Isolation of the gene for murine glucose-6-phosphatase, the enzyme deficient in glycogen storage disease type 1A.

Glycogen storage disease (GSD) type 1a (von Gierke disease) is caused by a deficiency in glucose-6-phosphatase, the key enzyme in glucose homeostasis catalyzing the terminal step in gluconeogenesis and glycogenolysis. Despite its clinical importance, this membrane-bound enzyme has eluded molecular characterization. Here we report the cloning and characterization of a murine glucose-6-phosphatase cDNA by screening a mouse liver cDNA library differentially with mRNA populations representing the normal and the albino deletion mouse known to express markedly reduced glucose-6-phosphatase activity. Additionally, we identified the gene that consists of 5 exons. Biochemical analyses indicate that the in vitro expressed enzyme is indistinguishable from mouse liver microsomal glucose-6-phosphatase exhibiting essentially identical kinetic constants, latency, thermal lability, and vanadate sensitivity. The characterization of the murine glucose-6-phosphatase gene opens the way for studying the molecular basis of GSD type 1a in humans and its etiology in an animal model.

clinical and biochemical investigations, G6Pase has eluded molecular characterization due primarily to its tight association with the endoplasmic reticulum (ER) and nuclear membranes (3). Characterization of the G6Pase protein and gene is critical for understanding the molecular basis of GSD type l a and for the development of novel therapeutic approaches to this genetic disorder.
To isolate cDNAs encoding GGPase, we took advantage of a n albino deletion mutant mouse that is known to express markedly reduced levels of G6Pase activity (4). The primary defect of this mutant mouse is the loss of the fumarylacetoacetate hydrolase gene located around the albino locus on chromosome 7 ( 5 ) . Fumarylacetoacetate hydrolase is the final enzyme in the tyrosine degradation pathway, and a deficiency of this enzyme leads to the accumulation of toxic tyrosine metabolites resulting in reduced expression of a group of liver-specific proteins, including G6Pase ( 5 , 6). Newborn homozygous deletion mice develop hypoglycemia shortly after birth, correlating with undetectable levels of G6Pase activity (4). In the present study, we have isolated and characterized a full-length cDNA (pmG6Pase) encoding murine liver microsomal G6Pase by screening a normal mouse liver cDNA library differentially (6) with probes representing mRNA populations from the normal and the albino deletion mutant mouse. Moreover, we have characterized the murine G6Pase transcription unit that spans approximately 10 kilobases and consists of 5 exons.

MATERIALS AND METHODS Library Screening and Characterization of cDNA and Genomic
Clones-A cDNA library in A g t l O representing wild-type homozygote (cCh/cch) mouse liver mRNA was screened differentially (6) with probes representing the mRNA populations from the wild-type and the albino deletion mutant mouse. pmG6Pase-1 that contains nucleotides 12-2259 of the murine G6Pase cDNA was one of the differentially expressed genes extensively characterized. The murine G6Pase gene was obtained by screening a mouse liver genomic library in Lambda Dash (Stratagene) with the pmG6Pase-l probe. The cDNA and genomic inserts of murine G6Pase were subcloned into pGEM (Promega Biotech, Madison, WI) vectors for further characterization. Both strands of the cDNA and genomic clones were sequenced by the Sanger dideoxy chain termination method (71, and the genomic sequences were compared with cDNA sequences to identify intron-exon junctions.
In Vitro Danscrzption and Danslation-In vitro transcription-translation of the pmG6Pase-l cDNA was performed using the TnT coupled reticulocyte lysate system obtained from Promega. The pmG6Pase-1 cDNA was analyzed in both sense and antisense orientations, and the in vitro synthesized proteins were analyzed by 10% polyacrylamide-SDS gel electrophoresis and fluorography.
Expression in COS-I Cells and Isolation of Microsomal Membranes-Nucleotides 12-1814 of the pmG6Pase cDNA (pSVLmGGPase), which contains the entire coding region at nucleotides 83-1153, was subcloned in a pSVL vector (Pharmacia LKE3 Biotechnology Inc.) and transfected into COS-1 cells by the DEAE-dextradchloroquine method ( 8 ) . Mock transfections of COS-1 cultures with the pSVL vector were used as controls.
Microsomal membranes were isolated by the method of Burchell et al. (9) either from Swiss Webster mice that had been fasted overnight or from freshly prepared homogenates of pSVLmG6Pase-transfected COS-1 cells. Disrupted microsomal membranes were prepared by incubating intact membranes in 0.2% deoxycholate for 20 min at 0 "C. The latency or intactness of microsomal preparations was assessed by assaying mannose-6-phosphohydrolysis in intact versus detergent-disrupted microsomes (10).
Phosphohydrolase and Phosphotransferase Assays-Phosphohydrolase activity was determined essentially as described by Burchell et al. (9). Phosphotransferase activity was determined by a modification of the method described by Jorgenson and Nordlie (11). The reaction mix-ture (10 pl) contained 100 mM HEPES buffer, pH 6.5, 50 mM glucose, ~-[ U -~~] g l u c o s e (los cpdreaction, 256 mCi/mmol, ICN Biochemicals, Irvine, CA), 4 m carbamyl-P, and deoxycholate-disrupted microsomal proteins. After incubation at 30 "C for 10 min, reactions were stopped by heating at 80 "C for 5 min. The samples were then centrifuged at 10,000 x g for 5 min, and 2-4 pl of supernatant was applied to a polyethyleneimine cellulose plate (J. T. Baker, Inc.). Glucose-6-P was separated from glucose by thin-layer chromatography developed in water. Spots were quantitated on a n AMBIS radioanalytic imaging system (San Diego, CAI. Northern Hybridization Analysis-RNA was isolated by the guanidinium thiocyanatdCsC1 method (12), separated by electrophoresis in 1.2% agarose gels containing 2.2 M formaldehyde (13), and transferred to Nytran membranes (Schleicher & Schuell). The hybridization and washing conditions were performed as previously described (6).

RESULTS AND DISCUSSION
Using differential cDNA screening of a mouse liver cDNA library, a group of liver genes affected in lethal albino mice has been isolated and characterized. Nucleotide sequence analysis of a 2248-bp cDNA (nucleotides 12-2259, designated pmG6Pase-1) revealed an open reading frame of 1071 nucleotides that encodes a 357-amino acid polypeptide (Fig. lA). The pmG6Pase-1 cDNA probe was used to screen a Lambda Dash mouse genomic library, and a genomic clone containing the entire murine G6Pase transcription unit was isolated and extensively characterized. The structural organization of the murine G6Pase transcription unit ( Fig. 1  Comparison of the nucleotide and deduced amino acid sequences of pmG6Pase cDNA with that in the data bases indicated no significant identity to any sequence reported to date. The predicted pmG6Pase-encoded protein has a calculated molecular mass of 40 kDa and contains an ER protein retention signal, motifs for protein glycosylation, and several membranespanning segments. This suggested that this cDNA may encode microsomal MPase, a glycoprotein of 35-36.5 kDa (14,15). The identity of the cDNA was confirmed by performing detailed biochemical studies after transient expression of the pSVLmG6Pase plasmid (nucleotides 12-1814 of the pmG6Pase cDNA) in COS-1 cells.
The WPase, produced by translation of murine G6Pase mRNA synthesized by in vitro transcription of the pmG6Pase-1 cDNA, migrated on a SDS-polyacrylamide gel as a 34-kDa polypeptide (Fig. s), which is considerably smaller than the predicted molecular size of 40 kDa. The anomalous electrophoretic mobility of the murine G6Pase protein may reflect the extremely hydrophobic nature of the encoded polypeptide. The hydropathy index analysis (16, 17) predicts that the G6Pase protein contains six putative membrane-spanning segments (Fig. 2 B ) .
The ER localization of murine G6Pase is predicted by the presence of two lysines, positioned 3 and 4 amino acids from the carboxyl terminus of the deduced protein (Fig. lA), a consensus motif for the retention of proteins in the ER (18). In addition, three potential asparagine-linked glycosylation sites are present in the predicted murine G6Pase polypeptide at amino acids [96][97][98][203][204][205][276][277][278], suggesting that G6Pase is a glycoprotein.
To demonstrate the functional identity of the pmG6Pase cDNA, detailed biochemical studies were performed in microsomal preparations of pSVLmG6Pase-transfected COS-1 cells and compared with activities in microsomes isolated from adult mouse livers. COS-1 cells, which are monkey cells transformed with a replication origin-defective SV40 DNA molecule, produce high levels of the SV40 large T antigen in culture (19). The pSVL vector, which contains a SV40 replication origin, is engineered to express genes efficiently after transient transfection in 21).
Hepatic G6Pase is known to exhibit latency, referring to the portion of enzymatic activity that is not expressed unless the microsomes are disrupted (3). Microsomal G6Pase has varying degrees of latency depending on the substrate utilized. Both glucose-6-P and mannose-6-P are rapidly hydrolyzed in disrupted microsomes; only glucose-6-P is hydrolyzed in intact microsomes (22). Therefore, mannose-6-P phosphohydrolysis in intact uersus detergent-disrupted microsomes is used to measure the latency or intactness of microsomal preparations. Latencies for mannose-6-P phosphohydrolysis are 95% or greater in rat liver microsomes (3) and are about 40-54% in microsomes derived from isolated rat hepatocytes or hepatoma cells (11). In agreement with values reported for microsomal G6Pase of the rat, mannose-6-P phosphohydrolase activity in microsomes isolated from mouse livers and pSVLmG6Pase-transfected COS-1 cells exhibited latency values of 97 and 50%, respectively (Table I). Therefore, microsomes of cultured cells have similarly reduced latencies. Hepatic G6Pase is characterized by its high thermal lability; G6Pase is completely inactivated by incubating the microsomal preparation (or homogenate) at pH 5.0 for 10 min at 37 "C (23). Under the same conditions, the majority of nonspecific phosphatases are still capable of hydrolyzing glucose-6-P. Incubation of microsomes isolated from pSVLmG6Pase-transfected COS-1 cells or adult mouse livers at 37 "C for 10 min at pH 5.0 abolished glucose-6-P phosphohydrolase activity (Table I), demonstrating that the expressed enzyme and mouse liver microsomal G6Pase have similar thermal lability.
The pH profiles of glucose-6-P phosphohydrolase activity in disrupted microsomes prepared from pSVLmG6Pase-transfected COS-1 cells and adult mouse liver were virtually identical (Fig. 3A). The pH optimum is close to 6.5 in both preparations, in agreement with the pH profile obtained for rat liver microsomal G6Pase (22).
Kinetic studies of phosphohydrolysis, with either glucose-6-P or mannose-6-P as the substrate, were performed with microsomes isolated from pSVLmG6Pase-transfected COS-1 cells and adult mouse livers. It was previously reported for rat that the K, values of glucose-6-P hydrolysis in intact microsomes are higher than in disrupted microsomes (10,241. In this study, we also observed a higher K, value for glucose-6-P hydrolysis for intact (3.3 mM) uersus disrupted (0.68 mM) microsomes of adult mouse livers, demonstrating the similarities between mouse and rat microsomes. We observed little increase in the K , value for glucose-6-P hydrolysis using intact microsomes from pSVLmG6Pase-transfected cells because of the reduced latency (Table I). Consequently, the kinetic parameters for G6Pase in transfected cells were determined for deoxycholatedisrupted microsomal preparations. The K , values for glucose-6-P and mannose-6-P hydrolysis were indistinguishable between microsomes of pSVLmG6Pase-transfected COS-l cells and adult mouse livers (Table I). Moreover, these values were similar to those reported for rat microsomal glucose-6-P (10,241 and mannose-6-P (10) phosphohydrolysis. Additionally, the Characteristics of microsomal G6Pase activity in pSVLmG6Pase-transfected COS-1 cells and adult mouse livers Latency for mannose-6-P (5 mM) hydrolysis, defined as (1intact/ disrupted) x 100, was performed in microsomes prepared from two independent batches of pSVLmG6Pase-transfected COS-1 cells and three different adult mouse livers. Activities were the average of three sus disrupted microsomes, in pSVLmG6Pase-or mock-transfected determinations. The specific phosphohydrolase activities of intact uer-COS-1 cells, were 157.7/313.0 and 1.6/3.3 nmol/midmg of microsomal protein, respectively. The specific phosphohydrolase activities in intact and disrupted microsomes of adult mouse livers were 11.2 and 349.2 nmol/midmg of microsomal protein, respectively. Thermal stability was determined by assaying glucose-6-P phosphohydrolase activity in deoxycholate (O.2%)-disrupted microsomal membranes, before and aRer incubation a t 37 "C for 10 min at pH 5.0, and refers to enzyme activities remaining after heat treatment. In the phosphohydrolase assay, the K,,, and Ki values are expressed in mM; V, , as pmol/midmg of microsomal protein. K,,, and V, , values represent the mean f S.E. The values for phosphotransferase activities (pmoVmidmg of microsomal protein) were obtained from the linear region of enzyme concentration curves. Mock transfected COS-1 cells exhibited <1% of the total thermal-sensitive phosphotransferase activities.  (Table I) and are in agreement with those reported previously for rat liver microsomes (10). Vanadate is a potent inhibitor of glucose-6-P phosphohydrolase activity (25). Microsomal preparations from both pSVLmG6Pase-transfected COS-1 cells and adult mouse livers were equally sensitive to vanadate, giving nearly identical inhibition curves of glucose-6-P hydrolysis (Fig. 3B). Moreover, vanadate was a competitive inhibitor of glucose-6-P phosphohydrolysis in both microsomal preparations (Fig. 3, C (1.5 PM) reported for permeable hepatocytes or rat microsomes (25). The reason for this discrepancy is unknown. However, a similar vanadate inhibition curve to that for the mouse G6Pase reported here was observed for commercially obtained crude microsomal preparations of rabbit G6Pase obtained from Sigma (data not shown).
In addition to displaying phosphohydrolytic activity, G6Pase is capable of catalyzing the formation of glucose-6-P from glucose and a variety of phosphate substrate donors (3). Phosphotransferase activities in microsomes of pSVLmG6Pasetransfected COS-1 cells and adult mouse livers were evaluated using carbamyl-P and glucose as substrates (Table I). Similar specific transferase activities were observed in both microsomal preparations, in good agreement with the value (0.294 pmol/min/mg of microsomal protein) reported for carbamyl-P glucose phosphotransferase activity in rat hepatocyte microsomes (11).
The expression of G6Pase mRNA was examined in livers of normal and albino deletion mice by Northern blot hybridization analysis (Fig. 4). As expected, G6Pase mRNA was detected only in normal mouse liver; little or no G6Pase transcripts were detected in the liver of the albino deletion mouse. Both liver with either an antisense pmG6Pase-1 probe (A) or an antisense probe containing nucleotides 730-1820 of the pmG6Pase cDNA ( B ) . The filters were rehybridized either with a transferrin (TF) or a heat shock protein 84 (HSP84) probe, which was used as an internal standard. and kidney are known to express high levels of G6Pase enzyme activity (1)(2)(3). G6Pase mRNA was evident in the liver and kidney but was not detectable in testes, brain, muscle, or lung, demonstrating that G6Pase mRNA expression is also restricted to the liver and kidney (Fig. 4).
Two models for GGPase catalysis have been proposed to account for the relationship between G6Pase and the membrane with which it is intimately associated. The conformation-substrate-transport model (26) proposes that G6Pase represents a single membrane channel protein capable of both transport and catalytic functions. The translocase-catalytic unit model (3,101 proposes that G6Pase is a multicomponent complex consisting of a G6Pase catalytic unit and associated translocases. The translocase-catalytic unit model has been used to explain the existence of GSD type la, lb, IC, and Id patients, which correspond to defects in GGPase, the putative glucose-6-P translocase, phosphatdpyrophosphate translocase, and glucose translocase (1-3, 27, 28). The absence of mutations in the G6Pase gene in GSD type lb, IC, or Id patients will support the translocase-catalytic unit model, suggesting the existence of translocases.
The present study demonstrates that the pmG6Pase cDNA encodes the multifunctional GGPase, the enzyme deficient in GSD type l a in humans. Cloning and characterization of the murine gene will greatly facilitate the isolation of the human G6Pase gene and the identification of mutations in GSD type l a patients. Accordingly, understanding the molecular basis for GSD type la and the development of new therapies is now possible. Finally, an animal model for studying the etiology of GSD type la in humans can be easily established by homologous gene targeting to disrupt the murine G6Pase gene.