The GH130 Family of Mannoside Phosphorylases Contains Glycoside Hydrolases That Target β-1,2-Mannosidic Linkages in Candida Mannan

Background: A cohort of a family of mannose phosphorylases lack phosphate binding residues, suggesting that they display non-phosphorylase activities. Results: The non-phosphorylase enzymes were shown to be β-mannosidases. Conclusion: Replacing basic phosphate binding residues with carboxylic amino acids converts mannoside phosphorylases into glycoside hydrolases. Significance: Functional phylogeny can be used to distinguish between closely related glycan phosphorylases and glycoside hydrolases.

The microbial recycling of complex glycans is an important biological process that plays a central role in the carbon cycle. The process is also of significant industrial interest particularly in the biofuel and biorefinery sectors (1). The cleavage of the glycosidic bonds in complex glycans is primarily mediated by glycoside hydrolases, although polysaccharide lyases and, to a lesser extent, glycan phosphorylases contribute to the degradative process (2). These carbohydrate active enzymes or CAZymes are grouped into sequence-based families in the CAZy database (3). Although enzymes in the same family may display different substrate specificities, the fold, catalytic mechanism, and catalytic apparatus is conserved in the vast majority of the 133 glycoside hydrolase (GH) 3

families (3).
Mannose-containing glycans are important components of the secondary cell walls of many plants. These plant mannans are homopolymers of ␤-1,4-Man units that can be decorated at O 6 with ␣-galactose residues and are thus further defined as galactomannans (Fig. 1A). In glucomannans, the backbone consists of random sequences of ␤-1,4-linked Man and Glc residues (4). The core of mammalian N-glycans contains a conserved Man-␤1,4-GlcNAc linkage, whereas ␣-linked Man units are commonly found elsewhere in these structures (5) (Fig. 1B). The proteins in yeast cell walls contain particularly complex N-glycans referred to as ␣-mannans in which the core N-glycan is extended by ϳ200 ␣-1,6-Man units that are also decorated with ␣-linked Man side chains (6). In some yeasts, such as the human pathogen Candida albicans, ␤-1,2-mannose residues cap the ␣-mannans side chains, and extended ␤-1,2-Man chains are attached to the ␣-mannan via phosphate bridges (Fig. 1C). ␤-1,2-Man units that play an important role in detection of C. albicans by the innate immune system via the C-type lectin, galectin-3 (7,8). ␤-1,2-Mannans also fulfill an important storage role in Leishmania parasites (9). Plant cell wall mannans are hydrolyzed by GH5, GH26, and GH113 ␤-mannanases, whereas ␤-1,4-mannosides are targeted primarily by enzymes in GH2 and GH5 (see Ref. 2 for review). These enzymes display a (␤/␣) 8 fold and cleave glycosidic linkages by a double displacement mechanism leading to retention of anomeric configuration (10 -12).
GH130 contains exo-acting ␤-mannoside phosphorylases that cleave the ␤-1,2and ␤-1,4-mannosyl linkages between Man and Man, Glc, or GlcNAc residues (13)(14)(15)(16)(17)(18). The phosphate nucleophile attacks C1 of the Man in the active site below the ␣-face of the pyranose ring leading to phosphorolysis. The single displacement mechanism displayed by these phosphorylases leads to inversion of anomeric configuration (16). The phosphate is positioned in the active site through polar interactions with three basic residues that are highly conserved within GH130 (19,20). A subset of GH130 members, however, lacks these basic residues, and it was proposed that these enzymes may function as glycoside hydrolases cleaving mannosidic bonds through a hydrolytic reaction (18).
To test the GH130 glycoside hydrolase hypothesis, the structure and biochemical properties of a representative GH130 enzyme were determined. The enzyme, BT3780, which lacks the canonical phosphate-binding basic residues, is up-regulated by the dominant human gut bacterium Bacteroides thetaiotaomicron (21) in response to yeast mannan. BT3780 is encoded by a polysaccharide utilization locus (PUL) in the B. thetaiotaomicron genome that orchestrates the depolymerization of yeast mannan. A similar PUL encodes BACOVA_03624, one of seven GH130 proteins of the human gut symbiont Bacteroides ovatus ATCC 8483. The crystal structure of BACOVA_03624 was released in the Protein Data Bank (PDB) in 2011 (code 3QC2), but to date no function had been attributed to this protein. Thus, a potential substrate for these enzymes is the Man-␤1,4-GlcNAc linkage at the base of N-glycans or the ␤-1,2-Man units that cap the side chain of some fungal ␣-mannans. The data, presented here, showed that BT3780 and BACOVA_03624 are exo-acting glycoside hydrolases that cleave Man-␤1,2-Man linkages using water, not phosphate, as the nucleophile in a single displacement reaction. The three-dimensional structure of the enzymes in concert with function phylogeny identified motifs within GH130 that delineate glycoside hydrolases and enzymes, both hydrolases and phosphorylases, that target ␤-1,2-mannosidic linkages.

Materials and Methods
Cloning, Expression, and Purification of Bacteroides GH130 Enzymes-BT3780 was amplified from B. thetaiotaomicron genomic DNA and cloned into pET28a with an N-terminal His 6 tag using NheI and XhoI restriction sites. To generate DNA encoding, the sequence of the protein was used as template for gene synthesis with codon optimization for Escherichia coli heterologous production (Biomatik, Cambridge, Canada) and was subsequently cloned into the pET28a vector. The two genes were expressed in E. coli BL21 cells transformed with the appropriate recombinant plasmids. The recombinant E. coli strains were cultured in Luria broth supplemented with 50 g/ml kanamycin. Cultures cells were grown at 37°C to midlog phase and induced with 1 mM isopropyl ␤-D-1-thiogalactopyranoside at 16°C overnight. Cells were harvested by centrifugation 5000 rpm for 5 min and resuspended in 20 mM Tris-HCl buffer, pH 8.0, containing 300 mM NaCl (buffer A). Cells were lysed by sonication, and the cell-free extract was recovered by centrifugation at 13,000 rpm for 30 min.  BT3780 was purified from the cell-free extract using immobilized metal affinity chromatography using Talon TM , a cobalt-based matrix. Proteins were eluted from the column in buffer A containing 100 mM imidazole. For crystallization trials, immobilized metal affinity chromatography-purified protein was concentrated and further purified by gel filtration chromatography using a Superdex S200 16/600 column equilibrated in Buffer A.
Enzyme Assays-All enzyme assays unless otherwise stated were carried out in 20 mM Na-Hepes buffer, pH 7.5., containing 100 mM NaCl. Assays were carried out with 1 M BT3780 against 1 mg/ml substrate at 37°C for up to 16 h. Aliquots were taken over a 16-h time course, and samples and products were assessed by TLC and high pressure anion exchange chromatography (HPAEC) with pulsed amperometric detection. Sugars were separated on a Carbopac PA200 guard and analytical column in an isocratic program of 100 mM sodium hydroxide. Sugars were detected using the carbohydrate standard quad waveform for electrochemical detection at a gold working electrode with an Ag/AgCl pH reference electrode. Kinetic parameters were determined using the D-mannose detection kit from Megazyme International, measuring the release of mannose at absorbance of 340 nm. To determine kinetic parameters, 2 M of BT3780 was assayed against varying concentrations of polysaccharide or oligosaccharides between 0.1 and 2 mM. Mannose release was measured, and the values were plotted using linear regression giving k cat /K m as the slope of the line. Mutants were assessed for activity against C. albicans mannan at 1 mg/ml with varying enzyme concentrations between 1 and 200 M. All assays were carried out in triplicate.
NMR Spectroscopy-Reaction buffer 10ϫ (1ϫ, 20 mM sodium phosphate buffer, pH 7.5, containing 100 mM NaCl) and C. albicans mannan were freeze dried and resuspended in D 2 O twice prior to the experiment. BT3870 was transferred to reaction buffer in D 2 O by extensive buffer exchange. Initial spectra were recorded of 900 l of 10 mg ml Ϫ1 C. albicans mannan in reaction buffer before initiating the reaction by the addition of 100 l of BT3780 (final concentration, 30 M). 1 H NMR spectra were recorded in D 2 O on a Bruker Avance III HD 500 MHz NMR spectrometer operating at 500.15 MHz at regular intervals. The chemical shift is quoted in ppm relative to tetrameth-ylsilane, and each spectrum was acquired with 16 scans. Spectra of mannose and mannose-1-phosphate (2.5 mM in reaction buffer) standards were also recorded.
Crystallography-BT3780 purified by immobilized metal affinity chromatography and size exclusion chromatography was concentrated in a 30-kDa cutoff centrifugal concentrator and buffer-exchanged into H 2 O prior to crystallization. Crystals were obtained in 2.2 M NH 4 SO 4 , with 0.3 M mannose in 96-well sitting drop TTP Labtech plates (200-nl drops) with BT3780, at 15 mg/ml. Crystals were cryoprotected with saturated NH 4 SO 4 , and data were collected at Diamond Light Source on Beamline I04-1 ( 0.92 Å) at 100 K. The data were integrated with XDS (23) and scaled with Aimless (24). Five percent of observations were randomly selected for the R free set. Space groups were determined using Pointless (25). The phase problem was solved by molecular replacement using the program Molrep (47) and the PDB search model 3QC2. Initial phases were used in the program Buccaneer (25) to automatically build the model. Solvent molecules were added using COOT (26) and checked manually. The model underwent cycles of model building in COOT (26) and refinement in Refmac (27). All other computing used the CCP4 suite of programs (28).The model was validated using MolProbity (29), and data statistics and refinement details are reported in Table 1. The model was deposited in the PDB and given the code 5A7V. Structure representations were made in PyMOL (version 1.7.4, Schrödinger).

Results
Biochemical Properties of BT3780 -The genome of B. thetaiotaomicron contains three PULs: MAN-PUL1, MAN-PUL2, and MAN-PUL3, which orchestrate the degradation of yeast ␣-mannan (21). Within MAN-PUL2 is a gene encoding BT3780, a member of CAZy family GH130. This family currently comprises ␤-D-mannoside phosphorylases that mediate bond cleavage through a single displacement mechanism with phosphate as the nucleophile, resulting in the inversion of anomeric configuration and the generation of ␣-mannose-1-phosphate. Thus, the possible linkages in Saccharomyces cerevisiae ␣-mannan targeted by BT3780 are Man-␤1,4-GlcNAc (possibly Man-␤1,4-GlcNAc-␤1,4-GlcNAc) and ␣-Man-1-phosphate (where cleavage would then be associated with a reverse phosphorolysis reaction). The data presented in Table 2 show that BT3780 displayed no exo-or endo-activity against these linkages in the presence or absence of phosphate. Yeast and fungal ␣-mannans are not all identical, and in the C. albicans ␣-mannan the mannose side chains are capped by one or more ␤-1,2-mannose units, which are therefore also potential substrates for BT3780. Incubation of C. albicans ␣-mannan with BT3780 released a product that co-migrates with mannose on HPAEC and TLC (Fig. 2), suggesting that the enzyme indeed targets these ␤1,2-mannose units. Furthermore, the inability of BT3780 to release mannose from S. cerevisiae mannan, which lacks the capping ␤-Man residues, is also consistent with this proposed specificity for ␤1,2-mannosyl linkages. The specificity of BT3780 for ␤-1,2-Man linkages was further supported by the observation that 212 and 620 mol of mannose were released from 0.5 mg of C. albicans ␣-mannan by the side chain cleaving ␣-mannosidase, BT3774 (21), before and after treatment of the glycan with BT3780, respectively. Several additional lines of evidence showed that BT3780 generated mannose and not mannose-1-phosphate. Thus, the activity of BT3780 could be monitored by standard mannose detection kits in which mannose-1-phosphate is not a substrate for the linker enzymes (Megazyme International D-mannose/D-fructose/D-glucose assay kit); NMR spectra of the sugar released by BT3780 revealed H 1 signals corresponding to both the ␣ and ␤ anomers of mannose, whereas the signal for the anomeric hydrogen in mannose-1-phosphate, with a chemical shift of 5.28 ppm, was absent (Fig. 3). Thus, the enzyme does not medi-ate phosphorolysis or reverse phosphorolysis reactions and therefore is not a glycoside phosphorylase. These data instead show that BT3780 is an exo-acting glycoside hydrolase that hydrolyzes ␤1,2-mannosidic linkages. The observation that the enzyme displayed no detectable activity against mannoselinked ␤1,4-linked to mannose or glucose (in mannans and glucomannans, respectively) or GlcNAc (Table 2) confirms its tight specificity for ␤-1,2-mannosidic linkages.
To explore further the specificity of BT3780, the catalytic activity of the enzyme against ␤1,2-mannooligosaccharides was determined. The data in Table 2 show that the enzyme exhibited low activity against the oligosaccharides with a K m that was too high to quantify, indicating weak affinity for these substrates. Given that the BT3780 displayed similar k cat /K m values for oligosaccharides with a degree of polymerization ranging from 2 to 4, the enzyme appears to contain only two subsites: Ϫ1 and ϩ1. Phosphate did not influence the hydrolytic activity of BT3780, and mannose-1-phosphate did not participate in reverse phosphorolysis reactions with a range of sugars to generate ␤-linked disaccharides (data not shown). GH130 mannose-phosphorylases invert the anomeric configuration of the phosphorylated mannose residue generating ␣-mannose-1phosphate (16). NMR analysis of the reaction products generated by BT3780 was used to determine whether the enzyme also hydrolyzed mannosidic linkages through a single displacement (inverting) mechanism. The data in Fig. 3 show that the initial  product was ␣-D-mannose, which subsequently mutarotated to a 2:1 ratio of the ␣and ␤-anomers of the sugar. C. albicans mannan polysaccharide was used as the substrate for the NMR experiment, which contributes a number of other features to the NMR spectra, including a peak at 4.99 ppm. The increase of this peak with incubation time likely corresponds to ␣-mannose side chains that are not capped by ␤-1,2 mannosides, which accumulate as digestion with BT3780 proceeds (30). These data show that both the glycoside phosphorylases and the glycoside hydrolases in family GH130 mediate bond cleavage though a single displacement inverting mechanism.
Crystal Structure of BT3780 -To explore the structural basis for the catalytic activity and specificity of BT3780, the crystal structure of the enzyme was determined. Residues 17-384 (the first 16 residues is a cleaved signal peptide, and residue 383 is the C-terminal Pro) were built into the 1.35 Å electron density map (Table 1 and Fig. 4A). The ␤-mannosidase displayed a fivebladed ␤-propeller fold. Each blade consists of a ␤-sheet generally comprising four antiparallel strands, although blade 5 only contained three ␤-strands. Blade 4 contains a loop insertion in ␤-strand 1 extending from Cys 269 to Ala 273 . The blades are arranged radially around the central axis and are strongly twisted. The ␤-sheets from the five blades pack face to face, with hydrophobic interactions and hydrogen bonds as observed for other ␤-propeller proteins. The propeller has a cylindrical shape with diameter and height of ϳ40 Å and is present as a monomer in solution (data not shown). Most ␤-propeller proteins are "closed" by the completion of the C-terminal fourstranded sheet through incorporation of a strand from the N terminus, or vice versa, colloquially termed "molecular Velcro." This is believed to provide considerable stabilization to the fold (31). Non-Velcroed propellers are rare, having primarily been described only for the seven-bladed prolyloligopeptidase, where the resultant flexibility is believed to facilitate substrate transfer (32). In BT3780 blades 1 and 5 are derived exclusively from N-and C-terminal sequences, respectively, and thus, no classical "Velcro" is present. There are, however, three hydrogen bonds between the N-and C-terminal blades. Of more significance is the long N-terminal loop extending from residues 17 to 70, which makes numerous polar contacts with the C-terminal blade and thus confers considerable stabilization of the protein fold.
BT3780 exhibits structural similarity to a range of proteins that display a five-bladed ␤-propeller fold, including several members of GH130. The closest homolog used as molecular replacement model is with the B. ovatus enzyme BACOVA_03624 (PDB code 3QC2, z score of 66, and root mean square deviation of 0.4 Å over 357 C␣ atoms with sequence identity of 86%), which was previously shown to crystalize as a monomer. Both proteins are very similar, with identical arrangement of ␤-strands in the five blades. Inspection of the secondary structure of the only cytoplasmic GH130 Bacteroides fragilis ␤-1,4-mannosylglucose phosphorylase BF0772 (PDB code 3WAS) reveals an extended ␣-helix at the N and C termini. These helices mediate protomer interactions in the B. fragilis enzyme leading to a hexameric structure (20). The absence of these helices in BT3780 and BACOVA_03624 explains why these ␤-1,2-mannosidases are monomers, although in the GH130 Man-␤1,4-GlcNAc phosphorylase UhgbMP, the hexameric structure is mediated by interactions between the ␤-propellers (19).  1 H NMR spectra were recorded in D 2 O on a Bruker Avance III HD 500 MHz NMR spectrometer operating at 500.15 MHz. The chemical shift is quoted in ppm relative to tetramethylsilane, and each spectrum was acquired with 16 scans. The spectrum of 10 mg/ml C. albicans mannan in 20 mM sodium phosphate, pH 7.5, 100 mM NaCl, was recorded prior to and after addition of 30 M BT3780, at the time points indicated on the graph. A peak at the chemical shift corresponding to the H-1␣ proton of mannose (5.10 ppm) was detected after 2 min, and mutarotation to the ␤-anomer was observed subsequently (4.82 ppm), indicating that the reaction proceeds with inversion of anomeric configuration. Spectra of mannose and mannose-1-phosphate (anomeric hydrogen 5.28 ppm) in reaction buffer are also shown.

JOURNAL OF BIOLOGICAL CHEMISTRY 25027
Active Site of BT3780 -An extended pocket is located in the center of the ␤-propeller with all five blades contributing to the topology of the pocket. The pocket houses two mannose residues (the crystal structure of BT3780 was obtained in the presence of mannose), and its location corresponds to the active site of BF0772 and thus fulfills the same function in the B. thetaiotaomicron enzyme (Fig. 4B). The mannose residue in the ϩ1 subsite (Man2) adopts a classic 4 C 1 chair conformation. In the Ϫ1 subsite (active site) (33); however, the mannose (Man1) is in a B 2,5 / O S 2 conformation, typical of the geometry adopted by the oxocarbenium ion transition state in ␤-mannanases and ␤-mannosidases (Fig. 4C) (34,35). The interactions between the enzyme and products bound at the Ϫ1 and ϩ1 subsites are detailed in Fig. 5  A is a schematic of BT3780 revealing the five-bladed propellerfold ramp colored blue at the N terminus to red at the C terminus. The ligand is shown in background as stick representation with green carbon atoms, red oxygen atoms, and blue nitrogen atoms. B shows a surface representation of BT3780 with mannose residues bound in the active site pocket comprising the ϩ1 and Ϫ1 subsites. The view is rotated 180°along a horizontal axis compared with A (ligand in foreground). The surface is represented semitransparent in gray. The secondary structure elements are represented in red for ␣-helices, yellow for ␤-strands, and green for loops. C shows the electron density map (2F o Ϫ F c ) of the two mannose residues at 1.5 . The electron density is shown in blue. the mannan side chains and thus expose the ␣-linked mannosidic linkages to attack by ␣-mannosidases such as BT3774 (see above). The importance of these active site residues in the activity of BT3780 was explored by site-directed mutagenesis. The data in Table 3 show that Asp 142 , Lys 199 , and Asp 363 all played a critical role in catalysis because substitution of these residues led to complete loss in mannosidase activity. The 40-fold reduction in k cat /K m of N74A against Candida mannan, compared with the wild type enzyme, indicates that Asn 74 also contributes to the activity of BT3780. R89A, however, was only 7-fold less active than wild type BT3780, and thus, although Arg 89 is highly conserved (see below) and apparently makes several interactions with substrate at the ϩ1 subsite, this residue makes little contribution to the catalytic efficiency of the enzyme.
The Catalytic Apparatus of BT3780 -A distinctive feature of the active site of BT3780 is the lack of residues that could easily fulfill the role of a classical catalytic general acid. Asp 142 is the only potential catalytic acid because it is invariant in GH130 (see below) and is closest to the scissile bond. The residue, however, is too distant (4.8 Å) to make direct polar contact with the glycosidic oxygen (Fig. 5). It is possible that this acidic residue donates a proton to the glycosidic oxygen via solvent. It has been proposed that the equivalent residue in the Man-␤1,4-Glc and Man-␤1,4-GlcNAc phosphorylases BF0772 and UhgbMP (Asp 131 and Asp 104 , respectively) activates a proton relay culminating in the protonation of the glycosidic oxygen (19,20). In this mechanism the glycosidic oxygen abstracts the proton from O 3 , which in turn receives a proton from O␦2 of the aspartate. Support for this proposal is provided by BF0772 and UhgbMP, in which O 3 of Man1 is within hydrogen bonding distance to O 4 of the Glc and GlcNAc, respectively, which occupy the ϩ1 subsite. The importance of the carboxylate of the aspartate is demonstrated by the catalytically inactive mutant D142N. It should be emphasized that to date only a direct interaction between the catalytic acid and the glycosidic oxygen has been observed. It should be noted, however, that in other glycoside hydrolases the substrate can provide direct catalytic nucleophilic assistance, typically the carbonyl of the C2 acetamido group in GH18 and GH20 N-acetylglucosaminidases (36) but also the O 2 of the active site mannose in GH99 endo-␣1,2mannosidases (37). Thus, the provision of catalytic groups by the substrate is not without precedent. The environment of Asp 142 in BT3780 is not obviously apolar, and there is no residue that could act as a direct pK a modulator of the aspartate. Thus, the identity of the catalytic acid of BT3780 remains opaque. It is possible that BT3780 lacks a canonical catalytic acid, which may explain the very modest activity displayed by the enzyme.
In GH130 mannoside phosphorylases, the phosphate acts as the nucleophile mounting a direct nucleophilic attack on the anomeric carbon. In inverting glycoside hydrolases, an activated water molecule acts as the catalytic nucleophile. In BT3780 the glutamates Glu 227 and Glu 268 lie on the appropriate face of Man1 to facilitate nucleophilic attack by water at the anomeric center. The carboxylates of both Glu 227 and Glu 268 hydrogen bond to the ␣-linked O 1 hydroxyl of the mannoside with distances of ϳ2.7 Å (Fig. 5), and either could potentially activate a water molecule in the single-displacement mechanism. A similar pair of carboxylic amino acids also hydrogen bond to the catalytic nucleophilic water of inverting GH67 ␣-glucuronidases, making the assignment of the catalytic base difficult (38). The mutation of both glutamates in BT3780 to Gln resulted in a relatively modest reduction in catalytic activity, 30-fold for Glu 227 and 200-fold for Glu 268 . Substitution of the catalytic residues of glycoside hydrolases generally results in an effectively inactive enzyme (39,40); however, mutation of the candidate catalytic base of several inverting glycanases did not result in complete loss of enzymatic activity enzymatic activity (41,42). It is possible that the Glu 227 /Glu 268 pair influences the position and pK a of the other, and thus both residues may contribute to the function of the catalytic base. Mutation of either of these residues could increase the polarity of the other, enabling it to fulfill a catalytic base function. Indeed, in retaining glycanases mutating the catalytic nucleophile or occupation of the active site with substrate results in the deprotonation of the catalytic acid/base, which thus functions exclusively as a base (43,44). The redundant identity of the catalytic base in BT3780 is consistent with the complete loss in activity only when both glutamates were substituted for glutamine (Table 3).
Functional Phylogeny-This report provides functional insight into the phylogeny of GH130. In a previous study this family was classified into two distinct subfamilies, GH130_1 and GH130_2, whereas the other sequences were too heterogeneous to be grouped into a single subfamily and were thus defined as GH130_NC (18). To date, the enzymes characterized from GH130_1 and GH130_2 are mannoside phosphorylases. In these enzymes, phosphate, which comprises the catalytic nucleophile, is positioned below the ␣-face of the Ϫ1 Man and makes polar interactions with three basic residues (Fig. 6). Members of the GH130_NC grouping, which includes BT3780 and BACOVA_03624, lacked the three basic residues that interact with phosphate in mannoside phosphorylases, and it was proposed that these enzymes might be glycoside hydrolases (18). The data published here are entirely consistent with this hypothesis. To evaluate further this proposal, the biochemical properties of another member of GH130_NC, BACOVA_03624 from B. ovatus, were evaluated. The data show that the enzyme is also a ␤-1,2-mannosidase (Table 2), supporting the functional classification of the GH130_NC grouping as ␤-mannosidases. Apart from the acidic/basic residue substitution, the other residues that bind to the Ϫ1 Man in BT3780 are conserved in mannoside phosphorylases (Figs. 6A  and 7). This suggests that the boat conformation of the bound

Relative catalytic activity of mutants of BT3780
The activity of the enzymes were determined using 1 mg/ml C. albicans mannan as the substrate.

Variant of BT3780
Relative activity mannose within GH130 enzymes is independent of hydrolytic or phosphorolytic cleavage of the glycosidic bond. Another distinguishing feature between GH130_NC and the two GH130 subfamilies GH130_1 and GH130_2, is a tyrosine/ glutamate substitution adjacent to the putative catalytic acid. It was previously proposed that the glutamate was the catalytic base (18), when no GH130 structure in complex with substrates was available. The residue in BT3780 (Glu 141 ), however, interacts with O 4 and O 6 of mannose at the ϩ1 subsite, and these hydrogen bonds replace the partial hydrophobic platform afforded by the tyrosine in GH130_1/2 enzymes (Fig. 6B). This glutamate is thus not in an appropriate position to act as the catalytic base.
In general, the ϩ1 subsite of GH130 enzymes are not highly conserved. This likely reflects the different specificities displayed by this family in which the scissile linkage can be ␤-1,2 or ␤-1,4 and the ϩ1 sugar Glc, Man, or GlcNAc. This is exemplified by BT3780 and BF0772, which target ␤-1,2-Man-Man and ␤-1,4-Man-Glc linkages, respectively. Consequently, the sugars in the ϩ1 subsite of these two enzymes are in a perpendicular orientation and rotated with respect to each other. Nevertheless Arg 89 , which interacts with O 3 and O 4 of Man2 in BT3780, is conserved in BF0772 where the basic residue makes hydrogen bonds with equatorial O 2 and O 3 of Glc (Fig. 6B) and explains why the B. fragilis enzyme does not hydrolyze Man-␤-1,4-Man linkages. The other residue that makes polar contacts with Glc at the ϩ1 subsite of BF0772 is Arg 94 . Although this amino acid is conserved in BT3780 (Arg 101 ), the basic residue does not interact with Man2 and is not invariant in GH130 (Fig. 7).
A key specificity determinant in BT3780 is likely to be Lys 199 . The lysine is the only residue that interacts with O 2 of Man2 (Fig. 5) and thus is likely to confer specificity for Man-␤1,2-Man linkages. Indeed, Lys 199 is conserved in the other three enzymes that are known to target Man-␤1,2-Man linkages (Fig. 7). These enzymes display mannosidase (BACOVA_03624) ( Table 2) and mannoside phosphorylase (Teth514_1788 and Teth514_1789) activities (13). The basic residue, however, is not conserved in mannoside phophorylases that target ␤1,4-Man linkages, irrespective of the nature of the ϩ1 sugar (Fig. 7).

Discussion
This report supports the hypothesis that the GH130 family contains glycoside hydrolases in addition to mannoside phosphorylases. The two characterized GH130 mannosidases target Man-␤1,2-Man linkages. The predicted GH130 glycoside hydrolases, based on substitution of basic residues with glutamates, also contain the lysine ␤1,2-Man specificity determinant. It remains unclear whether other GH130 glycoside hydrolases display additional specificities. The catalytic efficiency of BT3780 is modest, ϳ1000-fold lower than typical glycosidases. It is possible that this reflects in part the absence of a canonical catalytic acid leading to a low k cat . In a recent study, it was shown that B. thetaiotaomicron degrades yeast mannan through a selfish mechanism, which requires slow acting surface enzymes exemplified by the ␤1,2-mannosidase described here (21). It is also possible that although BT3780 is active on the Man-␤1,2-Man oligosaccharides of C. albicans mannan, its true substrate is a different fungal cell wall, which contains ␤-Man in an alternative context. Indeed, it is also possible that other enzymes contribute to the degradation of C. albicans mannan. Although few data are available on the fine structure of the cell wall mannans of other gut fungi, it should be noted that the Agaricus brasiliensis cell wall contains sulfated ␤1,3-Glc-␤1,2-Man (45), whereas Hericium erinaceus produces ␤-1,3-branched-␤-1,2-mannan (46). Although the biological rationale for both ␤-mannosidases and ␤-mannoside phosphorylases in GH130 is unclear, it should be noted that the phosphorylases are cytoplasmic, whereas the hydrolases are secreted. These different cellular locations suggest that the mannosidases target complex substrates that cannot be imported into the periplasm, whereas the phosphorylases, using intracellular inorganic phosphate, cleave and activate the glycone sugar enabling the phosphorylated molecule to enter cytoplasmic metabolic pathways. Based on the presence of signal peptide, the absence of the three basic residues that interact with phosphate in mannoside phosphorylases, conservation of the motifs comprising the predicted catalytic base of ␤-mannosidases, and of the lysine that confers linkage specificity, it is thus now possible to predict the catalytic mechanism and the ␤-mannoside substrates of the hundreds of uncharacterized enzymes within GH130. Because many of these enzymes, exemplified by BT3780 and BACOVA_03624, are highly prevalent in the human gut microbiome, these data provide insight into glycan utilization in this microbial ecosystem.