A glucose-6-phosphate hydrolase, widely expressed outside the liver, can explain age-dependent resolution of hypoglycemia in glycogen storage disease type Ia.

A fine control of the blood glucose level is essential to avoid hyper- or hypo-glycemic shocks associated with many metabolic disorders, including diabetes mellitus and type I glycogen storage disease. Between meals, the primary source of blood glucose is gluconeogenesis and glycogenolysis. In the final step of both pathways, glucose-6-phosphate (G6P) is hydrolyzed to glucose by the glucose-6-phosphatase (G6Pase) complex. Because G6Pase (renamed G6Pase-alpha) is primarily expressed only in the liver, kidney, and intestine, it has implied that most other tissues cannot contribute to interprandial blood glucose homeostasis. We demonstrate that a novel, widely expressed G6Pase-related protein, PAP2.8/UGRP, renamed here G6Pase-beta, is an acid-labile, vanadate-sensitive, endoplasmic reticulum-associated phosphohydrolase, like G6Pase-alpha. Both enzymes have the same active site structure, exhibit a similar Km toward G6P, but the Vmax of G6Pase-alpha is approximately 6-fold greater than that of G6Pase-beta. Most importantly, G6Pase-beta couples with the G6P transporter to form an active G6Pase complex that can hydrolyze G6P to glucose. Our findings challenge the current dogma that only liver, kidney, and intestine can contribute to blood glucose homeostasis and explain why type Ia glycogen storage disease patients, lacking a functional liver/kidney/intestine G6Pase complex, are still capable of endogenous glucose production.

Blood glucose homeostasis between meals is maintained by endogenous hepatic and renal glucose production via glycogenolysis and gluconeogenesis. In the terminal stages of both pathways, glucose-6-phosphate (G6P) 1 is hydrolyzed to glucose by the glucose-6-phosphatase (G6Pase) complex embedded in the membranes of the endoplasmic reticulum (ER) (reviewed in Refs. 1 and 2). This complex is composed of a G6P transporter (G6PT) that transports G6P from the cytoplasm into the lumen of the ER and a G6Pase catalytic subunit that hydrolyzes the G6P to glucose. Disruption of either component disturbs glucose homeostasis and results in glycogen storage disease type Ia (GSD-Ia, G6Pase deficiency) or type Ib (GSD-Ib, G6PT deficiency). The G6PT and G6Pase activities are co-dependent (3,4), and both GSD-Ia and GSD-Ib patients manifest the symptoms of failed G6P hydrolysis, namely hypoglycemia, growth retardation, hepatomegaly, nephromegaly, hyperlipidemia, hyperuricemia, and lactic acidemia (1,2).
Despite disruption of the G6Pase complex in GSD-I patients, several studies (5)(6)(7) indicate that these patients are still capable of producing glucose, implying there are alternative pathways for endogenous glucose production. G6PT is expressed ubiquitously (8), but G6Pase expression is considered to be restricted to the liver, kidney, and intestine (9,10). Therefore, we searched the ENSEMBL data base for a G6Pase homolog expressed outside of the liver, kidney, and intestine. We identified a theoretical phosphatidic acid phosphatase, PAP2.8, encoded within locus 92579 (accession number XM045901) on chromosome 17q21.31, which, by electronic Northern, is expressed ubiquitously (11). Recently this protein was described by Martin et al. (12) as a ubiquitously expressed G6Pase catalytic subunit-related protein (UGRP) lacking phosphohydrolase activity.
Human PAP2.8/UGRP is a hydrophobic protein sharing 36% amino acid sequence identity with human G6Pase. Sequencetopology alignment with G6Pase predicts that it should contain nine transmembrane helices and an active hydrolytic site lying on the luminal side of the ER (12)(13)(14)(15). Therefore, the reported absence of phosphohydrolase activity was unexpected. The activity assays used by Martin et al. (12) were performed on cell lysates in which PAP2.8/UGRP had been expressed transiently using a plasmid-based vector. We wondered whether the low level of UGRP expression in this system limited the sensitivity of the phosphohydrolase assay. In this study, we adopted a recombinant adenoviral vector-mediated expression system to increase the level of expression of PAP2.8/ UGRP and enhance the sensitivity of the G6P hydrolase assay. We demonstrate that PAP2.8/UGRP is a phosphohydrolase that can couple with G6PT to form an active G6Pase complex. We now propose naming PAP2.8/UGRP as G6Pase-␤, to differentiate it from the liver/kidney/intestine G6Pase that we rename G6Pase-␣. Our results provide new insights into glucose homeostasis and explain why patients genetically deficient in the G6Pase-␣ complex are still capable of endogenous glucose production.
Expression in COS-1 Cells and Western-blot Analysis-COS-1 cells in 150-cm 2 flasks were grown at 37°C in HEPES-buffered Dulbecco's modified minimal essential medium supplemented with 4% fetal bovine serum and infected with recombinant virus. The infected cultures were used to isolate microsomes for G6P uptake analysis after incubation at 37°C for 24 h and for phosphohydrolase and Western-blot analyses after incubation at 37°C for 48 h.
For Western blot analysis, COS-1 cell lysates were separated by electrophoresis through a 13% polyacrylamide-SDS gel, blotted onto polyvinylidene fluoride membranes (Millipore Corp., Billerica, MA), incubated either with a monoclonal antibody against the FLAG epitope (Scientific Imaging Systems, Eastman Kodak Co.) or a polyclonal antibody against human G6Pase-␣ (14). The antigen-antibody complex was visualized as described previously (14).
Phosphohydrolase and G6P Uptake Analyses-Phosphohydrolase activity was determined essentially as described previously (13). Reaction mixtures (100 l) contained 50 mM cacodylate buffer, pH 6.5, 10 mM G6P, and appropriate amounts of cell homogenates or microsomal preparations were incubated at either 30°C or 37°C for 10 min. Disrupted microsomal membranes were prepared by incubating intact microsomes in 0.2% deoxycholate for 20 min at 0°C. Nonspecific phosphatase activity was estimated by pre-incubating disrupted microsomal preparations at pH 5 for 10 min at 37°C to inactivate the acid-labile G6Pase-␣ and G6Pase-␤.
Statistical analysis using the unpaired t test was performed with the GraphPad Prism program (GraphPad Software, San Diego, CA). Data are presented as the mean Ϯ S.E.
For intracellular staining, the cells were blocked with 2% normal horse serum in PBS for 20 min at room temperature, incubated with a monoclonal anti-FLAG antibody in PBS containing 0.1% saponin for 1 h, followed by goat fluorescein isothiocyanate-conjugated anti-mouse IgG (Sigma) in PBS containing 0.1% saponin for 1 h. After washing in PBS containing 0.1% saponin, the cells were mounted with an anti-fade, water-based mounting medium (Vector Lab, Burlingame, GA) and analyzed under a laser scanning confocal fluorescence microscope (Leica TCS-4D DMIRBE, Heidelberg, Germany). Staining of the calreticulin ER marker was performed similarly using rabbit anti-calreticulin antibody (Affinity BioReagents, Golden, CO) and TRITC-conjugated goat anti-rabbit IgG (Sigma). Excitation wavelengths of 488 (for fluorescein isothiocyanate) and 568 (for TRITC) mm were used to generate fluorescence emission in green and red, respectively. Co-localization of green fluorescent G6Pase-␣ or G6Pase-␤ and red fluorescent calretculin is reflected by yellow fluorescence.
G6Pase-␤ Is Localized in the Endoplasmic Reticulum-G6Pase-␣ is a nine-transmembrane domain protein embedded in the ER membrane, oriented with its active site facing into the lumen (13,14). The protein contains a carboxyl-terminal KKXX motif, responsible for the retention of type I transmembrane proteins in the ER (18), and has a latent activity that is released upon disruption of the microsomes (10). In contrast, G6Pase-␤ lacks any apparent ER retention motif (19,20), although Ad-G6Pase-␤-infected microsomes do exhibit latency similar to Ad-G6Pase-␣ (Fig. 1B). To confirm the ER retention of G6Pase-␤, FLAG-tagged constructs of pSVL-G6Pase-␤ or pSVL-G6Pase-␣ were transfected into COS-1 cells and visualized by double immunostaining for the FLAG-tag and ERmarker protein calreticulin (21). As expected from the microsomal localization of the activities, both G6Pase-␣ and G6Pase-␤ co-localize with calreticulin (Fig. 3), confirming their retention within the ER.
The Active Site Residues Are Conserved between G6Pase-␤ and G6Pase-␣-Functional analysis of missense and codon deletion mutations in the G6Pase-␣ gene of GSD-Ia patients (22) as well as active site analysis (14) has identified 43 amino acid residues essential for G6Pase catalysis (Fig. 4). Thirty-five of these residues, including three active site residues, are conserved between mammalian G6Pase-␣ and G6Pase-␤ (Fig. 4). The active site of G6Pase-␣ is composed of residues Arg 83 , His 119 , and His 176 (14), and the analogous residues in G6Pase-␤ appear to be Arg 79 , His 114 , and His 167 (Fig. 4). To demonstrate the importance of these residues in G6Pase-␤, mutations known to disrupt the active site in G6Pase-␣ (14) were introduced into G6Pase-␤. Mutants R79A, H114A, or H167A were constructed in Ad-G6Pase-␤-3FLAG, expressed transiently in COS-1 cells, and assayed for expression by Western blot and G6P phosphohydrolase activity (Fig. 5). Each mutant is expressed as efficiently as the wild-type G6Pase-␤ protein but is devoid of G6P hydrolytic activity. Therefore, Arg 79 , His 114 , and His 167 are components of the active site of G6Pase-␤.

G6Pase-␤ Is a Functional Phosphohydrolase 47101 DISCUSSION
In humans, blood glucose levels must be maintained between 70 -110 mg/dl to avoid the complications of hyper-and hypoglycemia. Although insulin plays major roles in regulating blood glucose levels (23), blood glucose homeostasis between meals depends on gluconeogenesis and glycogenolysis. In the terminal step common to both pathways, G6P is hydrolyzed to glucose by the G6Pase complex, which is composed of a G6P transporter coupled to a G6P phosphohydrolase (1,2). Deficiencies in the G6Pase complex disrupt glucose production via gluconeogenesis and glycogenolysis, resulting in the metabolic disorder GSD-I, which is characterized by hypoglycemia, growth retardation, hepatomegaly, nephromegaly, hyperlipidemia, hyperuricemia, and lactic acidemia (1,2). Currently, only the liver, kidney, and intestine are considered to be involved in inter-prandial glucose homeostasis. This view is based on the absence of G6Pase-␣, an activity essential to the final stage of gluconeogenesis and glycogenolysis, outside of these organs. We now show that other tissues may be able to contribute to blood glucose homeostasis through the activity of PAP2.8/UGRP (11,12), a G6Pase-related protein expressed ubiquitously, that we rename G6Pase-␤.
An initial report on G6Pase-␤ ascribed no phosphohydrolase activity to the protein (12). By improving the transient expression of the gene with an adenovirus-based expression system, we have demonstrated that, like the hepatic/renal/intestinal G6Pase (now renamed G6Pase-␣), G6Pase-␤ is a G6P hydrolase. The two proteins are very similar. Unlike nonspecific phosphatases, both G6Pase-␣ and G6Pase-␤ are acid-labile. They are also inhibited readily by vanadate, although G6Pase-␤ shows a greater sensitivity to the inhibitor than G6Pase-␣. The activities share a common pH optimum of 6.5 but differ in their temperature profiles, G6Pase-␣ having a broad optimum centered at 30°C, whereas G6Pase-␤ has a sharper profile centered at 37°C. Both exhibit a similar K m toward G6P, suggesting that they could play similar physiological roles in cells, but the V max of G6Pase-␣ is ϳ6-fold greater than that of G6Pase-␤. Changing any of the three residues critical to the active site in G6Pase-␣ destroys the activity of G6Pase-␤, suggesting both phosphohydrolases share a similar structure in their active sites and confirming that G6Pase-␤ is the source of the phosphohydrolase activity being measured.
To participate in the final step of gluconeogenesis and glycogenolysis, the hydrolysis of G6P to glucose and phosphate, a G6P hydrolase must be able to couple functionally with the ubiquitously expressed ER transmembrane protein, G6PT (1). For this reason, nonspecific phosphatases are not able to contribute to glucose homeostasis. Both G6Pase-␣ and G6Pase-␤ are localized in the ER. When G6Pase-␤ is co-expressed with G6PT, the proteins couple to transport G6P from the cellular cytoplasm into the ER lumen. Although the rate of G6P accumulation mediated by the G6Pase-␤/G6PT complex is ϳ25% that of the G6Pase-␣/G6PT complex, it is still a significant accumulation, showing a broader pH optimum but a similar temperature dependence as G6Pase-␣/G6PT.
Although G6Pase-␤ has only ϳ12% of the activity of G6Pase-␣, the demonstration of a significant, specific G6P hydrolytic activity that can couple to G6PT outside of the liver raises interesting questions about the ability of other tissues to cycle glucose and contribute to blood glucose homeostasis. Of particular interest is the muscle, which expresses an elevated level of G6Pase-␤ (12). Although the liver stores the most glycogen weight for weight of any tissue, about 90 g in a 70-kg male (24,25), the muscle, by its sheer mass (40 -45% of the wet body weight) (24 -27), is the largest reservoir of body glycogen, storing about 300 g of glycogen per 70-kg male. It is interesting to speculate that some of this glycogen reservoir may be cycled into blood glucose by G6Pase-␤. Without further study, it is difficult to judge how important this might be to glucose homeostasis, because in muscle, cytoplasmic G6P has multiple fates, which include hydrolysis to glucose, glycogenesis, energy production via glycolysis, entry into the pentose phosphate pathway, and lipid synthesis. Studies of G6Pase-␣-deficient patients or mice (3) may hold interesting clues.
Monitoring of GSD-Ia patients, who lack a functional G6Pase-␣, clearly demonstrates that there is a significant supply of endogenous glucose production from an unknown source. Young GSD-Ia patients show significant hypoglycemia, but with age, the endogenous glucose production rate improves, starting from 50% of normal in young GSD-Ia patients to 67-FIG. 5. G6P phosphohydrolase activity and Western blot analysis of deoxycholate-disrupted COS-1 lysates infected with wildtype and active site mutants of Ad-G6Pase-␤-3FLAG. Phosphohydrolase activity was assayed at pH 6.5 and 37°C, and the results are given as mean Ϯ S.E.
FIG. 6. G6Pase-␤ couples with G6PT to mediate microsomal G6P uptake. A, time course of microsomal G6P uptake activity. B, microsomal G6P uptake activity in COS-1 cells infected with Ad-G6Pase-␤ (25 pfu/cell), Ad-G6Pase-␣ (25 pfu/cell), Ad-G6PT (50 pfu/cell), or co-infected with 25 pfu/cell of either Ad-G6Pase-␤ or Ad-G6Pase-␣ and 50 pfu/cell of Ad-G6PT. C, pH dependence of microsomal G6P uptake activity. For both the time course and pH-dependence, the cells were co-infected at the same time with 50 pfu/cell of Ad-G6PT and either 25 pfu/cell of Ad-G6Pase-␤ or 25 pfu/cell of Ad-G6Pase-␣. G6P uptake was performed at pH 6.5 and 30°C. The results are given as mean Ϯ S.E. 100% of normal in adult GSD-Ia patients (5)(6)(7). Because the muscle mass is only ϳ20% of the body weight of a newborn but improves through adolescence (36% muscle mass) to adulthood (40 -45%) (27), it is reasonable to suggest that the muscle G6Pase-␤/G6PT complex may be the source of some or all of the extra blood glucose. Indeed preliminary experiments examining the G6Pase activity in muscle derived from G6Pase-␣deficient (3) and G6PT-deficient (28) mice are promising in this respect. 2 Because G6Pase-␤ is well expressed in kidney (12), another potential source of the glucose is the kidney, although this seems less significant because the kidney weighs only ϳ17% of the wet weight of the liver (25) and has a significantly lower glycogen content weight for weight.
Our findings with G6Pase-␤ challenge the current dogma of glucose homeostasis and suggest that glucose recycling through glycogen or other G6P-mediated pathways may take place in a much wider range of tissues than previously envisaged. The proof of this rests in detailed in vivo measurements of glucose recycling, which are currently in progress. The importance of such a finding depends on the physiological impact of the loss of G6Pase-␤ activity. We are currently addressing the physiological importance of this through the use of knockout and over-expression mouse strains.