Identification of catalytic residues of Ca2+-independent 1,2-alpha-D-mannosidase from Aspergillus saitoi by site-directed mutagenesis.

The roles of six conserved active carboxylic acids in the catalytic mechanism of Aspergillus saitoi 1,2-alpha-d-mannosidase were studied by site-directed mutagenesis and kinetic analyses. We estimate that Glu-124 is a catalytic residue based on the drastic decrease of kcat values of the E124Q and E124D mutant enzyme. Glu-124 may work as an acid catalyst, since the pH dependence of its mutants affected the basic limb. D269N and E411Q were catalytically inactive, while D269E and E411D showed considerable activity. This indicated that the negative charges at these points are essential for the enzymatic activity and that none of these residues can be a base catalyst in the normal sense. Km values of E273D, E414D, and E474D mutants were greatly increased to 17-31-fold wild type enzyme, and the kcat values were decreased, suggesting that each of them is a binding site of the substrate. Ca2+, essential for the mammalian and yeast enzymes, is not required for the enzymatic activity of A. saitoi 1,2-alpha-d-mannosidase. EDTA inhibits the Ca2+-free 1,2-alpha-d-mannosidase as a competitive inhibitor, not as a chelator. We deduce that the Glu-124 residue of A. saitoi 1,2-alpha-d-mannosidase is directly involved in the catalytic mechanism as an acid catalyst, whereas no usual catalytic base is directly involved. Ca2+ is not essential for the activity. The catalytic mechanism of 1,2-alpha-d-mannosidase may deviate from that typical glycosyl hydrolase.

Glycosyl hydrolases are classified as retaining and inverting (4 -6). The inverting reaction occurs via a single displacement mechanism with inversion of anomeric configuration. The reaction usually involves two carboxylic acids. These residues are located ϳ9 Å apart on average. One acts as a general acid catalyst, donating a proton to the glycosidic oxygen of the scissile bond. The other acts as a base catalyst, activating water for nucleophilic attack at the anomeric carbon. The re-taining reaction proceeds via a double displacement mechanism with retention of anomeric configuration. The nucleophilic carboxylic acid residue attacks the glycosyl oxygen. A covalent glycosyl enzyme intermediate is formed and hydrolyzed with general acid-base catalytic assistance. In the retaining enzymes, the two carboxylic acid residues are ϳ5.5 Å apart.
Recently the three-dimensional structures of 1,2-␣-D-mannosidase from Penicillium citrinum were determined by Lobsanov et al. (7). The three-dimensional structures complexed with an inhibitor kifunensin and 1-deoxymannojirimycin demonstrated that both inhibitors bind to the protein at the bottom of the cavity in an unusual 1 C 4 conformation (4C1 is energetically more stable than 1C4). The inhibitor binding did not undergo major conformational changes. Three carboxylic acids (Glu-122, Asp-267, and Glu-409) were potentially involved in the catalytic mechanism. These correspond to Glu-124, Asp-269, and Glu-411 in the A. saitoi enzyme. The four other highly conserved carboxylic residues were also shown (the equivalent A. saitoi numbering is in parentheses): Glu-271 (Glu-273), Glu-412 (Glu-414), Glu-472 (Glu-474), and Glu-502 (Glu-504). These four residues seemed to be too distant from the substrate to be directly involved in catalytic action. They either are buried at the bottom of the active site or interact with atoms on the inhibitor far from the anomeric C-1. A molecule of Ca 2ϩ is located at the bottom of the active site cavity. Glu-271 (Glu-273), Glu-409 (Glu-411), Glu-412 (Glu-414), and Glu-472 (Glu-474) are hydrogen-bonded to the water molecules that coordinate the Ca 2ϩ . The distance of the Ca 2ϩ and an inhibitor/ substrate suggests that they can directly interact with each other. Moreover, one of the candidates for catalytic residues, Glu-409 (Glu-411), coordinates the Ca 2ϩ via a water molecule. The Ca 2ϩ is essential for the enzyme activity of the mammalian and yeast 1,2-␣-mannosidases (8 -14). These suggest that Ca 2ϩ may be directly involved in the catalytic reaction of 1,2-␣-Dmannosidases. The amino acid sequence of A. saitoi 1,2-␣-Dmannosidase (15) is 70% identical with that of P. citrinum enzyme (16), and their substrate specificity to a high mannose type oligosaccharide is the same (2,3,17). Structures of the active center are believed to be conserved among A. saitoi and P. citrinum enzymes because of their high homology. Although we previously demonstrated the probable roles of these acidic residues (18), their catalytic residues have not yet been determined experimentally.
In this study, we performed site-directed mutagenesis to determine the functional role of catalytic residues in the 1,2-␣-D-mannosidase from A. saitoi. Ca 2ϩ , potentially involved in the catalysis, was also analyzed with atomic absorption spectrophotometry. To probe the roles of the three active site carboxylic acids, Glu-124, Asp-269, and Glu-411, in greater detail, we generated the alanine mutants of these residues and assayed * 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. these enzymes. Rescue of activity by external anion was expected to provide valuable insight into the identity of the general base catalyst as well as confirmation of the identity of the general acid catalyst (19 -24).
Enzyme Assay-The purified wild type and mutant enzymes were assayed for 1,2-␣-D-mannosidase activity with Man-␣1,2-Man-OMe as a substrate. Mannose from Man-␣1,2-Man-OMe released by the enzymic reaction was stained by the Somogyi-Nelson (28,29) method. One katal of 1,2-␣-D-mannosidase activity was defined as the amount of enzyme required to liberate 1 mol of mannose from Man-␣1,2-Man-OMe per second at 30°C and pH 5.0.
The initial rates of 1,2-␣-D-mannosidase activity were determined for a Man-␣1,2-Man-OMe substrate at 30°C and pH 5.0 in 50 mM acetate buffer using 8 substrate-concentrations ranging from 0.5ϫ to 2ϫ K m . Values of the catalytic coefficient (k cat ) and Michaelis constant (K m ) values were determined by fitting initial rates as a function of substrate concentration to the direct linear plot of Cornish-Bowden.
Chemical Rescue Methodology-Activities of E124A, D269N, and E411A 1,2-␣-D-mannosidase mutants catalyzing Man 6 GlcNAc 2 -PA to Man 5 GlcNAc 2 -PA in the presence of external nucleophilic anion at pH 5.0 were analyzed by HPLC analysis. Ten pmol of Man 6 GlcNAc 2 -PA (Takara Shuzo Co., Kyoto) with or without 3 M sodium azide was added to the enzyme solution. After 72 h at 30°C, the reaction mixture was stopped by heating to 100°C for 3 min and analyzed on HPLC using TSKgel Amide-80 column (4.6 ϫ 250 mm, Tosoh Corp., Tokyo). The solvent and elution conditions were described by Kondo et al. (30).
Ca 2ϩ Binding Treatment-The purified recombinant A. saitoi 1,2-␣-D-mannosidase was used as a Ca 2ϩ -free wild type enzyme. NaCl was added to the purified enzyme solution in the final concentration of 1 M. The solution was dialyzed against 10 mM sodium acetate buffer, pH 5.0, containing 50 M CaCl 2 for 16 h at 4°C and then dialyzed against the same buffer without CaCl 2 . To analyze the content of Ca 2ϩ binding to the enzyme, the solution was applied to atomic absorption spectrophotometry (Shimadzu model AA-660 (P/N206 -1000-02) apparatus).
Thermal Stability Studies-The stability of the Ca 2ϩ -free and Ca 2ϩbound 1,2-␣-D-mannosidase against thermal-induced unfolding was studied. The unfolding transition curves were obtained by a circular dichroism measurement done in a Jasco J-720 spectropolarimeter. The protein solution was dialyzed against 50 mM formate-KOH buffer, pH 4.0, then diluted in the same buffer to adjust the A 280 to 0.2-0.3. Path length of the optical cuvette was 2 and 10 mm for the measurement of the far and near UV regions, respectively. The fraction of protein in the unfolding state (f U ) was determined using the equation where T is the ellipticity at temperature T, and N and D represent the ellipticity for the native and denatured protein, respectively.

Specific Activities of Mutant and Wild Type Enzyme-Seven
carboxylic acid residues to be studied were constructed in the cloned A. saitoi 1,2-␣-D-mannosidase gene (msdS) by site-directed mutagenesis. Recombinant enzymes were purified homogeneously on SDS-PAGE gels (Fig. 1A) and identified as A. saitoi 1,2-␣-D-mannosidase by Western blot analysis (Fig.  1B). Purified forms of all recombinant enzymes were assayed by the Somogyi-Nelson method (28,29). No activity was detected under standard assay conditions for any of the mutant enzymes except E504Q mutant, and the specific activity could only be determined after extended incubation, typically at 60 -90 min (Table I). From Table I, it appears obvious that the mutations generated resulted in enzymes with significantly reduced ability to hydrolyze Man-␣1,2-Man-OMe, as their specific activity is at least 10 3 -fold less than that of the wild type enzyme. The specific activities of D269N and E411Q mutants in particular were not detected completely. The results of several experiments suggest that there is no significant contamination by wild type 1,2-␣-D-mannosidase in the low activity of mutant enzymes (D269N and E411Q).
Kinetic Analysis-Kinetic characterization of E124Q, E124D, D269E, E273D, E411D, E414D, and E474D mutants and wild type 1,2-␣-D-mannosidase was performed at pH 5.0 using Man-␣1,2-Man-OMe as a substrate. Results of these kinetic experiments are shown in Table II. Kinetic parameters for D269N and E411Q mutants could not be measured, since the reaction rates were below the detectable limit. The k cat value for E124Q mutant was 0.0078 s Ϫ1 , almost 0.3% that of the wild type enzyme. E124D mutant enzyme also decreased the k cat value (0.9% of wild type enzyme). The K m values of E124Q and E124D mutant enzymes were essentially unchanged. Compared with wild type 1,2-␣-D-mannosidase, the D269E and E411D mutant enzymes showed 17-and 28-fold increases in K m values, respectively, whereas the k cat values were slightly lower than that of wild type enzyme. The K m values of E273D, E414D, and E474D mutant enzymes were greatly increased to 23-31-fold that of wild type enzyme, and the k cat values were decreased (1-15% of wild type enzyme). Fig. 2 shows the pH dependence of the reaction rate for Man-␣1,2-Man-OMe hydrolysis for wild type 1,2-␣-Dmannosidase and E124A and E124D mutant enzymes. The optimum pH of E124D and E124A mutant enzymes shifted from 5.0 to 4.0 and 4.5, respectively. The activity of E124A mutant enzyme was found to be lower than that of wild type enzyme over the whole pH range studied, but the effect was more pronounced for the basic limb.

pH-Activity Profiles of E124A and E124D Mutant 1,2-␣-D-Mannosidase-
Effects of an External Nucleophile on Alanine Mutant Enzymes-Mutation (Glu or Asp to Ala) at the general base catalyst creates a cavity in the active site that can accommodate a small external nucleophile. Sodium salts of azide at 3 M were used as external nucleophiles in the enzyme-catalyzed hydrolysis of Man 6 GlcNAc 2 -PA sugar chain. The reaction mixture was analyzed by HPLC (Fig. 3). The activity of wild type 1,2-␣-D-mannosidase was inhibited by a high concentration of azide. The E124A, D269N, and E411A mutant enzymes were inactive in the absence of azide, as a peak corresponding to a Man 5 GlcNAc 2 -PA was not detected. In the presence of azide, these mutant enzymes could not be reactivated.
Atomic Absorption Spectrophotometric Analysis and Kinetic Analysis of Wild Type 1,2-␣-D-Mannosidase-Concentrations of Ca 2ϩ were determined by atomic absorption spectrophotometric analysis. The purified recombinant wild type 1,2-␣-D-mannosidase almost completely did not contain Ca 2ϩ (data not shown). Other divalent metal cations including Mg 2ϩ , Mn 2ϩ , Co 2ϩ , Cu 2ϩ , and Zn 2ϩ were also not detected. After Ca 2ϩ treatment (see "Experimental Procedures"), Ca 2ϩ content was determined as 0.9 mol/mol of wild type 1,2-␣-mannosidase. Ki- netic parameters of Ca 2ϩ -free and Ca 2ϩ binding 1,2-␣-Dmannosidase were determined (Table III); the k cat and K m values were little affected by the Ca 2ϩ binding.
EDTA Inhibition-EDTA inhibited the Ca 2ϩ -free wild type 1,2-␣-D-mannosidase. The inhibition was investigated by fixed concentrations of EDTA in the assay buffer and by varying the concentrations of substrate. A Lineweaver-Burk plot showed a pattern consistent with competitive inhibition (Fig. 4) and gave a K i of 0.91 mM for EDTA.
Effect of Ca 2ϩ Binding on 1,2-␣-D-Mannosidase Thermal Stability-The spectra of Ca 2ϩ -free and Ca 2ϩ -bound 1,2-␣-D-mannosidase at far and near UV region were consistent with each other (data not shown). The Ca 2ϩ binding did not undergo a large conformational change. Thermally induced loss of secondary and tertiary structure was monitored at 222 and 291 nm, respectively (Fig. 5). T m , the temperature at which half of the molecules are unfolded, for Ca 2ϩ -free 1,2-␣-D-mannosidase was 56 and 58°C at 222 and 291 nm, respectively. The Ca 2ϩ binding contributed to the thermal stability increasing T m to 58 and 62°C at 222 and 291 nm, respectively.

Identification of Catalytic Amino Acid
Residues-Drastic reductions in the values of the catalytic coefficient (k cat ) were observed in the E124Q and E124D mutant enzymes (Table II). This may result from significant impairment of the chemical steps in the catalytic reaction. The K m value was essentially unchanged from wild type enzyme. It is, thus, clear that Glu-124 participates in glucosidic bond hydrolysis. E124D shifted the optimum pH to be more acidic (Fig. 2). This result is shown by the lower pK a of Asp than that of Glu and supports the assumption that Glu-124 is directly involved in the catalytic mechanism. It could not be determined whether the Glu-124 residue is a carboxylate group (COO Ϫ ) or carboxyl group (COOH). The pH dependence of E124A affected the basic limb, which was generally assumed to be due to the general acid catalyst. An external nucleophile sodium azide did not reactivate the E124A mutant enzyme (Fig. 3). Thus the Glu-124 may act as an acid catalyst. The mechanism of the all carbohydrate hydrolases usually involves a pair of carboxylic acid residues. In P. citrinum 1,2-␣-D-mannosidase, two of the three residues (Glu-124, Asp-269, and Glu-411 in A. saitoi enzyme) are believed to be potentially catalytic residues (7). Glu-124 was suggested to act as an acid catalyst as discussed above. The most likely candidates for the other catalytic residues are Asp-269 and Glu-411. However, D269E and E411D did not show a considerable decrease in the k cat values (Table II). Recently, a base catalyst of an inverting ␤-amylase (EC 3.2.1.2) from Bacillus cereus var. mycoides was determined by the chemical rescue methodology for a catalytic site mutant (E367A) by azide (19). In the case of A. saitoi 1,2-␣-D-mannosidase an external nucleophile did not reactivate the D269N and E411A mutant enzymes (Fig. 3). Neither Asp-269 nor Glu-411 residues can be a base catalyst in the normal sense. Although the D269N and E411Q mutant enzymes were catalytically inactive, the D269E and E411D mutant enzymes showed considerable activity. The results indicate that the carboxyl groups at these positions may be required for the 1,2-␣-D-mannosidase activity. Structural and modeling studies on cellobiohydrolase Cel6A (EC 3.2.1.91) from Trichoderma reesei suggest that the catalytic mechanism may not directly involve a catalytic base (31). The catalytic mechanism of 1,2-␣-D-mannosidase deviates from the typical glycosyl hydrolase.
Role of Glu-273, Glu-414, and Glu-474 Residues-E273D,     (Table II). The reduction of the k cat value is assumed to be due to the destabilization of the sugar ring. It has become increasingly clear that ring distortion at the Ϫ1 site is crucial in the catalytic mechanism of cellobiohydrolase (32,33). The catalytic reaction of ␣-amylase from Pseudomonas stutzeri involves three acidic residues; two residues act as a proton donor and a proton acceptor. The other residue works to tightly bind the substrate, giving a twisted and deformed conformation of the glucose ring at position Ϫ1 (34). Glu-273, Glu-414, and Glu-474 may bind the substrate to change the conformation of the mannose ring into more reactive one. The three-dimensional structure of P. citrinum 1,2-␣-D-mannosidase indicates that Glu-271, Glu-412, and Glu-472 residues (Glu-273, Glu-414, and Glu-474 in A. saitoi enzyme) in the active site are located at the bottom of the cavity and are not directly involved in the catalytic mechanism. It is demonstrated that these residues are hydrogen-bonded to the Ca 2ϩ via water molecules, and the Ca 2ϩ is involved in the stability of a substrate. But the recombinant wild type A. saitoi 1,2-␣-Dmannosidase contained no Ca 2ϩ and other divalent metal cations. Despite the Ca 2ϩ having no effect on the 1,2-␣-mannosidase activity (Table III), its binding sites were conserved in A. saitoi 1,2-␣-D-mannosidase. This also indicates that the drastic decrease of the activities of E273D, E414D, and E474D mutants (Table IV) did not parallel the loss of the Ca 2ϩ binding ability. These residues are involved in the substrate binding sites.
Ca 2ϩ Binding-It was reported earlier that the recombinant A. saitoi 1,2-␣-D-mannosidase has a molecule of Ca 2ϩ (25). In this study, the recombinant wild type A. saitoi 1,2-␣-D-mannosidase contained no Ca 2ϩ . The purified wild type enzyme did not bind Ca 2ϩ with an addition of CaCl 2 (data not shown). We found that the Ca 2ϩ -free enzyme bound 1 mol of Ca 2ϩ per 1 mol of protein after the addition of a high concentration of NaCl. It was also demonstrated that the Ca 2ϩ binding site is conserved (Table IV). 1,2-␣-Mannosidase generally requires a divalent metal cation for the activity (13,14,(35)(36)(37). In the case of A. saitoi 1,2-␣-D-mannosidase, the Ca 2ϩ had no effect on either the k cat or the K m value (Table III). Thus, it is clear that Ca 2ϩ is not essential for the activity of A. saitoi 1,2-␣-D-mannosidase, although the Ca 2ϩ binding site is conserved. EDTA was shown to behave as a competitive inhibitor of A. saitoi 1,2-␣-D-mannosidase (Fig. 4); this indicates that the enzyme contains no divalent metal cation. 1-Deoxymannojirimycin inhibits 1,2-␣mannosidase as a substrate analog. However, EDTA is not a substrate analog. Tris is a competitive inhibitor for 1,2-␣-mannosidase from rabbit liver (14). Tris and other buffers containing primary hydroxyl groups substantially decreased its activity (38,39). Carboxyl groups of EDTA may have an affinity for the active site of A. saitoi 1,2-␣-D-mannosidase. A molecule of Ca 2ϩ contributed to thermal stability (Fig. 5). The role of Ca 2ϩ for A. saitoi 1,2-␣-D-mannosidase is to protect the enzyme against thermal denaturation, and it was observed to have the same role in Saccharomyces cerevisiae-processing 1,2-␣-man-nosidase (12). Although Ca 2ϩ is essential for S. cerevisiae enzyme activity (12,37), A. saitoi 1,2-␣-D-mannosidase did not require Ca 2ϩ for the activity (Table III). This is the first report demonstrating that 1,2-␣-D-mannosidase requires no divalent metal cation for its activity.