Liver glucose-6 phosphatase (EC 3.1.3.9) (Glc6Pase)⁶ is a key enzyme in glucose production, catalyzing the terminal step of both gluconeogenic and glycogenolytic pathways. Until recently, it was commonly thought that only the level of glucose-6 phosphate could modulate the flux of substrate through Glc6Pase. However, Newgard et al. (1984) suggested the existence of a short-term regulation mechanism of the enzyme after glucose ingestion. It was also proposed that Glc6Pase could be regulated by insulin (Gardner et al. 1993) or by various nutrient metabolites such as proline (Bode et al. 1992), -ketoglutarate (Minassian et al. 1994), unsaturated fatty acids (UFA) (Mithieux et al. 1993) and long-chain fatty acyl-CoA esters (Mithieux and Zitoun 1996; see Mithieux 1997 for a recent review). Furthermore, recent data from our laboratory demonstrated that the enzyme activity is inhibited during the postprandial period (Minassian et al. 1995). In the latter report, it was suggested that insulin alone could not account for the inhibition and that a metabolite present in liver homogenate could be the inhibitor. Previously, Grant and Burchell (1989) reported the inhibition of Glc6Pase by glycogen in vitro. More recently, Liu et al. (1993) have reported that the inhibitory effect is not due to glycogen per se but could be dependent on a low molecular weight compound (<5000 Da) removable from crude glycogen by passage through an anion exchange column. The purpose of this study was to identify this metabolite.

MATERIALS AND METHODS

Glycogen was purified from liver by alkaline solubilization followed by repeated ethanol precipitation and then dried (Liu et al. 1993). It was quantified according to Keppler and Decker (1974). Glycogen and glycogen-associated fractions were prepared from crude glycogen samples as described by Liu et al. (1993) with slight modifications. Glycogen in suspension in water (100 mmol/L equivalent glucosyl) was passed through a Dowex 1 anion exchange column (200-400 mesh, Sigma, La Verpillière, France). The column was first eluted with water (neutral eluate, which contained the bulk glycogen) and then with 30 mmol/L HCl and 1 mol/L HCl (acidic eluates). Acidic fractions were evaporated to dryness to remove HCl and resuspended in an equivalent volume of 20 mmol/L Tris-HCl, pH 7.3 by sonication. Glycogen reconstituted with UFA was prepared by mixing chromatographed glycogen with increasing amounts of a mixture of pure UFA (Sigma) with the following composition: 30% oleic acid [18:1(n-9)], 25% arachidonic acid [20:4(n-6)] and 45% docosahexanoic acid [22:6(n-3)] (mass ratio). The latter was added to the neutral eluate just after chromatography and the mixture was subjected to three ultrasonic pulses.

RESULTS

Thin-layer chromatography analysis of total lipids extracted from glycogen demonstrated that no complex lipids were present. On the other hand, glycogen contained substantial amounts of free fatty acids (FFA) (not shown). Gas chromatography analysis demonstrated that glycogen contained several long-chain saturated and unsaturated fatty acids (UFA) in substantial amounts (Fig. 2). Most of them, including arachidonic acid [20:4 (n-6)] and other UFA, are efficient inhibitors of Glc6Pase activity (Mithieux et al. 1993). On the other hand, glycogen contained only low amounts of FFA after chromatography (Fig. 2). This indicated that UFA could be the Glc6Pase inhibitor associated with glycogen.

Glycogen contained 4.9 ± 0.7 nmol total FFA/µmol glucose residue (mean ± SEM, n = 4) before filtration (this represented 0.8 ± 0.07 g/100 g); 1.3 ± 0.7 nmol were still present in the neutral filtrate, and 1.2 ± 0.6 and 1.1 ± 0.2 nmol were retained by the column and eluted by 30 mmol/L and 1 mol/L HCl, respectively (Fig. 3). The proportion of UFA was ~0.43 in the nonchromatographed glycogen fraction and 0.85 in the two eluates. The greater proportion of saturated fatty acids was efficiently bound to the column but could not be eluted in acidic conditions. We calculated that the final concentration of UFA in the Glc6Pase assay medium in the presence of acidic eluates (conditions of Fig. 1), was ~20 µmol/L, e.g., a concentration that is expected to lower Glc6Pase activity by 20-30% (Mithieux et al. 1993).

Reconstituted glycogen (30 mmol/L equivalent glucosyl residue) inhibited Glc6Pase in a manner dependent on the amount of UFA added (Fig. 4). Before chromatography, glycogen contained ~2.1 nmol UFA/µmol glucosyl residue, and ~1 nmol UFA was still present in chromatographed glycogen (see Fig. 3). Chromatographed glycogen reconstituted with 1.2 nmol UFA/µmol glucosyl, which should thus contain UFA amounts very close to those found in glycogen before chromatography, inhibited Glc6Pase to a similar extent (compare Fig. 1 and 4).

DISCUSSION

It should be pointed out that chromatographed glycogen still contains UFA, but has no Glc6Pase inhibitory activity. This strongly suggests that this particular fraction of UFA must be strongly bound to glycogen because it is not exchangeable upon ion-chromatography. It is therefore not unexpected that it might not inhibit Glc6Pase because we have previously reported that UFA bound to bovine serum albumin do not inhibit Glc6Pase (Mithieux et al. 1993). On the other hand, the fraction of UFA that is exchangeable by chromatography likely is more loosely bound to glycogen, and can be released to inhibit Glc6Pase and participate to the regulation of the enzyme activity in vivo.

Taking into account the subcompartmentalization of the cytosolic glucose-6 phosphate pool demonstrated by Christ and Yungermann (1987) it appears that two nonmiscible glucose-6 phosphate pools can be distinguished; one is the result of glycogenolysis, whereas the other is due to gluconeogenesis (GNG). As a consequence, the glycogen particles should be in close contact with Glc6Pase in the smooth endoplasmic reticulum membrane, in such a way that neither glucose-6 phosphate formed by glycogenolysis could escape from the "glycogen subcompartment," nor could glucose-6 phosphate supplied by GNG enter in (Christ and Yungermann 1987). The presence of UFA within the glycogen granule could therefore play an important role in the regulation of Glc6Pase in vivo, especially at the level of those molecules involved in glycogenolysis. The latter proposal fits very well with the emerging concept of autoregulation of hepatic glucose production (HGP) and with an important role for fatty acids in this autoregulation. It has been established that GNG could be either decreased (Puhakainen et al. 1991) or increased (Jenssen et al. 1990) with no change in overall hepatic glucose output, strongly suggesting that glycogenolysis was concomitantly increased (or decreased, respectively) to compensate for the lower (or higher) supply of glucose by GNG. This demonstrated that reciprocal autoregulatory mechanisms exist between both pathways, tending to maintain HGP constant in spite of the possible variations in one or the other flux (Jenssen et al. 1990, Puhakainen et al. 1991). Free fatty acids are a crucial regulator of GNG because they stimulate it in a concentration-dependent manner through their intrahepatic oxidation (Williamson et al. 1969). Very interestingly, the same phenomenon of upholding HGP was evidenced when GNG was either decreased because of the fall of circulating fatty acids induced by the lipolytic inhibitor acipimox (Puhakainen and Yki-Järvinen 1993), or increased through the elevation of plasma FFA concentration by means of intralipid/heparin infusion (Clore et al. 1991). On the basis of the assumption that the intracellular pool of fatty acids co-compartmentalized with glycogen and that fated to oxidation both undergo parallel variations, UFA might be a key factor involved in autoregulation of HGP. On the one hand, hepatic glucose output might be maintained at high fatty acid levels. This could be explained because an inhibition of glycogenolysis (due to Glc6Pase inhibition by UFA) could be concomitant with a stimulation of GNG (through FFA oxidation). On the other hand, HGP should not vary greatly at low fatty acid levels. In this case, low "unstimulated" gluconeogonic fluxes could be compensated by high "unrepressed" glycogenolytic fluxes.

In conclusion, the results reported here strongly suggest that substantial amounts of FFA may be associated to the glycogen granule in vivo. Two classes of FFA could exist. The one, strongly bound to glycogen, might not inhibit Glc6Pase. The other, loosely bound, might be released to inhibit Glc6Pase as a result of the presence of UFA. The presence of FFA associated to glycogen could play a crucial role in the phenomena of regulation of Glc6Pase activity and of HGP in various situations.