induction by glucagon and repression by insulin is consistent with its role in blood glucose homoeostasis. Its induction by glucose [12–16] could be seen as counterintuitive . However, an additional function of G6pc is to buffer the hepatocellular G6P concentration at elevated glucose levels [31,47]. This function is best supported by the changes in G6P concentration during acute inhibition of G6P hydrolysis with the chlorogenic derivative S4048 [35,36]. In conditions of negligible glycogenolysis and gluconeogenic flux, S4048 has no effect on G6P levels at 5 mM glucose, but it markedly elevates G6P levels at higher glucose concentrations . The induction of G6pc by glucose [12–16] or by S4048 [35,36] can therefore be rationalized as a mechanism to prevent excessive build up of phosphorylated intermediates. In the present study, we identified F2,6P2 as a critical metabolite that is essential for glucose recruitment of ChREBP to the G6pc promoter and induction of G6pc.
F2,6P2 was discovered through a search for the regulator of phosphofructokinase 1 that mediates the inhibition of glycolysis by glucagon . F2,6P2 is a ‘dead-end’ metabolite that is synthesized from F6P and degraded to F6P by a bifunctional protein with distinct kinase and bisphosphatase catalytic sites [1,42]. Formation of F2,6P2 by the liver isoform (PFKFB1) of PFK2/FBP2 is regulated by allosteric mechanisms involving H6P and other metabolites [1,46] and by PKA-mediated phosphorylation of Ser32  which increases the bisphosphatase/kinase ratio, resulting in depletion of F2,6P2. Dephosphorylation of pSer32 is catalysed by a PP2A activated by X5P . The hepatocyte concentration of F2,6P2 is very sensitive to small changes in concentration of glucose or cAMP . F2,6P2 thus integrates both substrate and hormone signals . At basal cAMP concentrations, the hepatic concentration of F2,6P2 parallels the H6P concentration. Several lines of evidence from the present study support involvement of F2,6P2 in substrate-induced activation of ChREBP. First, induction of G6pc with glucose or DHA which enters the glycolytic/gluconeogenic pathway at the triose phosphate level, correlates strongly with elevation of F2,6P2. Secondly, incubation with 2-DG which is associated with elevation of 2-DG6P (H6P), but not F2,6P2, levels does not induce G6pc. Thirdly, selective depletion of F2,6P2 with a bisphosphatase-active kinase-deficient variant of the bifunctional protein abolishes glucose-induced ChREBP translocation, recruitment of ChREBP to the G6pc promoter and G6pc mRNA induction without lowering either G6P or X5P levels. Conversely overexpression of wild-type PFK2/FBP2 elevates F2,6P2 and enhances G6pc expression at elevated glucose levels without modulating the G6P content. Fourthly, induction of G6pc by xylitol and DHA was abolished by selective depletion of F2,6P2 despite elevated X5P and H6P levels. The involvement of F2,6P2 cannot be explained by its established function as activator of glycolysis  because the lowering of F2,6P2 levels and G6pc mRNA in incubations with DHA was not associated with changes in glycolysis. This therefore points to a role for F2,6P2 in regulating expression of ChREBP target genes that is independent of changes in glycolysis. The mechanism by which F2,6P2 regulates ChREBP remains to be determined. There is currently no evidence that glucokinase is a direct ChREBP target . Thus the role of F2,6P2 in the glucose-repression of glucokinase suggests either that the involvement of F2,6P2 in transcriptional regulation is not confined to ChREBP or alternatively that glucokinase may be an indirect ChREBP target. Further work is required to distinguish between these possibilities.
The present evidence supporting an essential role for F2,6P2 in ChREBP recruitment to the G6pc and Pklr promoters does not exclude the involvement of additional metabolites acting either co-ordinately with F2,6P2 or at a more distal site as suggested by the transcriptional activation of the GAL4–ChREBP fusion protein by 2-DG . A number of points are relevant in this context. First, since F6P is a substrate for the formation of F2,6P2, elevation of F2,6P2 levels during glucose stimulation is contingent on adequate substrate levels of F6P, which is in equilibrium with G6P. Likewise overexpression of PFK-WT raises F2,6P2 levels and G6pc mRNA at 25 mM glucose, but has a negligible effect on either parameter at 5 mM glucose. Thus the possibility that F2,6P2 acts by an allosteric mechanism on ChREBP in conjunction with F6P or another metabolite linked to it cannot be excluded. We found no evidence for induction of G6pc or Pklr by 2-DG at either 5 or 25 mM glucose, in agreement with previous findings on Pklr in hepatocytes . A key question is how this is reconciled with the demonstration that 2-DG transcriptionally activates the GAL4–ChREBP system in INS1 823/13 β-cells and ChREBP target genes in HepG2 cells [25,26]. A possible explanation is that both F2,6P2 and H6P are required for transcriptional activation of ChREBP and that H6P is limiting in proliferating islet β-cells , whereas F2,6P2 is limiting in hepatocytes. The liver isoform of PFK2/FBP2 (PFKFB1) has a high bisphosphatase/kinase ratio , whereas the islet isoform has a low ratio . This is consistent with differences in dynamics of H6P and F2,6P2 in hepatocytes and INS1 cells. The most compelling data from the present study in support of mechanisms additional to F2,6P2 are the marked induction of G6pc by xylitol in the presence of 25 mM glucose (Figures 5E and 5F). This indicates the involvement of additional mechanism(s) linked to xylitol metabolism. Further work is required to identify these mechanisms.
Uyeda and colleagues proposed a role for X5P in regulating F2,6P2 by dephosphorylation of PFK2/FBP2-Ser32  and ChREBP activation by dephosphorylation of Ser626 and Thr666 . In support of the former mechanism, we have demonstrated that both xylitol and elevated glucose levels counteract the phosphorylation of Ser32 by low glucagon concentration, consistent with metabolite activation of a phosphatase . However, in the absence of glucagon, the elevation of F2,6P2 caused by glucose did not involve dephosphorylation of Ser32. This supports a role for allosteric mechanisms  in the elevation of F2,6P2. Thus PFK2/FBP2 functions as the phosphometabolite sensor that generates F2,6P2 in response to elevated F6P and/or other phosphorylated intermediates. The raised F2,6P2 levels causes ChREBP recruitment to the promoter of its target genes.
A previous paper reported a role for PFKFB3 in the control of cell proliferation . A role for ChREBP in redirecting carbohydrate metabolism towards nucleotide synthesis during cell proliferation has also been reported . Further work is required to understand the functional links between F2,6P2, PFK2/FBP2 and ChREBP.
Catherine Arden, Susan Tudhope, John Petrie, Ziad Al-Oanzi and Kirsty Cullen performed experiments and analysis and contributed to interpretation and discussion. Alex Lange and Howard Towle provided essential reagents and contributed to the discussion. Loranne Agius designed the study and wrote the paper.
The work was supported by the Medical Research Council [grant number G0501543/ID 76338] and Diabetes UK [grant number 07/0003488]. J.L.P. was supported by a Diabetes UK Studentship [grant number 07/0003559] and Z.H.A.-O. was supported by Al-Jouf University, Sakaka, Saudi Arabia.
Abbreviations: ChIP, chromatin immunoprecipitation; ChREBP, carbohydrate-response element-binding protein; 2-DG, 2-deoxyglucose; 2-DG6P, 2-deoxyglucose 6-phosphate; DHA, dihydroxyacetone; FBPI, fructose bisphosphatase inhibitor; F2,6P2, fructose 2,6-bisphosphate; F6P, fructose 6-phosphate; Gck, glucokinase; GFAT, glutamine:fructose-6-phosphate amidotransferase; G6P, glucose 6-phosphate; G6pc, glucose-6-phosphatase catalytic subunit; H6P, hexose 6-phosphate; MEM, minimal essential medium; Mlx, Max-like protein X; NAG, N-acetylglucosamine; PFK2/FBP2, 6-phosphofructo-2-kinase–fructose-2,6-bisphosphatase; PFK-KD, kinase-deficient PFK2/FBP2; PFK-WT, wild-type PFK2/FBP2; PKA, protein kinase A; Pklr, pyruvate kinase liver/red blood cells; PP2A, type 2A protein phosphatase; RT, reverse transcription; TXNIP, thioredoxin-interacting protein; X5P, xylulose 5-phosphate
- © The Authors Journal compilation © 2012 Biochemical Society