Crystal Structure of Glycoside Hydrolase Family 20 Lacto-N-biosidase from Soil Bacterium Streptomyces sp. Strain 142

Abstract

Lacto-N-biosidase hydrolyzes the β-GlcNAc or β-GalNAc bond of sugar chains to release lacto-N-biose I (Gal-β1,3-GlcNAc) or galacto-N-biose (Gal-β1,3-GalNAc) from the non-reducing end. Typical substrates for lacto-N-biosidase include type I oligosaccharides contained in human breast milk, such as lacto-N-tetraose. Lacto-N-biosidases have recently received significant attention because of their potential to synthesize milk oligosaccharides. Bifidobacterial lacto-N-biosidases belonging to glycoside hydrolase families 20 and 136 have been studied. The GH20 lacto-N-biosidases utilize a substrate-associated hydrolysis mechanism. LnbB from Bifidobacterium bifidum is the only lacto-N-biosidase with reported crystal structures in GH20. In this study, the crystal structure of the lacto-N-biosidase from Streptomyces sp. strain 142 (StrLNBase) was solved in a complex with lacto-N-biose and galacto-N-biose. The stabilizing residue, which recognizes the nitrogen atom of the N-acetyl group of the −1 subsite, and the catalytic acid/base residue, were determined to be D304 and E305, respectively. The structure of StrLNBase is similar to that of LnbB; however, in the complex with galacto-N-biose, there were two structures exhibiting different sugar conformations. A phylogenetic analysis revealed that lacto-N-biosidases discovered in the soil bacteria Streptomyces spp. and human gut bacteria Bifidobacterium spp. may be divided into two separate groups, which suggests that they evolved divergently.

Abbreviations

GH20, glycoside hydrolase family 20; LNB, lacto-N-biose I; LNBase, lacto-N-biosidase; LNT, lacto-N-tetraose; TLC, thin-layer chromatography

INTRODUCTION

Lacto-N-biosidase (LNBase) hydrolyzes the central β-1,3-glycosidic bond of lacto-N-tetraose (LNT; Galβ1-3GlcNAcβ1-3Galβ1-4Glc) and releases lacto-N-biose I (LNB; Gal-β1,3-GlcNAc) from the non-reducing end [1, 2, 3]. LNBase is classified into the glycoside hydrolase (GH) families 20 and 136. A GH20 LNBase was first discovered from the soil bacterium Streptomyces sp. strain 142 (StrLNBase) [2]. LNBase in the same GH family was discovered in Bifidobacterium bifidum (LnbB) [3]. LnbB (Accession No. ABZ78855.1) and StrLNBase (Accession No. AAD10478.1) share 38 % amino acid sequence identity over the entire sequence. The first crystal structure of the GH20 LNBases was determined using the catalytic domain (residues 41-663) of LnbB, which consists of 1,112 amino acids (PDB ID: 4H04 and 4JAW) [4]. At the present time, the crystal structure of LnbB is the only available three-dimensional structure of the GH20 LNBases. The catalytic domain of LnbB adopts a (β/α)₈ barrel fold and the hydrolysis reaction occurs via a substrate-assisted anomer-retaining mechanism (β→β), in which the N-acetyl group of GlcNAc acts as a nucleophile [4]. StrLNBase is commercially available from Takara Bio, Inc. (Shiga, Japan) (Product code: TKR 4456). LNBases can be used for discriminating between type I and type II sugar chains which compose Gal-β1,3-GlcNAc mainly found in human milk and Gal-β1,4-GlcNAc mainly found in other animal milk, respectively, as they only cleave type I sugar chains. StrLNBase also hydrolyzes β-1,4-glycosidic bonds in branched pyridylamino (PA)-oligosaccharides (Fig. 1) [1, 2]. GH20 LNBases also cleave β-1,3/4-glycosidic bonds in the sugar chain of phospholipid gangliosides to release LNB and galacto-N-biose (GNB; Gal-β1,3-GalNAc) [1, 2, 5].

Fig. 1. Structures of the oligosaccharide substrates and specificity of StrLNBase reported by Sano et al. [1, 2].

　Glycan structure presentation according to Symbol Nomenclature for Glycans (SNFG) (https://www.ncbi.nlm.nih.gov/glycans/snfg.html).

GH20 comprises LNBase (EC 3.2.1.140), β-N-acetylhexosaminidases (HexNAcases; EC 3.2.1.52), which catalyze hydrolytic removal of GlcNAc/GalNAc from the non-reducing end and 6-sulfo-β-N-acetylhexosaminidases (EC 3.2.1.−). Recently, a GH20 enzyme from Treponema denticola was shown to display activities on both type I and type II sugar chains, pNP-LacNAc and pNP-LNB, which is the first report of N-acetyllactosaminidase (LacNAcase) activity, despite at a low activity [6].

GH20 LNBase has attracted attention as an enzyme used for the synthesis of LNT, which is the primary core structure of human milk oligosaccharides [7]. For example, the enzymatic synthesis of LNT at 3.7 % yield using the transglycosylation reaction of the LNBase from Aureobacterium sp. L-101 with p-nitrophenyl (pNP)-LNB (donor) and lactose (acceptor) was reported in 1999 [8]. After the structural information for LnbB was reported, various site-directed mutants at the active site residues of LNBase were generated and used for LNT synthesis. The transglycosylation yields of the W394F and W394A mutants of LnbB were 32 % and 16.6 %, respectively [9]. Mutagenesis-driven, high-yield LNT synthesis was achieved based on the structural and functional information of LnbB, which identified the active site residues responsible for substrate recognition and catalysis. In this study, we solved the structure of GH20 StrLNBase, which has not yet been achieved since its discovery in 1993, and compared its structure and function with that of LnbB.

EXPERIMENTAL MATERIALS AND METHODS

Enzyme production and purification

The expression plasmid was designed based on the amino acid sequence (Accession No. AAD10478.1) lacking the N-terminal signal peptide (1-30) as predicted by SignalP 5.0. A codon-optimized, pET28a-based plasmid DNA was synthesized by GenScript Biotech Corporation (Tokyo, Japan). To improve the protein solubility, the coding sequence of the expression plasmid was transferred into pCold-proS2 (Takara Bio Inc.). The expression plasmid was then introduced into Escherichia coli BL21(DE3)/pRARE2. The transformants were cultured in lysogeny broth containing 100 μg/mL ampicillin at 37 °C for 3 h until reaching an O.D. of 0.5, then cooled at 15 °C for 30 min. Isopropyl 1-thio-β-D-galactopyranoside was added to a final concentration of 0.1 mM to induce protein expression. Following incubation at 15 °C for 24 h, the cells were harvested by centrifugation and suspended in 50 mM HEPES-NaOH (pH 8.0). Cell extracts were obtained by sonication and centrifuged to remove cell debris. The protein was purified using the cOmplete His-Tag Purification Resin (Roche Ltd., Basel, Switzerland). The sample solution was exchanged for imidazole-free HEPES buffer and concentrated using a 30 kDa cut-off Amicon Ultra (Millipore Corporation, Burlington, MA, United States). The protein solution was treated with HRV 3C Protease (Takara Bio Inc.) to remove the ProS2 solubilization tag (23 kDa) as well as the His₆-tag. The tag-free protein was collected in flow-through fractions from the cOmplete His-Tag Purification Resin. The protein was further purified by Superdex 200 pg 16/60 column chromatography (Cytiva, Tokyo, Japan) and the concentration was determined using a Nanodrop instrument based on the absorbance at 280 nm. Theoretical molar absorption coefficients (1.669 M⁻¹cm⁻¹) were also used for purified protein samples to check for consistency.

Crystallography

Crystals complexed with LNB were obtained at 20 °C using the sitting drop vapor diffusion method with microseeding. A protein solution containing 5.0 mg/mL StrLNBase and 10 mM LNB (0.6 μL), a reservoir solution containing 0.3 M NH₄F, 10 % (w/v) PEG3350 (0.4 μL), and a crushed crystal seed solution diluted 1000-fold (0.2 μL) were mixed to prepare crystallization drops. Crystals complexed with GNB (GNB complex 1) were obtained by mixing a protein solution containing 5.0 mg/mL StrLNBase and 10 mM GNB (0.6 μL), a reservoir solution [0.3 M NH₄F, 12.5 % (w/v) PEG3350] (0.4 μL), and a seed solution (0.2 μL). For crystals complexed with GNB (GNB complex 2), the concentration of NH₄F in the reservoir solution was changed to 0.25 M, and the seed solution was diluted 100-fold. LNB was kindly provided by Prof. Takane Katayama and Prof. Motomitsu Kitaoka. GNB was purchased from Biosynth Ltd. (Berkshire, UK). Before data collection, the crystals were cryoprotected by the reservoir solution supplemented with 20 % (v/v) PEG400 and 20 % (v/v) MPD for the LNB and GNB complexes, respectively. The crystals were flash-cooled by dipping into liquid nitrogen. Diffraction data were collected at 100 K using beamlines (BL-1A and AR NW-12A) at the Photon Factory of the High Energy Accelerator Research Organization (KEK, Tsukuba, Japan) with exposure time = 0.1 s, oscillation width = 0.1 degree, and number of frames = 1,800. Diffraction images were processed using XDS [10]. Molecular replacement was done using Molrep [11]. An LnbB structure (PDB ID: 4H04) was used as a template. Manual model rebuilding and refinement were achieved using Coot and REFMAC5 [12, 13]. Molecular graphic images were prepared using PyMOL [14].

LNBase assay

LNBase activity was measured using p-nitrophenyl-β-LNB (pNP-LNB; Toronto Research Chemicals Inc., Toronto, Canada) or p-nitrophenyl-β-GNB (pNP-GNB; Tokyo Chemical Industry Co., Ltd., Tokyo, Japan) as a substrate. The standard assay mixture (50 μL) contained the substrate dissolved in 50 mM HEPES-NaOH (pH 7.0) buffer and an appropriate enzyme concentration. The amount of p-nitrophenol released was measured by absorbance at 400 nm. The reaction was continuously monitored at 25 °C and pH 7.0 (ε = 7,560 M⁻¹cm⁻¹). Curve fitting was performed in R.

For the lacto-N-biosidase activity assay of the StrLNBase, thin-layer chromatography (TLC) analysis was used to detect GNB released from GA1 (ELICITYL S.A., Crolles, France) after incubation at room temperature for 19 hours in a reaction solution containing 20 U/mL or 2 U/mL LNBase, 2 mM GA1, and sodium phosphate buffer at pH 6.0. The solvent consisted of 1-butanol/acetic acid/water (2:1:1 by volume) and diphenylamine-phosphoric acid reagent for detection as described previously [15]. As standards, 10 mM GA1, 5 mM GNB and 5 mM lactose were spotted with the reaction samples.

Phylogenetic analysis of GH20 LNBases

Sequences for classifying GH20 LNBase were retrieved from the NCBI database. To examine the closest phylogenetic relatives of GH20 LNBase, the sequences were queried against GenBank's nonredundant database with BLASTP using StrLNBase and LnbB protein sequences. The top 50 hits for each sequence within Streptomyces spp. were retrieved with > 50 % identities. Furthermore, excluding Streptomyces spp., the other top 50 hits for GH20 LNBase sequences were retrieved with > 43 % identities. A phylogenetic tree of full-length amino acid sequences was constructed using MEGAX software using the neighbor joining method after sequence alignment using MUSCLE [16]. The percentage of replicate trees, in which the associated taxa clustered together in the bootstrap test (1,000 replicates), is shown next to the branches.

RESULTS AND DISCUSSION

Substrate specificity and kinetic parameters

GA1 (Galβ1-3GlcNAcβ1-4Galβ1-4Glc) is a sphingolipid present in human cells. LnbB cleaves the β-1,4 glycosidic bond of GA1 [5]. Although StrLNBase has been shown to cleave the β-1,4 glycosidic bond of oligosaccharide i in Fig. 1 [2], its activity toward GA1 has not been reported. The hydrolysis activity of StrLNBase on GA1 was evaluated by TLC and compared with that of LnbB as a positive control (Fig. 2). For LnbB, there were no GA1 spots with either 2 U/mL or 20 U/mL enzyme, whereas GNB and lactose spots (degradation products) were also observed, suggesting that GA1 was completely degraded even at low enzyme concentrations. On the other hand, treatment with StrLNBase resulted in a remaining GA1 spot at 20 U/mL, and the GA1 spot was more apparent at 2 U/mL. The degradation product of StrLNBase showed a GNB spot, which was slightly lighter at 2 U/mL. This suggests that StrLNBase hydrolyzes GalNAc-β1,4-Gal; however, the activity was weaker compared with that of LnbB.

Fig. 2. TLC analysis of the enzymatic activity of StrLNBase and LnbB liberating GNB and lactose from GA1 (Gal-β1,3-GalNAc-β1,4-Gal-β1,4-Glc) by cleavage of the central β1,4 glycosidic bond.

　The enzyme solution (2 U/mL or 20 U/mL) was incubated with 2 mM GA1 in sodium phosphate buffer (pH 6.0) at room temperature for 19 hours.

Next, we compared the kinetic parameters of LnbB and StrLNBase against pNP-β-LNB and pNP-β-GNB (Table 1 and Fig. S1; see J. Appl. Glycosci. Web site). Although the K_m values were too low to measure for both enzymes because there are no plot below 0.01 mM pNP substrates used in these measurements, the k_cat values of StrLNBase for both substrates were slightly lower compared with those of LnbB. The optimum pH of StrLNBase is 5.5, although it was measured at pH 7.0 and 25 °C for continuous measurements, so there is still a possibility that the k_cat of LnbB could be comparable to that of LnbB if measured at pH5.5 and 30 °C, optimum conditions. Sano et al. reported that the relative activity of StrLNBase for a GNB-containing PA-tetrasaccharide was lower at 0.061 % relative activity compared with 100 % relative activity of PA-LNT (Fig. 1, compounds e and a, respectively) [1, 2]. However, the k_cat values for the two pNP substrates were similar in this study (Table 1).

Table 1. Kinetic parameters of StrLNBase.

	pNP-β-LNB			pNP-β-GNB
	K m (mM)	k cat (s-1)	U/mg	K m (mM)	k cat (s-1)	U/mg
LnbBa (pH 6.0)	< 0.05	16	7.7	< 0.05	4.9	2.5
StrLNBase (pH 7.0)	< 0.05	3.3 ± 0.10	3.0	< 0.05	2.7 ± 0.09	2.5

^aGotoh A et al., Carbohydrate Research, 2015, performed at 30 °C and pH 6.0

Crystal structures

The crystal structure of StrLNBase complexed with LNB was determined at 1.60 Å resolution (Table 2). StrLNBase consists of an N-terminal α/β domain, a catalytic (β/α)₈-barrel domain, and a C-terminal β-trefoil domain (Fig. 3A). This domain architecture was similar to that of LnbB (Fig. S2A; see J. Appl. Glycosci. Web site); however, GH20 β-N-acetylhexosaminidases lack the C-terminal domain. At the active site, an LNB molecule was bound as an α-anomer, even though GH20 adopts a “β to β” type anomer-retaining mechanism (Fig. 3B). This contrasts with the previous LnbB structure, in which an LNB molecule was bound to an β-anomer [4]. The N-acetyl group of LNB was in a bent conformation and its oxygen atom was located below the anomeric C1 atom of GlcNAc. The bent conformation of the N-acetyl group has also been observed in LnbB and other GH enzymes adopting a substrate-assisted mechanism [4]. E305 and D304 are highly conserved catalytic residues in GH20 enzymes. E305 is located above the N-acetyl group of LNB. This residue functions as an acid/base catalytic residue; however, it does not form a hydrogen bond with the O1-hydroxy of GlcNAc because of the “wrong” anomer configuration of α-LNB in the crystal structure. D304 forms a hydrogen bond with the nitrogen atom of the N-acetyl group. This interaction is suitable for the stabilization of the reaction intermediate LNB-oxazoline [17]. Taken together, the active site structure of StrLNBase was similar to that of LnbB (Fig. S2B; see J. Appl. Glycosci. Web site).

Table 2. Refinement and statistics.

	StrLNBase LNB complex	StrLNBase GNB complex 1	StrLNBase GNB complex 2
Data collection statistics
PDB ID	8HVB	8HVC	8HVD
Beamline	PF BL-1A	PF-AR NW-12A	PF-AR NW-12A
Wavelength (Å)	1.052000	1.00000	1.00000
Space group	P1 21 1	P1 21 1	P1 21 1
Unit cell parameters
a, b, c (Å)	53.970, 51.277, 86.530	53.918, 50.852, 86.052	52.779, 60.825, 88.232
α, β, γ (°)	90.000, 98.080, 90.000	90.000, 97.918, 90.000	90.000, 98.186, 90.000
Resolution (Å)a	48.50-1.60 (1.63-1.60)	48.35-1.58 (1.61-1.58)	47.94-1.41 (1.43-1.41)
Total reflections	435,975	213,120 (10,690)	364,876 (16,650)
Unique reflections	62,043	63,448 (3,121)	107,633 (5,349)
Completeness (%)a	100.0 (100.0)	99.9 (100.0)	99.9 (100.0)
Multiplicitya	7.0 (7.1)	3.4 (3.4)	3.4 (3.1)
Mean I/σ (I)a	9.2 (3.2)	8.1 (1.8)	12.4 (2.1)
R merge (%)a	13.6 (92.4)	8.9 (57.7)	5.1 (48.2)
CC1/2a	0.995 (0.916)	0.983 (0.803)	0.995 (0.784)
Refinement statistics
Resolution range (Å)	48.55-1.60	48.40-1.58	47.99-1.41
No. of reflections all/free	62,029 / 3,015	63,432 / 3,109	107,610 / 5,355
R factor/Rfree (%)	14.9 / 18.0	17.0 / 20.0	16.5 / 19.3
r.m.s from ideal values
Bond lengths (Å)	0.0120	0.0112	0.0232
Bond angles (°)	1.714	1.712	1.754
Ramachandran plot (%)
Favored	97.16	97.67	97.33
Allowed	2.50	2.17	2.50
Outlier	0.33	0.17	0.17
Multimetric validation
Clashscore	2.07	3.14	1.64
RMS bonds (Å)	0.0150	0.0148	0.0113
RMS angles (°)	1.85	1.89	1.75
Molprobity score	1.13	1.35	0.95
Cremer-Pople parameter 　(ϕ (°), θ (°), Q)
Gal	9.301, 8.486, 0.616	6.199, 4.453, 0.595	11.041, 6.100, 0.593
GlcNAc/GalNAc	351.984, 15.019, 0.527	269.585, 43,104, 0.526	123.975, 5.774, 0.608

^aValues in parentheses are for the highest resolution shell.

Fig. 3. Crystal structures of StrLNBase.

　(A) Overall structure, and the active sites of (B) LNB complex, (C) GNB complex 1, and (D) GNB complex 2 with polder map (5σ). Catalytic residues and residues forming hydrogen bonds are shown in magenta and green, respectively. Hydrogen bonds are shown as yellow dashed lines.

Two different crystal structures of StrLNBase complexed with GNB were determined at 1.58 Å and 1.41 Å resolution (Table 2). Crystals for the two complex structures (GNB complex 1 and GNB complex 2) were grown under very similar conditions. The GNB molecule in complex 1 was a β anomer and the N-acetyl group was present in a bent conformation (Fig. 3C). The two catalytic residues in GNB complex 1 were in the same position as that in the LNB complex structure. In contrast, the GNB molecule in complex 2 was an α anomer, and the N-acetyl group was in an extended conformation, indicating that it was in an inactive state (Fig. 3D). The two catalytic residues, E305 and D304, were located far from the O1-hydroxy and N-acetyl groups of GNB and water molecules were present in the gap between them. Y406 forms a hydrogen bond with the α anomer O1 oxygen atom of GNB, suggesting that this hydrogen bond mimics the interaction that stabilizes the oxazoline intermediate. There were no particular changes in the other active site residues interacting with GNB. We collected six crystallographic data sets of co-crystals with GNB under virtually the same crystallization conditions, from which the structures of complexes 1 and 2 were obtained. The remaining four crystals contained nearly identical to the GNB complex 2 structure (data not shown). This suggests that crystals in the inactive state predominantly grew under these conditions.

Cremer-Pople parameters that measure sugar ring puckering conformations [18] were calculated for LNB/GNB in the StrLNBase structures (Table 2). Gal residues in subsite −2 were all in a stable chair (⁴C₁) conformation (θ < 10°). On the other hand, subsite −1 sugar in GHs acting on β-glycosides often adopt a distorted (non-⁴C₁) conformation to circumvent steric hindrance for nucleophilic attack [19]. For GHs utilizing the substrate-assisted mechanism, the bent conformation of the N-acetyl group also results in sugar ring distortion [20]. GalNAc in the GNB complex 2 structure was in a typical ⁴C₁ conformation (θ = 5.8°) because it was in an inactive form with extended N-acetyl group and displaced catalytic residues (Fig. 3D). GlcNAc in the LNB complex structure was in a slightly distorted ⁴C₁ conformation (θ = 15°), likely because of steric hindrance between the α-anomeric O1 hydroxy and the bent N-acetyl groups (Fig. 3B). GalNAc in GNB complex 1, in which the catalytic center structures is in an active state with β-anomeric O1 hydroxy and the bent N-acetyl groups (Fig. 3C), adopts an ⁴H₅ conformation (φ = 270°, θ = 43°), which is close to the ⁴E conformation of GlcNAc in the LNB complex structure of LnbB (φ ~ 250°, θ ~ 60°) [4]. The results indicate that the transition state of StrLNBase is also a ⁴E-like conformation as suggested for GH20 and other GH enzymes that utilize a substrate-assisted mechanism [21, 22, 23].

Phylogenetic analysis of GH20 LNBases

GH20 LNBases have only been characterized for enzymes isolated from Bifidobacterium and Streptomyces, which belong to the Phylum Actinomycetota (previously known as the Phylum Actinobacteria). Gotoh et al. suggested that these two bacterial groups evolved GH20 LNBases differently because their amino acid sequence identity is 38 % at a maximum [5]. In the present study, we compared the two GH20 LNBase groups. A phylogenetic tree of putative GH20 LNBases was constructed by including protein sequences from Streptomyces spp. group (StrLNBase and its homologs with > 50 % identities) and Bifidobacterium spp. group (LnbB and its homologs with > 43 % identities), and 50 protein sequences form other bacterial species including uncultured sources (Fig. 4). The sequences from the Streptomyces spp. group had average length of 640 amino acids corresponding to the core three-domain architecture of the GH20 LNBase (Fig. 2A). In contrast, sequences from the Bifidobacterium spp. group, which primarily consists of those from B. bifidum, have an average length of 1,100 amino acids. Consistent with LnbB, the bifidobacterial sequences have C-terminal extensions, such as carbohydrate-binding module (CBM) family 32 (CBM32), bacterial Ig-like domain, and a transmembrane region [3]. The large protein architecture, which consists of the catalytic domain, CBMs, and a C-terminal transmembrane region, is often found in GHs from B. bifidum [24]. This protein architecture is believed to play an important role in the substrate binding as a membrane-anchored extracellular multidomain enzyme. The shorter LNBases (~640 amino acids in length) were also found in various soil bacteria, such as the genera Kribella, Microbacterium, Agromyces, Cellulosimicrobium, Arthrobacter, and Actinomyces (Fig. 4). Taken together with the phylogenetic separation and the difference of domain architecture, the two groups of LNBases appear to have evolved quite differently in the human intestinal tract and environmental soil. Interestingly, the Streptomyces sp. 142 strain also exhibits α-1,3/4-fucosidase activity [25]. It is unclear whether the soil bacteria contain the enzymes that specifically degrade human glycans. This is likely because the bacterial glycosidases contribute to the degradation of animal bodies in the ground and return them to the soil.

Fig. 4. A phylogenetic tree of putative GH20 LNBases.

　Protein sequences from Streptomyces spp. group (green), Bifidobacterium spp. group (blue), and other soil bacteria (uncolored) are shown. The maximum sequence identity between the Streptomyces spp. and Bifidobacterium spp. groups is 38 %. The tree was constructed by the maximum likelihood method. StrLNBase and LnbB are indicated by red and black arrows, respectively.

CONFLICTS OF INTEREST

The authors declare that they have no conflict of interest.

ACKNOWLEDGMENTS

We would like to thank Professor Takane Katayama at Kyoto University and Professor Motomitsu Kitaoka at Niigata University for their assistance and for providing LNB. We also appreciate Dr. Takatoshi Arakawa and Dr. Toma Kashima at the University of Tokyo for their support and discussion. We would also like to thank Dr. Mutsumi Sano, Dr. Kumi Hayakawa, Dr. Ikunoshin Kato, and the research group at Takara Shuzo Co., Ltd. for allowing us to freely conduct structural analysis research on StrLNBase. This work was supported in part by the Kato Memorial Bioscience Foundation, JST ACT-X (JPMJAX21BN) and JSPS-KAKENHI (21K15025).

REFERENCES

• [1]  Sano M, Hayakawa K, Kato I. Purification and characterization of an enzyme releasing lacto-N-biose from oligosaccharides with type 1 chain. J Biol Chem. 1993; 268: 18560-6.
• [2]  Sano M, Hayakawa K, Kato I. An enzyme releasing lacto-N-biose from oligosaccharides. Proc Natl Acad Sci USA. 1992; 89: 8512-6.
• [3]  Wada J, Ando T, Kiyohara M, Ashida H, Kitaoka M, Yamaguchi M, et al. Bifidobacterium bifidum lacto-N-biosidase, a critical enzyme for the degradation of human milk oligosaccharides with a type 1 structure. Appl Environ Microbiol. 2008; 74: 3996-4004.
• [4]  Ito T, Katayama T, Hattie M, Sakurama H, Wada J, Suzuki R, et al. Crystal structures of a glycoside hydrolase family 20 lacto-N-biosidase from Bifidobacterium bifidum. J Biol Chem. 2013; 288: 11795-806.
• [5]  Gotoh A, Katoh T, Sugiyama Y, Kurihara S, Honda Y, Sakurama H, et al. Novel substrate specificities of two lacto-N-biosidases towards beta-linked galacto-N-biose-containing oligosaccharides of globo H, Gb5, and GA1. Carbohydr Res. 2015; 408: 18-24.
• [6]  Vuillemin M, Muschiol J, Zhang Y, Holck J, Barrett K, Morth JP, et al. Discovery of lacto-N-biosidases and a novel N-acetyllactosaminidase activity in the CAZy family GH20: functional diversity and structural insights. ChemBioChem. 2025; 26: e202400710.
• [7]  Zhu Y, Luo G, Wan L, Meng J, Lee SY, Mu W. Physiological effects, biosynthesis, and derivatization of key human milk tetrasaccharides, lacto-N-tetraose, and lacto-N-neotetraose. Crit Rev Biotechnol. 2022; 42: 578-96.
• [8]  Murata T, Inukai T, Suzuki M, Yamagishi M, Usui T. Facile enzymatic conversion of lactose into lacto-N-tetraose and lacto-N-neotetraose. Glycoconj J. 1999; 16: 189-95.
• [9]  Castejón-Vilatersana M, Faijes M, Planas A. Transglycosylation activity of engineered Bifidobacterium lacto-N-biosidase mutants at donor subsites for lacto-N-tetraose synthesis. Int J Mol Sci. 2021; 22: 3230.
• [10]  Kabsch W. XDS. Acta Crystallogr D Biol Crystallogr. 2010; 66: 125-32.
• [11]  Vagin A, Teplyakov A. Molecular replacement with MOLREP. Acta Crystallogr D Biol Crystallogr. 2010; 66: 22-5.
• [12]  Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr. 2010; 66: 486-501.
• [13]  Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr D Biol Crystallogr. 1997; 53: 240-55.
• [14]  PyMOL [Internet]. New York: Schrödinger, Inc.; Available from: http://www.pymol.org/pymol
• [15]  Anderson K, Li SC, Li YT. Diphenylamine-aniline-phosphoric acid reagent, a versatile spray reagent for revealing glycoconjugates on thin-layer chromatography plates. Anal Biochem. 2000; 287: 337-9.
• [16]  Stecher G, Tamura K, Kumar S. Molecular evolutionary genetics analysis (MEGA) for macOS. Mol Biol Evol. 2020; 37: 1237-9.
• [17]  Mark BL, Vocadlo DJ, Knapp S, Triggs-Raine BL, Withers SG, James MNG. Crystallographic evidence for substrate-assisted catalysis in a bacterial β-hexosaminidase. J Biol Chem. 2001; 276: 10330-7.
• [18]  Cremer D, Pople JA. General definition of ring puckering coordinates. J Am Chem Soc. 1975; 97: 1354-8.
• [19]  Davies GJ, Ducros VM-A, Varrot A, Zechel DL. Mapping the conformational itinerary of β-glycosidases by X-ray crystallography. Biochem Soc Trans. 2003; 31: 523-7.
• [20]  Vocadlo DJ, Withers SG. Detailed comparative analysis of the catalytic mechanisms of β-N-acetylglucosaminidases from families 3 and 20 of glycoside hydrolases. Biochemistry. 2005; 44: 12809-18.
• [21]  Jitonnom J, Lee VS, Nimmanpipug P, Rowlands HA, Mulholland AJ. Quantum mechanics/molecular mechanics modeling of substrate-assisted catalysis in family 18 chitinases: Conformational changes and the role of Asp142 in catalysis in ChiB. Biochemistry. 2011; 50: 4697-711.
• [22]  Sumida T, Ishii R, Yanagisawa T, Yokoyama S, Ito M. Molecular cloning and crystal structural analysis of a novel β-N-acetylhexosaminidase from Paenibacillus sp. TS12 capable of degrading glycosphingolipids. J Mol Biol. 2009; 392: 87-99.
• [23]  Rao FV, Dorfmueller HC, Villa F, Allwood M, Eggleston IM, Van Aalten DMF. Structural insights into the mechanism and inhibition of eukaryotic O-GlcNAc hydrolysis. EMBO J. 2006; 25: 1569-78.
• [24]  Katoh T, Ojima MN, Sakanaka M, Ashida H, Gotoh A, Katayama T. Enzymatic adaptation of Bifidobacterium bifidum to host glycans, viewed from glycoside hydrolyases and carbohydrate-binding modules. Microorganisms. 2020; 8: 481.
• [25]  Sano M, Hayakawa K, Kato I. Purification and characterization of α-L-fucosidase from Streptomyces species. J Biol Chem. 1992; 267: 1522-7.