INTRODUCTION

The hyperthermophilic archaeon Pyrococcus furiosus is an obligately anaerobic heterotroph, which grows optimally at 98-100 °C (1). It employs a fermentative type of metabolism (2), using polysaccharides, such as starch, glycogen, and pullulan (3), or disaccharides, such as maltose (3) and cellobiose (4), as carbon and energy sources. In order to utilize the different carbohydrates, P. furiosus synthesizes several intracellular and extracellular glycosyl hydrolases. Specifically, -amylase (5), amylopullulanase (6), -glucosidase (7), and -glucosidase (4) activities have been purified and characterized. The -amylase, amylopullulanase, and -glucosidase are believed to work cooperatively to degrade -linked polysaccharides, such as starch, glycogen, or pullulan (8). The endo-acting, -specific amylase and amylopullulanase degrade -linked polysaccharides to di- and trisaccharides (5, 6). -Glucosidase presumably hydrolyzes these shorter oligosaccharides to glucose for use in a novel Embden-Meyerhof pathway (8, 9). Although P. furiosus cannot grow on cellulose or carboxymethylcellulose (4), it is not clear whether P. furiosus can utilize other -linked complex carbohydrates as growth substrates. To date, no endo-acting, -specific glycosyl hydrolases, such as cellulases, xylanases, or mannanases, have been identified in P. furiosus. However, when P. furiosus is grown on 5 mM cellobiose, a cell density of 7 × 10⁸ cells/ml has been reported (4). Apparently, cellobiose is transported across the cell membrane and hydrolyzed to glucose by the intracellular -glucosidase (9). Thus, the - and -glucosidases may play similar roles in the degradation of polysaccharides for the nutritional requirements of P. furiosus.

In addition to the physiological role of these glycosyl hydrolases within P. furiosus, it is also interesting to examine their relationship to similar enzymes from the other domains of life. This can be done on the basis of substrate specificity. However, many glycosyl hydrolases have a broad range of specificities (10). Henrissat (10) proposed an alternate and complementary classification scheme for glycosyl hydrolases based on amino acid sequence similarities. For example, glycosyl hydrolase family 1 is composed of exo-acting, -specific enzymes with similar amino acid sequences. Based on substrate specificity, enzymes in this family have been characterized as -glucosidases (EC), -galactosidases (EC), phospho--gluco/galactosidases (EC/85), lactase-phlorizin hydrolases (EC/62), and thioglucosidases (EC). Family 1 glycosyl hydrolases provide a favorable framework for comparative studies of mesophilic and thermophilic enzymes for a number of reasons. First, the enzymes in this family function over a wide range of temperatures from mesophilic (11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 24, 25, 26) to moderately thermophilic (27, 28, 29, 30, 31) to hyperthermophilic (4, 32, 33). Second, enzymes in this family have been isolated from all three domains (bacteria, eucarya, and archaea), allowing the analysis of possible evolutionary relationships. Finally, crystal structures have been determined for some family 1 enzymes (34, 35), facilitating structural comparisons among these enzymes.

For this report, crude cell extracts of P. furiosus were examined for the presence of exo-acting glycosyl hydrolases. In addition to the -glucosidase (7) and -glucosidase (4) reported previously, a -mannosidase activity was isolated and characterized in relation to the other glycosyl hydrolases of P. furiosus. In order to investigate the molecular basis for substrate specificity differences between the -mannosidase and the -glucosidase, the genes for both of these enzymes were isolated from a genomic library by expression screening. A search for homology between the deduced amino acid sequences of the -mannosidase and the -glucosidase and other glycosyl hydrolases was completed. Based on the relative catalytic efficiencies of the two enzymes, it is likely that they play different physiological roles in P. furiosus. Several possible functions for the -mannosidase are discussed.

MATERIALS AND METHODS

Purification of -Mannosidase from P. furiosus

P. furiosus (DSM 3638) was grown on maltose-based medium in a 600-liter batch fermentor, and cell-free extract was prepared as described previously (36, 37). All purification steps were carried out at room temperature using an FPLC system (Pharmacia Biotech Inc.). The purification protocol for -mannosidase was as follows.

The cell-free extract was applied directly to a column (10 × 20 cm) of DEAE-Sepharose (Pharmacia). After washing the column with 7 liters of buffer (50 mM Tris/HCl, pH 8, containing 2 mM sodium dithionite, 10% glycerol (v/v)), the adsorbed proteins were eluted with a linear gradient from 0 mM to 115 mM NaCl (90 ml) and 115 mM to 376 mM NaCl (5000 ml) at 12 ml/min. -Mannosidase activity eluted between 264 mM and 288 mM NaCl. These fractions were pooled and concentrated using an Amicon Ultrafiltration Cell 202 with a YM10 membrane (Amicon, Beverly, MA) and a pressure of 55 p.s.i.g.

The concentrated fractions from the previous column were equilibrated to 50 mM sodium phosphate buffer, pH 7.0, containing 243 g/liter ammonium sulfate (buffer A). About 10% of the equilibrate pool (268 ml) was applied to a column (5 cm × 50 cm) of phenyl-Sepharose 650 M (Toso Haas, Montgomeryville, PA) previously equilibrated with buffer A. The column was washed with 600 ml 100% buffer A followed by 1000 ml of 25% buffer A. The remaining adsorbed proteins were eluted with a 2000-ml linear gradient from 25% to 0% buffer A. -Mannosidase activity eluted at 0% buffer A. Fractions containing -mannosidase activity were combined, concentrated as described above, and equilibrated to 25 mM potassium phosphate buffer, pH 7.0.

The concentrated pool from the previous column was applied at 10 ml/min to a column (5 cm × 30 cm) of hydroxyapatite (Calbiochem, Sunnyvale, CA) previously equilibrated with 25 mM potassium phosphate buffer, pH 7.0. After the column was washed with 900 ml of 25 mM potassium phosphate, adsorbed proteins were eluted with a 600-ml linear gradient from 25 mM to 100 mM, a 400-ml linear gradient from 100 mM to 250 mM, and a 300-ml linear gradient from 250 mM to 500 mM potassium phosphate buffer, pH 7.0. -Mannosidase activity eluted between 110 and 135 mM potassium phosphate buffer. Active fractions were pooled, concentrated, and equilibrated to 100 mM sodium phosphate buffer, pH 7.0.

The concentrated pool from the previous column was applied to a Pharmacia HiLoad 16/60 Superdex 200 gel filtration column (V_o = 39.3 ml; V_t = 120.6 ml) pre-equilibrated to 100 mM sodium phosphate buffer, pH 7.0. The column was developed at 0.5 ml/min. -Mannosidase activity eluted as a symmetrical peak at 75.0 ml.

Enzyme Assays

-Mannosidase activity was assayed routinely using 1.0 mM para-nitrophenol (pNp)¹ substrate equilibrated to the desired temperature and pH. After equilibrating the sample to the desired temperature and pH in a heat block containing silicone oil (Dow Corning Corp., Midland, MI), the reaction was initiated by adding equilibrated substrate to the sample in an Eppendorf tube (U.S.A. Scientific Products, Milton Keynes, United Kingdom). The reaction was stopped at 5 min by chilling on ice. After cooling, the liquid in the Eppendorf tube was transferred to a microtiter plate (Corning, Corning, NY). The increase in absorbance at 405 nm as a result of pNp liberation was measured using an EL 340 Microplate Bio-Kinetics Reader (Bio-Tek^TM Instruments, Winooski, VT). All activities were corrected for thermal degradation of the pNp substrate, which was below 0.5% of the enzyme hydrolysis rate. Absorbances were converted to concentrations of pNp using standards of known concentration. All assays were performed in triplicate. One unit of glycosidase activity was defined as that amount of enzyme required to catalyze the release of 1.0 µmol of pNp/min.

The purified enzyme was also tested for amylase activity. A standard reaction mixture contained 17 g/liter soluble starch equilibrated to the desired temperature and pH. The procedure was the same as above. The reaction was followed using the method of Laderman et al. (5). One unit of amylase activity was defined as that amount of enzyme hydrolyzing 1 mg of starch/min. All assays were performed in triplicate.

Total Protein Assays

Total protein concentration was determined using a BCA protein assay reagent kit (Pierce). Samples were diluted to the linear range (where A₅₉₅ = 0.1-1.0) and incubated with reagent at 50 °C for 30 min in a sealed microtiter plate. The absorbance at 595 nm was determined using an EL 340 Microplate Bio-Kinetics Reader with albumin as the standard (Sigma).

Electrophoresis and Activity Staining

Isoelectric focusing was carried out on a Phast System (Pharmacia), according to manufacturer's protocols. Native- and SDS-PAGE were performed using standard procedures (38). High molecular mass (Boehringer Mannheim) and broad pI standards (Pharmacia) were used for PAGE and isoelectric focusing, respectively. For -mannosidase activity staining, non-fixed gels were incubated at 95 °C for several minutes in 100 mM sodium phosphate buffer, pH 7.0, containing 1.0 mM ManpNp. Upon the appearance of a yellow band, the gel was marked at that location.

Estimation of Molecular Mass

The P. furiosus -mannosidase was treated with the bifunctional reagent dimethyl suberimidate (Sigma) according to Davies and Stark (39). The homogeneous enzyme (3.0 mg/ml) in 200 mM triethanolamine/HCl, pH 8.5, was mixed (in various proportions) with dimethyl suberimidate (1 mg/ml) in the same buffer and incubated for 3 h at 25 °C. The reaction was stopped, and the proteins were denatured by incubation at 90 °C for 60 min in the presence of 1% SDS and 1% -mercaptoethanol as described by Pisani et al. (33) and subjected to 10% SDS-PAGE. Aldolase (Boehringer Mannheim) was used as a cross-linking control as described by Pisani et al. (33).

Kinetic Constants and Substrate Specificity

Kinetics parameters of P. furiosus -mannosidase were determined using standard reaction mixtures, containing either ManpNp or GlcpNp. The reactions were performed at 95 °C. The release of pNp was measured as described above using different initial concentrations of substrate (0.05-10 mM). All assays were performed in triplicate. Values for the maximal reaction velocity (V_max) and the Michaelis-Menten constant (K_m) were determined from Lineweaver-Burk plots. Substrate specificity was determined using the standard reaction mixture, except that alternate substrates to ManpNp were used. Depending on the substrate, either the amount of pNp released or the amount of starch degraded was measured after a 5-min incubation at 95 °C.

Thermostability

For thermostability determination, the homogeneous enzyme was incubated in Eppendorf tubes submersed in oil baths, at 90, 100, and 110 °C. The samples were covered with Ampliwax (Perkin-Elmer) to prevent evaporation. At appropriate time intervals, aliquots were withdrawn and tested for -mannosidase activity at 95 °C in a standard reaction mixture.

Sequencing

All DNA sequencing reactions were performed using either the Perkin-Elmer Applied Biosystems dye primer or dye terminator cycle sequencing kits and a model 377 automated DNA sequencer (Perkin-Elmer). Sequences were aligned and edited using the program Sequencher (Genecodes, Ann Arbor, MI).

Purified -mannosidase and -glucosidase were denatured and run on 12.5% polyacrylamide using standard procedures (38). Protein was electroblotted to a polyvinylidene difluoride membrane and Ponceau S-stained (38). N-terminal Edman degradation was performed on single bands with approximate molecular masses of 60 and 58 kDa for the -mannosidase and -glucosidase, respectively, using a liquid phase sequencer (Applied Biosystems model 477).

Expression Screening

The F factor Fkan from E. coli strain CSH118 (40) was introduced into the phophnlac E. coli strain BW14893 (41). A library prepared from randomly sheared genomic P. furiosus DNA was obtained from M. Snead (Recombinant BioCatalysis, Inc., La Jolla, CA) and was introduced into BW14893 Fkan. Cells were plated on 100-mm LB plates containing 100 µg/ml ampicillin, 80 µg/ml methicillin, and 1 mM isopropyl -D-thiogalactopyranoside at a density of greater than 1000 colonies/plate. Colony lifts were performed using Millipore HATF membrane filters. Transformation plates were returned to the 37 °C incubator after the filter-lift to regenerate colonies. The transferred colonies were lysed with chloroform vapor in 150-mm glass Petri dishes. The filters containing lysed colonies were transferred to 100-mm glass Petri dishes containing Whatman 3MM filter paper saturated with Z buffer (40) and either 1 mg/ml X-glu (Diagnostic Chemicals Ltd, Oxford, CT) or 1 mg/ml X-gal (ChemBridge Corp., Northbrook, IL). The dishes were incubated at 80-85 °C. ``Positives'' were observed as blue spots on the filter membranes. Approximately 20-30 positives/plate were observed. One positive clone from the X-gal assay was purified by restreaking cells from the original regenerated plate. Several other positives from both assays were recovered by transforming DNA isolated from the blue spots on the filter lifts into electrocompetent E. coli DH10B cells. The filter-lift assay was repeated on transformation plates to identify ``positives.'' LB medium containing 100 µg/ml ampicillin was inoculated with repurified positives and incubated at 37 °C overnight. Plasmid DNA was isolated from these cultures,and the plasmid insert was sequenced. The partial sequences of three clones (two from X-glu, one from X-gal) that contained inserts revealed that two of the clones overlapped (one from X-gal, one from X-glu) and one was unique (X-glu).

RESULTS

Purification of -Mannosidase

Fractions from DEAE-Sepharose Fast Flow anion-exchange chromatography were assayed for -amylase, -glucosidase, -glucosidase, and -mannosidase activities (Fig. 1). Three peaks of -glycosidase activity eluted at 0, 0.26-0.28, and 0.33-0.36 M NaCl, respectively. The peaks at 0 and 0.33-0.36 M NaCl had an identical substrate specificity to that reported for the -glucosidase (4). The peak of -glycosidase activity that eluted between 0.26 and 0.28 M NaCl showed different relative specific activities on several aryl glycosides than the previously reported glycosyl hydrolases from P. furiosus (4, 5, 6, 7). The purification procedure for this -glycosidase activity is shown in Table I.

Table I. Purification of the -mannosidase from P. furiosus -Mannosidase activity was determined at 95 °C, using 1.0 mM ManpNp as substrate. Purification step Volume Protein Specific activity Total activity -Fold Recovery ml mg/ml units/mg units % Cell extract 134 15.1 0.044 89.0 1 100 DEAE 268 1.75 0.058 27.2 1.3 31 HIC 27 1.25 0.62 20.9 14 24 HAP 1.3 0.68 5.3 4.69 120 5.3 Superdex 200 1.22 0.07 31.1 2.66 700 3.0

The purified enzyme was tested for its substrate specificity. Table II shows the activity of the enzyme toward several aryl-glycosides. The substrate specificities of the P. furiosus -glucosidase and -glucosidase are reported for comparison. The new enzyme exhibited the highest specific activity with ManpNp as substrate and, therefore, was characterized as a -mannosidase. The -mannosidase did not hydrolyze the -glycosidic linkages of GlcpNp or GalpNp, nor did it degrade starch.

Table II. Substrate specificities of purified -mannosidase, -glucosidase, and -glucosidase from P. furiosus Activity was determined by measuring the release of pNp by absorbance at 405 nm. pNp-sugars were used at 1 mM concentrations. Assays of -mannosidase-specific activity were performed at 95 °C in 100 mM sodium phosphate buffer, pH 7.0. Assays of -glucosidase-specific activity were performed at 90 °C in 100 mM sodium citrate buffer, pH 5.5 (4). Assays of -glucosidase-specific activity with GlcpNp were performed at 108 °C in 100 mM sodium phosphate buffer, pH 5.5 (7), and with ManpNp, GlcpNp, GalpNp, and XylpNp at 95 °C in 100 mM sodium phosphate buffer, pH 7.0.  Substrate  -Mannosidase  -Glucosidase  -Glucosidase Specific activity Relative activity Specific activity Relative activity Specific activity Relative activity units/mg % units/mg % units/mg % ManpNp 31.1 100 16 3.6 0 0 GalpNp 19.2 61.8 153 34.3 0 0 GlcpNp 1.4 4.7 446 100 0 0 XylpNp 1.3 4.2 41 9.2 0 0 GlcpNp 0 0 0 0 287 100

The purified enzyme displayed optimal activity at 105 °C (Fig. 2) and pH 7.4 (Fig. 3). An isoelectric point of 6.9 was determined from an activity-stained isoelectric focusing gel (data not shown). The molecular mass of denatured -mannosidase was approximately 60 kDa as determined from SDS-PAGE. When the -mannosidase was treated with dimethyl suberimidate (at the enzyme/bifunctional reagent molar ratio of 1:300), four protein bands were noted after SDS-PAGE, corresponding to molecular masses of 60, 140, 180, and 220 kDa (Fig. 4). These results indicate that the P. furiosus -mannosidase in its native conformation is a tetramer consisting of four identical subunits similar to the S. solfataricus glycosidases (32, 33) and the P. furiosus -glucosidase (4), which were also determined to be homotetramers.

Effect of temperature on the activity of purified -mannosidase from P. furiosus.

Effect of pH of the activity of purified -mannosidase from P. furiosus.

The rate dependence on substrate concentration followed Michaelis-Menten kinetics. From Lineweaver-Burk plots, K_m and V_max values of 0.79 mM and 31.1 units/mg were determined with ManpNp as substrate (Table III). In addition, the same analysis with GlcpNp as substrate was used to determine K_m and V_max values of 2.9 mM and 14.8 units/mg. Assuming that the smallest catalytic unit of the -mannosidase was 1 monomer unit with a molecular mass of 60 kDa, turnover numbers (k_cat values) of 40 and 5.3 s¹ were calculated for ManpNp and GlcpNp, respectively. Table III provides a comparison between the kinetic rate constants of the -mannosidase and -glucosidase from P. furiosus. The -mannosidase had a significantly lower catalytic efficiency for the hydrolysis of -glycosidic bonds than the -glucosidase. The -mannosidase had a higher K_m and lower V_max with GlcpNp as substrate than with ManpNp, indicating both a more specific binding and a more efficient cleavage of the glycosidic linkage when mannose is the terminal, non-reducing moiety.

Table III. Kinetic constants for the -mannosidase and -glucosidase from P. furiosus Lineweaver-Burk plots were used to determine the kinetic constants. Activity of the -mannosidase was determined using different initial concentrations of ManpNp and GlcpNp in 100 mM sodium phosphate, pH 7. The amount of pNp released was measured following a 5-min incubation at 95 °C. Activity of -mannosidase was determined at 95 °C, and activity of the -glucosidase was determined using GlcpNp in 100 mM sodium citrate, pH 5.5, at 90 °C (4). ND, kinetic analysis with this substrate was not performed. Kinetic parameter Substrate  -Mannosidase  -Glucosidase Km (mM) GlcpNp 2.93 0.15 ManpNp 0.79 ND Vmax (units/mg) GlcpNp 14.8 700 ManpNp 31.1 25.2a kcat/Km (s1 mM1) GlcpNp 5.3 4,500 ManpNp 40 ND a  Calculated as 3.6% of Vmax for GlcpNp (4).

The thermostability of purified -mannosidase was measured at 90 °C and 100 °C (Fig. 5). At 90 °C, the -mannosidase showed appreciable thermostability, with almost no loss of activity after 24 h and a half-life of greater than 60 h. At 100 °C, the half-life diminished to 77 min. At 110 °C, however, the enzyme had a half-life of less than 15 min (data not shown).

Thermostability of P. furiosus -mannosidase at 90 and 100 °C.

The nucleotide and deduced amino acid sequences for the P. furiosus -mannosidase and -glucosidase were determined. The sequence for the -glucosidase was identical to that reported previously (43). The NH₂ termini of the -mannosidase and the -glucosidase purified from P. furiosus were determined by Edman degradation to be MFPEKFLXGVAQXGFQXEMGD and MKFPKNFMF, respectively, and were identical to the deduced amino acid sequences. The deduced amino acid sequences of the -mannosidase and the -glucosidase were 46.5% identical (Fig. 6). The -mannosidase and -glucosidase amino acid sequences were 19.0 and 17.9% identical to the predicted amino acid sequence of the -amylase (42), respectively. In addition, the -mannosidase and -glucosidase NH₂-terminal amino acid sequences shared no homology with the 13 amino acid residues at the NH₂ terminus of the -glucosidase (data not shown) from P. furiosus.

Comparison of the amino acid sequences of the -mannosidase and -glucosidase from P. furiosus.

The deduced amino acid sequences for the -mannosidase and the -glucosidase were similar to the sequences for other glycosyl hydrolases (Table IV). On this basis, both the -mannosidase and the -glucosidase were classified as family 1 glycosyl hydrolases. The -mannosidase and the -glucosidase sequences showed the greatest homology with the -glycosidases from two Sulfolobus solfataricus strains (44, 45). The P. furiosus enzymes were more distantly related to the family 1 glycosyl hydrolases from bacteria and eucarya. Clustering of family 1 glycosyl hydrolases into three groups has previously been reported (31). One group is composed of bacterial and eucaryal -glycosidases, a second group contains the bacterial phospho--glycosidases, and a third group contains the archaeal -glycosidases. The P. furiosus -mannosidase and -glucosidase both contained the conserved active site residues found in other family 1 enzymes. Two conserved carboxylates (residues 210 and 414 of P. furiosus -mannosidase) presumably act as the acid/base and nucleophile, respectively, in the mechanism of glycoside-linkage hydrolysis (31, 46, 47).

Table IV. Comparison of amino acid sequences among family 1 glycosyl hydrolases Identity is determined as the number of identical amino acid residues in each pairwise comparison, expressed as percentage of the total number of amino acid residues compared. The GAP program of GCG (Genetics Computer Group, Inc., Madison, WI) was used with a gap weight of 3.0 and a gap length weight of 0.1. The following abbreviations are used: Pfum, P. furiosus -mannosidase; Pfug, P. furiosus -glucosidase; Sso1, Sulfolobus solfataricus DSM 1616 -galactosidase (44); Sso2, S. solfataricus MT-4 -galactosidase (45); Cs, Caldocellum saccharolyticum -glucosidase A (81); Tm, Thermotoga maritima -glucosidase A (28); Ct, Clostridium thermocellum -glucosidase A (31); Bpa, Bacillus polymyxa -glucosidase A (11); Bpb, B. polymyxa -glucosidase B (11); Lb, Lactobacillus casei phospho--galactosidase (14); Sr, Streptococcus rochei phospho--galactosidase (21); Bn, Brassica napus myrosinase (20); Sa, Sinapsis alba myrosinase (24); Hs, Homo sapiens lactase-phlorizin hydrolase (17); Oc, Oryctolagus cuniculus lactase-phlorizin hydrolase (17); Ec, Escherichia coli phospho--glucosidase (26). Enzyme % Identity Pfum Pfug Sso1 Sso2 Cs Tm Ct Bpa Bpb Bn Sa Hs Oc Lb Sr Ec Pfum 46.5 46.3 45.0 33.7 34.4 31.3 33.6 32.2 29.1 29.8 30.0 31.3 32.4 32.2 25.6 Pfug 53.5 53.9 33.7 34.3 29.6 30.6 30.2 27.7 32.1 26.2 27.0 29.7 32.0 25.0 Sso1 72.0 28.8 32.2 30.9 26.1 22.1 27.1 22.7 29.2 27.7 23.4 30.8 22.7 Sso2 30.1 33.8 31.3 27.2 29.7 29.3 30.0 29.0 27.6 22.6 34.4 33.5 Cs 49.8 51.6 43.3 41.5 36.0 36.4 38.7 36.6 39.5 35.5 35.7 Tm 53.2 46.7 42.9 39.7 38.8 47.4 44.2 37.7 31.7 32.5 Ct 49.6 43.9 36.9 36.3 40.2 37.6 38.4 35.0 34.2 Bpa 44.4 35.4 35.0 37.2 34.8 36.2 31.9 33.9 Bpb 33.9 34.7 36.7 37.6 34.4 33.2 30.1 Bn 91.2 40.3 38.6 32.9 29.6 24.3 Sa 40.1 39.2 32.4 29.7 25.7 Hs 85.5 29.9 30.5 28.1 Oc 28.9 30.3 27.4 Lb 29.3 33.5 Sr 33.9

DISCUSSION

Relative to -glucosidases, there are few reports on the purification and characterization of -mannosidases. The -mannosidases (EC) from bacteria, including Enterococcus casseliflavus (48) and Bacillus sp. AM001 (49), and lower eucarya, including Aspergillus niger (50) and Aspergillus awamori (51), degrade the terminal, non-reducing -mannopyranoside linkages of mannan. The physiological role of the -mannosidases from higher eucarya is different. These enzymes degrade the terminal, non-reducing -mannopyranoside linkages of glycoproteins (52, 53, 54). Genetic disorders associated with -mannosidase deficiency have been described in a number of mammals (55, 56, 57, 58). In humans, the absence of -mannosidase results in the deleterious storage of the disaccharide Man1-4GlcNAc (57, 58, 59). The gene for the bovine -mannosidase has been sequenced (60) and the deduced amino acid sequence is similar to sequences of family 2 glycosyl hydrolases.

The P. furiosus -mannosidase is distinct from the other glycosyl hydrolases from P. furiosus. The substrate specificity and predicted amino acid sequence of the -mannosidase are significantly different from those of the -glucosidase, -glucosidase, and -amylase previously purified from P. furiosus. Based on both substrate specificity and amino acid sequence, the -mannosidase and -glucosidase are the most closely related. Both of these enzymes are exo-acting, -specific glycosyl hydrolases that release the terminal, non-reducing sugars from -glycosidic bonds. The -mannosidase is most active with mannose as the terminal non-reducing sugar, while the -glucosidase has its highest specific activity with glucose in this location (4). This suggests a difference in the way that the two enzymes interact with the hydroxyl on C-2 of the terminal, non-reducing sugar.

Although the presence of two similar enzymes within P. furiosus might appear to be an unnecessary metabolic burden, several other organisms, including some thermophiles, contain the genes for two or more exo-acting, -specific glycosyl hydrolases. For example, Bacillus polymyxa contains two genes, bglA and bglB, encoding different family 1 -glucosidases (11). The deduced amino acid sequences are 44.7% identical, but the enzymes have distinctly different biochemical characteristics. -Glucosidase A is intracellular and cleaves cellobiose through pyrophosphate-mediated hydrolysis, while -glucosidase B is extracellular and cleaves cellobiose without cofactors (11). The thermophilic bacterium Clostridium thermocellum also has two -glucosidase genes, bglA and bglB (31). The proteins encoded by these genes are only 21.7% identical. The sequence for -glucosidase A is similar to family 1 glycosyl hydrolases (31), while -glucosidase B is similar to family 3 glycosyl hydrolases, which include -glucosidases from fungi and rumen bacteria (61). The thermophilic bacterium Thermotoga maritima may have as many as four different exo-acting, -specific glycosyl hydrolases, including a -xylosidase (27), a -galactosidase (28), and possibly two -glucosidases (28). The gene for one -glucosidase (bglA) has been sequenced, and the deduced amino acid sequence is similar to family 1 glycosyl hydrolases. Two -glucosidase-encoding genes have been sequenced from the thermoacidophilic archaeon S. solfataricus. Although these genes may have come from different strains of S. solfataricus (32), they are both homologous to family 1 glycosyl hydrolases.

The high degree of homology between the sequences of the -mannosidase and -glucosidase from P. furiosus suggests that the enzymes may be evolutionarily related. Gene duplication is frequently observed among glycosyl hydrolases (10). It has been proposed that the enzyme produced from the original gene copy would continue hydrolyzing the original substrate, while duplicate gene copies could constitute templates for constructing enzymes with activity directed to a new, but stereochemically similar, substrate (10). The divergence of glycosyl hydrolases to acquire new specificities is not unexpected, given the stereochemical resemblance among pyranoside substrates. It is not clear whether the two enzymes from P. furiosus represent a case of gene duplication, and, if so, which one was the predecessor.

All glycosyl hydrolases are believed to act by a general acid catalysis mechanism in which two amino acid residues participate in the hydrolysis of the glycosidic bond (62). For family 1 glycosyl hydrolases, the two catalytic residues are both glutamic acid residues (positions 210 and 414, P. furiosus -mannosidase numbering) that are strictly conserved (47). The glutamic acid closer to the N terminus functions as the acid/base (i.e. proton donor) (31, 63) and the other glutamic acid acts as the nucleophile (45, 64). It has been suggested that all family 1 glycosyl hydrolases have an 8-fold /-barrel structure (48, 64). Structural and sequence comparisons of family 1 glycosyl hydrolases indicate that the two conserved carboxylates in these enzymes occur at the C-terminal ends of -strands 4 and 7 (47). Similar structure and catalytic residues have been observed for glycosyl hydrolase families 2, 5, 10, 17, 30, 35, 39, and 42 and are the basis for the classification of these families into a superfamily (47, 65).

Some family 1 glycosyl hydrolases also have glycosyl transferase activities (4, 13, 14, 15, 32). The S. solfataricus -glucosidase has been implicated in the glycosylation of membrane lipid components (32). The -glucosidases from both P. furiosus and S. solfataricus have been used for a variety of synthetic glycosyl transferase reactions (68, 69). Similarly, the P. furiosus -mannosidase may be involved in the biosynthesis of intracellular components including proteins, membrane components, or other compounds. Mannose-conjugated glycolipids have been identified in a number of archaea (70, 71, 72). In addition, the major glycolipids of closely related halophilic euryarchaeota are built from a basic diglycoside, Man1-4Glc (72, 73, 74, 75, 76, 77, 78). A novel osmoprotectant, 2-O--D-mannosylglycerate, was recently discovered in both thermophilic bacteria (79) and archaea, including P. furiosus (80). When P. furiosus is grown at supraoptimal salt concentrations, 2-O--D-mannosylglycerate becomes the predominant intracellular solute (80). 2-O--D-Mannosylglycerate may play a role in osmoprotection (80), or it may be an activated precursor in the synthesis of certain membrane glycolipids in response to salt stress. We have begun experiments to determine if the -mannosidase has glycosyl transferase activities and if it is involved in the synthesis of this novel compound.