Research article

A role for PFK-2/FBPase-2, as distinct from fructose 2,6-bisphosphate, in regulation of insulin secretion in pancreatic β-cells

Catherine Arden, Laura J. Hampson, Guo C. Huang, James A. M. Shaw, Ali Aldibbiat, Graham Holliman, Derek Manas, Salmaan Khan, Alex J. Lange, Loranne Agius


PFK-2/FBPase-2 (6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase) catalyses the formation and degradation of fructose 2,6-P2 (fructose 2,6-bisphosphate) and is also a glucokinase-binding protein. The role of fructose 2,6-P2 in regulating glucose metabolism and insulin secretion in pancreatic β-cells is unresolved. We down-regulated the endogenous isoforms of PFK-2/FBPase-2 with siRNA (small interfering RNA) and expressed KA (kinase active) and KD (kinase deficient) variants to distinguish between the role of PFK-2/FBPase-2 protein and the role of its product, fructose 2,6-P2, in regulating β-cell function. Human islets expressed the PFKFB2 (the gene encoding isoform 2 of the PFK2/FBPase2 protein) and PFKFB3 (the gene encoding isoform 3 of the PFK2/FBPase2 protein) isoforms and mouse islets expressed PFKFB2 at the mRNA level [RT–PCR (reverse transcription–PCR)]. Rat islets expressed PFKFB2 lacking the C-terminal phosphorylation sites. The glucose-responsive MIN6 and INS1E cell lines expressed PFKFB2 and PFKFB3. PFK-2 activity and the cell content of fructose 2,6-P2 were increased by elevated glucose concentration and during pharmacological activation of AMPK (AMP-activated protein kinase), which also increased insulin secretion. Partial down-regulation of endogenous PFKFB2 and PFKFB3 in INS1E by siRNA decreased PFK-2/FBPase-2 protein, fructose 2,6-P2 content, glucokinase activity and glucoseinduced insulin secretion. Selective down-regulation of glucose-induced fructose 2,6-P2 in the absence of down-regulation of PFK-2/FBPase-2 protein, using a KD PFK-2/FBPase-2 variant, resulted in sustained glycolysis and elevated glucose-induced insulin secretion, indicating an over-riding role of PFK-2/FBPase-2 protein, as distinct from its product fructose 2,6-P2, in potentiating glucose-induced insulin secretion. Whereas down-regulation of PFK-2/FBPase-2 decreased glucokinase activity, overexpression of PFK-2/FBPase-2 only affected glucokinase distribution. It is concluded that PFK-2/FBPase-2 protein rather than its product fructose 2,6-P2 is the over-riding determinant of glucose-induced insulin secretion through regulation of glucokinase activity or subcellular targeting.

  • β-cell
  • fructose 2,6-bisphosphate
  • glucokinase
  • insulin
  • islets
  • 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase (PFK-2/FBPase-2)


PFK-2/FBPase-2 (6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase) is a bifunctional enzyme that catalyses the synthesis and degradation of fructose 2,6-P2 (fructose 2,6-bisphosphate), a potent allosteric activator of the glycolytic enzyme, PFK-1 (phosphofructokinase-1) [1,2]. The N-terminal half acts as a kinase to catalyse the ATP-dependent phosphorylation of fructose 6-P (fructose 6-phosphate) to fructose 2,6-P2, while the C-terminal half acts as a bisphosphatase [2]. An unequivocal role for fructose 2,6-P2 as a regulator of glycolysis is established for a number of tissues including liver and heart [1,3,4], but whether fructose 2,6-P2 has a physiological role in control of glycolysis and insulin secretion in pancreatic islets remains unclear [57].

There are four isoenzymes of PFK-2/FBPase-2, designated liver, heart, brain and testis isoforms, encoded by genes PFKFB1, PFKFB2, PFKFB3 and PFKFB4 respectively [2]. Differential splicing of these genes gives rise to isoforms that share conservation of the catalytic core but differ in the length of the N- and C-terminal regions that contain phosphorylation sites that regulate the kinase or bisphosphatase activity. The PFKFB1 gene encodes the liver and skeletal-muscle isoforms, which differ in the length of the N-terminus. The liver isoform is regulated by phosphorylation of Ser-32 by PKA (protein kinase A; also called cAMP-dependent protein kinase), which causes inhibition of the kinase and activation of the bisphosphatase, resulting in depletion of fructose 2,6-P2 and suppression of glycolysis [8]. Dephosphorylation of P-Ser-32 is catalysed by a type 2A phosphatase that is activated by xylulose 5-P. The increase in fructose 2,6-P2 caused by elevated glucose is explained by an increase in xylulose 5-P, which activates the phosphatase and dephosphorylates PFK-2/FBPase-2 [9]. The muscle isoform lacks the first 32 residues and is not regulated by cAMP [2]. The heart-type isoform encoded by PFKFB2 and the brain/inducible isoform encoded by PFKFB3 are activated (increase in kinase/bisphosphatase activity) by phosphorylation of Ser-466 (heart) or Ser-461 (brain) by AMPK (AMP-activated protein kinase), resulting in increased fructose 2,6-P2 [1012]. The heart isoform also has additional phosphorylation sites at the N-terminus (Ser-94) and C-terminus (Ser-475 and Ser-483) that can be phosphorylated by protein kinase C, calmodulin-dependent kinases and kinases activated by insulin and growth factor signalling [2]. Various splice variants have been reported for both PFKFB2 and PFKFB3 in human and rat tissue, which differ in the C-terminal domain and therefore in their regulatory properties [1316].

Two independent studies on rat pancreatic islets have reported the expression of an isoform of the PFKFB2 gene that does not contain the Ser-466 phosphorylation site [17,18]. A more recent study by Baltrusch et al. [18] also reported the presence of a novel splice variant of PFKFB2, although as a minor form. Malaisse et al. [5,6] reported that glucose caused a large increase in fructose 2,6-P2 in rat islets similar to the increment in hepatocytes, except that the effect was more rapid. Since islets do not express the liver isoform of PFK-2/FBPase-2, the glucose effect has been suggested to be due to an increase in the concentration of 6-phosphogluconate, which inhibits the bisphosphatase activity [17]. A later study on rat islets reported much more modest effects of glucose on fructose 2,6-P2 [7].

Work by Baltrusch et al. identified PFK-2/FBPase-2 as a glucokinase-binding protein [18] and subsequent work in β-cells focused on the possible role of PFK-2/FBPase-2 in regulating glucokinase activity [19,20]. However, the question of whether fructose 2,6-P2 has a role in pancreatic β-cells in control of glucose metabolism and insulin secretion through activation of PFK-1 or other mechanisms, as reported for other tissues [21], has not been addressed.

In the present study, we investigated the role of endogenous PFK-2/FBPase-2 in control of insulin secretion by use of siRNA (small interfering RNA) to down-regulate the endogenous isoforms, and we used KA (kinase active) and KD (kinase deficient) variants to distinguish between a role for fructose 2,6-P2 as distinct from a non-catalytic role for PFK-2/FBPase-2 protein in regulating β-cell function.


Isolation of pancreatic islets

Pancreatic islets were isolated from male C57/BL6 mice (12–16 weeks old) and male Wistar rats (body weight 180–220 g). The pancreas was injected with 1.5 mg/ml liberase (Roche) in Ringer phosphate buffer (10 mM Hepes, pH 7.4, 90 mM NaCl, 5 mM NaHCO3, 4.8 mM KCl, 0.7 mM KH2PO4, 0.6 mM MgSO4, 2.5 mM CaCl2, 0.1% BSA and 5.5 mM glucose) and shaken in the same medium for 6 min with further digestion with 0.75 mg/ml liberase for 2–6 min. After washing with enzyme-free buffer (Ringer phosphate buffer not containing liberase), islets were picked and mRNA was extracted from 50 (rat) or 100 (mouse) islets by using micro-spin columns (Qiagen). Human islets were isolated from pancreases retrieved from heart-beating deceased human donors following ethical approval and informed consent from the donors' relatives. Cold ischaemic time was <9 h. Islets were isolated at King's College Islet Isolation Facility, London, U.K. [22] and transported to Newcastle University in Islet Transport Medium [CMRL 1066-Supplemented (Cellgro, Herndon, VA, U.S.A.) containing 5% human serum albumin]. Islet viability was >90% and purity >70%. Islet RNA was extracted using a GeneElute Mammalian Total RNA extraction kit (Sigma–Aldrich) and RNasin® Plus RNase Inhibitor (Promega) to prevent RNA degradation.

Cell lines and hepatocyte culture

MIN6 (p20–27) and INS1E (p110–153) cells were cultured as described previously [23]. Hepatocytes were isolated from male Wistar rats and were suspended in MEM (minimal essential medium) containing 5% (v/v) newborn calf serum (Invitrogen) and seeded on to multiwell plates [24].

Treatment with adenoviral vectors

Adenoviral vectors for expression of KD and KA variants of PFK-2/FBPase-2 and glucokinase were described previously [21,25]. Adenoviral stocks were prepared by replication in HEK-293 cells (human embryonic kidney cells) and tested for their effects on enzyme activity (glucokinase) or fructose 2,6-P2 (PFK-2/FBPase-2 variants). They were then used at titres that altered glucokinase or fructose 2,6-P2 levels by up to 4-fold relative to endogenous levels. Cells were treated with adenoviral vectors (2–2.5-fold increments) at either 2 h (hepatocytes) or 24 h (MIN6) after plating, for 2 h (hepatocytes) or 6 h (MIN6). Cells were then cultured for 24 h to allow protein expression.

Treatment with siRNA

INS1E cells were transfected either with scrambled siRNA (Allstar negative control; Qiagen) or with siRNA targeted against PFKFB2 and PFKFB3 isoforms [rat PFKFB2 (SI01960014) and rat PFKFB3 (SI01960035); Qiagen] using Lipofectamine™ 2000 (Invitrogen). Cells were incubated for 72 h to allow for silencing.

Real-time RT–PCR

Cellular RNA was extracted using TRIzol® (Invitrogen) and treated with DNaseI (Roche, Lewes, East Sussex, U.K.). Single-strand cDNA was synthesized from 1 μg of total RNA with random hexamers and SuperScript II (Invitrogen). Real-time RT–PCR was performed in a total volume of 10 μl containing 50 ng of reverse-transcribed total RNA and 5 ng of forward and reverse primers (Table 1). The reactions were carried out in capillaries in a Light Cycler (Roche) with initial denaturation at 95 °C for 10 min followed by 40 cycles consisting of 95 °C for 15 s, 60 °C for 7 s and 72 °C for 15 s. The relative amount of PFKFB mRNA was corrected relative to the 18S RNA. The amplification of the correct PCR fragments was confirmed by DNA sequencing.

View this table:
Table 1 Primer sequences for real-time RT–PCR

Insulin secretion and glycolysis

MIN6 and INS1E cells were cultured in 24-well plates and washed with Krebs–Ringer buffer [26]. They were then pre-incubated for 30 min at 37 °C in glucose-free Krebs–Ringer buffer, followed by a 1 h incubation at the indicated glucose concentration with [3-3H]glucose (2 μCi/ml) for determination of glycolysis and insulin secretion. Insulin was determined in the medium using a Rat Insulin ELISA kit (Mercodia). For hepatocytes, glycolysis was determined as described previously [24].

Determination of metabolites and enzyme activity

For determination of fructose 2,6-P2, MIN6 and INS1E were extracted into 0.05 or 0.15 M NaOH and hepatocytes in 0.1 M NaOH and heated for 5 min at 80 °C. Fructose 2,6-P2 was determined as described in [27]. For determination of ATP, cells were extracted in 3% (w/v) HClO4 and ATP was determined fluorimetrically [28]. Glucokinase activity in MIN6 was determined in 13000 g supernatants of sonicated extracts as described in [29]. For elution in digitonin, MIN6 cells were incubated with 0.04 mg/ml digitonin in 150 mM KCl, 3 mM Hepes and 2 mM dithiothreitol for 4 min and glucokinase activity was determined on the digitonin extract [29]. PFK-2 activity was determined as described in [30].

Western blotting

Proteins were fractionated by SDS/PAGE and transferred on to nitrocellulose [26]. Membranes were then blotted for glucokinase immunoreactivity with rabbit antibody against human glucokinase residues 318–405 (Santa Cruz Biotechnology) and PFK-2/FBPase-2 immunoreactivity with an antibody to the bisphosphatase domain raised in chicken [18] and AMPK-Thr172(P) (New England Biolabs).

Statistical analysis

Results are expressed as means±S.E.M. based on 3–6 individual experiments except for Figures 4(A), 8(C) and 8(D), which represents replicates within a single experiment. Statistical analysis was by either ANOVA followed by the Bonferroni test or paired t test using the PRISM (GraphPad) analysis program.


mRNA expression of PFK-2/FBPase-2 isoforms in islets and β-cell lines

Previous studies on rat islets reported the expression of either the heart [17] or brain [18] isoforms of PFK-2/FBPase-2 lacking the C-terminal phosphorylation sites. However, the isoforms expressed in human and mouse islets or the MIN6 and INS1E glucose-responsive β-cell lines have not been reported. Using isoform-specific primers, we determined the expression of PFKFB1–3 in the cell lines and in rat (Figures 1A–1C), mouse (Figures 1D–1F) and human (Figure 1H) islets. RNAs from rat and mouse liver, heart and brain were used as positive controls for the rat and mouse primers. Rat islets expressed PFKFB2, with isoforms PFKFB1 and PFKFB3 being almost undetectable (Figures 1A–1C), while mouse islets also expressed predominantly PFKFB2, with low levels of PFKFB1 and PFKFB3 (Figures 1D–1F). Human islets expressed PFKFB2 and PFKFB3, with isoform PFKFB1 being almost undetectable (Figure 1H). Using primers to the C-terminal phosphorylation site Ser-466 of PFKFB2, we confirmed that both mouse and human islets express this C-terminal region. However, rat islets did not express the region around the Ser-466 phosphorylation site (Figures 1G and 1H), consistent with previous findings [17,18]. INS1E (rat, Figures 1A–1C) and MIN6 (mouse, Figures 1D–1F) cells expressed mainly PFKFB2 and PFKFB3. In MIN6 cells as in mouse islets, the PFKFB2 C-terminal Ser-466 phosphorylation site was expressed at high levels (compared with heart). INS1E cells expressed high levels of PFKFB2 mRNA compared with heart tissue (6–8-fold, Figure 1B) but low levels of the C-terminal Ser-466 region. Increasing passage number of INS1E (p110–153) and MIN6 (p19–35) cells had negligible effect on the PFKFB expression profile.

Figure 1 Expression of PFKFB isoforms in rat, mouse and human islets and MIN6 and INS1E cells

RNA was extracted from islets and cell lines and expression of PFKFB1, PFKFB2 and PFKFB3 mRNA was determined by real-time RT–PCR. (AC) Primers against rat PFKFB1 (A), PFKFB2 (B) and PFKFB3 (C). (DF) Primers against mouse PFKFB1 (D), PFKFB2 (E) and PFKFB3 (F). Results are expressed as PFKFB/18S RNA ratio (n=3). (G) Primers against the C-terminus of rat and mouse PFKFB2. Results are representative of two experiments. (H) Primers against human PFKFB1, PFKFB2, PFKFB2 (C-terminus) and PFKFB3. Results represent four human islet preparations.

Increase in fructose 2,6-P2 by pharmacological activation of AMPK

Tissues that express the heart or brain isoforms show elevation of fructose 2,6-P2 levels during pharmacological activation of AMPK with oligomycin (which inhibits ATP synthetase) or AICAR {5-amino-4-imidazolecarboxamide 1-β-D-ribofuranoside; which is phosphorylated to ZMP [AICAR monophosphate (5-aminoimidazole-4-carboxamide-1-β-D-furanosyl 5′-monophosphate)], an AMP analogue} as a result of phosphorylation of Ser-466 (heart isoform) or Ser-461 (brain isoform), which increases the kinase activity of PFK-2/FBPase-2 [10,11]. We therefore tested whether these pharmacological activators of AMPK affect fructose 2,6-P2 levels in MIN6 and INS1E cells. Preliminary studies testing the effects of AICAR (100–500 μM) showed a concentration-dependent increase in fructose 2,6-P2 (control, 100%; 100 μM, 114±14; 200 μM, 133±9; 500 μM, 200±19). At the highest concentration tested, AICAR (500 μM) also increased cellular ATP (untreated, 6.5±0.3; AICAR, 8.4±0.6 nmol/mg of protein, n=5, P<0.005). However, 500 μM AICAR decreased glycolysis (80±11% relative to the control). Subsequent experiments on glycolysis and insulin secretion were therefore performed at AICAR concentrations up to 200 μM.

Incubation of MIN6 with 0.5 μM oligomycin for 2–10 min or with 200 μM AICAR for 1 h increased the phosphorylation of AMPK-Thr-172, consistent with activation of the enzyme (Figures 2A and 2B). This was associated with a biphasic increase in fructose 2,6-P2 after addition of oligomycin (Figure 2A) and a sustained increase during incubation with AICAR (Figure 2B). AICAR (200 μM) increased insulin secretion at 25 mM glucose but had no effect on glycolysis (Figures 2C and 2D). The increase in fructose 2,6-P2 by AICAR in MIN6 cells was associated with an increase in kinase activity of PFK-2/FBPase-2 (Figure 2E). This is consistent with expression of the C-terminal phosphorylation sites of the heart and/or brain isoforms and contrasts with the suppression of PFK-2/FBPase-2 kinase activity by AICAR in hepatocytes (Figure 2F), which express PFKFB1. The fructose 2,6-P2 content was also increased in mouse islets by AICAR (Figure 3A) and in INS1E cells with AICAR (1.4-fold) and with oligomycin (1.2-fold at 2 min, P<0.05).

Figure 2 Effects of oligomycin and AICAR on fructose 2,6-P2 content

(A) MIN6 cells were incubated with 50 μM oligomycin for 2, 5 or 10 min at 5 mM glucose. Immunoblot for AMPK-Thr172(P) (upper panel) and fructose 2,6-P2 content (F26P2), expressed as a percentage of the control (n=5). *P<0.05, relative to no oligomycin. (BD) MIN6 cells were incubated without or with 50, 100 and 200 μM AICAR for 1 h at 5 or 25 mM glucose. (B) AMPK-Thr172(P) and fructose 2,6-P2. (C) Glycolysis; (D) insulin secretion. Results are expressed as percentage of the 5 mM glucose control. Control values are: fructose 2,6-P2, 6.9±1.26 pmol/mg of protein; glycolysis, 128±16 nmol/h per mg of protein; and insulin secretion, 1.5±0.4 μg/mg of protein, n=4, *P<0.05. **P<0.01, relative to 5 mM glucose control. #P<0.05, relative to 25 mM glucose control. (E, F) PFK-2 activity determined after 1 h incubation without or with 500 μM AICAR. (E) MIN6 cells; (F) hepatocytes, n=4, *P<0.05. **P<0.01, effect of AICAR.

Figure 3 Effects of glucose on fructose 2,6-P2

(A) Mouse and rat islets were incubated with 5 or 25 mM glucose and without or with 200 μM AICAR for 1 h; n=3 islet preparations. Fructose 2,6-P2 (F26P2) is expressed as fmol per islet. (B) MIN6 and INS1E cells were incubated with varying glucose concentrations (5–35 mM) for 1 h. Fructose 2,6-P2 is expressed as pmol/mg of protein; n=4. *P<0.05, **P<0.01, ***P<0.005 relative to 5 mM glucose.

Glucose-dependent regulation of fructose 2,6-P2 in MIN6 and INS1E cells

The fructose 2,6-P2 content was glucose-responsive in islets, MIN6 and INS1E cells. In rat and mouse islets, it was increased by 57–69% at 25 mM compared with 5 mM glucose (Figure 3A). Fructose 2,6-P2 was 2-fold higher in MIN6 than in INS1E cells at 5 mM glucose and it was increased by 50% and 3-fold respectively at 35 mM glucose (Figure 3B). In MIN6 cells, the kinase activity of PFK-2/FBPase-2 was slightly but significantly (P<0.05) higher at 25 mM compared with 5 mM glucose (5 mM, 4.0±0.3; 25 mM, 4.9±0.3, n=8), suggesting that the effect of glucose is at least in part due to covalent modification of PFK-2/FBPase-2.

Because 6-phosphogluconate is an inhibitor of the bisphosphatase activity of PFK-2/FBPase-2 and is elevated by high glucose concentration [17], we next tested the effects of epiandrosterone, an inhibitor of glucose 6-P (glucose 6-phosphate) dehydrogenase. Epiandrosterone caused a concentration-dependent suppression of fructose 2,6-P2, in MIN6 and INS1E cells (Figures 4A and 4B), consistent with a possible role for 6-phosphogluconate or downstream metabolites of the pentose phosphate pathway in increasing fructose 2,6-P2, and it also suppressed glucose-induced insulin secretion (Figure 4C). At 25 mM glucose, epiandrosterone suppressed but did not totally abolish the increase in fructose 2,6-P2 caused by AICAR (Figure 4D). Kinase activity of PFK-2/FBPase-2 measured under parallel conditions as in Figure 4(D) (control, 4.2±0.2; epiandrosterone, 3.9±0.2; AICAR 6.0±0.4; AICAR+epiandrosterone, 4.2±0.4, n=4) was decreased by the inhibitor in the presence (P<0.05) but not in the absence of AICAR. The latter is consistent with a role for 6-phosphogluconate in suppressing fructose 2,6-P2 by competitive inhibition of the bisphosphatase activity of PFK-2/FBPase-2 [17].

Figure 4 Epiandrosterone suppresses fructose 2,6-P2 in MIN6 and INS1E cells

MIN6 (A) and INS1E (B) cells were incubated for 1 h with various epiandrosterone concentrations at 5 or 25 mM glucose. Fructose 2,6-P2 (F26P2) is expressed as pmol/mg of protein; n=1 (A) or 2 (B). (C) Insulin secretion was determined in MIN6 cells incubated for 1 h without or with 50 μM epiandrosterone at 5 or 25 mM glucose and is expressed as percentage of 5 mM glucose. Control values: 1.5±0.2 μg/mg of protein; n=2. (D) Fructose 2,6-P2 was determined in MIN6 cells after 1 h incubation without or with 25 μM epiandrosterone or 500 μM AICAR at 25 mM glucose, n=4, *P<0.05, ***P<0.005, effect of AICAR; ###P<0.005, effect of epiandrosterone.

The role of endogenous PFK-2/FBPase-2 in pancreatic β-cells

To date, studies investigating the role of PFK-2/FBPase-2 in regulating glucokinase activity in β-cells have relied on PFK-2/FBPase-2 overexpression in cells expressing either endogenous or overexpressed glucokinase [19,20]. To determine the role of endogenous PFK-2/FBPase-2 in pancreatic β-cells, we used siRNA against PFKFB2 and PFKFB3 to down-regulate endogenous enzyme. Silencing was confirmed by a 50% decrease in the mRNA levels of PFKFB2 and PFKFB3 (Figure 5A) and by a 40–50% decrease in immunoreactivity to PFK-2/FBPase-2 and in fructose 2,6-P2 content (Figures 5B and 5C). Combined down-regulation of PFKFB2 and PFKFB3 suppressed insulin secretion at 25 mM glucose and also glucokinase activity. Rates of glycolysis and glucokinase immunoreactivity were not significantly changed (Figures 5D–5F).

Figure 5 Silencing of PFK-2/FBPase-2 in INS1E cells

INS1E cells were transfected with either scrambled (negative) or siRNA against PFKFB2 and PFKFB3 (FB2+FB3) and cultured for 72 h. (A) PFKFB2 and PFKFB3 mRNA. (B) PFK-2/FBPase-2 protein determined by immunoblotting and densitometry. (C) Fructose 2,6-P2 at 5 and 25 mM glucose. (D) Insulin secretion. (E) Glycolysis. (F) Glucokinase activity and protein determined in the 13000 g supernatant after sonication. Results are expressed as percentage of negative siRNA control. Control values: fructose 2,6-P2, 5.8±1.0 pmol/mg of protein; insulin secretion, 1.3±0.2 μg/mg of protein; glycolysis, 102±18 nmol/h per mg of protein; glucokinase activity, 2.7±0.2 m-units/mg of protein; n=4. Immunoblots are representative of four experiments. *P<0.05, **P<0.01, ***P<0.005, relative to 5 mM control. #P<0.05, ##P<0.01, ###P<0.005, relative to negative siRNA control.

The role of fructose 2,6-P2 in pancreatic β-cells

To distinguish between a role for PFK-2/FBPase-2 protein as distinct from fructose 2,6-P2 in regulating glucokinase activity and insulin secretion, we used adenoviral vectors expressing either KA or KD variants of PFK-2/FBPase-2 to up-regulate and down-regulate fructose 2,6-P2 respectively. We validated these variants in hepatocytes, where the role of fructose 2,6-P2 in the control of glycolysis is well established. Titrated expression of KA-PFK-2/FBPase-2 resulted in a 2-fold increase in fructose 2,6-P2 and in a small increase in glycolysis (Figures 6A and 6B), whereas expression of KD-PFK-2/FBPase-2 decreased fructose 2,6-P2 and glycolysis (Figures 6C and 6D) similarly to glucagon (Figure 6E). This confirms that suppression of glycolysis by glucagon is due to suppression of fructose 2,6-P2.

Figure 6 Alteration of fructose 2,6-P2 in hepatocytes by expression of KA-PFK-2/FBPase-2 and KD-PFK-2/FBPase-2

KA-PFK-2/FBPase-2 (KA; A, B) and KD-PFK-2/FBPase-2 (KD; C, D) were expressed in hepatocytes by incubation for 2 h with increasing titres of adenoviral vectors (1–3). After overnight culture, fructose 2,6-P2 (A, C) and glycolysis (B, D) were determined after 1 h incubation with 25 mM glucose with glucagon (100 nM) where indicated in (C, D). (E) Fructose 2,6-P2 content (A, C) compared with glycolysis (B, D). Results are expressed as a percentage of control. Control values: fructose 2,6-P2, 132±21 pmol/mg of protein; glycolysis, 98±26 nmol/h per mg of protein, n=4 (A, B) or 5 (C, D). *P<0.05, ***P<0.005, relative to control.

Titrated expression of KA-PFK-2/FBPase-2 in MIN6 cells increased fructose 2,6-P2 during incubation with 5 mM glucose and it increased glycolysis and insulin secretion (Figures 7A–7C). Cellular ATP was unchanged in cells expressing KA-PFK-2/FBPase-2 (n=2; results not shown). Titration with the KD-PFK-2/FBPase-2 lowered fructose 2,6-P2 by 50% at 15 mM glucose (Figure 7D). However, unlike in hepatocytes, glycolysis showed a small apparent increase (Figure 7E). Insulin secretion was similarly increased by the two variants despite the converse changes in fructose 2,6-P2 (Figure 7F).

Figure 7 Alteration of fructose 2,6-P2 in MIN6 cells by expression of KA-PFK-2/FBPase-2 and KD-PFK-2/FBPase-2

KA-PFK-2/FBPase-2 (AC) and KD-PFK-2/FBPase-2 (DF) were expressed in MIN6 cells by incubation with increasing titres (1–3) of adenovirus for 6 h. After overnight culture, cells were incubated for 1 h with the glucose concentration indicated (AC, 25 and 5 mM; DF, 5 and 15 mM) for determination of fructose 2,6-P2 (A, D), glycolysis (B, E) and insulin secretion (C, F). Results are expressed as a percentage of the control. Control values: 5 mM (AC): fructose 2,6-P2, 7.1±0.8 pmol/mg of protein; glycolysis, 133±22 nmol/h per mg of protein; insulin secretion, 1.8±0.4 μg/mg of protein. 15 mM (DF): fructose 2,6-P2, 17.6±3.3 pmol/mg of protein; glycolysis, 208±43 nmol/h per mg of protein; insulin secretion, 2.6±0.2 μg/mg of protein; n=5 (AC) or 6 (DF). *P<0.05, **P<0.01, ***P<0.005, compared with the control.

Activation of glucokinase by PFK-2/FBPase-2 in pancreatic β-cells

We determined whether an increase in glucokinase activity could explain the increase in glucose-induced insulin secretion by the KA and KD variants of PFK-2/FBPase-2. Glucokinase activity in the cytosol fraction (13000 g supernatant of sonicated extracts) was not increased in cells overexpressing KA-PFK-2/FBPase-2 (Figure 8A) or KD-PFK-2/FBPase-2 (results not shown). However, using a digitonin permeabilization assay that elutes the free or unbound enzyme, there was a higher glucokinase activity eluted from cells overexpressing KA-PFK-2/FBPase-2 (Figure 8B). Since expression of PFK-2/FBPase-2 increases glucokinase activity in cytosol extracts of MIN6 cells expressing glucokinase–GFP (green fluorescent protein) chimaeras but not in untransfected MIN6 cells [26], we determined the effects of separate and combined overexpression of glucokinase and KA-PFK-2/FBPase-2 in MIN6 cells (Figure 8C). Whereas overexpression of either enzyme alone had little effect on glucose-induced insulin secretion, combined overexpression of both proteins stimulated insulin secretion, similar to the effect of a glucokinase activator (Figure 8D).

Figure 8 Effect of expression of PFK-2/FBPase-2 on glucokinase activity

(A, B) KA-PFK-2/FBPase-2 was expressed in MIN6 cells as in Figure 7. After overnight culture, cells were extracted either by sonication and centrifugation at 13000 g for 10 min (A) or by permeabilization with 0.04 mg/ml digitonin for 4 min (B) and glucokinase activity was determined; n=5 (A) or 4 (B). (C, D) KA-PFK-2/FBPase-2 and rat liver glucokinase were expressed in MIN6 cells by incubation with adenoviral vectors for 6 h. After overnight culture, glucokinase activity was determined in the sonicated supernatant (C) and insulin secretion (D) was determined at 5 or 25 mM glucose in the absence or presence of 10 μM GKA (glucokinase activator) [27]. n=1 based on three replicates within an individual experiment representative of two experiments. Results are expressed as a percentage of the control. Control values: (A) glucokinase activity=1.49±0.3 m-units/mg of protein; (C) glucokinase activity=1.02 m-units/mg of protein; (D) insulin secretion=0.51±0.06 μg/mg of protein; **P<0.01, relative to the control.


Three key findings emerged from the present study. First, the cell content of fructose 2,6-P2 in β-cells is regulated by glucose and by activation of AMPK. The latter was associated with activation of the kinase activity of PFK-2/FBPase-2 and is consistent with the expression of PFKFB2 and PFKFB3 isoforms. Secondly, in pancreatic β-cells, endogenous PFK-2/FBPase-2 has a critical role in regulation of cell function as shown by the decrease in total glucokinase activity and glucose-induced insulin secretion during partial down-regulation of endogenous PFK-2/FBPase-2 isoforms by siRNA. Thirdly, by using KD and KA variants of PFK-2/FBPase-2, we demonstrate an over-riding role of PFK-2/FBPase-2 protein as distinct from its catalytic activity in regulating glucokinase activity. This does not rule out an additional role for fructose 2,6-P2 in the control of metabolic oscillations and/or pulsatile insulin secretion.

Several allosteric effectors are involved in the regulation of PFK-1 and thereby glycolysis. They include both inhibitors (ATP, citrate and phosphoenolpyruvate) and activators (ADP, AMP, fructose 1,6-P2, glucose 1,6-P2, 6-phosphogluconate and fructose 2,6-P2). Of these, fructose 2,6-P2 is the most potent effector with an activation constant in the nanomolar or micromolar range depending on the concentrations of other effectors [1]. The relative sensitivity of PFK-1 to these effectors is isoform-dependent, with the M (muscle)-isoform, having a relatively higher affinity for fructose 1,6-P2 and lower affinity for fructose 2,6-P2 compared with the L (liver)- and C (platelet)-isoforms [31,32]. Rat pancreatic islets express predominantly the C isoform with lower levels of M- and L-isoforms [33], and both deficiency in M-isoform and overexpression of the L-isoform are associated with impaired insulin secretion, suggesting a critical role for the different isoforms in islets [34,35].

The regulation of the cell content of fructose 2,6-P2 is dependent on the PFKFB isoforms expressed. In liver, which expresses PFKFB1, the fructose 2,6-P2 content is regulated by cAMP [8] and by glucose through a xylulose 5-phosphate-dependent type 2A phosphatase [9], whereas in heart and monocytes, which express PFKFB2 and PFKFB3, it is regulated by AMPK activity [10,11]. In the latter tissues, the stimulation of glycolysis by anoxia could be attributed to activation of PFK-1 either by AMP or by the elevated fructose 2,6-P2. The specific contribution of the latter mechanism has not been determined [11]. We show in the present study that in mouse islets and pancreatic β-cell lines, the fructose 2,6-P2 content is increased by glucose and by pharmacological activation of AMPK. The latter was associated with activation of the kinase activity of PFK-2/FBPase-2 and is consistent with expression in mouse islets and β-cell lines of PFKFB2 and PFKFB3. The lack of concomitant stimulation of glycolysis during incubation with AICAR, despite activation of AMPK, could be due to either inhibition of PFK-1 by ZMP as shown for the FTO2B hepatoma [36] or to inhibition of glucose phosphorylation [37]. In hepatocytes, AICAR inhibits glucose phosphorylation by an AMPK-independent mechanism involving inhibition of glucokinase translocation from the nucleus to the cytoplasm [38] and/or inhibition of glucokinase activity by ZMP [37]. In pancreatic β-cells, glucokinase does not shuttle between the nucleus and cytoplasm as in hepatocytes. However, inhibition of glucokinase by ZMP [37] cannot be firmly excluded.

In pancreatic β-cells, both stimulation and inhibition of insulin secretion during activation of AMPK with AICAR has been reported [3942]. In the present study, AICAR stimulated insulin secretion and increased cellular ATP, despite the lack of stimulation of glycolysis. It is possible that the variable effects of AICAR on insulin secretion in different studies may reflect the net changes in ATP/ADP. Differences in cell culture conditions such as the folate content of the medium, which affects the metabolism of ZMP, may account for variability in cellular ATP with AICAR treatment [36].

We found that rat islets express PFKFB2 lacking the C-terminal AMPK motif, consistent with previous findings of a truncated heart isoform [17,18]. The apparent species difference between rat islets and murine and human islets is surprising. We cannot rule out age-dependent differences in isoform expression as shown for PFK-1 [43]. This might in part explain the wide variation in the rat islet content of fructose 2,6-P2 reported previously [57]. Although the truncated PFK-2FB2 lacks the terminal AMPK motif (Ser-466), it has two putative AMPK motifs (Ser-58 and Ser-86) with a hydrophobic residue at P-5 and a basic residue at P-3/-4, consistent with these sites being potential substrates for both AMPK and PKA [44], which causes activation of the heart isoform [45].

The present finding that the fructose 2,6-P2 content of β-cells is elevated during activation of AMPK, raises the question of the possible role of this mechanism in β-cells. A tentative hypothesis is that the increase in fructose 2,6-P2 may have a role in pulsatile insulin secretion, which is attributed to oscillations in glycolysis and in the ATP/ADP ratio [4648]. Pulsatile insulin secretion occurs both in vivo and in perfused islets and clonal β-cells in vitro and is attributed to oscillations in the ATP/ADP ratio and in metabolites of glycolysis. Metabolic oscillations are not unique to β-cells but also occur in other cell types and in muscle and heart extracts [49]. The oscillations in cell extracts involve converse fluctuations in hexose monophosphates (glucose 6-P and fructose 6-P) and fructose 1,6-P2, indicating repetitive activation and inactivation of PFK-1. The inactivation phase is preceded by the rise in ATP/ADP, whereas the activation of PFK-1 is preceded by the decline in ATP/ADP ratio. The latter has been conventionally explained in skeletal muscle by the rise in fructose 1,6-P2 which is a high-affinity activator of the muscle PFK-1 isoform [49]. It can be hypothesized that in tissues expressing the PFKFB2 and PFKFB3 isoforms such as islets (which also express the C and L PFK-1 isoforms), the metabolic oscillations may involve activation of AMPK during the decline in ATP/ADP, resulting in an increase in fructose 2,6-P2 and activation of the C-/L-isoforms of PFK-1. Loss of pulsatile insulin secretion in both human Type 2 diabetes and animal models is well documented [46], and may be due to age-dependent changes in the expression of either PFK-1 [34,35,43] or PFK-2/FBPase-2 isoforms.

The present study provides evidence for control of fructose 2,6-P2 by glucose (25 mM versus 5 mM) in pancreatic islets and in MIN6 and INS1E cell lines. Glucose may impact the fructose 2,6-P2 content by covalent modification of PFK-2/FBPase-2 or through changes in concentrations of metabolites that act as activators or inhibitors of either the kinase or bisphosphatase activity. Elevated glucose may cause a decrease in AMPK activity [40,41], which would be expected to decrease the kinase activity of PFK-2/FBPase-2 and thereby the cell content of fructose 2,6-P2. However, in the present study, elevated glucose caused an increase in kinase activity of PFK-2/FBPase-2. This suggests that glucose causes covalent modification of PFK-2/FBPase-2, by a mechanism that over-rides any effect resulting from glucose-induced suppression of AMPK [40,41]. The increase in fructose 2,6-P2 caused by glucose can be explained by both covalent modification of PFK-2/FBPase-2, as shown by the increase in kinase activity and by an increased concentration of 6-phosphogluconate (a competitive inhibitor of the bisphosphatase activity) as a result of increased flux through the pentose phosphate pathway [17]. The marked suppression of fructose 2,6-P2 with epiandrosterone, in the absence of a corresponding decrease in kinase activity of PFK-2/FBPase-2, is consistent with a role for 6-phosphogluconate as an inhibitor of the bisphosphatase.

The higher fructose 2,6-P2 content in the cell lines compared with islets (10–25 pmol/mg versus 3 pmol/mg) may be related to the elevated flux through the pentose phosphate pathway in proliferating cells. Since 6-phosphogluconate is an activator of PFK-1 [31,32] as well as an inhibitor of the bisphosphatase [17], the apparent refractoriness of glycolysis to fluctuations in fructose 2,6-P2 with the KD variant in MIN6 cells could be due to the elevated concentration of 6-phosphogluconate, which may substitute for fructose 2,6-P2 in the regulation of PFK-1 or to altered expression of PFK-1 isoforms [33].

By using siRNA to down-regulate endogenous PFKFB2 and PFKFB3, the present study demonstrates for the first time the critical role of PFK-2/FBPase-2 in β-cell function. Down-regulation of PFK-2/FBPase-2 protein by 50% resulted in a decrease in fructose 2,6-P2, in total glucokinase activity and glucose-induced insulin secretion. The lack of suppression of glucokinase activity or insulin secretion during down-regulation of fructose 2,6-P2 with the KD variant rules out a role for fructose 2,6-P2 in explaining the effects of down-regulation of PFK-2/FBPase-2. Two mechanisms can be considered for the role of PFK-2/FBPase-2 in insulin secretion: either it determines subcellular targeting of glucokinase or it enhances (or stabilizes) glucokinase catalytic activity. The increase in glucokinase activity in the digitonin-elutable fraction (although not in the cytosol fraction) suggests that overexpressed PFK-2/FBPase-2 may affect the subcellular distribution of glucokinase. In pancreatic β-cells, glucokinase associates with insulin granules and microtubules [23,50]. The co-localization of glucokinase and PFK-2/FBPase-2 with insulin granules as shown by both subcellular fractionation and fluorescence imaging [20,23] may play a role in determining the subcellular distribution of glucokinase. The lack of increase in total glucokinase activity when KA-PFK-2/FBPase-2 was overexpressed can be explained by a saturating role for the endogenous PFK-2/FBPase-2 protein in stabilizing glucokinase. Previous work showed that overexpression of PFK-2/FBPase-2 increases glucokinase activity in MIN6 cells expressing GFP–glucokinase chimaeras but not in non-transfected cells expressing only endogenous glucokinase [26]. This was explained by a role for endogenous PFK-2/FBPase-2 in stabilizing endogenous glucokinase as in hepatocytes [24]. The present study demonstrates that overexpression of either glucokinase or PFK-2/FBPase-2 alone has a negligible effect on insulin secretion. However, combined overexpression of both proteins stimulates insulin secretion (similar to a glucokinase activator) and also enhances glucokinase activity relative to overexpression of glucokinase alone. This supports a model for co-ordinate control of insulin secretion by glucokinase and PFK-2/FBPase-2, consistent with the findings from the siRNA experiments, where suppression of PFK-2/FBPase-2 led to a decrease in total glucokinase activity and in glucose-induced insulin secretion.


This work was supported by project and equipment grants from Diabetes UK (RDO1/0002364 and RD05/0003049) and MRC (G100348 and G0501543) to L. A. and C. A. We thank Professor S. Amiel (Department of Diabetes, Endocrinology and Internal Medicine, King's College London School of Medicine, London, U.K.) for human islets provided as part of a collaborative research project (J. A. M. S., A. A., G. C. H. and D. M.) funded by the Diabetes Foundation including salary support for A. A. We thank Dr Claus Wollheim (Department of Cell Physiology and Metabolism, Univeristy Medical Center, Geneva, Switzerland) and Dr Jun-Ichi Miyazaki (Division of Stem Cell Regulation Research, Osaka University Graduate School of Medicine, Suita, Osaka, Japan) for the gift of the cell lines.

Abbreviations: AICAR, 5-amino-4-imidazolecarboxamide 1-β-D-ribofuranoside; AMPK, AMP-activated protein kinase; fructose 6-P, fructose 6-phosphate; fructose, 2,6-P2, fructose 2,6-bisphosphate; GFP, green fluorescent protein; glucose, 6-P, glucose 6-phosphate; KA, kinase active; KD, kinase deficient; PFK-1, phosphofructokinase-1; PFK-2/FBPase-2, 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase; PKA, protein kinase A; RT–PCR, reverse transcription–PCR; siRNA, small interfering RNA; ZMP, AICAR monophosphate (5-aminoimidazole-4-carboxamide-1-β-D-furanosyl 5′-monophosphate)


View Abstract