Identification of the Catalytic Residues of Bifunctional Glycogen Debranching Enzyme*

Eukaryotic glycogen debranching enzyme (GDE) possesses two different catalytic activities (oligo-1,431,4glucantransferase/amylo-1,6-glucosidase) on a single polypeptide chain. To elucidate the structure-function relationship of GDE, the catalytic residues of yeast GDE were determined by site-directed mutagenesis. Asp-535, Glu-564, and Asp-670 on the N-terminal half and Asp1086 and Asp-1147 on the C-terminal half were chosen by the multiple sequence alignment or the comparison of hydrophobic cluster architectures among related enzymes. The five mutant enzymes, D535N, E564Q, D670N, D1086N, and D1147N were constructed. The mutant enzymes showed the same purification profiles as that of wild-type enzyme on b-CD-Sepharose-6B affinity chromatography. All the mutant enzymes possessed either transferase activity or glucosidase activity. Three mutants, D535N, E564Q, and D670N, lost transferase activity but retained glucosidase activity. In contrast, D1086N and D1147N lost glucosidase activity but retained transferase activity. Furthermore, the kinetic parameters of each mutant enzyme exhibiting either the glucosidase activity or transferase activity did not vary markedly from the activities exhibited by the wild-type enzyme. These results strongly indicate that the two activities of GDE, transferase and glucosidase, are independent and located at different sites on the polypeptide chain.

residue from the branched chain of glycogen is first transferred to the main chain by the transferase to expose the ␣-1,6-glucosyl stub that is in turn hydrolyzed by the glucosidase, thus allowing glycogen phosphorylase to degrade the linearized ␣- (1,4) polymer. Genetic deficiency of the enzyme in human Type III glycogen storage disease (GSD-III or Cori's disease) is characterized by hepatomegaly, hypoglycemia, variable myopathy, and cardiomyopathy (4,5).
GDE has been purified and characterized from rabbit (2, 3) and yeast (6,7). Inhibitors specifically affecting either the transferase or the glucosidase activity provided evidence that each of the two activities occur at distinct catalytic sites (8). Liu et al. (9) showed that the transferase activity was irreversibly inactivated by carbodiimide in the presence of amines without affecting the glucosidase activity and concluded the existence of two distinct active sites, although the locations were not defined. The amino acid sequence analysis of rabbit GDE indicated that the N-terminal half may encompass the transferase activity, leaving the glucosidase activity at the C-terminal half (10).
Several facts point to the resemblance of yeast GDE to the mammalian GDEs. The yeast, human, and rabbit GDEs have 1536, 1515, and 1555 amino acid residues, respectively, as deduced from nucleotide sequences (10 -12). The yeast GDE N-terminal half also possesses four conserved sequences in the ␣-amylase family, and the C-terminal half displays about 50% identity with the C-terminal half of other mammalian GDEs. Foremost, yeast GDE exhibits the same transferase and glucosidase action toward glycogen and branched cyclodextrins as those of mammalian GDEs (7). Therefore, yeast GDE can serve as a model for the eukaryotic GDE for elucidating the structure-function relationship.
To determine the location of the active sites of both the transferase and glucosidase activities, we constructed several mutant yeast GDEs by site-directed mutagenesis. From the analysis of the enzymatic activities, the amino acid residues involved in the catalysis of the transferase and glucosidase of yeast GDE were identified.
Site-directed Mutagenesis-Plasmids harboring the GDE gene with a * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Present address: Nara Prefecture Institute of Public Health, Ohmori-cho, Nara city, Nara 630-8131, Japan.
single-site mutation were constructed with the QuickChange site-directed mutagenesis kit. The mutagenesis primers were synthesized by the TaKaRa Syuzo Custom service. DNA sequencing via the dideoxy chain-terminating method (13) with dsDNA serving as a template confirms the site and type of mutations occurring on the gene.
Expression and Purification of Wild-type and Mutant GDEs-The expression and purification of the wild-type and mutant GDEs were performed as described previously (11). E. coli JM 105 harboring recombinant plasmid bearing the wild-type or mutant GDE gene was grown in LB medium containing 50 g/ml of ampicillin at 25°C to an A 600 ϭ 0.6 before isopropyl-␤-D-thiogalactopyranoside (IPTG) was added to the final concentration of 0.02 mM. Induction of the enzymes was carried out for 12 h at 25°C, and the cells were harvested by centrifugation at 5,500 ϫ g for 5 min. The collected cell pellet was resuspended in 20 mM Tris-HCl buffer (pH 7.5) and sonicated. To the supernatant obtained by centrifuging the sonicated samples at 17,000 ϫ g for 20 min was added ammonium sulfate to a final concentration of 0.7 M. The crude lysate was then applied to a ␤-CD-Sepharose-6B affinity column previously equilibrated with 20 mM Tris-HCl buffer, pH 7.5, 0.7 M ammonium sulfate. The column was washed with the same buffer repeatedly before eluting the enzyme with a linear gradient of ammonium sulfate (0.7-0 M) in 20 mM Tris-HCl buffer (pH 7.5). Homogeneity of the enzymes was analyzed by SDS-polyacrylamide gel electrophoresis, and the protein was visualized by staining with Coomassie Brilliant Blue R-250. The protein concentration was determined with the BCA protein assay reagent kit with bovine serum albumin as the standard (14).
HPLC Analysis of the Sugar Products-Transferase and glucosidase activities of the mutant enzymes were analyzed by HPLC using the branched cyclodextrins Glc-␤-CD and Glc 5 -␤-CD as substrates, respectively (7). A 10-l reaction mixture contains 25 mM citrate buffer, pH 6.5, 3 g of purified enzyme and 10 mM Glc-␤-CD or Glc 5 -␤-CD substrate. The reaction was carried out at 37°C for 1 or 5 h, and the enzyme was inactivated by 30 s of boiling over a water bath. The reaction mixture was then 5-fold-diluted with distilled water and filtered through ultrafree-MC (30,000 cutoff filter, Millipore). An aliquot of the filtrate was injected into the HPLC (BioLC, Dionex, Sunnyvale, CA) and allowed to adsorb on a CarboPac PA-1 column (250 ϫ 4 mm) for 3 min under an eluent of 150 mM NaOH. The mixtures of branched cyclodextrin products were eluted by increasing linearly the sodium acetate concentration in 150 mM NaOH from 0 -33% for 45 min at the flow rate of 1.0 ml/min. The sugar was detected with a pulsed amperometric detector (PAD-II) with a gold-working electrode and triplepulsed amperometry.
Enzymatic Assays for Transferase and Glucosidase Activities-The kinetic parameters for transferase activity were measured according to the method of Tabata and Dohi (15) using maltopentaose as a substrate. The reaction mixture (25 l) containing varied concentrations of maltopentaose in 40 mM phosphate buffer (pH 6.5), and 5 l of enzyme solution was incubated at 37°C for 10 min. The products formed were analyzed by HPLC as described above. The velocity of the transferase activity at 37°C, pH 6.5, was expressed as the sum of the products maltose and maltotriose (nmol/min).
Glucosidase activity was determined by measuring the amount of glucose released from the substrate Glc-␤-CD as described previously (11). The reaction mixture (50 l) contained 50 mM citrate buffer (pH 6.5) and an appropriate amount of Glc-␤-CD. The reaction was initiated by the addition of 10 l of enzyme solution and was carried out at 37°C for 10 min. The enzyme reaction was terminated with 30 s of boiling over a water bath. The released glucose was determined spectrophoto-metrically by monitoring the reduction of NADP using hexokinase and glucose 6-phosphate dehydrogenase (16). One unit of activity was defined as the amount of enzyme catalyzing the release of 1 mol of glucose/min at 37°C, pH 6.5. Table I, alignment of the amino acid sequences of human muscle (12), rabbit muscle (10), and yeast (11) GDE revealed the presence of four consensus sequences commonly found in the ␣-amylase family or glycosyl hydrolase family 13 (17) on the N-terminal half. It has been suggested that the consensus sequences II, III, and IV in the ␣-amylase family contain three conserved acidic amino acids (two Asp and one Glu) (18). Jespersen et al. (19) suggested that the corresponding carboxyl residues in human GDE are involved in the transferase activity. Therefore, amino acid residues, Asp-535, Glu-564, and Asp-670, which are located in the consensus sequences II, III, and IV of yeast GDE, respectively, were chosen as the target for site-directed mutagenesis to determine their role in transferase activity.

Analysis of the Amino Acid Sequences for the Prediction of Amino Acids Essential for GDE Activity-As shown in
Liu et al. (10) suggested that the C-terminal half might  encompass the glucosidase activity. However, the amino acid sequence of yeast GDE (11) showed that its C-terminal half had no significant homology with other amylolytic enzymes except for the mammalian GDEs (10,12). Likewise, the amino acid sequence on the C-terminal half of GDE and ␣-1,6-glucosidase (isomaltase) from the same origin Saccharomyces cerevisiae D-346 showed no significant homology. Because the carboxyl group is proposed to be involved in the catalysis of many glycosyl hydrolases (20), we analyzed the secondary structural features of the carboxyl residues in the C-terminal half of the yeast GDE by hydrophobic cluster analysis (HCA). The oligo-1,6-glucosidase from Bacillus cereus was used as a model enzyme because its catalytic residues have been identified by x-ray crystallographic analysis (21) (Fig. 1), and it acts on the ␣-1,6-glucosidic linkage similar to that of the glucosidase of yeast GDE. Indeed, the HCA plot indicated a significant similarity to the secondary structural features of the C-terminal half of yeast GDE and oligo-1,6-glucosidase from B. cereus. Among the acidic amino acids, Asp-1086 and Asp-1147 of yeast GDE showed similar patterns in the HCA plot as Asp-206 in the conserved region II and Glu-230 in the conserved region III, the two catalytic residues of oligo-1,6-glucosidase (Fig. 1). Furthermore, these residues were found well conserved among the amino acid sequences on the C-terminal half of human, rabbit, and yeast GDEs. The Asp-1086 and Asp-1147 were tagged as possible candidates for the active sites of amylo-1,6-glucosidase activity of the yeast GDE.

Amino Acid Substitution by Site-directed Mutagenesis and the Expression and Purification of Mutant GDEs-The
Asp-535, Glu-564, and Asp-670 on the N-terminal half, and Asp-1086 and Asp-1147 on the C-terminal half were substituted to their respective amides by site-directed mutagenesis. Sequence analysis of the mutated GDE genes confirmed the desired mutations without any second-site mutations. All the mutant enzymes were highly expressed and have the same elution profile on the ␤-CD-Sepharose-6B affinity chromatography as that of the wild-type enzyme. In addition, all the purified mutant GDEs were homogeneous and have the same molecular mass as the wild-type enzyme.
Characterization of the Mutant GDEs-The glucosidase and transferase activities of mutant yeast GDEs were determined separately by HPLC using two types of branched cyclodextrin as substrates, Glc-␤-CD and Glc 5 -␤-CD, respectively (Figs. 2 and 3). The purified wild-type enzyme transfers maltosyl and maltotriosyl residues from one Glc 5 -␤-CD to another to form Glc 7 -, Glc 8 -␤-CD and Glc 2 -, Glc 3 -␤-CD. These branched ␤-CDs produced by the transferase reaction are also substrates for the transferase, accordingly Glc-␤-CD, Glc 2 -, -Glc 8 -␤-CD arise as products by the transferase action and then Glc-␤-CD is hydrolyzed to glucose and ␤-CD by the glucosidase action (Fig. 2, a  and b, W). HPLC profiles of mutants D535N, E564Q, and D670N gave only a single peak corresponding to substrate Glc 5 -␤-CD (Fig. 2a, N-M), indicating that these mutants are deficient in transferase activity. In contrast, the HPLC profiles of mutants GDE D1086N and D1147N showed that in addition to the peak corresponding to the substrate Glc 5 -␤-CD, products of various branched cyclodextrins, Glc-␤-CD, Glc 2 -␤-CD, Glc 3 -␤-CD, Glc 7 -␤-CD, and Glc 8 -␤-CD were also detected (Fig. 2b, C-M), inferring that an active transferase is expressed in mutants GDE D1086N and D1147N. In addition, the products of the reaction mixture containing these mutant enzymes showed no glucose or ␤-CD (Fig. 2b, C-M), indicative of glucosidasedeficient mutant GDEs.
Using Glc-␤-CD as a substrate, the glucosidase activity of GDE was also analyzed by HPLC. The transferase-deficient mutant enzymes D535N, E564Q, and D670N indeed gave hydrolysis products of glucose and ␤-CD from the substrate Glc-␤-CD similar to the wild-type enzyme (Fig. 3a), whereas the glucosidase-deficient mutant enzymes D1086N and D1147N failed to hydrolyze the substrate Glc-␤-CD (Fig. 3b).
The HPLC analyses indicate that the mutants retained either the transferase or the glucosidase activity (Table II), depending on the site of mutation each possessed. However, the question remains whether the retained activity of each mutant varied from that of the wild-type GDE. Therefore, we then compared the kinetic parameters of each mutant and wild-type enzyme.
Kinetic Parameters of Mutant GDEs-The glucosidase activity of the transferase-deficient mutants, D535N, E564Q, and D670N, was measured using Glc-␤-CD as a substrate, whereas the transferase activity of the glucosidase-deficient mutants, D1086N and D1147N, was measured using maltopentaose as a substrate (see "Experimental Procedures"). As shown in Table  III, kinetic parameters for the glucosidase activity of the transferase-deficient mutants were at the same level as that of the wild-type enzyme. In addition, K m values for the transferase  D535N (D564Q, D670N). b, HPLC profile of the sugar products released by the action of mutant enzyme D1086 (D1147N). Symbols are as described in Fig. 2. activity of glucosidase-deficient mutants were also similar to that of wild-type enzyme (D1086N, D1147N, wild-type: 10.0, 10.6, 10.8 mM, respectively). These results indicated that each mutant GDE catalyzes its respective reaction similar to the wild-type, without being influenced by the loss of other functions.

DISCUSSION
The five mutants D535N, E564Q, D670N, D1086N, and D1147N were produced to discriminate the individuality of the two catalytic sites of GDE as earlier proposed (8 -11). The identical purification profiles of these mutant enzymes on ␤-CD-Sepharose-6B affinity column chromatography with the wild-type enzyme indicate that the mutant GDEs have the same binding affinity to ␤-CD, a substrate analogue, as the wild type. These data suggest that the mutation incurred did not influence the conformation of the enzyme. As shown in Table II, the mutant GDEs possessed either the transferase activity or the glucosidase activity. Mutants D535N, E564Q, and D670N exhibit only the glucosidase activity, whereas D1086N and D1147N mutants showed only the transferase activity. Moreover, the kinetic parameters of the mutant enzyme activity were similar to that exhibited by the wild type. It is noteworthy that Asp-535, Glu-564, and Asp-670 are located at the N-terminal half of the amino acid sequence, whereas Asp-1086 and Asp-1147 are at the C-terminal half. We therefore conclude that the active sites of the transferase and glucosidase of the yeast GDE are independent of each other. This study agrees well with the enzyme inhibitor studies and amino acid sequence analysis of rabbit GDE (8 -10) suggesting that the two activities take place at two distinct catalytic sites of the enzyme molecule. Though Teste et al. (22) from their analysis of the yeast GDE deletion mutant suggested recently that the C-terminal half is indispensable for both activities, we believe that this is not the case. Rather, the C-terminal half of GDE might take part only in substrate binding, a function that was lost in the deletion mutant because of possible structural changes brought by deletion.
The Asp-535, Glu-564, and Asp-670 residues proposed to comprise the active site of the transferase of yeast GDE are situated in consensus sequences II, III, and IV of the N-terminal half, respectively, similar to the two Asp and one Glu (Asp-229, Glu-257, Asp-328) residues identified as the catalytic residues of cyclodextrin glucanotransferase, CGTase (23)(24)(25). The Asp-229 of CGTase and Asp-549 of rabbit GDE had been confirmed to work as a catalytic nucleophile of the transferase (26,27). Because Asp-535 of yeast GDE corresponds to Asp-229 of CGTase and Asp-549 of rabbit GDE in the amino acid sequence alignment (Table I), the possibility exists that Asp-535 may play the same role in transferase action. It has been suggested that the GDE transferase is an ␣-retaining glycosidase (9) and likewise a CGTase. Therefore, the transferase of yeast GDE may employ a double displacement mechanism to process ␣-linked glucose polymers as demonstrated by CGTase (28).
Spectrochemical analysis of the glucosidase reaction revealed that the yeast GDE hydrolyzed the ␣-1,6-glucosidic linkage and released a ␤-anomer of glucose from Glc-␤-CD, an inverted configuration of the expected product (data not shown). Similarly, such an ␣-inverting mechanism was also observed with the glucosidase of rabbit GDE (9). In the case of the ␣-inverting glycosidase belonging to the glycosyl hydrolase family 15 (29), the combination of differential labeling and site-directed mutagenesis or the crystal structure analysis identified two carboxyl residues of glucoamylase as the catalytic acid/catalytic base (30 -32). Hence, Asp-1086 and Asp-1147 of the yeast GDE located at the C-terminal half may act as a general acid catalyst or general base catalyst for the glucosidase reaction. However, structural conformation of the active sites may differ between the glucosidase of yeast GDE and glucoamylase, because while the glucoamylase was inactivated by the modification induced by the inhibitor of water-soluble carbodiimide (33)(34)(35), the glucosidase of yeast GDE was not affected (data not shown).
In this study, we clearly demonstrated that the transferase and glucosidase of yeast GDE are well discriminated at two distinct sites of a single polypeptide chain, the transferase located at the N-terminal half and the glucosidase at the Cterminal half. (Fig. 4). Because the catalytic mechanism and primary structural characteristics of the transferase of yeast GDE is similar to the enzymes belonging to the ␣-amylase family, the N-terminal half of the yeast GDE protein may also be folded in the form of a ␣/␤ 8 -barrel structure. On the other hand, though glucosidase at the C-terminal half of yeast GDE exhibited a catalytic mechanism (␣-inverting) in good agree-  Asp1147-Asn Active Inactive a Activity against Glc 5 -␤-CD as a substrate. b Activity against Glc-␤-CD as a substrate. ment with that of glucoamylase, their primary structural characteristics indicate a possible difference of structural conformation between the two types of glucosidases. This study thereby provides clear evidence of the distinctiveness of the active sites of the transferase and glucosidase of yeast GDE and may serve as a basis for the in depth understanding of the structurefunction relationship of bifunctional glycogen debranching enzymes.