A novel role for glucose 6-phosphatase in the small intestine in the control of glucose homeostasis.

Glucose 6-phosphatase (EC 3.1.3.9, Glc-6-Pase) is a crucial enzyme in the control of glucose homeostasis. It catalyzes the last biochemical reaction of gluconeogenesis and glycogenolysis, i.e. the hydrolysis of glucose 6-phosphate (Glc-6-P) into glucose and Pi. Glc-6-Pase is therefore unique in that it confers upon the tissues in which it is expressed the capacity to release glucose into the blood. It has long been recognized that the liver and the kidney possess this capacity because both tissues express high levels of Glc-6-Pase (1–4). Glc-6-Pase is a hydrophobic protein that is embedded within endoplasmic reticulum membrane (1–4). The reason for this localization is not clear.

Before Glc-6-Pase cDNA was isolated, detection of Glc-6-Pase gene expression was based on the assay of enzyme activity. Because numerous phosphatases other than Glc-6-Pase hydrolyze Glc-6-P (e.g. alkaline phosphatase), the accurate measurement of nonspecific Glc-6-P-phosphohydrolase is crucial, especially in the small intestine (SI). This tissue contains high levels of alkaline phosphatase, which generates a high, nonspecific background (Fig. 1A). Thus, conflicting reports of Glc-6-Pase in the rat SI range from nil (5) to levels comparable with those in the liver (6). In addition to difficulties in quantifying nonspecific activity in the SI, reports of Glc-6-Pase activity in the SI have been further confounded by an anteroposterior expression gradient in this tissue (see below) and by the inactivation of Glc-6-Pase during the isolation of intestinal microsomes. 2 This latter complication may be due to the presence of an endogenous Glc-6-Pase inhibitor in the SI (7).
The availability of Glc-6-Pase cDNA, isolated from mouse and human tissues (8,9), has greatly helped resolve discordant reports of Glc-6-Pase expression in tissues other than the liver and kidney (10). Using RT-PCR, Northern blotting, and the highly specific Glc-6-Pase assay described in Fig. 1, we first reported in 1997 (11) that Glc-6-Pase mRNA and protein are present in the SI in the rat and human species (Fig. 1). In contrast, Glc-6-Pase mRNA could not be detected in numerous other tissues previously identified as expressing it, including the brain and pancreas (both whole tissue and isolated islets) (10,12,13). In addition, there is a decreasing gradient of Glc-6-Pase gene expression from the duodenum to the distal jejunum in the rat (14), whereas the gene is expressed up to the ileum in humans (12). Fig. 1B, which shows the non-quantitative RT-PCR data, may be misleading. Quantitative analysis by Northern blot revealed that the amount of Glc-6-Pase mRNA is about 7-8 times less abundant in the jejunum than in the liver (12). Moreover, the specific Glc-6-Pase activity, i.e. about 1 mol/min/g in the mid-jejunum (Fig. 1A), is lower than noted in the liver or kidney. It is noteworthy that the endoplasmic reticulum membrane contains a translocase that is required for Glc-6-Pase activity (15). The glucose-6-phosphate translocase is putatively involved in the translocation of Glc-6-P from the cytosol to the hydrolytic site of Glc-6-Pase, located at the luminal side of the membrane. However, in addition to being expressed in the SI, this protein is also found in numerous nongluconeogenic tissues (Fig. 1B). It has been suggested that it may serve to feed an ubiquitously expressed intraluminal hexose-6-phosphate dehydrogenase, the role of which is to provide the reducing equivalents needed for several reductases, such as 11␤-hydroxysteroid dehydrogenase-1 or glutathione reductase, that are expressed in the lumen of the endoplasmic reticulum (see Ref. 16 for review).

Control of Glc-6-Pase Gene Expression in Small Intestine
A key question is whether the Glc-6-Pase gene in the SI is regulated by insulin as it is in the liver and kidney (10). Both Glc-6-Pase mRNA and enzymatic activity are markedly induced in the fasted or diabetic states (12,14). These expression parameters are rapidly normalized by re-feeding or insulin treatment, respectively (12). It is noteworthy that expression of the gene is present in the ileum of fasted rats though not in streptozotocin-treated diabetic rats. In addition, Glc-6-Pase gene expression is strongly enhanced from the proximal SI up to the distal ileum in nursing rat pups (18). Nursing during the neonatal period represents a particular hypoinsulinemic fed state (because of relative deficiency of mother's milk in glucidic nutrients) that is typified by enhanced gluconeogenesis (17). This suggests that non-insulin-regulated control mechanisms (e.g. nutrition) may be crucial in regulating Glc-6-Pase gene expression in the SI (12).
Glc-6-Pase gene transcription may be related to the expression of a set of intestinal genes that all share both hepatic nuclear factor 1 (HNF1) and CDX binding sites in a very proximal region of the promoter (19). CDX1 and -2 are nuclear factors related to the "caudal" gene of Drosophila, which governs the anteroposterior axis of development, and are exclusively expressed in the intestine (CDX1 and -2) and the endocrine pancreas (CDX2) in mammals (20,21). CDX1 transactivates the Glc-6-Pase promoter by forming a ternary complex with both the transcription machinery and the TATA box of the gene, whereas CDX2 antagonizes the transactivating effect of CDX1 (22). Because CDX1 expression appears to be predominant in regard to CDX2 in the proximal intestine, these factors might play a role in the anteroposterior gradient of expression of Glc-6-Pase in the SI (22). Additionally, this dual regulation might be related to the absence of Glc-6-Pase in pancreatic islets, which express the CDX2 isotype exclusively (21). Thus, the extent to and mechanism by which the nutrition-and hormone-sensitive expression of Glc-6-Pase is dependent on CDX1/CDX2 merits further study.

Glc-6-Pase Gene Expression Confers Gluconeogenic Capacity on Small Intestine
Because the SI utilizes glucose very actively, the assessment of its participation in glucose production has required the simultaneous partitioning of the release and uptake of glucose by this tissue. This has been successfully accomplished using a combination of arteriovenous glucose balance and 3-[ 3 H]glucose dilution techniques previously used in similar studies of the kidney (23). Though glucose production by the rat SI could not be demonstrated in either the fed or short term fasted state (24 h), it progressively increases after 24 h of fasting, where it accounts for 8.6 Ϯ 1.5 and 16.9 Ϯ 4.0 mol/kg/min, i.e. about 20 and 35% of total endogenous glucose production (EGP), after 48 and 72 h of fasting, respectively (14,24). It has also been estimated that glucose release from the SI may represent about one-third of EGP in streptozotocin-treated diabetic rats (14,24). These determinations must, however, be regarded with caution. A major difficulty in these studies relates to the accuracy of tracer analysis (23,24). One must also bear in mind that the glucose release surpasses the glucose uptake in the SI only in diabetic rats (24), whereas the release only compensates the uptake in 72-h fasted rats (14) and is even slightly lower than the uptake in 48-h fasted rats (24). It is noteworthy that a portion of the glucose produced by the SI in diabetic rats may originate from glycogen stores, the latter being markedly increased in this situation (24). Consistent with glycogen storage in the SI and liver of diabetic animals (14), a glucose-phosphorylating activity of Glc-6-Pase may occur at high glucose concentration, as previously suggested by Robert Nordlie (25). This hypothesis derived mainly from in vitro experiments. To our knowledge, it has not been supported by definitive in vivo experimentation. Interestingly, glucose production from the SI, like its hepatic counterpart, may be rapidly blunted in rats by insulin infusion during the euglycemic-hyperinsulinemic clamp (24). Incorporation studies of 14 C-and 13 C-labeled gluconeogenic precursors into glucose have indicated that glutamine (for about 80%) and glycerol (for the residual 20%) may account for the total glucose synthesized by the SI (24). In contrast, lactate and alanine, which are the major gluconeogenic substrates in the liver, are not incorporated into glucose in the SI. These findings have implicated glutaminase, alanine aminotransferase (ALT) (because glutamate dehydrogenase is not of quantitative importance in the SI (see Ref. 26 as a review)), glycerol kinase, and glycerol-3-phosphate dehydrogenase (G3PDH) as critical enzymes in gluconeogenesis in the SI (Fig. 2).
It is noteworthy that gene expression studies support the specific selection of gluconeogenic precursors by the SI. Pyruvate carboxylase (PC) is a key anaplerotic enzyme involved in the removal of the citric acid cycle intermediates for glyceroneogenesis and lipogenesis. In the liver, it also plays an important role in gluconeogenesis, accounting for the incorporation of both alanine and lactate after conversion to pyruvate by ALT and lactate dehydrogenase (LDH), respectively, into glucose. Accordingly, the V max of PC is doubled in the liver of rats in hypoinsulinemic states (24). Conversely, the activity of PC has been suggested to be decreased by at least 80% in the SI of rats under similar conditions, although the mechanism of this suppression would merit further characterization (24). This is in keeping with the fact that alanine and lactate are not incorporated into glucose formed in the SI (24). In contrast to PC, during hypoinsulinemia there is a strong induction of cytosolic phosphoenolpyruvate carboxykinase (PEPCK-C) in the SI and in the liver (27). In the liver, PEPCK-C uses oxaloacetate, produced either from pyruvate by PC or from substrates cycling through the Krebs cycle to generate phosphoenolpyruvate. This is also the fate of glutamine after conversion to glutamate and ␣-ketoglutarate (Fig. 2). Therefore, the coordinate inverse regulation of PC and PEPCK-C in the SI fits well with the use of glutamine as a major gluconeogenic substrate and is an explanation for the lack of utilization of alanine and lactate for FIG. 1. Expression of the Glc-6-Pase gene in the rat small intestine. A, assay of specific Glc-6-Pase activity in the liver, kidney, jejunum, and skeletal muscle. The assay is based on the specific inactivation of the enzyme upon brief exposure to acidic pH at 37°C (38). Frozen samples from fed rats were powdered and homogenized by sonication in 20 mM Hepes, 0.25 M sucrose, pH 7.3 (100 g of wet tissue/ml). One-half of each homogenate was brought to pH 5.0 by addition of 1 M HCl and incubated for 10 min at 37°C. The pH was then neutralized to 7.3 by addition of 1 M NaOH. Treated (light gray bars) and non-treated homogenates (dark gray bars) were assayed for phosphohydrolase activity toward glucose-6 phosphate under conditions of maximal velocity (20 mM Glc-6-P) (see Ref. 12 for more details). The results are expressed as mean Ϯ S.E. (n ϭ 5). *, significantly different from non-treated, p Ͻ 0.05 (Student's t test for paired data). B, specific amplification of Glc-6-Pase and glucose-6-phosphate translocase (G6PT) mRNAs by reverse transcription/polymerase chain reaction in various tissues from fed rats. M, markers; WAT, white adipose tissue. Purification of total RNA and conditions of RT-PCR were described in detail in Refs. 12 and 10. C, immunodetection of Glc-6-Pase protein in the SI mucosa from 48-h fasted rats (confocal fluorescence microscopy).
Minireview: Novel Role for Glucose 6-Phosphatase 44232 gluconeogenesis (24,27). Glycerol has a lesser role than glutamine as a glucose precursor in the SI. This is in agreement with the fact that the expression of the gene for glycerol kinase is not induced in hypoinsulinemia (24).
A recent study of gene expression in the SI in the early phases of fasting (i.e. between 0 and 48 h) has highlighted the respective roles of Glc-6-Pase and PEPCK-C in the initiation of gluconeogenesis in this tissue. It is noteworthy that the rapid induction of the PEPCK-C gene (its enzymatic activity is already maximally induced at 24 h of fasting, whereas Glc-6-Pase activity is not yet increased) is not sufficient to trigger detectable glucose production by the SI (14). In contrast, glucose production from the SI accounts for 20% of EGP at 48 h of fasting, a time at which both enzymes exhibit increased activity (14,24). This suggests that Glc-6-Pase may be more crucial than PEPCK-C for the initiation of gluconeogenesis in the SI.
It must be emphasized that gluconeogenesis from glutamine and from glycerol is unique in that, unlike gluconeogenesis from lactate and alanine, it is an exergonic process. The reason is that both gluconeogenic pathways involve reactions generating reducing equivalents, e.g. catalyzed by ␣-ketoglutarate dehydrogenase (NADH ϩ H ϩ ) and succinate dehydrogenase (FADH 2 ) for glutamine or glycerol-3-phosphate dehydrogenase (NADH ϩ H ϩ ) for glycerol. Assuming the aerobic oxidation of reducing equivalents, the net yield represents 4 ATP per glucose produced from either precursor, taking into account the ATP requirement for the urea synthesis in the case of glutamine. The increasing utilization of glutamine and glycerol by gluconeogenesis in the SI in periods of energy depletion such as fasting is thus highly favorable for the energetics of the whole body. It is also noteworthy that, although both glucose and glutamine are actively utilized by the SI, they are only partially oxidized (26,28,29). This is in line with their major metabolic fates described below. In fact, in fasted rats the SI derives at least 50% of the energy requirement from ketone bodies (26,28,30).

Key Role of Small Intestine in Distribution of Gluconeogenic Substrates within Splanchnic Bed
The key roles of the kidney and/or small intestine in glucose supply during hypoinsulinemic states or under conditions where the liver is absent, e.g. during the anhepatic phase of liver transplantation in humans, has been discussed in a recent review (31). In strong agreement with the latter concept, it has been emphasized that the total gluconeogenic flux through PEPCK-C is decreased only by one-third in mice with liverspecific knock-out of the gene (32), suggesting a major role for extrahepatic PEPCK-C in the systemic glucose supply in these animals (31). We would like to stress that the role of the SI in whole body gluconeogenesis, especially within the splanchnic bed, is likely to be more than just that of a glucose contributor. It is indeed striking that although glutamine is the most abundant plasma amino acid and is considered a key gluconeogenic precursor (29,33), the liver is unable to utilize glutamine for gluconeogenesis. This is due to the expression in hepatocytes of a liver-type glutaminase that has a very low affinity for glutamine (33). This strikingly differs from enterocytes, which express high levels of the high affinity, kidney-type glutaminase (26,33). In fact, enterocytes actively utilize both glutamine and Minireview: Novel Role for Glucose 6-Phosphatase 44233 glucose (26). This favors alanine synthesis from pyruvate via ALT (26,29). Alanine, in turn, continues to the liver via the portal vein where it is converted to glucose. It is likely that the induction of gluconeogenesis from glutamine in the SI, because of the ALT-dependent coupling between glutamate and pyruvate utilization, may synergize with the production of alanine for the liver (see Fig. 2). Moreover, the reduction of PC activity (24) and the very low pyruvate dehydrogenase activity in the enterocyte (26) may also divert the pyruvate pool from either gluconeogenesis or oxidation, respectively, increasing flux through ALT and/or LDH. The flux through the latter enzyme is facilitated by the fact that the kinetics of LDH isoforms expressed in the SI (rich in M relative to L polypeptide chains) favors the pyruvate to lactate conversion, because LDH type M is not inhibited by increased pyruvate concentrations (34). The following observations support this proposal. 1) Glucose oxidation is virtually abolished, favoring lactate production in enterocytes in the fasted state (30); and 2) both the uptake of glutamine and the release of alanine are concomitantly increased during the processing of a glutamine load in long term versus short term fasting in dogs (35). It is worth emphasizing that the rapid induction of transcription of the gene for PEPCK-C in the SI after a 24-h fast might enhance the supply of gluconeogenic substrates furnished to the liver even without induction of Glc-6-Pase activity and of glucose production. PEPCK-C is a major cataplerotic enzyme that is involved in the disposal of citric acid cycle intermediates. It is thus expected that the increased flux via PEPCK-C may result in the enhanced synthesis of phosphoenolpyruvate, which can be converted to pyruvate via pyruvate kinase, thereby increasing pyruvate synthesis (Fig. 3). This is in agreement with the previously suggested role of PEPCK-C in the replenishment of the pyruvate pool in the SI (14,26,29). It has also been suggested that other enzymes also synthesize pyruvate, i.e. malic enzyme and oxaloacetate decarboxylase (26). However, the relative importance of these pathways remains uncertain (26). Altogether, these processes may function to increase the lactate and alanine supply to the liver from the earliest times of fasting (see Fig. 3). In strong support of this hypothesis is the observation that a substantial proportion of glutamine carbons is almost equally distributed between alanine and lactate in enterocytes and/or isolated intestine from short term fasted rats (26,28).

Conclusion and Future Directions
It is becoming an accepted view that the kidney and/or the SI may augment the liver in controlling glucose homeostasis in various situations, e.g. during fasting (31). We emphasize here that gluconeogenesis in the SI plays an important role in glucose balance. The strong induction of gluconeogenic genes in the SI during suckling in the rat suggests that gluconeogenesis by the SI also occurs in the fed state (18). The recent suggestion, from studies of the Glut2 knock-out mouse (36) and Glc-6-Pase-deficient patients (37), that Glc-6-Pase might be involved in the absorption of glucose from the SI lumen is also in keeping with a key role for intestinal Glc-6-Pase in the fed state. One may anticipate that important new insights in the role of the intestinal Glc-6-Pase gene in these different aspects of whole body glucose metabolism will soon be available, e.g. from intestine-specific gene knock-out mice and/or mice that overexpress the gene in that tissue. Because Glc-6-Pase and other gluconeogenic enzymes are all expressed in the human SI (12,27), deciphering the role of Glc-6-Pase in this tissue could constitute a significant breakthrough in the understanding of glucose homeostasis in humans.