α-N-Acetylgalactosaminidase from Infant-associated Bifidobacteria Belonging to Novel Glycoside Hydrolase Family 129 Is Implicated in Alternative Mucin Degradation Pathway*

Background: The degradation pathway of the intestinal mucin by bifidobacteria is poorly understood. Results: A novel α-N-acetylgalactosaminidase, NagBb, was identified from Bifidobacterium bifidum JCM 1254. Conclusion: NagBb might be involved in intracellular degradation of Tn antigen (GalNAcα1-Ser/Thr). Significance: NagBb represents a novel glycoside hydrolase family 129 in the CAZy database. Bifidobacteria inhabit the lower intestine of mammals including humans where the mucin gel layer forms a space for commensal bacteria. We previously identified that infant-associated bifidobacteria possess an extracellular membrane-bound endo-α-N-acetylgalactosaminidase (EngBF) that may be involved in degradation and assimilation of mucin-type oligosaccharides. However, EngBF is highly specific for core-1-type O-glycan (Galβ1–3GalNAcα1-Ser/Thr), also called T antigen, which is mainly attached onto gastroduodenal mucins. By contrast, core-3-type O-glycans (GlcNAcβ1–3GalNAcα1-Ser/Thr) are predominantly found on the mucins in the intestines. Here, we identified a novel α-N-acetylgalactosaminidase (NagBb) from Bifidobacterium bifidum JCM 1254 that hydrolyzes the Tn antigen (GalNAcα1-Ser/Thr). Sialyl and galactosyl core-3 (Galβ1–3/4GlcNAcβ1–3(Neu5Acα2–6)GalNAcα1-Ser/Thr), a major tetrasaccharide structure on MUC2 mucin primarily secreted from goblet cells in human sigmoid colon, can be serially hydrolyzed into Tn antigen by previously identified bifidobacterial extracellular glycosidases such as α-sialidase (SiaBb2), lacto-N-biosidase (LnbB), β-galactosidase (BbgIII), and β-N-acetylhexosaminidases (BbhI and BbhII). Because NagBb is an intracellular enzyme without an N-terminal secretion signal sequence, it is likely involved in intracellular degradation and assimilation of Tn antigen-containing polypeptides, which might be incorporated through unknown transporters. Thus, bifidobacteria possess two distinct pathways for assimilation of O-glycans on gastroduodenal and intestinal mucins. NagBb homologs are conserved in infant-associated bifidobacteria, suggesting a significant role for their adaptation within the infant gut, and they were found to form a new glycoside hydrolase family 129.


Bifidobacteria inhabit the lower intestine of mammals including humans where the mucin gel layer forms a space for commensal bacteria. We previously identified that infant-associated bifidobacteria possess an extracellular membrane-bound endo-␣-N-acetylgalactosaminidase (EngBF) that may be involved in degradation and assimilation of mucin-type oligosaccharides.
However, EngBF is highly specific for core-1-type O-glycan (Gal␤1-3GalNAc␣1-Ser/Thr), also called T antigen, which is mainly attached onto gastroduodenal mucins. By contrast, core-3-type O-glycans (GlcNAc␤1-3GalNAc␣1-Ser/Thr) are predominantly found on the mucins in the intestines. Here, we identified a novel ␣-N-acetylgalactosaminidase (NagBb) from Bifidobacterium bifidum JCM 1254 that hydrolyzes the Tn antigen (GalNAc␣1-Ser/Thr). Sialyl and galactosyl core-3 (Gal␤1-3/4GlcNAc␤1-3(Neu5Ac␣2-6)GalNAc␣1-Ser/Thr), a major tetrasaccharide structure on MUC2 mucin primarily secreted from goblet cells in human sigmoid colon, can be serially hydrolyzed into Tn antigen by previously identified bifidobacterial extracellular glycosidases such as ␣-sialidase (SiaBb2), lacto-N-biosidase (LnbB), ␤-galactosidase (BbgIII), and ␤-Nacetylhexosaminidases (BbhI and BbhII). Because NagBb is an intracellular enzyme without an N-terminal secretion signal sequence, it is likely involved in intracellular degradation and assimilation of Tn antigen-containing polypeptides, which might be incorporated through unknown transporters. Thus, bifidobacteria possess two distinct pathways for assimilation of O-glycans on gastroduodenal and intestinal mucins. NagBb homologs are conserved in infant-associated bifidobacteria, suggesting a significant role for their adaptation within the infant gut, and they were found to form a new glycoside hydrolase family 129.
Bifidobacteria naturally occur in human intestines and are predominant in those of newborn infants. They give various beneficial effects to the host such as stimulation of the immune response, prevention of the growth of pathogenic enterobacteria, and suppression of inflammatory and allergic responses; therefore, they are recognized as probiotics (1). Because they mainly reside in the lower intestines where the sugars are highly limited, they possess various glycosidases to hydrolyze indigestible oligosaccharides and glycoconjugates. Endo-␣-Nacetylgalactosaminidase (EC 3.2.1.97) acts on the O-glycosidic linkage of GalNAc␣1-O-Ser/Thr in the core structures of mucin-type oligosaccharides to release oligosaccharides from glycoproteins. Several pathogenic and non-pathogenic bacteria possess enzymes with this type of activity (2)(3)(4)(5). Previously, we identified the gene engBF encoding endo-␣-N-acetylgalactosaminidase from Bifidobacterium longum subsp. longum JCM 1217 and subsequently found that the homologous genes are conserved in various infant-associated bifidobacteria (6). The EngBF and its homologs in bacteria were classified into a new glycoside hydrolase (GH) 3 family, 101, in the CAZy database (7). EngBF showed relatively strict specificity toward the core-1-type O-glycan (Gal␤1-3GalNAc␣1-Ser/Thr) compared with the other GH101 enzymes (8 -10). The ␤1,3-linked Gal in the core-1 structure is known to be highly resistant to common ␤-galactosidases from general intestinal bacteria. However, bifidobacteria can release the core-1 disaccharide from mucin glycoproteins through the action of the extracellular membrane-bound EngBF, incorporate it into the cytosol through a specific ATP-binding cassette-type transporter on the plasma membrane (11)(12)(13), and finally phosphorolyze it by a specific phosphorylase in the cytosol (14). These findings suggest that bifidobacteria selectively grow in the intestines by assimilating mucin-type oligosaccharides utilizing a unique degradation pathway. In fact, the isomeric disaccharide Gal␤1-3GlcNAc, a building unit of the type-1 chain in human milk oligosaccharides, which can be recognized by the same transporter and phosphorylase, specifically enhanced the growth of bifidobacteria in vitro (15).
Mucins are secreted from specific mucin-producing epithelial cells in mucosal tissues and form a gel layer to protect the epithelia from pathogenic bacteria and viruses, digestive enzymes, and acidic digestive juice. In the gastrointestinal tract, the stomach and duodenum mainly secrete MUC5AC and MUC6, whereas the small and large intestines secrete MUC2, MUC3, and MUC4. It has been reported that MUC2 is the major mucin secreted from goblet cells in human colon (16) and that a swollen layer of MUC2 supplies the space for commensal bacteria (17,18). The structures of O-glycans on mucin are quite different depending on the source of mucosal tissues. Gastric and duodenal mucins generally contain the core-1 and the core-2 (Gal␤1-3(GlcNAc␤1-6)GalNAc␣1-Ser/Thr) structures. Recent studies revealed that MUC2 in the sigmoid colon mainly contains the core-3 structure (GlcNAc␤1-3GalNAc␣1-Ser/Thr) (19). As mentioned above, EngBF acts on the core-1 structure with a 300 times higher rate than core-3; thus, we speculate that EngBF is involved in degradation of gastroduodenal mucins, which are released and transported from the stomach and duodenum.
To better understand the degradation and assimilation of the intestinal mucin by bifidobacteria, we searched the genome of Bifidobacterium bifidum JCM 1254 using the sequence of EngBF as a query. As a result, we found an uncharacterized gene encoding a putative intracellular protein showing very slight similarity to EngBF. Heterologous expression of this protein revealed that it showed glycosidase activity with substrate specificity distinct from EngBF. Interestingly, its putative homologs were conserved in infant-associated bifidobacteria, and they were found to form a novel GH family.

EXPERIMENTAL PROCEDURES
Culture and Genome Sequence of B. bifidum JCM 1254-B. bifidum JCM 1254 was cultured on Gifu Anaerobic Medium (GAM) broth (Nissui Pharmaceutical, Japan) for 16 h at 37°C under anaerobic conditions using Anaeropack-Anaero (Mitsubishi Gas Chemical, Japan). Genomic DNA was extracted from bacterial cells using standard methods. Draft sequencing of the genome of B. bifidum JCM 1254 was performed using a Genome Sequencer 20 System (Roche Applied Science). The details will be reported elsewhere.
Expression and Purification of Recombinant NagBb-The full-length DNA fragment encoding the 634-amino acid polypeptide of NagBb was amplified by high fidelity PCR using PrimeSTAR HS DNA polymerase (Takara Bio, Japan), genomic DNA from B. bifidum JCM 1254 as a template, and the following primers: forward, 5Ј-ggaattcgatgatgcaattcaccatgtcc; and reverse, 5Ј-actcgagatgagaaccgttcgctgcta. The PCR product was digested with EcoRI and XhoI and ligated into the corresponding sites of pET23b(ϩ) (Novagen, Germany) to express C-terminal His 6 -tagged protein. The Escherichia coli BL21(DE3)⌬lacZ strain that lacks ␤-galactosidase activity was transformed with the constructed plasmid and cultured in Luria-Bertani liquid medium containing 100 g/ml ampicillin at 25°C. When the optical density at 600 nm (A 600 ) reached 0.5, isopropyl ␤-D-1-thiogalactopyranoside was added to the culture to a final concentration of 0.5 mM. After additional culture for 5 h, cells were harvested and lysed in BugBuster Protein Extraction Reagent (Novagen) to obtain cell-free extract. His 6tagged NagBb was purified by a HisTrap HP column (1 ml; GE Healthcare) followed by Superdex 200 10/300 GL (GE Healthcare) gel filtration in line with the ÄKTA Explorer (GE Healthcare). Active fractions were collected, concentrated, and desalted using an Amicon Ultracel-10K (EMD Millipore, Germany). Protein concentrations were determined by the BCA protein assay reagent (Thermo Scientific-Pierce) using bovine serum albumin as a standard.
Stereochemical Analysis-GalNAc␣1-pNP (2 mM) was incubated with NagBb in 20 mM sodium acetate buffer (pH 5.0) at 37°C. The reaction mixtures were immediately analyzed by normal-phase HPLC using a TSKgel Amide 80 column (4.6 ϫ 250 mm; Tosoh, Japan). Elution was carried out at 25°C using acetonitrile/water (3:1, v/v) as a solvent at a flow rate of 1.5 ml/min, and the absorbance was monitored at 214 nm.
Transglycosylation Reaction-NagBb (35 ng/l) was incubated with 3 mM GalNAc␣1-pNP as a donor in the presence of 100 mM Ser as an acceptor in 50 mM sodium acetate buffer (pH 5.0) at 37°C. The reaction mixture was analyzed by the Prominence amino acid analyzer equipped with a postcolumn fluorescence labeling system using o-phthalaldehyde and a Shim-pack Amino-Na column (Shimadzu, Japan) as described previously (26).
To confirm the expression of NagBb in the bifidobacterial cells, we incubated GalNAc␣1-pNP with the cell-free extracts of B. bifidum JCM 1254 cultured in the presence of several types of sugars. The activities were detected in all extracts; however, the activities were ϳ2-fold higher in the lysates from the cells cultured in the presence of GalNAc or Gal␤1-3GalNAc than in the absence of GalNAc (Table 1). Because EngBF (8) and GH36 ␣-galactosidase MelA/Aga2 (27,28) never act on GalNAc␣1-pNP, this result suggests that the nagBb gene is constitutively expressed and that its expression is enhanced in the presence of GalNAc. No activity was detected in the culture supernatants and the non-disrupted cell suspensions, indicating that NagBb is expressed intracellularly as predicted by its amino acid sequence.
Substrate Specificity of NagBb-To characterize the detailed substrate specificity of NagBb, we incubated the recombinant enzyme expressed in E. coli with various natural substrates containing ␣-linked GalNAc and analyzed the reaction mixtures by TLC (Fig. 1). NagBb completely hydrolyzed GalNAc␣1-Ser, the minimum structure of the Tn antigen. The enzyme acted very weakly on GalNAc␣1-UDP and GalNAc␣1-3Gal␤1-4Glc but not at all on Neu5Ac␣2-6GalNAc␣1-Ser (sialyl Tn antigen) and GalNAc␣1-3(Fuc␣1-2)Gal (blood group A trisaccharide). The K m and k cat values for GalNAc␣1-Ser were estimated to be 2.2 mM and 47.6 s Ϫ1 , respectively, but those for other substrates could not be determined accurately due to high K m values and low solubility. Therefore, only k cat /K m values were estimated ( Table 2). The k cat /K m value for GalNAc␣1-Ser was ϳ10 times higher than that for GalNAc␣1-pNP.
Next, we investigated endo-type activities of NagBb using synthetic pNP-substrates with various core structures of mucin-type O-glycans. Hydrolyses were monitored by measuring released para-nitrophenol (Table 2) and also by detecting released oligosaccharides (Fig. 1). As mentioned above, NagBb hydrolyzed core-1 structure (Gal␤1-3GalNAc␣1-pNP) 27 times slower than GalNAc␣1-pNP. In addition, NagBb very slowly hydrolyzed core-3 (GlcNAc␤1-3GalNAc␣1-pNP)-and core-8 (Gal␣1-3GalNAc␣1-pNP)-type structures but not other core structures. Core-7 (GalNAc␣1-6GalNAc␣1-pNP) may be sequentially hydrolyzed from the non-reducing termi- nus by exo-␣-N-acetylgalactosaminidase activity because the disaccharide was never detected. From these results, we concluded that the natural substrate for NagBb occurring in the intestines is essentially the Tn antigen. General Properties of NagBb-General enzyme properties of NagBb were determined using GalNAc␣1-pNP as a substrate. The enzyme was stable over the pH range of 3.0 -11.0 and most active at pH 5.0 Ϯ 0.5 (supplemental Fig. S3, A and B). The optimal temperature for activity was found to be 55°C, and the enzyme was stable up to 45°C (supplemental Fig. S3, C and D). These pH and thermal profiles of NagBb provide sufficient enzyme property for functioning in human intestines. The enzyme activity was inhibited in the presence of several metal ions at 5 mM such as Mn 2ϩ (reduced to 75%), Zn 2ϩ (30%), and Cu 2ϩ (7%) but was not affected by Mg 2ϩ , Ca 2ϩ , Co 2ϩ , Ni 2ϩ , and EDTA (supplemental Fig. S4A). However, the enzyme was stabilized by the addition of Mn 2ϩ (supplemental Fig. S4B) as observed for EngBF, an enzyme that has been shown to contain four Mn 2ϩ ions in the crystal (29).
Catalytic Mechanism and Critical Residues of NagBb-To determine the stereochemical course of the hydrolysis catalyzed by NagBb, GalNAc␣1-pNP was incubated with NagBb, and the anomeric configuration of the released GalNAc was analyzed by normal-phase HPLC (Fig. 2). Generally, an ␣-anomeric sugar is eluted from a normal-phase column slightly faster than the corresponding ␤-anomer (30, 31). After a 2-min a Kinetic parameters were determined by quantifying the released GalNAc using an HPLC system. Data represent mean Ϯ S.D. (n ϭ 3). b Not determined. c Sequentially hydrolyzed from non-reducing terminus by exo activity. enzyme reaction, almost exclusively GalNAc␣OH was detected (␣/␤ ratio, 94:6). Then the ratio of the ␣and ␤-anomers of GalNAc gradually changed to reach an equilibrium (␣/␤ ratio, 55:45) due to the slow mutarotation of GalNAc␣OH. This data confirmed that NagBb catalyzes the hydrolysis reaction via the retaining mechanism. We previously identified the catalytic residues of EngBF by docking analysis of its three-dimensional structure with Gal␤1-3GalNAc (29). Alignment of NagBb and EngBF revealed that several critical residues are aligned: i.e. Asp-435 in NagBb corresponds to the catalytic nucleophile Asp-789 in EngBF, and Asp-330 in NagBb corresponds to "fixer" Asp-682, the third essential residue in EngBF (supplemental Fig. S1). Asp-682 in EngBF is an important residue forming hydrogen bonds with O4 and O6 of GalNAc and O6 of Gal. However, the catalytic acid/base residue corresponding to Glu-822 in EngBF was not found in NagBb. To predict the three-dimensional structure of NagBb, we analyzed its sequence by the remote homology-based fold recognition method using the Phyre version 2 server (32). Amino acid residues 172-630 in NagBb could be modeled using GH13 ␣-amylase 1 (TVAI) from Thermoactinomyces vulgaris R-47 (Protein Data Bank code 1JI1) (33), whose active center is similar to that of EngBF, as a template. Asp-435 and Asp-330 in NagBb are located at positions close to the previously identified nucleophile and fixer, respectively, of TVAI and EngBF (supplemental Fig. S5). Both D435A and D330A NagBb mutations completely lost their catalytic activity toward GalNAc␣1-pNP similarly to what was observed for the corresponding EngBF mutants, suggesting that these residues function as nucleophile and fixer, respectively.
Transglycosylation Reaction-Some of the GH101 enzymes including EngBF catalyze a transglycosylation reaction in which the released Gal␤1-3GalNAc is transferred to the hydroxyl group of a suitable acceptor via the same anomeric linkage because they are retaining glycosidases (5,34,35). To test whether NagBb catalyzes a transglycosylation reaction, we incubated 3 mM GalNAc␣1-pNP and 35 ng/l NagBb in the presence of 100 mM Ser as an acceptor. The reaction mixture was analyzed by an amino acid analyzer equipped with a postcolumn amino group labeling system. In the presence of active enzyme, a new product was detected at the same retention time as that of the standard GalNAc␣1-Ser (Fig. 3, B and C). The product was not detected when NagBb was preheated (Fig. 3A), indicating that the new product was enzymatically synthesized by the transglycosylation reaction. Quantitatively, 0.19 mM transglycosylation product was yielded in the reaction mixture; this corresponded to approximately 6.3% of the initial concentration of the donor substrate. However, due to the limitation of solubility, we could not add the donor substrate over 3 mM concentration in the transglycosylation reaction mixture. Therefore, we tried to use GalNAc␣1-DMT, which shows much higher solubility than GalNAc␣1-pNP, as a donor. The k cat /K m value of NagBb for GalNAc␣1-DMT was determined to be 0.98 s Ϫ1 mM Ϫ1 , which is less than half of the value for GalNAc␣1-pNP (Table 2). When 3 mM GalNAc␣1-pNP in the transglycosylation reaction mixture was replaced with the same concentration of GalNAc␣1-DMT, the amount of GalNAc␣1-Ser was reduced to one-third (0.06 mM), and the transglycosy-lation yield was 2.1%. However, we could increase the GalNAc␣1-DMT concentration to 10, 20, and 30 mM, obtaining 0.22, 0.45, and 0.49 mM transglycosylation product, respectively. The transglycosylation yields against donor substrate were ϳ2.2, 2.2, and 1.6%, respectively. Because GalNAc␣1-DMT can be synthesized easily in water solution without any protection of hydroxyl groups (23), it may become a useful donor substrate for large scale transglycosylation reactions.
Phylogenetic Analysis of NagBb and Its Homologs-Database searches revealed that NagBb homologs are present in the genome of several Bifidobacterium species such as another strain of B. bifidum, PRL2010 (36); B. longum subsp. longum NCC2705 (37); B. longum subsp. infantis ATCC 15697 (38); and Bifidobacterium breve DSM20213, which are frequently found species in the intestines of newborn infants. These homologs are similar in size (627-634 amino acids), and all are predicted to be intracellular soluble enzymes. Three species, B. longum subsp. longum, B. longum subsp. infantis, and B. breve, possess highly similar homologs (96 -99% identities in amino acid sequences), and NagBb showed 76 -78% identities with these homologs. Possible NagBb homologs were found in several other enterobacteria such as Clostridium scindens and Eubacterium biforme (57 and 44% identities, respectively) but not in common species such as Bacteroides fragilis, Bacteroides thetaiotaomicron, and Clostridium perfringens. The predicted extracellular lipoproteins were also found as distant relatives in Victivallis vadensis and Candidatus Solibacter usitatus (33 and 25% identities, respectively). The phylogenetic tree of these close homologs and related proteins including conserved domains of GH101 enzymes was constructed using the neighbor-joining method (Fig. 4). Although NagBb homologs and GH101 enzymes may share a common ancestral protein, these groups can be clearly divided. Thus, we propose that NagBb and its homologs should be assigned to a new family GH 129.
the normal condition, the Tn antigen could be uncovered on MUC2. Before or after the generation of the Tn structure, core proteins may be cleaved by some kind of protease and then incorporated into the cytosol of bifidobacteria and degraded by NagBb. The transporter responsible for Tn antigen-containing peptides is currently unknown. In the vicinity of the nagBb locus, there are putative ATP-binding cassette transporter genes, one of which may be a candidate for the transporter of Tn antigen-containing peptides. NagBb homologs are conserved in infant-associated bifidobacteria, i.e. B. longum subsp. longum, B. longum subsp. infantis, and B. breve. Thus, this novel degradation pathway for core-3 O-glycan is likely important for growth and adaptation of bifidobacteria in the gut of newborn infants. Because the NagBb homolog has never been found in pathogenic and opportunistic infectious bacteria, this novel pathway could be a promising target for the development of a highly selective bifidogenic factor, a so-called "prebiotic." GalNAc␣1-Ser or ␣-GalNAc-containing glycosides might be better prebiotics, and the transglycosylation activity of NagBb may be exploited for their productions.
Exo-␣-N-acetylgalactosaminidases (␣-N-acetylgalactosaminidases; EC 3.2.1.49) have been found in GH27, GH36, and GH109 families. GH27 contains eukaryotic lysosomal ␣-Nacetylgalactosaminidases, which have generally broad specificity. A fungal enzyme, NagA, from Acremonium sp., a member of GH27, acts on both blood group A substance and Tn antigen (45). AagA from C. perfringens is the only ␣-N-acetylgalactosaminidase in GH36 and acts on blood group A substance (46). Both GH27 and GH36 contain many ␣-galactosidases and share a common catalytic mechanism (47). GH109 exclusively contains ␣-N-acetylgalactosaminidases that display an unusual mechanism involving NAD ϩ and have high activity toward blood group A substance (48). So far as we know, the enzymes belonging to these three GH families do not show any endotype activity. NagBb does not show significant sequence similarity with these enzymes, and its substrate specificity is quite different; therefore, we are identifying it as a novel GH129 exo-/endo-␣-N-acetylgalactosaminidase.