Discovery of β-1,4-d-Mannosyl-N-acetyl-d-glucosamine Phosphorylase Involved in the Metabolism of N-Glycans*

Background: N-Glycans are metabolized by sequential glycoside hydrolase-catalyzed reactions. Results: A phosphorylase encoded in a gene cluster involved in N-glycan metabolism in the genome of Bacteroides thetaiotaomicron catalyzed reversible phosphorolysis of β-1,4-d-mannosyl-N-acetyl-d-glucosamine. Conclusion: An N-glycan metabolic pathway containing a unique phosphorylase was discovered. Significance: B. thetaiotaomicron efficiently utilizes the energy of ATP via a phosphorylase-dependent metabolic pathway. A gene cluster involved in N-glycan metabolism was identified in the genome of Bacteroides thetaiotaomicron VPI-5482. This gene cluster encodes a major facilitator superfamily transporter, a starch utilization system-like transporter consisting of a TonB-dependent oligosaccharide transporter and an outer membrane lipoprotein, four glycoside hydrolases (α-mannosidase, β-N-acetylhexosaminidase, exo-α-sialidase, and endo-β-N-acetylglucosaminidase), and a phosphorylase (BT1033) with unknown function. It was demonstrated that BT1033 catalyzed the reversible phosphorolysis of β-1,4-d-mannosyl-N-acetyl-d-glucosamine in a typical sequential Bi Bi mechanism. These results indicate that BT1033 plays a crucial role as a key enzyme in the N-glycan catabolism where β-1,4-d-mannosyl-N-acetyl-d-glucosamine is liberated from N-glycans by sequential glycoside hydrolase-catalyzed reactions, transported into the cell, and intracellularly converted into α-d-mannose 1-phosphate and N-acetyl-d-glucosamine. In addition, intestinal anaerobic bacteria such as Bacteroides fragilis, Bacteroides helcogenes, Bacteroides salanitronis, Bacteroides vulgatus, Prevotella denticola, Prevotella dentalis, Prevotella melaninogenica, Parabacteroides distasonis, and Alistipes finegoldii were also suggested to possess the similar metabolic pathway for N-glycans. A notable feature of the new metabolic pathway for N-glycans is the more efficient use of ATP-stored energy, in comparison with the conventional pathway where β-mannosidase and ATP-dependent hexokinase participate, because it is possible to directly phosphorylate the d-mannose residue of β-1,4-d-mannosyl-N-acetyl-d-glucosamine to enter glycolysis. This is the first report of a metabolic pathway for N-glycans that includes a phosphorylase. We propose 4-O-β-d-mannopyranosyl-N-acetyl-d-glucosamine:phosphate α-d-mannosyltransferase as the systematic name and β-1,4-d-mannosyl-N-acetyl-d-glucosamine phosphorylase as the short name for BT1033.

In this study, we identified a gene cluster including a gene encoding a novel GH130 phosphorylase, BT1033, for the energy-efficient metabolism of complex type N-glycans in the genome of the human gut bacterium B. thetaiotaomicron VPI-5482. BT1033 phosphorylase is a key enzyme in a new metabolic pathway of N-glycans, allowing an alternative to the ␤-mannosidase-containing conventional metabolic pathway. BT1033 plays a crucial role to convert the ManGlcNAc-transported from the periplasmic space into the cytoplasm by the major facilitator superfamily transporter-into ␣-D-mannose 1-phosphate (␣-Man1P) and N-acetyl-D-glucosamine.

EXPERIMENTAL PROCEDURES
Sequence Analysis-Similarity searches were performed at the Swiss Institute of Bioinformatics using the basic local alignment search tool (BLAST) network service. The National Center for Biotechnology Information (NCBI) BLASTP tool was used to search the Swiss-Prot/TrEMBL database (22). Prediction of protein localization and signal peptide was conducted using PSORTb version 3.0.2 (23) and SignalP 4.1 server (24), respectively.
Cloning, Expression, and Purification-The gene encoding BT1033 (GenBank TM accession number AAO76140.1) was amplified by PCR from genomic DNA of B. thetaiotaomicron VPI-5482, using KOD-plus DNA polymerase (Toyobo, Osaka, Japan) with the following oligonucleotides based on the genome sequence (GenBank TM accession number AE015928) (2): 5Ј-ggaattccatatgaataagattcaaattc-3Ј as the forward primer containing an NdeI site (underlined) and 5Ј-tttctcgaggataatgctcgttcgttttg-3Ј as the reverse primer containing an XhoI site (underlined). The amplified gene was purified using a FastGene Gel/PCR extraction kit (Nippon Genetics Co., Tokyo, Japan), digested by NdeI and XhoI (New England Biolabs, Beverly, MA), and inserted into pET24a (ϩ) (Novagen, Madison, WI) to encode a His 6 tag fusion at the C terminus of the recombinant protein. The expression plasmid was propagated in Escherichia coli DH5␣ (Toyobo), purified by a FastGene Plasmid Mini Kit (Nippon Genetics Co.), and verified by sequencing (Operon Biotechnologies, Tokyo, Japan). An E. coli BL21 (DE3) (Novagen) transformant harboring the expression plasmid was grown at 37°C in 200 ml of Luria-Bertani medium (1% tryptone, 0.5% yeast extract, and 0.5% NaCl) containing 50 g/ml kanamycin, until the absorbance reached 0.6 at 600 nm. The expression was induced by 0.1 mM isopropyl ␤-D-thiogalactopyranoside and continued at 18°C for 24 h. The cells were harvested by centrifugation at 20,000 ϫ g for 20 min and suspended in 50 mM HEPES-NaOH buffer (pH 7.0) containing 500 mM NaCl (buffer A). The suspended cells were disrupted by sonication (Branson sonifier 250A; Branson Ultrasonics, Emerson Japan, Kanagawa, Japan). The supernatant collected by centrifugation at 20,000 ϫ g for 20 min was applied to a HisTrap HP column (GE Healthcare), equilibrated with buffer A containing 10 mM imidazole, using Ä KTA Prime (GE Healthcare). After washing with buffer A containing 22 mM imidazole and subsequent elution using a 22-400 mM imidazole linear gradient in buffer A, fractions containing recombinant protein (BT1033) were pooled, dialyzed against 10 mM HEPES-NaOH buffer (pH 7.0), and concentrated (AMICON Ultra-15 filter; Millipore, Billerica, MA). The protein concentration was determined spectrophotometrically at 280 nm using a theoretical extinction coefficient of ⑀ ϭ 37,823 M Ϫ1 cm Ϫ1 , based on the amino acid sequence (25). The molecular mass of purified BT1033 was estimated by SDS-PAGE (Mini-PROTEAN Tetra electrophoresis system; Bio-Rad) and by gel filtration (HiLoad 26/600 Superdex, 200 pg; GE Healthcare) equilibrated with 10 mM HEPES-NaOH buffer (pH 7.0) containing 150 mM NaCl at a flow rate of 0.5 ml/min, using Marker Proteins for molecular Weight Determination on High Pressure Liquid Chromatography (Oriental Yeast Co., Tokyo, Japan) as standards.
Measurement of Synthetic Activity-The synthetic activity was routinely determined by measuring the increase in inorganic phosphate (P i ) using a reaction mixture containing 10 mM ␣-Man1P (␣-Man1P bis(cyclohexylammonium) salt; Sigma-Aldrich) and 10 mM GlcNAc (Wako Pure Chemicals, Osaka, Japan) in 40 mM sodium acetate buffer (pH 5.5) at 30°C by following the method of Lowry and Lopez (26) as described previously (27). One unit of the synthetic activity was defined as the amount of enzyme that catalyzed the release of 1 mol of P i per min under the above conditions.
Structural Determination-Reaction products for structural determination were generated in 500 l of reaction mixture (pH 5.5) containing BT1033 (1.5 and 3.7 M GlcNAc and GlcNAc 2 , respectively), 50 mM ␣-Man1P, and 50 mM GlcNAc or GlcNAc 2 . The reaction mixtures were incubated at 30°C for 24 h, followed by desalting using Amberlite MB-3 (Organo, Tokyo, Japan). The reaction products were purified using an HPLC system (Prominence; Shimadzu, Kyoto, Japan) equipped with a Shodex Asahipak NH2P-50 4E column (4.6-mm internal diameter ϫ 25 cm; Showa Denko KK, Tokyo, Japan) at 30°C under a constant flow (1.0 ml/min) of 75% acetonitrile in water as the mobile phase. Fractions containing the reaction products were collected, followed by lyophilization. The amounts of products obtained were 4 and 5 mg from GlcNAc and GlcNAc 2 as the acceptors, respectively. One-dimensional ( 1 H and 13 C) and two-dimensional (double-quantum filtered correlation spectroscopy, heteronuclear single-quantum coherence, and heteronuclear multiple-bond correlation) NMR spectra of the product were acquired in D 2 O with 2-methyl-2-propanol as an internal standard using a Bruker Avance 800 spectrometer (Bruker Biospin, Rheinstetten, Germany). Proton signals were assigned based on the double-quantum filtered correlation spectra. 13 C signals were assigned using the heteronuclear single-quantum coherence spectra, based on the assignment of the proton signals. The linkage position of each disaccharide was determined by detecting the inter-ring cross-peaks in each heteronuclear multiple-bond correlation spectrum.
Measurement of Phosphorolytic Activity-The substrates for phosphorolysis of BT1033 were generated in 5 ml of reaction mixture (pH 5.5) containing 15 M BT1033, 500 mM ␣-Man1P, and 500 mM GlcNAc or 500 mM GlcNAc 2 . After incubation at 30°C for 24 h, the reaction mixtures were desalted using Amberlite MB-3 and loaded onto a Toyopearl HW-40S column (50-mm internal diameter ϫ 950 mm; Tosoh) equilibrated with distilled water at a flow rate of 1.0 ml/min. Fractions containing the reaction products were collected, followed by lyophilization. The amounts of products obtained were 123 and 40 mg from GlcNAc and GlcNAc 2 as the acceptors, respectively. The phosphorolytic activity was routinely determined by quantifying the ␣-Man1P released during a phosphorolytic reaction in 40 mM sodium acetate buffer (pH 5.5) containing 10 mM substrate and 10 mM P i at 30°C by the colorimetric method as described previously (28). One unit of the phosphorolytic activity was defined as the amount of enzyme that catalyzed the liberation of 1 mol of ␣-Man1P from the substrates per min under the above conditions.
Temperature and pH Profile-The effects of pH on the phosphorolytic and synthetic activities using 212 nM BT1033 were measured under the standard conditions described above, by substituting 40 mM sodium acetate buffer (pH 5.5) with the following 40 mM buffers: sodium citrate (pH 3.0 -5.5), bis-(2-hydroxyethyl)aminotris(hydroxymethyl)methane-HCl (pH 5.5-7.0), HEPES-NaOH (pH 7.0 -8.5), and glycine-NaOH (pH 8.5-10.5). The thermal and pH stabilities were evaluated by measuring the residual synthetic activity under the standard conditions after incubation of BT1033 (7 and 11 M to study thermal and pH stabilities, respectively) at a temperature range of 30 -90°C for 15 min in 67 mM sodium acetate buffer (pH 5.5) and in the various pH values at 4°C for 24 h, respectively.
Kinetic Analysis-The initial velocities of the phosphorolytic reaction with ManGlcNAc were determined under the standard conditions with 42 nM BT1033 and a combination of initial concentrations of ManGlcNAc (0.5, 1.0, 2.0, 3.0, 5.0, and 10 mM) and P i (0.1, 0.2, 0.3, 0.5, 1.0, and 2.0 mM). The kinetic parameters were calculated by curve-fitting the experimental data to theoretical Equation 1 for a sequential Bi Bi mechanism using GraFit version 7.0.2 (Erithacus Software Ltd., London, UK).

RESULTS
Prediction of the Enzymatic Function of BT1033-A gene cluster involved in complex type N-glycan metabolism was identified in the genome of B. thetaiotaomicron VPI-5482 (Fig. 1B). Based on assignments using BLASTP (22) (supplemental Table  S1), the gene cluster contains nine unidirectionally transcribed ORFs (Fig. 1B) that encode four glycoside hydrolases (GH92 ␣-mannosidase (BT1032), GH20 ␤-N-acetylhexosaminidase (BT1035), exo-␣-sialidase (BT1036), and GH18 ␤-1,4-D-Mannosyl-N-acetyl-D-glucosamine Phosphorylase endo-␤-N-acetylglucosaminidase (BT1038)), a major facilitator superfamily transporter (BT1034), a hypothetical protein with unknown function (BT1037), an outer membrane lipoprotein (BT1039) and a TonB-dependent oligosaccharide transporter (BT1040) constituting a Sus-like protein, and a GH130 phosphorylase (BT1033). The sequence analysis including prediction of protein localization based on PSORTb version 3.0.2 (23) and signal peptide identification using version SignalP 4.1 (24) suggests that BT1033 plays a role in the intracellular phosphorolysis of ManGlcNAc liberated from complex type N-glycans by sequential glycoside hydrolase-catalyzed reactions in the periplasmic space and transported into the cytoplasm. However, there have been no reports on the N-glycans metabolism that a phosphorylase participates. In this study, BT1033 was recombinantly expressed in E. coli BL21 (DE3) to investigate the detail enzymatic properties, as described below.
Preparation of Recombinant BT1033-Recombinant BT1033 was purified by nickel chelate affinity chromatography with a yield of 24 mg from the cell lysate of a 200-ml culture. Purified BT1033 migrated in SDS-PAGE as a single protein band with an estimated size of 35 kDa in agreement with the theoretical molecular mass of 37,823. However, the molecular mass was estimated by gel filtration to be 147 kDa, indicating that BT1033 is a homotetramer in solution, whereas R. albus GH130 ␤-1,4-D-mannosyl-D-glucose phosphorylase (EC 2.4.1.281) and ␤-1,4-D-mannooligosaccharide phosphorylase (EC 2.4.1.-) showing 39 and 70% sequence similarities with BT1033, respectively, have been reported to be homodimeric and homohexameric, respectively (21).
Synthetic Reaction Catalyzed by BT1033-The acceptor specificity in the synthetic reaction was examined using various carbohydrate acceptor candidates (see "Experimental Procedures") together with ␣-Man1P as the donor. BT1033 utilized GlcNAc and GlcNAc 2 as the suitable acceptors, with specific activities of 41 and 8 units/mg, respectively. Each synthetic reaction gave a single product from GlcNAc and GlcNAc 2 . The products from GlcNAc and GlcNAc 2 were identified by 1 H and 13 C NMR spectroscopic analysis to be the corresponding ␤-1,4-D-mannopyranosyl-␤-1,4-N-acetyl-D-glucosamine (ManGlcNAc) (supplemental Fig. S1) and ␤-1,4-D-mannopyranosyl-␤-1,4-N-acetyl-D-glucosaminyl-␤-1,4-N-acetyl-D-glucosamine (ManGlcNAc 2 ) (supplemental Fig. S2), respectively. In addition BT1033 showed weak synthetic activities with D-glucose and D-mannose. Therefore, the kinetic parameters for the four acceptors (GlcNAc, GlcNAc 2 , D-glucose, and D-mannose) were determined to investigate the accepter preference of BT1033 in the presence of ␣-Man1P as the donor ( Table 1). The K m value for GlcNAc was in the millimolar range and was 12 times lower than that for GlcNAc 2 . However, the k cat value for GlcNAc was in the same range with that for GlcNAc 2 . These results indicate that the catalytic efficiency (k cat /K m ) values for GlcNAc and GlcNAc 2 mainly depend on the K m values. In addition, the fact that the k cat /K m value for GlcNAc was 140 -200 times greater than those for the other acceptors indicates that GlcNAc is the most effective acceptor for BT1033. Based on the acceptor specificity and considering that the kinetic parameters for ␣-Man1P are in the same ranges as those of the other inverting phosphorylases for their specific donors (20, 21, 27, 30 -36), we here propose 4-O-␤-D-mannopyranosyl-N-acetyl-D-glucosamine:phosphate ␣-D-mannosyltransferase as the systematic name and ␤-1,4-D-mannosyl-N-acetyl-D-glucosamine phosphorylase as the short name for BT1033 (Fig. 2).
Phosphorolytic Reaction Catalyzed by BT1033-BT1033 catalyzed the phosphorolysis of ManGlcNAc and ManGlcNAc 2 with inversion of the anomeric configuration to release ␣-Man1P and GlcNAc or GlcNAc 2 , respectively. The specific activity values on ManGlcNAc and ManGlcNAc 2 were 258 and 30 units/mg of protein, respectively. In addition, BT1033 did not cleave the substrates in the absence of P i . Double reciprocal plots of the initial velocities against various initial concentrations of ManGlcNAc and P i gave a series of lines intersecting at a single point (Fig. 3). These results indicate that the phosphorolytic reactions on ManGlcNAc follow a sequential Bi Bi mechanism, similar to inverting phosphorylases (21, 27, 30 -35). In addition, BT1033 did not phosphorolyze ␤-1,4-D-mannosyl-D-glucose and ␤-1,4mannooligosaccharides (degree of polymerization of 2-6), known substrates of the GH130 ␤-1,4-D-mannosyl-D-glucose phosphorylase (20,21), and ␤-1,4-mannooligosaccharide phosphorylase (21), respectively. The kinetic parameters for ManGlcNAc and ManGlcNAc 2 are summarized in Table 2. The k cat /K m value for ManGlcNAc was 58 times greater than that for ManGlcNAc 2 . In addition, the parameters for ManGlcNAc are in the same ranges with those of the other inverting phosphorylases (21, 27, 30 -35), indicating that ManGlcNAc is the true substrate of BT1033.
Basic Properties of BT1033-BT1033 was stable up to 55°C during 15 min of incubation (Fig. 4A) and in the range of pH 4.5-10.5 at 4°C for 24 h (Fig. 4B). The optimum pH for both the phosphorolytic and synthetic reactions was pH 5.5 (Fig. 4C).  (Fig. 5A). As shown in Fig. 5B, synthesis of ManGlcNAc was observed along with a decrease in GlcNAc at

DISCUSSION
Enzymatic Function of BT1033-BT1033 has unique substrate specificity in the phosphorolytic reaction; it recognizes ManGlcNAc and ManGlcNAc 2 as the substructures of the N-glycans. The amino acid sequence of BT1033 shows 42, 39, and 79% similarities with those of B. fragilis and R. albus ␤-1,4-mannosyl-D-glucose phosphorylases and R. albus ␤-1,4mannooligosaccharide phosphorylase belonging to GH130 (supplemental Table S1), respectively, which are involved in the catabolism for plant polysaccharide ␤-1,4-mannan (20,21). To date, no three-dimensional structures of reported GH130 phosphorylases have been solved. In addition, there has been no report of enzymatic activities of four putative GH130 glycosidases from Bacteroides ovatus, B. thetaiotaomicron, Parabacteroides distasonis, and Thermotoga maritima, of which the three-dimensional structures are available (Protein Data Bank codes 3QC2, 3R67, 3TAW, and 1VKD, respectively). Furthermore, information regarding the amino acid residues involved in the substrate recognition or catalysis is not available because the substrate-enzyme complex structure has not been solved for GH130. However, together with the fact that BT1033 showed no phosphorolytic activity toward ␤-1,4-mannosyl-Dglucose, the substrate for the reported GH130 ␤-1,4-mannosyl-D-glucose phosphorylase (20,21), the much lower k cat /K m values of D-glucose and D-mannose substituted at C2 position of GlcNAc as an suitable acceptor in the synthetic reaction with strict ␤-1,4-regioselectivity clearly suggest that BT1033 recognizes the C2 N-acetyl group at subsite ϩ1. Regarding the recognition of the C2 N-acetyl group of the acceptor molecule at subsite ϩ1, a GH94 N,NЈ-diacetylchitobiose phosphorylase (EC 2.4.1.280) from Vibrio proteolyticus has been reported to recognize the C2 N-acetyl group of GlcNAc at subsite ϩ1 by interaction with the methyl group in a small hydrophobic pocket formed by Cys-493 and Val-631. In contrast, the space is occupied by a well conserved bulky Tyr residue in case of GH 94 cellobiose phosphorylase (EC 2.4.1.20) (37,38). In addition, BT1033 did not utilize N-acetyl-D-galactosamine with an axial hydroxyl group at the C4 position as the acceptor in the synthetic reaction, clearly indicating that the equatorial C4 hydroxyl group of GlcNAc is essential for acceptor binding at subsite ϩ1. Notably, GlcNAc 2 acted as the acceptor in the synthetic reaction, suggesting that BT1033 possesses a moderately large binding space to accommodate the oligosaccharide acceptor. Therefore, addition of an excess of the enzyme was confirmed to cause elongation of successive D-mannose residue at    It should be noted that the enzymatic properties of BT1033, such as strict regioselectivity and acceptor specificity, enable the enzymatic syntheses of ManGlcNAc and ManGlcNAc 2 , which are common core structures shared by all N-glycans such as high mannose, complex, and hybrid types (10). Stereoselec-tive formation of the ␤-mannosidic linkages is a considerable challenge in synthetic glycochemistry because the vicinal C2 hydroxyl group blocks access to the ␤-face because of its steric and polar effects (39). Therefore, the new synthetic method for core structure of N-glycans employing the synthetic reaction of this unique mannoside phosphorylase would be a strong tool for the efficient preparation of various N-glycans and glycoconjugates used in the research field of glycobiology and glycotechnology. In addition, ManGlcNAc was synthesized by a one-pot enzymatic reaction from sucrose and GlcNAc as the starting materials, which are industrially prepared and are available at reasonable cost (Fig. 5A). The details of ManGlcNAc synthesis are as follows: (i) sucrose is phosphorolyzed into ␣-D-glucose 1-phosphate and fructose by sucrose phosphorylase; (ii) ␣-Dglucose 1-phosphate is converted into ␣-Man1P via D-glucose 6-phosphate, D-fructose 6-phosphate, and D-mannose 6-phosphate by the sequential reactions of ␣-phosphoglucomutase, D-glucose 6-phosphate isomerase, D-mannose-6-phosphate isomerase, and ␣-phosphomannomutase; and (iii) ManGlcNAc is generated from ␣-Man1P and GlcNAc by BT1033 together with the release of P i . Because P i is recycled in the reaction, the overall reaction can be described as the transformation of sucrose and GlcNAc to ManGlcNAc and fructose by the concomitant action of the six enzymes in the presence of catalytic amounts of P i . Notably, this strategy circumvents the addition of costly ␣-Man1P and is compatible with large scale production. Moreover, starch and cellobiose/cellodextrin biomass are available as the starting materials by substituting sucrose phosphorylase with starch phosphorylase (EC 2.4.1.1) and cellobiose/cellodextrin phosphorylase (EC 2.4.1.20/EC 2.4.1.49), respectively, which catalyze the release of ␣-D-glucose 1-phosphate by the phosphorolysis. This one pot enzymatic approach using the inexpensive starting materials can be extended to include the production of a variety of valuable ␤-mannosides using ␤-1,4-D-mannosyl-D-glucose and ␤-1,4-mannooligosaccharide phosphorylases belonging to GH130 with known acceptor specificities in the synthetic reaction (21).
Physiological Role of BT1033-The complex type N-glycans had been considered to be metabolized in a conventional pathway where ␤-mannosidase and ATP-dependent hexokinase participate. In this study, a gene cluster including a gene encoding a ␤-1,4-D-mannosyl-N-acetyl-D-glucosamine phosphorylase (BT1033) for the energy-efficient metabolism of complex type N-glycans was identified in the genome of B. thetaiotaomicron VPI-5482. This is the first report of a metabolic pathway for complex type N-glycans in which a unique phosphorylase  ␤-1,4-D-Mannosyl-N-acetyl-D-glucosamine Phosphorylase participates. A possible metabolic pathway for complex type N-glycans in B. thetaiotaomicron is illustrated in Fig. 6. In the pathway, an oligosaccharide chain is cleaved from glycoprotein by endo-␤-N-acetylhexosaminidase (BT1038) and transported into periplasmic space by a Sus-like system that comprises one pair of outer membrane proteins homologous to SusC (BT1040) and SusD (BT1039) (40). The transported oligosaccharide chain is sequentially degraded by ␣-sialidases (BT1036), ␤-galactosidase, ␤-N-acetylhexosaminidase (BT1035), and ␣-mannosidase (BT1032). The resultant ManGlcNAc is transported into the cytoplasm by an major facilitator superfamily transporter (BT1034), followed by phosphorolysis by BT1033 to produce ␣-Man1P and N-acetyl-D-glucosamine. The ␣-Man1P released is converted into D-fructose 6-phosphate via D-mannose 6-phosphate by the sequential reaction of phosphomannomutases (BT1548 or BT3950) and D-mannose-6-phosphate isomerase (BT0373) to enter glycolysis. The GlcNAc released is also converted into D-fructose 6-phosphate via GlcNAc 6-phosphate and D-glucosamine 6-phosphate by sequential reaction of hexokinase (BT2430) and D-glucosamine 6-phosphate deaminase (BT0258, BT3587, or BT4127) to enter the glycolysis.
Thus, BT1033 plays a necessary role to phosphorolyze the intracellular ManGlcNAc, as an alternative to ␤-mannosidase in the conventional metabolic pathway for complex type N-glycans. Notably, intestinal anaerobes such as B. fragilis (16,17), Bacteroides helcogenes (13), Bacteroides salanitronis (14), Bacteroides vulgatus (15), Prevotella denticola, Prevotella dentalis, Prevotella melaninogenica, P. distasonis (15), and Alistipes finegoldii were also identified to possess the similar metabolic pathway for complex type N-glycans (Fig. 1B), according to sequence analyses using BLAST (22) and PSORTb (23). One notable feature of the new N-glycan metabolic pathway is that these intestinal anaerobes efficiently use the energy of ATP, because the D-mannose residue of ManGlcNAc can be directly phosphorylated without consuming ATP by ATP-dependent carbohydrate kinase. Hence, the energy-efficient strategy for N-glycan metabolism in which a unique ␤-1,4-D-mannosyl-Nacetyl-D-glucosamine phosphorylase participates would have provided intestinal anaerobic bacteria with evolutionary advantages because anaerobic respiration is in general less energyefficient than aerobic respiration.
Conclusions-A gene cluster involved in complex type N-glycan metabolism was identified in the genome of B. thetaiotaomicron VPI-5482. It was demonstrated that BT1033 encoded in the gene cluster catalyzed the reversible phosphorolysis of ManGlcNAc in a typical sequential Bi Bi mechanism. This is the first report of a metabolic pathway for complex type N-glycans in which a unique phosphorylase participates. In addition, several intestinal anaerobes were also identified to possess the similar metabolic pathway for the N-glycans. One notable feature of the new metabolic pathway for N-glycans is that these intestinal anaerobes efficiently use the energy of ATP, in comparison with a conventional pathway in which ␤-mannosidase and ATP-dependent hexokinase participate, because it is possible to directly phosphorylate the D-mannose residue of ManGlcNAc to enter glycolysis.