Metabolic Mechanism of Mannan in a Ruminal Bacterium, Ruminococcus albus, Involving Two Mannoside Phosphorylases and Cellobiose 2-Epimerase

Background: Characteristics of two 4-O-β-d-mannosyl-d-glucose phosphorylases from Ruminococcus albus were investigated. Results: One enzyme was specific for 4-O-β-d-mannosyl-d-glucose, as observed for the Bacteroides fragilis enzyme, but the other showed high activity toward mannooligosaccharides longer than β-1,4-mannobiose. Conclusion: Two phosphorylases play distinct roles in the metabolism of mannan. Significance: A new enzyme catalyzing the phosphorolysis of β-1,4-mannooligosaccharides was identified. Ruminococcus albus is a typical ruminal bacterium digesting cellulose and hemicellulose. Cellobiose 2-epimerase (CE; EC 5.1.3.11), which converts cellobiose to 4-O-β-d-glucosyl-d-mannose, is a particularly unique enzyme in R. albus, but its physiological function is unclear. Recently, a new metabolic pathway of mannan involving CE was postulated for another CE-producing bacterium, Bacteroides fragilis. In this pathway, β-1,4-mannobiose is epimerized to 4-O-β-d-mannosyl-d-glucose (Man-Glc) by CE, and Man-Glc is phosphorolyzed to α-d-mannosyl 1-phosphate (Man1P) and d-glucose by Man-Glc phosphorylase (MP; EC 2.4.1.281). Ruminococcus albus NE1 showed intracellular MP activity, and two MP isozymes, RaMP1 and RaMP2, were obtained from the cell-free extract. These enzymes were highly specific for the mannosyl residue at the non-reducing end of the substrate and catalyzed the phosphorolysis and synthesis of Man-Glc through a sequential Bi Bi mechanism. In a synthetic reaction, RaMP1 showed high activity only toward d-glucose and 6-deoxy-d-glucose in the presence of Man1P, whereas RaMP2 showed acceptor specificity significantly different from RaMP1. RaMP2 acted on d-glucose derivatives at the C2- and C3-positions, including deoxy- and deoxyfluoro-analogues and epimers, but not on those substituted at the C6-position. Furthermore, RaMP2 had high synthetic activity toward the following oligosaccharides: β-linked glucobioses, maltose, N,N′-diacetylchitobiose, and β-1,4-mannooligosaccharides. Particularly, β-1,4-mannooligosaccharides served as significantly better acceptor substrates for RaMP2 than d-glucose. In the phosphorolytic reactions, RaMP2 had weak activity toward β-1,4-mannobiose but efficiently degraded β-1,4-mannooligosaccharides longer than β-1,4-mannobiose. Consequently, RaMP2 is thought to catalyze the phosphorolysis of β-1,4-mannooligosaccharides longer than β-1,4-mannobiose to produce Man1P and β-1,4-mannobiose.

coside hydrolases (8,9). Thus, the inverting phosphorylases are classified into glycoside hydrolase (GH) families on the basis of the similarity of their amino acid sequences (10). Only limited information concerning MP from B. fragilis is available, but this enzyme was recently categorized into GH family 130 together with the putative proteins from the genome sequence of known bacterial species, the amino acid sequences of which are similar to that of B. fragilis MP. In this family, three-dimensional structures of Thermotoga maritima MSB8 TM1225 protein (Protein Data Bank code 1VKD), Bacteroides thetaiotaomicron VPI-5482 BT_4094 protein (Protein Data Bank code 3R67), and Parabacteroides distansonis ATCC 8503 BDI_3141 protein (Protein Data Bank code 3TAW) were elucidated, although their biochemical properties have not been examined thus far. These proteins are formed by five-bladed ␤-propeller folds, which are observed in the glycoside hydrolases belonging to GH families 32 (11,12), 43 (13)(14)(15)(16)(17)(18), and 68 (19,20).
R. albus 7 has several putative ␤-1,4-mannanase genes and two MP-like genes (Rumal_0099 and Rumal_0852) in the genome, indicating that this bacterium also has components necessary to metabolize mannan via epimerization and phosphorolysis, as observed in B. fragilis. Rumal_0852 and Rumal_0099 encode proteins with 59 and 27% sequence identity to B. fragilis MP, respectively. In this study, two MP isozymes from R. albus NE1, which is phylogenetically close to R. albus 7 (5), were characterized in detail, and the physiological functions of these enzymes in the metabolism of mannan are discussed.
Identification of MP Isozymes in the Cell-free Extract of R. albus NE1-MP activity of the cell-free extract of R. albus NE1, obtained from 1.5 liters of the culture fluid (5), was detected by thin layer chromatography (TLC). Six l of cell-free extract of R. albus NE1 was mixed with 2 l of 100 mM Man-Glc and 2 l of 50 mM reaction buffer (pH 7.0) and incubated at 37°C for 3 h. Sodium phosphate buffer or MES-NaOH buffer was used as reaction buffer. The reaction mixture of 1 l was analyzed by TLC, in which a developing solvent of acetonitrile/ ethyl acetate/1-propanol/water (85:20:50: 30, v/v/v/v) was used. The chromatogram was visualized by spraying a detection reagent (acetic acid/sulfuric acid/anisaldehyde, 100:2:1, v/v/v) and heating. The cell-free extract was subjected to DEAE-Sepharose CL-6B column chromatography (1.5 ϫ 9.0 cm; GE Healthcare), in which adsorbed protein was eluted by a linear gradient of 0 -0.5 M NaCl in 20 mM MES-NaOH buffer (pH 6.5). Enzyme activity was checked by TLC analysis as described above. Two active peaks (the enzymes eluted by high and low concentrations of NaCl were designated as RaMP1 and RaMP2, respectively) were further purified by butyl-Sepharose CL-6B (1.5 ϫ 9.0 cm; GE Healthcare), hydroxyapatite (1.5 ϫ 12 cm; Seikagaku, Tokyo), and Superdex 200 (1.6 ϫ 60 cm; GE Healthcare) column chromatography. Finally, the sample was separated by SDS-PAGE. The masses of the tryptic peptides derived from the ϳ40-kDa protein were measured and assigned to the theoretical values for Rumal_0099 and Rumal_0852 proteins, as described elsewhere (23).
Production and Purification of Recombinant RaMP1 and RaMP2-The genomic DNA of R. albus NE1 as the template, Primestar HS DNA polymerase (Takara Bio, Otsu, Japan), and the primers listed in Table 1 were used in the PCR. The RaMP1 and RaMP2 genes were amplified by PCR and cloned into the EcoRV site of pBluescript II SK(ϩ) vector (Stratagene, La Jolla, CA). The DNA sequences of the cloned genes were analyzed with the ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, CA). These plasmids were used as the templates in the PCR to construct the expression plasmids. NdeI and XhoI sites were introduced to the 5Ј and 3Ј termini of the target gene, respectively, by PCR (primers listed in Table 1) and cloned into the NdeI and XhoI sites of the pET-23a vector (Novagen, Darmstadt, Germany).
The transformants of Escherichia coli BL21 (DE3) harboring the expression plasmid of RaMP1 or RaMP2 were cultured in 0.5 and 1.5 liters of LB broth containing 100 g/ml ampicillin, respectively, at 37°C until the A 600 reached 0.6. The protein production was induced by the addition of isopropyl ␤-D-thiogalactoside at a final concentration of 0.1 mM, and the incubation was continued at 18°C for 16 h. The recombinant enzymes were purified from the E. coli cell extract. The concentrations of the purified enzymes were determined based on the concentration of each amino acid after acid hydrolysis (24). Enzyme Assay-A reaction mixture of 50 l consisting of the appropriate concentration of enzyme diluted with 20 mM MES-NaOH buffer (pH 6.5) containing 1 mg/ml bovine serum albumin, 100 mM sodium phosphate buffer (pH 6.5), and 2 mM Man-Glc was incubated at 37°C for 10 min. The enzyme reaction was terminated by boiling the mixture for 3 min after the addition of 25 l of 4 M Tris-HCl buffer (pH 7.0). The D-glucose produced was measured by the glucose oxidase-peroxidase method (25). One unit of enzyme activity was defined as the amount of enzyme producing 1 mol of D-glucose in 1 min under these conditions.
Temperature and pH stabilities were evaluated by the residual activity after temperature treatment (incubation at various temperatures for 20 min at pH 6.5) and pH treatment (incubation at various pH values for 24 h at 4°C), respectively. The values over which the enzymes retained more than 90% of their original activities were considered to be stable ranges.
The kinetic parameters for phosphorolysis and synthesis of Man-Glc were calculated from the initial velocities toward various concentrations of Man-Glc and inorganic phosphate and of Man1P and D-glucose, respectively, by fitting to the equation for a sequential Bi Bi mechanism (26). Non-linear regression was performed with Grafit version 7.0.2 (Erithacus Software, West Sussex, UK).
Product inhibition analysis was carried out to determine the order of substrate binding and product release. First, the rate of phosphorolysis catalyzed by rRaMP1 and rRaMP2 toward Man-Glc in the presence of 0 -2 mM Man1P was measured at varying concentrations of Man-Glc and 10 mM sodium phosphate buffer (pH 6.5) or at varying concentrations of sodium phosphate buffer (pH 6.5) and 2 mM Man-Glc. Then, in the case of rRaMP2, the phosphorolysis of Man-Glc was measured in the presence of 0 -100 mM cellobiose at varying concentrations of Man-Glc and 10 mM sodium phosphate buffer (pH 6.5) or at varying concentrations of sodium phosphate buffer (pH 6.5) and 2 mM Man-Glc. In the case of rRaMP1, initial velocities for the synthesis of Man-Glc were measured in the presence of 0 -20 mM Man-Glc at varying concentrations of Man1P and 50 mM D-glucose or at varying concentrations of D-glucose and 5 mM Man1P.
Structural Analysis of the Oligosaccharides Produced by the Synthetic Reaction-The oligosaccharides produced by recombinant RaMP1 (rRaMP1) were prepared as follows. A reaction mixture of 0.2 ml consisting of 493 nM rRaMP1, 50 mM MES-NaOH buffer (pH 6.5), 250 mM D-glucose or D-xylose, and 250 mM Man1P was incubated at 37°C for 3 h. The reaction products were purified by HPLC as described previously (28). The synthetic reaction of recombinant RaMP2 (rRaMP2) was performed in a reaction mixture of 1 ml, consisting of 79 nM rRaMP2, 50 mM MES-NaOH buffer (pH 6.5), 100 mM Man1P, and 100 mM acceptor substrate. The reaction was carried out at 37°C for 12 h. Cellobiose, ␤-1,4-mannobiose, and N,NЈdiacetylchitobiose were used as acceptors. The reaction products were purified by gel filtration column chromatography under the following conditions: column, Bio-Gel P-2 (1.8 ϫ 100 cm; Bio-Rad); eluent, water; flow rate, 2.5 ml/h; fraction volume, 3 ml.
Electrospray ionization-MS of the oligosaccharides produced was carried out using an Exactive mass spectrometer (Thermo Scientific, San Jose, CA). The samples were introduced by flow injection. Methanol was used as a mobile phase solvent. The positive ion was detected under the following conditions: spray voltage, 3.00 kV; capillary temperature, 300°C. Nuclear magnetic resonance (NMR) spectra were recorded in D 2 O (Wako Pure Chemical Industries) at 300 K using an ECP-400 (400 MHz; Jeol, Tokyo, Japan). Sodium 3-(trimethylsilyl)-1-propanesulfonate was used as the standard. A series of twodimensional homo-and heteronuclear correlated spectra (COSY, HSQC, HSQC-TOCSY, and HMBC) were obtained.

RESULTS
Identification of RaMP1 and RaMP2 in the Cell-free Extract of R. albus NE1-MP activity in the cell-free extract of R. albus NE1 was analyzed by TLC (Fig. 1). Man-Glc was degraded to D-glucose and Man1P only in the presence of inorganic phosphate, indicating that R. albus NE1 produces intracellular MP. R. albus MP was purified to identify the gene encoding it. MP activity was identified in two peaks, RaMP1 and RaMP2, eluted by high and low concentrations of NaCl, respectively, via anion exchange column chromatography using a DEAE-Sepharose CL-6B column (data not shown). Two MPs were further purified, and the masses of the tryptic peptides were analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (supplemental Fig. S1). The obtained masses of the digests of RaMP1 and RaMP2 matched with the theoretical values of the proteins encoded by Rumal_0852 and Rumal_0099 from R. albus 7, respectively. The analyzed tryptic fragments derived from RaMP1 and RaMP2 covered 27 and 36% of their entire sequences, respectively.
Production, Purification, and General Properties of Recombinant RaMP1 and RaMP2-The genes encoding RaMP1 and RaMP2 were obtained from the genomic DNA of R. albus NE1 by PCR. The sequences of the amplified genes were completely identical to the corresponding genes of R. albus 7. These genes were overexpressed in E. coli, and the recombinant enzymes were purified to homogeneity. From 0.5 and 1.5 liters of the culture broths of the E. coli transformants, 6.8 and 11.4 mg of the purified rRaMP1 and rRaMP2 (the specific activities were 55.2 and 2.03 units/mg) were obtained. RaMP1 and RaMP2 were 45 and 38 kDa on SDS-PAGE, respectively (data not shown), which coincided well with the theoretical masses from their amino acid sequences, but were 80 and 209 kDa on a gel filtration column, respectively. These results indicate that RaMP1 and RaMP2 exist as a homodimer and homohexamer in solution, respectively. Both RaMP1 and RaMP2 showed highest activity at pH 6.5, and their optimum temperatures were 50 and 45°C, respectively. RaMP1 retained original activity in a pH range of 4.5-10.5 and below 45°C, whereas RaMP2 was stable at pH 3.5-9.5 and below 40°C.
Kinetic Mechanisms for Phosphorolysis and Synthesis of Man-Glc by RaMP1 and RaMP2-Initial reaction velocities for phosphorolysis at various concentrations of Man-Glc and inorganic phosphate were measured to investigate the kinetic mechanisms of RaMP1 and RaMP2. The lines obtained for both enzymes from double reciprocal plots of 1/v versus 1/[Man-Glc] at various concentrations of inorganic phosphate were linear and crossed at a certain point (Fig. 2). This indicates that these enzymes catalyze the phosphorolysis of Man-Glc through a sequential Bi Bi mechanism involving the formation of a ternary complex. To investigate the orders of substrate binding and product release, product inhibition analysis was carried out (Fig. 3). Man1P acted as a competitive inhibitor of rRaMP1 and rRaMP2 against both Man-Glc and inorganic phosphate, as observed for several cellobiose phosphorylases (3,29), indicating that these substrates bind to the enzymes in random order, and Man1P can be the second product (the first substrate in the reverse reaction). Therefore, a possible kinetic mechanism for these enzymes is a random order Bi Bi mechanism or a randomordered Bi Bi mechanism (3,29). In the reverse reaction of rRaMP1, Man-Glc served as a competitive inhibitor against both Man1P and D-glucose, indicating that rRaMP1 catalyzes phosphorolysis of Man-Glc through a random order Bi Bi mechanism. In the case of rRaMP2, inhibition of phosphorolysis toward Man-Glc by cellobiose, which is not phosphorolyzed by rRaMP2 but acts as an acceptor substrate of this enzyme (described below), was analyzed, because a very high concentration of Man-Glc is required for the inhibition analysis of the synthetic reaction. Uncompetitive inhibition by cellobiose was observed when the concentration of inorganic phosphate was varied, presumably because of the formation of an enzymephosphate-cellobiose complex, which does not react or reacts very slowly. On the other hand, cellobiose competitively inhibited the phosphorolysis of Man-Glc at various concentrations of Man-Glc. This inhibition occurred by binding of cellobiose to ϩ1 and ϩ2 subsites or to Ϫ1 and ϩ1 subsites. The former case indicates that D-glucose (cellobiose) also can be the second product, and the kinetic mechanism of the phosphorolysis of Man-Glc by rRaMP2 is a random order Bi Bi mechanism as rRaMP1. The possibility of the latter case makes the order of release (binding) of Man1P and D-glucose (cellobiose) ambiguous. However, in the kinetic analysis of the synthetic reaction toward cellobiose and Man1P, no substrate inhibition by cellobiose was observed, indicating that cellobiose acts only as an acceptor substrate and does not bind to Ϫ1 and ϩ1 subsites. The kinetic parameters for the phosphorolysis and synthesis of Man-Glc are summarized in Table 2. In the phosphorolytic reaction of Man-Glc, rRaMP1 had a 4.8-fold higher k cat and 91-fold lower K m(MG) compared with rRaMP2. The K m(P i ) of rRaMP2 was considerably lower (12-fold) than that of rRaMP1.
Kinetic parameters for the synthesis of Man-Glc with rRaMP1 and rRaMP2 were determined from the initial reaction rates for various concentrations of Man1P and D-glucose ( Table  2). rRaMP1 had a 6.6-fold higher k cat value for the synthetic reaction than rRaMP2, as it did for the phosphorolytic reaction. The K m(Glc) value of rRaMP1 was 3.7-fold lower than that of rRaMP2, indicating that the ϩ1 subsite of RaMP1 is more suitable for binding the D-glucose moiety than RaMP2, consistent with the higher affinity of rRaMP1 toward Man-Glc than that of rRaMP2. The rRaMP2 K m(Man1P) value was 18-fold lower than that of rRaMP1, corresponding with the observation that RaMP2 has higher affinity toward inorganic phosphate than RaMP1 during the phosphorolysis of Man-Glc.
The internal consistency of the kinetic parameters of rRaMP1 and rRaMP2 was confirmed with the Haldane relationship for a sequential Bi Bi mechanism (K ϭ (k cat p K s(Man1P K mGlc )/ (k cat s K sMan-Glc K m(P i ) ), where k cat p and k cat s represent k cat for phosphorolytic and synthetic reactions, respectively) (30). Values of 0.435 Ϯ 0.062 and 0.658 Ϯ 0.137 were obtained for K of rRaMP1 and rRaMP2, respectively, from the kinetic parameters in Table 2. These values were similar in magnitude to the value 0.272 Ϯ 0.021 experimentally obtained with rRaMP1 at pH 6.5 at 37°C for the thermodynamic equilibrium constant (K) calculated from the following equation,

Substrate Specificity of RaMP1 and RaMP2 in Phosphorolytic
Reactions-D-Glucose was not detected in the reaction carried out in the presence of rRaMP1 or rRaMP2 and cellobiose or lactose even at a 10-fold higher enzyme concentration than used for the enzyme assay toward Man-Glc, indicating that these enzymes are highly specific to the mannosyl residue at the non-reducing end of the substrate.
Acceptor Specificity of RaMP1 in Synthetic Reaction-The acceptor specificity of rRaMP1 was investigated based on the initial reaction rates toward Man1P and various acceptor substrates. rRaMP1 showed a very narrow acceptor specificity. Almost all the oligosaccharides tested were not recognized as acceptor substrates. Only laminaribiose served as a poor acceptor substrate for rRaMP1. The apparent kinetic parameters, k cat(app) and K m(app) , for laminaribiose, determined in the presence of 10 mM Man1P, were 7.6-fold lower and 6.0-fold higher than those for D-glucose, respectively (Table 3). rRaMP1 was not active toward D-glucose derivatives with changes at the C2-position, and it was less active toward derivatives with changes at the C3-position compared with D-glucose. The monodeoxygenation at the 6-OH-position was less effective for RaMP1 than at the other positions. The k cat(app) and K m(app) values of rRaMP1 for 6-deoxy-D-glucose were 1.4-  DECEMBER 7, 2012 • VOLUME 287 • NUMBER 50 JOURNAL OF BIOLOGICAL CHEMISTRY 42393 and 4.6-fold higher than those for D-glucose, respectively. On the other hand, 6-deoxy-6-fluoro-D-glucose and D-xylose were considerably less favorable acceptor substrates than 6-deoxy-D-glucose. This enzyme had a 3.3-fold lower k cat(app) and 6.0-fold higher K m(app) for 6-deoxy-6-fluoro-Dglucose than those for D-glucose, and a saturation curve was not obtained for D-xylose because of its very high K m(app) . These results indicate that the 6-OH partially contributes to the enzyme-substrate interaction as a hydrogen bond donor and that the methylene group of D-glucose is also involved in stabilization of the transition state, presumably through the hydrophobic interaction.

Metabolic Mechanism of Mannan in R. albus
Structures of oligosaccharides produced by the synthetic reaction of rRaMP1 with D-glucose and D-xylose as acceptors were investigated. A single oligosaccharide was produced in each reaction, and 2.0 and 4.9 mg of product was obtained by HPLC, respectively. The structures of these oligosaccharides were elucidated by electrospray ionization-MS and NMR. The oligosaccharides produced from the reactions with D-glucose and D-xylose gave signals at 365.11 and 335.09 m/z [M ϩ Na] ϩ , respectively. Each oligosaccharide had a correlation peak between the 4C of the acceptor and the 1H of the D-mannosyl residue at the non-reducing end. The 1 J CH values for the D-mannosyl residues at the non-reducing ends of these oligosaccharides in 13 C NMR (insensitive nuclei enhanced by polarization transfer) were 161 Hz, indicating that D-mannosyl residues were bound to the acceptor substrates through a ␤-linkage (31). On the basis of these results, the oligosaccharides produced in the synthetic reactions in the presence of D-glucose and D-xylose were identified as Man-Glc and 4-O-␤-D-mannosyl-D-xylose, respectively. The chemical shifts of 1 H and 13 C NMR analyses are summarized in supplemental Table S1.
Acceptor Specificity of RaMP2 in Synthetic Reaction-rRaMP2 was not active when D-glucose was derivatized at the C6-position but showed very broad acceptor specificity (Table  3). rRaMP2 showed the highest k cat(app) /K m(app) toward N-acetyl-D-glucosamine among the 2-OH derivatives of D-glucose, although N-acetyl-D-glucosamine has a bulky acetamide group at the C2-position. All of the D-glucose derivatives with changes at the C3-position acted as acceptor substrates of rRaMP2. The k cat(app) and K m(app) values for these derivatives were considerably lower than for D-glucose. The K m(app) value for 3-deoxy-D-glucose was too low to determine the value. rRaMP2 had k cat(app) /K m(app) values for 1,5-anhydro-D-glucitol and methyl ␣-D-glucoside similar to that for D-glucose, and the k cat(app) /K m(app) for methyl ␤-D-glucoside was 2.1-fold higher than for D-glucose. All of the oligosaccharides acted as acceptor substrates of rRaMP2. The k cat(app) /K m(app) values for ␤-linked glucobioses were 1.4 -4.5-fold higher than for D-glucose due to low K m(app) values for these oligosaccharides. Cellobiose was the best acceptor substrate for rRaMP2 among the ␤-linked glucobioses in terms of its k cat(app) / K m(app) value, followed by gentiobiose. rRaMP2 also had a 3.4-fold lower K m(app) for N,NЈ-diacetylchitobiose. The k cat(app) /K m(app) for maltose was 50% of that for D-glucose; thus, the acceptor site of rRaMP1 efficiently accommodates even an ␣-linked oligosaccharide. D-Mannose was a poor acceptor substrate of rRaMP2 compared with D-glucose, but rRaMP2 showed 7.3-20-fold higher k cat(app) /K m(app) values for ␤-1,4-mannooligosaccharides than for D-glucose. The k cat(app) /K m(app) values for ␤-1,4-mannooligosaccharides increased depending on the degree of polymerization (DP) up to ␤-1,4-mannotetraose, whereas those for substrates longer than ␤-1,4-mannotetraose slightly decreased with increasing substrate chain length, although a lower K m(app) was observed for longer ␤-1,4-mannooligosaccharides.

DISCUSSION
In B. fragilis, the metabolic pathway of mannan has been postulated as follows: (i) hydrolysis of mannan by extracellular ␤-1,4-mannanase, (ii) epimerization of the resulting ␤-1,4mannobiose to Man-Glc by CE, and (iii) phosphorolysis of Man-Glc to D-glucose and Man1P by MP (7). Phosphomannose mutase and mannose-6-phosphate isomerase convert Man1P to fructose 6-phosphate, which is then metabolized through  glycolysis. R. albus NE1 is also a CE producer (5), and two MP homologous genes were encoded in the genome of R. albus 7. This bacterium also has putative proteins involved in the metabolism of mannan, eight extracellular ␤-1,4-mannanases belonging to GH family 26, a phosphomannomutase, phosphomannose isomerase, and sugar transport protein similar to sugar/cation symporter (Table 5). MP activity was detected in the cell-free extract of R. albus NE1, and two MP isozymes, the deduced amino acid sequences of which are completely identical to the corresponding genes of R. albus 7, were found. Consequently, these MP isozymes are predicted to be physiologi-cally functional, and R. albus NE1 (R. albus 7) is thought to degrade mannan through a similar pathway as postulated for B. fragilis (Fig. 5). Both rRaMP1 and rRaMP2 catalyzed phosphorolytic and synthetic reactions of Man-Glc through a random Bi Bi mechanism in contrast to other known inverting carbohydrate phosphorylases (3,(32)(33)(34). Both rRaMP1 and rRaMP2 specifically catalyzed the phosphorolysis of a ␤-1,4-mannosidic linkage at the non-reducing end of a substrate, but they had acceptor specificities that were clearly different from each other. Consistent with the acceptor specificity of B. fragilis MP, which