Liver Microsomal Glucose-6-phosphatase Is Competitively Inhibited by the Lipid Products of Phosphatidylinositol 3-Kinase*

We have studied the effect of various phospholipids on the activity of glucose-6-phosphatase (Glc6Pase) in untreated and detergent-treated rat liver microsomes. Glc6Pase is inhibited in the presence of phosphoinositides in a dose-dependent manner within a range of concentration 0.5–10 μm. The order of efficiency in untreated microsomes is: phosphatidylinositol (PI) 3,4,5P3 > PI3,4P2 = PI4,5P2 > PI3P = PI4P > PI. In contrast, Glc6Pase is not inhibited in the presence of phosphatidylserine, phosphatidylcholine, and phosphatidylethanolamine, diacylglycerol, and inositol 1,4,5-trisphosphate at concentrations up to 100 μm. The mechanism of Glc6Pase inhibition by PI4,5P2, PI3,4P2, and PI3,4,5P3 is competitive in both untreated and detergent-treated microsomes. In untreated microsomes, the K i for PI3,4,5P3 (1.7 ± 0.3 μm, mean ± S.D. n = 3) is significantly lower (p< 0.01) than that for PI3,4P2 (5.0 ± 0.8 μm) and for PI4,5P2 (4.7 ± 0.7 μm). In detergent-treated microsomes, Glc6Pase is less sensitive to the inhibition and there is no difference anymore among the K i values for the three compounds: 8.3 ± 0.8, 11.1 ± 0.5 and 8.9 ± 0.4 μm for PI3,4,5P3, PI3,4P2, and PI4,5P2, respectively. This inhibition phenomenon might be of special importance with regards to the insulin’s inhibition of hepatic glucose production.

Glucose-6-phosphatase (Glc6Pase) 1 is a crucial enzyme of glucose homeostasis since it catalyzes the ultimate biochemical reaction of both glycogenolysis and gluconeogenesis, e.g. the hydrolysis of glucose-6 phosphate (Glc6P) in glucose and P i (1). It allows the gluconeogenic tissues in which it is specifically expressed to release glucose in blood. Numerous data have been provided during the last years, strongly suggesting that Glc6Pase might play an important regulatory role in the glucose release by the liver and the kidney, through mechanisms involving either gene expression (2)(3)(4), and/or biochemical inhibitions of its enzymatic activity (Refs. 5-7, recently reviewed in Ref. 8).
In contrast with most of the other regulatory enzymes of liver glucose metabolism, Glc6Pase is a very hydrophobic protein localized within the reticulum endoplasmic membrane (1,8). This particular location within the cell is currently not explained. The observation that the kinetic properties of the enzyme are crucially dependent on the presence or the absence of the membrane has pointed out the importance of membrane lipids in the regulation of Glc6Pase activity in situ (8). In our previous works, we characterized the inhibition effect of unsaturated fatty acids (9) and of fatty acyl-CoA esters (10) on Glc6Pase activity. In the present work, we have addressed the question of the possible modulation of Glc6Pase activity by phospholipids themselves. Numerous previous studies have been devoted to the role of phospholipids in the stability and/or the activity of Glc6Pase by means of the effects induced by various phospholipases (11)(12)(13)(14)(15) or by phospholipid transfer proteins (15,16). None has specifically addressed the question of the effect of the phosphorylation of the inositol ring of phosphatidylinositol.Wehavepaidspecialattentionto3-phosphoinositides, the lipid products of phosphatidylinositol 3-kinase (PI3K) activity, since there is growing evidence to suggest that this enzyme is essential for insulin's regulation of glucose metabolism (see Ref. 17 as a recent review).
protein/ml, Lowry assay with bovine serum albumin as a standard) in all experiments.

RESULTS
Glucose-6-phosphatase, assayed in untreated microsomes, was not affected in the presence of PE or PS within the range of concentration of 0.5 to 10 M. In contrast, it was slightly inhibited in the presence of PI and PI3P, and strongly inhibited in the presence of PI3,4P 2 and PI3,4,5P 3 within the same concentration range (Fig. 1). The inhibition was dependent on the phosphorylated state of the inositol ring: the higher the phosphate number, the more pronounced the inhibitory effect. In some instances, higher lipid concentrations (20 and 50 M) were studied. The maximal inhibitory effect (75-80%) was reached from 5-10 M for PI3,4,5P 3 ( Fig. 1) and from 20 M for PI3,4P 2 (not shown). The highest inhibitory effects were obtained at 50 M for PI3P and PI (about 60 and 40% inhibition, respectively). The relative efficiencies of phosphoinositides were compared at a concentration producing half-maximal inhibition of Glc6Pase activity (for PI3,4,5P 3 ), e.g. 2 M. PI3,4,5P 3 was about 2.5 times more efficient than PI3,4P 2 and 10 -20 times more efficient than PI or PI3P. PI4P and PI4,5P 2 had similar inhibitory efficiencies as their respective isomers PI3P and PI3,4P 2 (not shown). The inhibitory effect was the same whether microsomes had been preincubated in the presence of the inhibitors for 15 min at 25°C or not. The dose-dependent inhibition effects were the same whether the inhibitors were added in the absence (Fig. 1) or in the presence of 20 M PS (final concentration in the assay). Since the nature of the fatty acid esterified in positions 1 and 2 of glycerol might influence the inhibition effect (3-phosphoinositides (dipalmitoyl-) from Matreya were synthetically obtained), D 3 -phosphoinositides were also synthesized in vitro from their natural precursors from Sigma (purified from bovine brain and likely enriched in arachidonic acid in position 2 of glycerol). However, both types of 3-phosphoinositides had very similar effects and the results presented in Fig. 1 constitute a mixing of both sets of results. Other compounds were tested up to concentrations of 100 M: neither PE, PS, PC, nor DAG or IP3 (e.g. the products of the hydrolysis of PI4,5P 2 by phospholipase C) inhibited Glc6Pase (not shown). In contrast, PA slightly inhibited Glc6Pase to the same extent as P i (not shown). After detergent treatment of microsomes, Glc6Pase was less sensitive to the inhibition by PI3,4,5P 3 . The enzyme was inhibited by 50% at 10 M (not shown). In addition, there was no clear difference with the inhibition effect induced by PI3,4P 2 and PI4,5P 2 (see below).
The kinetics of Glc6Pase in untreated and detergent-treated microsomes (velocity versus the substrate concentration) were hyperbolic in the absence and in the presence of the inhibitors (studied at concentration of 2, 5, and 10 M in untreated microsomes and 5, 10, and 20 M in untreated microsomes). The double reciprocal plots (Lineweaver-Burk plots) fitted straight lines in all cases and converged on the same point located on the ordinate axis (data not shown). This revealed a classical pure competitive inhibition mechanism. The respective K i values for PI3,4,5P 3 , PI3,4P 2 , and PI4,5P 2 were determined from Dixon plots (reciprocal of the velocity versus the inhibitor concentration). The K i of PI3,4,5P 3 was three times lower than the respective Kis of PI3,4P 2 and PI4,5P 2 for Glc6Pase in untreated microsomes, and five times lower than its corresponding value in detergent-treated microsomes (Table I). DISCUSSION In this paper, we report that Glc6Pase is specifically inhibited in vitro in the presence of minute amounts of a particular class of phospholipids, e.g. phosphoinositides. In contrast, it is unaffected upon addition of the other major phospholipids of the membrane, even at high concentrations. Noteworthy, the most potent effector is PI3,4,5P 3 , the major product of PI3K activity (20,21), which is three times more efficient than its lipid precursor (PI4,5P 2 ) to inhibit Glc6Pase in untreated microsomes. Although microsomes constitute artificially revesiculated membranes, in which some specific properties of Glc6Pase could have been lost, as compared with Glc6Pase assayed in homogenates from livers freeze-clamped in situ (22) or in filipin-permeabilized isolated hepatocytes (23,24), one may suppose that they represent an acceptable environment to characterize in vitro the inhibition mechanisms of Glc6Pase under a conformational state close to that in vivo. That the enzyme is less efficiently inhibited by PI3,4,5P 3 after detergent solubilization of the reticulum membrane, and has lost the difference in sensitivity between the inhibition by PI3,4,5P 3 and by PI3,4P 2 or PI4,5P 2 , may agree with this assumption.
Numerous hypotheses have been put forward in the past to account for the structural organization of Glc6Pase within the microsomal membrane, from the simplest (monomeric protein) to the most complex (system involving several distinct proteins). From recent results from us and others (reviewed in Ref. 8), we have suggested that Glc6Pase could be a unique transmembrane protein (which however might be either a homo-or a heteromultimer). The velocity and substrate specificity of the enzyme should be dramatically dependent on at least three

TABLE I Inhibition parameters of Glc6Pase by phosphoinositides
The K i values were obtained from the plots of the reciprocal of the velocity versus the concentration of the inhibitor (Dixon plots). The data obtained at each substrate concentration were linear and converged on the same intersection point located above the abscissa axis (in agreement with a competitive inhibition phenomenon). K i was given by the abscissa of this point. The results were obtained from three different preparation of microsomes and are expressed as means Ϯ S.D.
b Significantly different from corresponding value in untreated microsomes, p Ͻ 0.01.
Competitive Inhibition of Glucose-6-phosphatase by 3-Phosphoinositides different types of structural conformations, which could be related to the surrounding lipidic environment (8). Whether microsomes had been treated by detergent or not, the mechanism of inhibition of Glc6Pase by PI3,4,5P 3 , PI3,4P 2 , and PI4,5P 2 is purely competitive. This strongly suggests that the substrate site is involved as a site of interaction of the inhibitors. This is not unexpected given the obvious structural analogy between the phosphorylated inositol ring and Glc6P. However, it is important to note that the lipidic part of the molecule is required for this interaction to take place, since IP 3 has no inhibitory effect alone, even at concentrations by 10 times higher. This suggests that the association of the lipidic part of the inhibitors to some hydrophobic regions of the enzyme (either on the catalytic unit or on putative associated polypeptide(s) might stabilize the interaction between the phosphorylated ring and the catalytic site. This should strongly favor the formation of the enzyme-inhibitor complex (K i values are in the micromolar range) with regards to the enzyme-substrate complex (K m is in the millimolar range). That the K i values for the enzyme in detergent-treated microsomes are slightly higher than the corresponding ones in untreated microsomes suggests that the enzyme-inhibitor interaction is weakened upon the conformational change induced by the removal of surrounding lipids (8). Noteworthy, it has been reported that the treatment of rat liver microsomes by a phosphatidylinositol-specific phospholipase C induced a decrease of the Glc6Pase K m (14). This is in agreement with the elimination of a competitive inhibitor of the enzyme.
Numerous recent studies strongly suggest that PI3K is an essential enzyme in the signal transduction of insulin to glucose metabolism in various cell types (17). For example, there is a good correlation between the insulin's stimulation of PI3K activity and of glucose transport in isolated rat adipocytes (25) and between the inhibition of PI3K by specific inhibitors and the blockade of insulin stimulatory effects on glucose transporter translocation and glucose transport in 3T3-L1 (26) and isolated (27) adipocytes and in cultured myocytes and rat muscle (28). Also, insulin's stimulation of PI3K activity is strongly reduced in the insulin resistant liver of the diabetic ob/ob mouse (29). The availability of synthetic analogues of 3-phosphoinositides has allowed to identify recently a variety of binding protein (see Ref. 30 as a review). Among them, different isotypes of protein kinase C have been shown to be activated in the presence of PI3,4,5P 3 (31)(32)(33). However, the present work is the first report of the inhibition by PI3,4,5P 3 of a key enzyme in glucose metabolism. There has been compelling indirect evidence provided that the activity of this enzyme is negatively regulated by insulin (6,8,34). With regards to the intracellular location of Glc6Pase (endoplasmic reticulum), it has been strongly suggested that the specificity of insulin to stimulate glucose transport via PI3K activation in adipocytes is dependent on the subcellular targeting of PI3K to low density microsomes, from which glucose transporters translocate (35)(36)(37). This might obviate the physiological concern that total PI3,4,5P 3 is estimated to be by 10 times less abundant than its lipid precursor PI4,5P 2 in most stimulated cells (38), while having only a three times higher potency to inhibit Glc6Pase in vitro. If such a microsome PI3K-targeting process takes place in the liver upon insulin stimulation, the ratio of PI3,4,5P 3 to PI4,5P 2 could be substantially higher in the close vicinity of Glc6Pase than elsewhere in the cell. It is also possible that the difference in sensitivity between the respective effects induced by the two compounds be higher in situ than in isolated microsomes (see above). Another important difference between the in vivo and in vitro situation is the way of access of the inhibitor to the enzyme: outside the membrane (in vitro) versus inside the membrane (in vivo). The competitive inhibition mechanism of microsomal Glc6Pase by the main lipid product of PI3K reported here might therefore be of special biological meaning.