Enterococcus faecalis α1-2-mannosidase (EfMan-I): an efficient catalyst for glycoprotein N-glycan modification

Abstract

While multiple α1–2-mannosidases are necessary for glycoprotein N-glycan maturation in vertebrates, a single bacterial α1–2-mannosidase can be sufficient to cleave all α1–2-linked mannose residues in host glycoprotein N-glycans. We report here the characterization and crystal structure of a new α1–2-mannosidase (EfMan-I) from Enterococcus faecalis, a Gram-positive opportunistic human pathogen. EfMan-I catalyzes the cleavage of α1–2-mannose from not only oligomannoses but also high-mannose-type N-glycans on glycoproteins. Its 2.15 Å resolution crystal structure reveals a two-domain enzyme fold similar to other CAZy GH92 mannosidases. An unexpected potassium ion was observed bridging two domains near the active site. These findings support EfMan-I as an effective catalyst for in vitro N-glycan modification of glycoproteins with high-mannose-type N-glycans.

Introduction

N-Linked glycan modification is an important post-translational modification (PTM) of proteins that affects their conformation, solubility, stability, immunogenicity, and recognition by other molecules [1,2]. In fact, many therapeutic proteins are N-glycosylated glycoproteins and proper glycosylation is important for their pharmacokinetics, cellular distributions, and biological activities [3–5]. Glycan engineering of therapeutic glycoproteins has shown to improve their functions [6–8]. Among various methods [9–11] including glycoengineering of host biosynthetic pathways [12–14], chemoselective ligation of tagged proteins and glycans [10,11], total chemical synthesis by native chemical ligation [15], in vitro chemoenzymatic glycosylation remodeling [16], and in vitro enzymatic glycosylation of glycoproteins, the latter is an attractive strategy to produce glycan-defined glycoproteins [8].

In eukaryotes, N-glycosylation is a part of a glycoprotein quality control process in the endoplasmic reticulum (ER). Eukaryotic N-glycans usually belong to one of the three types of structures including oligomannose, complex, and hybrid. These structures share a common tri-mannose core Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ1–(Asn-X-Ser/Thr) but differ on the extension at the non-reducing end [17], with or without an additional core fucose [18], and the linkage of the core fucose [19]. Oligomannose-type (or high-mannose type) N-glycans can be considered as precursors for the biosynthesis of complex and hybrid-type N-glycans in glycoproteins. Early steps of such conversion involve trimming high-mannose N-glycans by α1–2-mannosidase-catalyzed reactions. In human, multiple α1–2-mannosidases, including endoplasmic reticulum (ER) α1–2-mannosidase I (ERManI) and three Golgi α1–2-mannosidases (GM1A, GM1B, and GMIC), are responsible for removing four α1–2-linked mannose from high-mannose-type N-glycan Man₉GlcNAc₂ to form Man₅GlcNAc₂ [20]. In comparison, α1–2-mannosidases capable of catalyzing the cleavage of all four α1–2-mannose residues in high-mannose type N-glycan Man₉GlcNAc₂ have been identified in a limited number of bacteria such as Bacteroides thetaiotaomicron VPI-5482 (Bt3990 and Bt2199) [21], Streptococcus pneumoniae (SpGH92) [22], and Cellulosimicrobium cellulans [23]. An exo-1,2-α-D-mannanase purified from Bacillus sp. M-90 has also been shown to catalyze the cleavage of all α1–2-linked mannose residues in Man₉GlcNAc₂ to form Man₅GlcNAc₂ [24]. All known exo-α1–2-mannosidases are mainly classified into three glycoside hydrolase (GH) families in the Carbohydrate Active Enzyme (CAZy) database (http://www.cazy.org) [25–27] including GH38, GH47, and GH92. While characterized mammalian ER and Golgi α1–2-mannosidases are in the CAZy GH47 family [20], most bacterial α1–2-mannosidases characterized so far are in the GH92 family, and lysosomal α-mannosidases are in the GH38 family.

Bacterial α1–2-mannosidases are attractive enzymes for in vitro enzymatic glycan modification of the high-mannose-type N-glycans in natural and recombinant glycoproteins to hybrid or complex-type N-glycans. They are also appealing targets for better understanding how host glycoprotein glycans can be utilized by commensals and pathogenic bacteria to shape microbiota community [28]. By sequence alignment and literature survey, we identified EF2217 from Enterococcus faecalis V583 [29], a member in the CAZy GH92 family, as a promising candidate for a novel and highly active bacterial α1–2-mannosidase. Enterococcus faecalis is a Gram-positive commensal bacterium and an opportunistic pathogen that can cause healthcare-associated infections (HAIs) including urinary tract infections, bacteremia, and infective endocarditis [30]. It was the first clinical vancomycin-resistant enterococcal isolate reported in the USA [31] and the full genomic DNA of strain V583 was reported in 2003 [29]. More importantly, the bacterium was shown to be able to use high-mannose-type N-glycans efficiently but not sialylated N-linked glycans for growth [32]. Two endo-β-N-acetylglucosaminidases from Enterococcus faecalis V583 (EF0114 and EF2683) were isolated and found to be able to cleave high-mannose N-glycan from glycoprotein to form a single N-linked N-acetylglucosamine (GlcNAc) residue [32–34]. Activities of exo-mannosidases in producing free mannose residues from the N-glycans released from glycoproteins were also observed [32] but not characterized. We describe here that EF2217 is an efficient α1–2-mannosidase (EfMan-I) catalyzing the cleavage of α1–2-linked mannose residues from not only synthetic mannosides and high-mannose N-glycans released from glycoproteins, but also high-mannose N-glycans on glycoproteins. The enzyme is thus an effective catalyst and a very useful tool for in vitro enzymatic modification of glycoproteins with high-mannose-type N-glycans. The crystal structure reveals that it shares high similarity with other bacterial GH92 α-mannosidases, with an N-terminal β-sandwich domain and a C-terminal α-helical domain. The active site lies in a cleft between the two domains and contains a catalytic calcium ion. An unexpected potassium ion was also observed bridging the two domains near the active site that may help position the two domains.

Materials and Methods

Bacterial strains, plasmids, and materials

E. coli electrocompetent DH5α and chemically competent BL21 (DE3) cells were purchased from Invitrogen (Carlsbad, CA). Genomic DNA of Enterococcus faecalis V583 (ATCC#700802D-5) was from American Type Culture Collection (ATCC, Manassas, VA, USA). Vector plasmids pET22b(+) was from Novagen (EMD Biosciences Inc. Madison, WI, USA). Restriction enzymes NdeI and XhoI were purchased from New England Biolabs, Inc. (Beverly, MA, USA). Nickel-nitrilotriacetic acid agarose (Ni²⁺-NTA agarose) was from Qiagen (Valencia, CA, USA). GeneJET Plasmid Miniprep kit, and GeneJET PCR Purification kit were from Thermo Scientific (San Diego, CA, USA). Herculase-enhanced DNA polymerase was from Stratagene (La Jolla, CA, USA). T4 DNA ligase and 1 kb DNA ladder were from Promega (Madison, WI, USA). para-Nitrophenyl-α-D-mannopyranoside (ManαpNP) was from Chem-Impex Int’L Inc. (Wood Dale, IL, USA). α2-Mannobiose, α3-mannobiose, and 3α,6α-mannopentaose (Man₅) were from Santa Cruz Biotechnology (Dallas, TX, USA). α6-Mannobiose, hexokinase from S. cerevisiae, phosphoglucose isomerase from baker’s yeast, and glucose-6-phosphate dehydrogenase from L. mesenteroides were from Sigma-Aldrich (Saint Louis, MO, USA). 1,3-α−1,6-α-D-Mannotriose (Man₃) was purchased from Carbosynth Limited (San Diego, CA, USA). E. coli phosphomannose isomerase was from Megazyme (Chicago, IL, USA).

Cloning of EfMan-I

EfMan-I was cloned as an C-His₆-tagged fusion protein in pET22b(+) vector using genomic DNA of Enterococcus faecalis V583 (ATCC#700802D-5) as the template for polymerase chain reactions (PCR). The primers used were: forward primer 5’-GATCCATATGAATATTCAAGCGATTGATACGCG-3’ (NdeI restriction site is italicized) and reverse primer 5’-ACGCCTCGAG TTTTTCCGTACTTAATGAAAACGG-3’ (XhoI restriction site is italicized). PCR was performed in a 50-μl reaction mixture containing genomic DNA (1 μg), forward and reverse primers (1 μM each), 10 × Herculase buffer (5 μl), dNTP mixture (1 mM), and 5 U (1 μl) of Herculase-enhanced DNA polymerase. The reaction mixture was subjected to 30 cycles of amplification with an annealing temperature of 52°C. The resulting PCR product was purified and digested with NdeI and XhoI restriction enzymes. The purified and digested PCR product was ligated with predigested pET22b(+) vector and transformed into electrocompetent E. coli DH5α cells. Selected clones were grown for minipreps and characterization by restriction mapping and DNA sequencing performed by Davis Sequencing Facility at the University of California-Davis.

Expression and purification of EfMan-I

E. coli strains were cultured in LB rich medium (10 g l⁻¹ tryptone, 5 g l⁻¹ yeast extract, and 10 g l⁻¹ NaCl) supplemented with ampicillin (100 μg ml⁻¹). Over-expression of EfMan-I was achieved by inducing the E. coli BL21 (DE3) cell culture with 0.15 mM of isopropyl-1-thio-D-galactopyranoside (IPTG) when the OD_{600 nm} of the culture reached 0.8–1.0 followed by incubation at 20°C for 20 h.

Bacterial cells were harvested by centrifugation at 4°C in a Sorvall Legend RT centrifuge with a hanging bucket rotor at 4000 × rpm for 2 h. Harvested cells were resuspended in lysis buffer (Tris-HCl buffer, 100 mM, pH 8.0 containing 0.1% Triton X-100) (20 ml for cells collected from one liter cell culture). Lysozyme (100 μg ml⁻¹) and DNase I (5 μg ml⁻¹) were added to the cell resuspension. The resulting mixture was incubated at 37°C for 1 h with shaking at 200 rpm. Cell lysate (supernatant) was obtained by centrifugation at 12000 × rpm for 15 min. Purification was carried out by loading the supernatant onto a Ni²⁺-NTA column pre-equilibrated with 10 column volumes of binding buffer (5 mM imidazole, 0.5 M NaCl, 50 mM Tris-HCl, pH 7.5). The column was washed with 10 column volumes of binding buffer and 10 column volumes of washing buffer (24 mM imidazole, 0.5 M NaCl, 50 mM Tris-HCl, pH 7.5). The target protein was eluted with Tris-HCl buffer (50 mM, pH 7.5) containing imidazole (200 mM) and NaCl (0.5 M). The fractions containing the purified enzymes were collected, dialyzed against Tris-HCl buffer (20 mM, pH 7.5) containing 10% glycerol with changing of dialysis buffer for 3 times, and stored at 4°C.

Substrate specificity studies of EfMan-I

Typical enzymatic assays were performed in a 10 μl reaction mixture in MES buffer (100 mM, pH 6.0) containing substrate (10 mM) and EfMan-I (1.0 μg or 10 μg). Substrates used were: ManαpNP, α2-mannobiose, α3-mannobiose, α6-mannobiose, Man₃ and Man₅. Reactions were allowed to proceed for 30 min, 18 h, or 24 h at room temperature (RM) or at 37°C and were analyzed by thin layer chromatography (TLC) in a developing solvent (n-PrOH : H₂O : NH₄OH = 5 : 2 : 1 or n-PrOH : H₂O : NH₄OH = 3 : 2 : 1).

pH Profile of EfMan-I

Standard enzymatic assays were carried out in duplicate in 384-well plates in a final volume of 20 μl in a buffer (100 mM) with a pH ranging from 3.0–9.0 containing ManαpNP (0.3 mM), CaCl₂ (4 mM), and EfMan-I (10.0 μg). Buffers used were: citric acid-Na₂HPO₄, pH 3.0–4.0; NaOAc/HOAc, pH 4.5; MES, pH 5.0–6.5, and Tris-HCl pH 7.0–9.0. Reactions were incubated at 37°C for 30 min and quenched by adding 40 μl of N-cyclohexyl-3-aminopropanesulfonic acid (CAPS) buffer (0.5 M, pH 10.5). The amounts of para-nitrophenolate formed were determined by measuring the A_{405 nm} of the reaction mixtures using a microplate reader. Reactions of GalβpNP with an excess amount of β-galactosidase were used as controls.

Effects of ethylenediaminetetraacetic acid (EDTA) and metal Ions on EfMan-I

EDTA (10 mM), CaCl₂, or MgCl₂ were added at different concentrations (1, 2, 4, 8, 10, 20 mM) to reactions in MES buffer (100 mM, pH 6.0) to analyze their effects on the mannosidase activity of EfMan-I (10.0 μg) against ManαpNP. Reactions without the addition of EDTA or metal ions were used as a control. The concentrations of the substrates and other reaction conditions were the same as described above for the pH profile assays.

Kinetic studies of EfMan-I

Kinetic assays were carried out using three different substrates including ManαpNP, α2-mannobiose, and RNaseB. When ManαpNP was used as the substrate, the assays were carried out in duplicates in 384-well plates in a total volume of 20 μl in MES buffer (100 mM, pH 6.0) containing CaCl₂ (4 mM), ManαpNP at different concentrations in the range of 0.1–10 mM (0.1 mM, 0.2 mM, 0.4 mM, 0.5 mM, 1 mM, 2 mM, 5 mM, 10 mM), and EfMan-I (14 μg). Reactions were incubated at 37°C for 10 min and quenched by adding 40 μl of CAPS buffer (0.5 M, pH 10.5). The amounts of para-nitrophenolate formed were analyzed by a microplate reader and compared to a pNP standard curve. Apparent kinetic parameters were obtained by fitting the averages to the Michaelis-Menten equation using Grafit 5.0.

When α2-mannobiose was used as the substrate, reactions were performed in duplicate at 37°C for 10 minutes in a total volume of 50 μl in MES buffer (100 mM, pH 6.0) containing varied concentrations of α2-mannobiose (0.1, 0.2, 0.5, 1, 2, 5 and 10 mM) and EfMan I (6.2 nM). Reactions were quenched by incubating the reaction mixtures at 98°C for 10 min. Precipitates were removed by centrifugation (11000 ×g, 5 min, 4°C). The concentrations of mannose released were determined by a coupled enzyme assay as described [21]. Briefly, reactions were carried out at 25°C in duplicate in a 384-well plate in the presence of Tris-HCl buffer (100 mM, pH 8.0), MES buffer (50 mM, 6.0), MgCl₂ (10 mM), NADP⁺ (1 mM), ATP (1 mM), hexokinase from S. cerevisiae (0.75 U, Sigma-Aldrich, H4502–1KU), E. coli phosphomannose isomerase (0.75 U, Megazyme, E-PMIEC), phosphoglucose isomerase from baker’s yeast (0.75 U, Sigma-Aldrich, P5381–1KU), and glucose-6-phosphate dehydrogenase from L. mesenteroides (0.75 U, Sigma-Aldrich, G9404–1KU). The absorbance of the NADPH formed was monitored by a microplate reader (Synergy HT, BioTek) at 340 nm in a 10-min interval and the maximum reading was recorded and compared to the mannose standard curve.

To use RNase B as the substrate, the effective percentage of α1–2-mannose-containing RNase B (44%) was determined by complete releasing of α1–2-linked mannose from RNase B (0.8 mM) by incubating with EfMan-I (2.5 uM) at 37°C for 4 hours followed by quantification of the released mannose using coupled enzyme assay as described above. Kinetics studies using RNase B as the EfMan-I substrate were carried out at 37°C for 10 minutes in a total volume of 50 μl in duplicate in the presence of MES buffer (100 mM, pH 6.0), EfMan-I (0.031 μM), and varied concentrations (0.1, 0.2, 0.3, 0.5, 1.0 and 2.5 mM) of RNase B. Reactions were quenched by incubating the reaction mixture at 98°C for 10 min. Precipitates were removed by centrifugation (11000 × g, 5 min, 4°C). The concentrations of mannose released were determined similarly as described above using the coupled enzyme assay [21].

¹H NMR-based time course studies of EfMan-I using ManαpNP as the substrate

The assay was carried out in an NMR tube in 80% D₂O/H₂O solution (0.6 ml) containing phosphate buffer (10 mM, pH 6.0), ManαpNP (10 mM), CaCl₂ (4 mM), and EfManI (0.75 mg). NMR spectra (16 scans for each time point) were recorded at different reaction time points (0, 5, 10, 20, 30, 40, 50, 60, 90, and 120 min) in an 800 MHz Bruker Avance III spectrometer at a temperature of 310 K. The H₂O signal at 4.705 ppm was suppressed.

MALDI-TOF MS analysis of N-glycans released from RNase B with or without EfMan-I treatment

RNase B or EfMan-I treated RNase B (1 mg each) sample was denatured by adding 0.5% SDS and dithiothreitol (DTT) (40 mM), The mixtures were incubated at 98°C for 10 min and then at room temperature for 5 min. The glycans were released using 6 μg of peptide N-glycosidase F (PNGase F) by incubating the reaction mixtures at 37°C for 24 h. The deglycosylated proteins were precipitated by adding three-fold volume of cold ethanol and chilling the mixtures at −80°C for 2 h. The mixtures were then centrifuged at 15000 rpm for 20 min and the supernatants containing the glycans were dried in speed vacuum for matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometry (MS) analysis.

High-performance liquid chromatography (HPLC) and MALDI-TOF MS analyses of 2-aminobenzamide (2-AB)-labeled N-glycans released from RNase B with or without EfMan-I treatment

2-AB solution (A) was prepared by adding 88 mg of 2-AB to 1 ml of DMSO/AcOH (7:3 by volume). The solution was stored at −20°C. Sodium cyanoborohydride solution (B) was freshly prepared by adding 64 mg of NaCNBH₃ to 1 ml of a solution mixture of DMSO/AcOH (7:3 by volume). To a dried free reducing glycan sample (0.01 to 1 mg), solution A (5 to 50 μl) and an equal volume of solution B were added. The mixture was heated at 65°C for 2 h. A 10-fold volume of acetonitrile was then added to precipitate the glycan-2AB conjugates. The mixture was cooled down at −20°C for 1 h and centrifuged at 13000 rpm for 2 min. The precipitate was collected and dissolved in H₂O (100–1000 μl). The obtained mixture was centrifuged at 13000 rpm for 2 min. The supernatant was collected, diluted, and the glycan-2AB conjugates were analyzed using high performance liquid chromatography (HPLC) with an AdvanceBio Glycan Map column (Agilent) using a slow gradient of Solvent A (ammonium formate, 100 mM, pH 4.5) and Solvent B (acetonitrile) and monitoring at A_{254 nm}.

Crystallization of EfMan-I

EfMan-I in Tris-HCl buffer (10 mM, pH 7.5) was concentrated to 10 mg ml⁻¹ using centrifugal filter units (EMD Millipore, Billerica, MA, USA) and crystallized by sitting drop vapor diffusion in a 1:1 ratio of protein and reservoir solution at 21°C. The reservoir solution contained NaOAc (0.1 M, pH 4.5) and NaH₂PO₄ (0.8 M)/K₂HPO₄ (1.2 M). The crystals were briefly soaked in the reservoir solution that also contained 30% glycerol prior to flash cooling in liquid nitrogen for X-ray diffraction data collection.

Data collection, model building, and refinement

X-Ray diffraction data were collected at Stanford Synchrotron Radiation Lightsource (SSRL) beamline 12–2 at 100 K. Diffraction data were indexed and integrated with XDS [35], and then scaled with XSCALE [36]. A complete diffraction data set was collected to 2.15 Å resolution with an R_merge of 9.6%. Data collection and structure refinement statistics are given in Table S1. The crystals belong to the orthorhombic space group C222₁ with unit cell parameters: a = 163.35 Å, b = 169.00 Å, c = 258.52 Å, α = β = γ = 90°, which contain four monomers per asymmetric unit (V_M = 2.70 ų/Da, solvent = 54.48%) [37]. The EfMan-I structure was solved by molecular replacement applying a model of the previously solved α-mannosidase structure from Streptococcus pneumoniae (PDBID: 5swi) using the program PHASER [38]. Atomic model building was carried out with the molecular graphics program COOT [39,40]. The structure was refined with the program REFMAC5 [41] part of the CCP4 package [42]. Noncrystallographic symmetry restraints were included during refinement. The final R-factor (15.5%) and R-free (18.9%) along with the quality of the models are listed in Table S1.

Results and Discussions:

Cloning and purification of EfMan-I

The full-length gene of EF2217 (2142 bp encoding 713 aa) was cloned from the genomic DNA of Enterococcus faecalis V583 into pET22b(+) vector. EfMan-I was expressed as a C-His₆-tagged fusion protein. Protein sequence alignment (Fig. S1 and Table S2) showed that EfMan-I shared 46.4% identity with SpGH92 from Streptococcus pneumoniae TIGR4 [22]. It also shared 34.1% and 32.1% identities, respectively, with Bt2199 and Bt3990 which are two α1–2-mannosidases from Bacteroides thetaiotaomicron VPI-5482 [21].

EfMan-I was expressed at 20°C as a soluble form in E. coli BL21 (DE3) cells by induction using 0.15 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG). Under these expression conditions, 85 mg of purified protein was routinely obtained from 1 liter of E. coli cell culture by one-step nickel nitrilotriacetic acid (Ni²⁺-NTA) affinity chromatography. This expression level was comparable or higher than that reported for Bt2199 and Bt3990 (>500 nM or 41 mg/liter culture) [21].

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis (Fig. 1) indicated that the apparent molecular weight of purified EfMan-I was about 73 kDa which was smaller than the calculated molecular weight (82.7 kDa). Nevertheless, N-terminal amino acid sequencing (Table S3) showed an intact N-terminus. The efficient one-step Ni²⁺-NTA affinity chromatography process indicated the presence of the C-His₆-tag in the recombinant EfMan-I introduced by cloning.

Fig. 1.

SDS-PAGE (10% Tris-Glycine gel) analysis for the expression and the purification of EfMan-I. Lanes: BI, whole cell extract before induction; AI, whole cell extract after induction; L, lysate after induction; PP, Ni²⁺-NTA column purified protein; M, protein markers (Bio-Rad precision Plus Protein Standards, 10–250 kDa).

Substrate specificity studies of EfMan-I

Substrate specificity studies of EfMan-I (Fig. 2) in MES buffer (100 mM, pH 6.0) using para-nitrophenyl-α-D-mannopyranoside (ManαpNP), α2-mannobiose, α3-mannobiose, α6-mannobiose, D-mannotriose (Man₃), or 3α,6α-mannopentaose (Man₅) as the substrate showed that α2-mannobiose was completely hydrolyzed by EfMan-I while ManαpNP was partially hydrolyzed and other compounds tested were not hydrolyzed by EfMan-I in a 30-min time frame. However, after a prolonged incubation time of 24 h, ManαpNP was completely hydrolyzed to form mannose and pNP (Fig. 2, lane 2). It was interesting to note that both α3-mannobiose and α6-mannobiose were also partially hydrolyzed by EfMan-I to form mannose (Fig. 2, lanes 6 and 8) and the hydrolytic activity of EfMan-I towards α3-mannobiose was slightly higher than that towards α6-mannobiose. Nevertheless, oligomannoses Man₃ and Man₅ were not suitable substrates for EfMan-I and they remained intact even after a long (18 h) incubation period with 10-fold more enzyme (10 μg) at room temperature or at 37°C (Fig. 2C). This suggested that EfMan-I only catalyzed unbranched α1–3- and α1–6-linked mannose with low efficiency. The weak activity on cleaving α1–6-linked mannose on an unbranched mannoside chain is similar to a reported soil bacterium Solitalea canadensis α-mannosidase (Sca6Man4191) [43]. Therefore, EfMan-I is an α-mannosidase with high efficiency in catalyzing the cleavage of α1–2-linked mannose and with weak activity towards unbranched α1–3- and α1–6-linked mannosides.

Fig. 2.

Substrate specificity studies of EfMan-I by thin-layer chromatography (TLC) analysis with reactions carried out at 37°C for (A) 30 min or (B) 24 h in the presence of EfMan-I (1.0 μg) (developing solvent used was n-PrOH:H₂O:NH₃·H₂O = 5:2:1, by volume). For (C), reactions were carried out in the presence of EfMan-I (10 μg) at room temperature (for 10a and 12a) or at 37°C (for 10b and 12b) for 18 h (developing solvent used was n-PrOH:H₂O:NH₃·H₂O = 3:2:1, by volume). Lanes: 1, mannose; 2, ManαpNP in the presence of EfMan-I; 3, ManαpNP; 4, α2-mannobiose in the presence of EfMan-I; 5, α2-mannobiose; 6, α3-mannobiose in the presence of EfMan-I; 7, α3-mannobiose; 8, α6-mannobiose in the presence of EfMan-I; 9, α6-mannobiose; 10, Man₃ in the presence of EfMan-I; 11, Man₃; 12, Man₅ in the presence of EfMan-I; 13, Man₅.

pH Profile and metal ion effects on EfMan-I

As shown in Fig. 3A, EfMan-I preferred a mild acidic reaction condition and its optimum activity was observed at pH 6.0. Medium activity was observed at pH 6.5 and in a pH range of 5.0–5.5. EfMan-I activity decreased drastically when the pH was at or below 4.5, or at or above 7.0. The effects of a metal ion such as Ca²⁺ or Mg²⁺ and a chelating agent ethylenediaminetetraacetic acid (EDTA) on the activity of EfMan-I in catalyzing the cleavage of ManαpNP were examined at pH 6.0. As shown in Fig. 3B, the activity of EfMan-I was abolished by the addition of EDTA while it was not significantly affected by the addition of Ca²⁺ or Mg²⁺ for up to 20 mM, indicating that a pre-existing metal ion bond to EfMan-I was important for its catalytic activity.

Fig. 3.

pH Profile (A) as well as metal ion and EDTA effects (B) on EfMan-I using ManαpNP as the substrate. For A, buffers used were: citric acid-Na₂HPO₄, pH 3.0–4.0; NaOAc/HOAc, pH 4.5; MES, pH 5.0–6.5, and Tris-HCl pH 7.0–9.0. For B, Black columns, MgCl₂; white columns, CaCl₂; light gray columns, EDTA; dark gray column, no metal and no EDTA control.

Kinetic studies of EfMan-I

Three substrates including ManαpNP, α2-mannobiose, and RNaseB were used to determine the apparent kinetics data for EfMan-I. As shown in Table 1, ManαpNP was the least efficient substrate among the three. Using ManαpNP as the substrate, the catalytic efficiency (k_cat/K_M) of EfMan-I was 4.7 × 10⁻² s⁻¹ mM⁻¹ or 2.82 min⁻¹ mM⁻¹ (Fig. S2). In comparison, the catalytic efficiencies of Bt2199 and Bt3990 were 0.84 and 1.3 min⁻¹ mM⁻¹, respectively [21]. EfMan-I was the most efficient in catalyzing the cleavage of α1–2-mannosyl linkage in α2-mannobiose with a catalytic efficient (k_cat/K_M) of 58 s⁻¹ mM⁻¹ (Fig. S3), which was more than 1200-fold higher than that of ManαpNP. The difference was contributed by both a higher turnover number (367-folder) and an improved binding (3.5-fold) for α2-mannobiose. RNase B was also an efficient substrate with a catalytic efficient (k_cat/K_M) of 22 s⁻¹ mM⁻¹ (Fig. S4), which was about 468-fold higher than that of ManαpNP.

Table 1.

Apparent kinetics data of EfMan-I.

Acceptor	kcat (s−1)	KM (mM)	kcat/KM (s−1 mM−1)
ManαpNP	0.49±0.03	11±1.7	4.7 × 10−2
α2-mannobiose	(1.8±0.1)×102	3.1±0.3	58
RNase B	42±3	1.9±0.2	22

¹H NMR-based time course studies of EfMan-I using ManαpNP as the substrate

¹H NMR-based time course studies (Fig. S5) showed that EfMan-I-catalyzed reaction follows an inversion mechanism similar to that described for Bt2199 and Bt3990 previously [21]. For example, β-mannose was observed as the main product of EfMan-I from ManαpNP substrate in the first 10 min. With a longer reaction time and the accumulation of β-mannose product, α-mannose appeared quickly due to a quick mutarotation process.

MALDI-TOF MS analysis of N-glycans released from RNase B with or without EfMan-I treatment

In order to test the activity of EfMan-I in catalyzing the cleavage of α1–2-mannose from high-mannose N-glycans on glycoprotein substrates, RNase B, a glycoprotein containing a single N-glycosylation, was treated with EfMan-I. The N-glycans of RNase B with or without EfMan-I-treatment were then released by PNGase F [44,45] and analyzed by MALDI-TOF mass spectrometry [46]. As shown in Fig. S6, without EfMan-I-treatment (RNase B control), Man₅GlcNAc₂, Man₆GlcNAc₂, Man₇GlcNAc₂, Man₈GlcNAc₂, and Man₉GlcNAc₂ were observed as the N-glycans released from RNase B by PNGase F (Fig. S6A). In comparison, with EfMan-I treatment, all Man_6–9GlcNAc₂ N-glycan structures were converted to Man₅GlcNAc₂ (Fig. S6B). These results confirmed that EfMan-I was specific for catalyzing the cleavage of α1–2-linked mannose residues on high-mannose-type N-glycans on glycoproteins.

HPLC and MALDI-TOF MS analyses of 2-AB labeled N-glycans released from RNase B with or without EfMan-I treatment

To quantitatively analyze the activity of EfMan-I against glycoprotein, the N-glycans of RNase B with or without EfMan-I treatment were released and labeled with 2-aminobenzamide (2-AB) [47]. As shown in Fig. 4 and Fig. S7, without EfMan-I treatment, 2-AB-labeled Man₅GlcNAc₂ (~50%), Man₆GlcNAc₂ (~37%), and Man_7–9GlcNAc₂ (~13%) were observed for the RNase B sample. With EfMan-I treatment, all Man_6–9GlcNAc₂ on RNase B were converted to Man₅GlcNAc₂.

Fig. 4.

HPLC analysis of 2-AB-labeled high-mannose N-linked glycans from RNase B without (A) and with (B) EfMan-I treatment; and schematic illustration of the reaction catalyzed by EfMan-I (C) [48].

Crystal structure of EfMan-I

EfMan-I crystallized in Space group C222₁ with four molecules per asymmetric unit. However, given the limited amount of surface area buried between monomers in the crystal packing, EfMan-I is most likely monomeric in solution. The electron density, at 2.15 Å resolution, clearly defined residues 1–712 in all four subunits in the asymmetric unit. The overall structure of EfMan-I contains two domains; an N-terminal 14-stranded antiparallel β-sandwich domain and a larger C-terminal helical domain with (α/α)₆ topology (Fig. 5A). This two-domain architecture is seen in homologous GH92 α-mannosidases [21,22,49] and other glycosidases [23,50–52]. The N-terminal β-sandwich displays some similarities to non-catalytic β-sandwich domains found in larger GH38 family glycosidases like Drosophila Golgi retaining α-mannosidases [53] and bovine lysosomal α-mannosidase [54]. In EfMan-I and in homologous GH92 family α-mannosidases, the active site lies in a cleft between the N-terminal β-sandwich domain and the C-terminal (α/α)₆ domain. It is interesting to note that the two-domain glycosidases are typically exo-glycosidases, while glycosidases with only a single (α/α)₆ domain are typically endo-glycosidases [55,56], suggesting the β-sandwich domain restricts the active site to only terminal sugars.

Fig. 5.

Overall structure of EfMan-I monomer (A), its active site (B) and potassium-binding site (C). Ribbon drawing of EfMan-I colored by a rainbow gradient according to sequence number, starting with blue at the N-terminus, and ending with red at the C-terminus. The active site is identified by the Ca²⁺ ion shown as a green sphere together with glycerol molecules (sticks) that interact with the Ca²⁺ ion shown as a green sphere together with the coordination ligands (water molecule ligands shown as smaller red spheres). Yellow dashed lines represent coordination interactions (between 2.6–2.9 Å). The two glycerol molecules modeled in the active site are shown as sticks with white-colored carbon atoms. A K⁺ ion binds to the peripheral of the (α/α)₆ domain is shown as a purple-colored sphere together with the coordination ligands (water molecule ligands shown as smaller red spheres). Only main chain atoms are shown for Thr597 and Tyr599.

The active site is located close to the center of the (α/α)₆ domain, and the N-terminal β-sandwich domain helps to shape part of the active site. Specifically, Trp68 of the N-terminal domain lies above the active site, and helps in binding substrate and substrate analogs as observed in other α-mannosidases [21,22]. Electron density clearly reveals that the active site of all four subunits in the asymmetric unit contains a bound metal with an octahedral coordination geometry. Although no divalent cations were included in protein purification or crystallization, the metal was modeled as a calcium ion because other GH92 family α-mannosidases require calcium for activity and calcium ions were observed in their structures [21,22,49]. The presence of a calcium ion in the crystal structure can explain the activity of EfMan-I observed without the addition of an external metal cation. The calcium is coordinated by the side chains of Asn562 (2.6 Å), Gln563 (2.4 Å), and Asp604 (2.4 Å) together with two water molecules (2.6 & 2.7 Å) (Fig. 5B). The sixth ligand comes from a “tube” of electron density, which was modeled as glycerol since it was added as a cryoprotectant. Additionally, electron density was observed near the calcium ion in the active site in all four monomers in the crystallographic asymmetric unit, which was also modeled as glycerol (Fig. 5B). Glycerol was also observed in the active site of other α-mannosidases (see below) [21–23].

Strong electron density was also observed adjacent to Asn20 that was modeled as a potassium ion since it was present in the crystallization conditions (Fig. 5C). Placement of sodium, also present in the crystallization conditions, resulted in strong positive difference electron density after subsequent rounds of refinement, confirming the likelihood of a more electron dense potassium ion at this location. Ligation distances and geometry are also similar to potassium ions in other protein structures [57]. The potassium ion binds on the surface of the protein at the interface between the N-terminal and C-terminal domains, where Asn20 Oδ1 coordinates (2.7 Å) together with carbonyl oxygens from Thr597 (2.8 Å) and Tyr599 (2.7 Å) (Fig. 5C). Two water molecules finish the coordination sphere resulting in a square pyramidal geometry. Although this potassium ion is also observed in all four monomers of the asymmetric unit, it is likely not directly involved in catalysis because it binds on the peripheral of the protein over 13 Å from the active site. However, the presence of a potassium ion situated at the interface between the two domains suggests the ion may play a role in helping position the proper orientation between the domains to shape the active site. This hypothesis still remains to be tested in future studies. In comparison, a sodium ion was modeled in a similar position in the Bacteroides thetaiotaomicron α-mannosidase Bt2199 (PDBID 2ww2) [21]. In Bt2199 structure, the ion is octahedrally coordinated by main chain carbonyl oxygens of Ser40, Tyr643, Gly645, and Asp648 with two waters completing the octahedral coordination.

Comparison of EfMan-I structure to other α-mannosidases

The closest EfMan-I homolog with a known structure is an α-mannosidase from human pathogen Streptococcus pneumoniae (SpGH92) [22] which has 47% sequence identity. The two crystal structures superimpose with an rmsd of 1.07 Å over 672 equivalent α-carbons (Fig. 6A). The only major deviation is in a loop centered around residue Met359 near the active site (Fig. 6A). In the EfMan-I structure, the loop adopts a conformation that closes in on the active site. However, in the SpGH92 structure, one of the four monomers in the asymmetric unit adopts two conformations of the loop, one resembles more closely the conformation of EfMan-I. The three other SpGH92 monomers in the asymmetric unit only adopt one conformation extending away from the active site as observed in Fig. 6A. The different conformations between these two structures that help shape the active site pocket may be important in substrate binding specificity. However, the corresponding loop in the α-mannosidases from Bacteroides thetaiotaomicron (the next closest homolog at 32–34% identity) displays (not shown) this loop in a closed conformation similar to EfMan-I [21,49].

Fig. 6.

Superposition of EfMan-I (salmon color) and SpGH92 (slate blue) structures (A) and their active sites (B). They superimpose with an rmsd of 1.07 Å over 672 equivalent α-carbons. Calcium ions in the active site are shown as green spheres, and the 359 loop near the active site is highlighted in bright red color for EfMan-I. The side chains of residues that coordinate the calcium ion are shown along with ordered glycerol molecules found in both mannosidase structures. Note that in each structure a glycerol coordinates the calcium ion and in the EfMan-I structure, another glycerol binds in a similar location to mannose observed in SpGH92, which is 3.4 Å below the indole ring of Trp68. Glu494 and Asp604 are proposed to be the respective general acid and general base in catalysis.

The active sites of the EfMan-I and SpGH92 structures are very similar with homologous residues coordinating the calcium ion (Fig. 6B). In the EfMan-I structure, two glycerol molecules are observed in the active site, one is involved in coordinating the calcium ion (−1 subsite), and another glycerol stacks up against Trp68 (+1 subsite). In the SpGH92 structure (PDBID 5swi), crystals were soaked with α1–2-mannobiose, which was hydrolyzed and only mannose was observed to bind to the +1 subsite location in a similar location to the secondary glycerol of EfMan-I (Fig. 6B). The −1 subsite of SpGH92 was occupied by a glycerol molecule that coordinated the calcium ion as also observed in EfMan-I. Previous α-mannosidase kinetic and structural analyses have proposed a general acid-base mechanism and identified catalytic residues [21]. These residues map to Glu494 and Asp604 in serving as the respective general acid and general base catalytic residues in EfMan-I (Fig. 6B). It is interesting to note that Asp604 also serves as a calcium ligand in the active site.

Several crystal structures have been determined for GH92 family α-mannosidases from the human gut microbe Bacteroides thetaiotaomicron (Bt) with various substrates and inhibitors bound in the active site [21,49]. These include Bt2119, Bt3990, Bt3130, and Bt3965, all of which have similar structures to EfMan-I. Bt3990, which has a 32% sequence identity to EfMan-I, superimposes onto EfMan-I with an rmsd of 1.29 Å over 696 equivalent α-carbons. The structure of Bt3990 was determined with a nonhydrolyzable disaccharide of thio-α2-mannobioside (PDBID 2ww3). Superposition of this active site onto EfMan-I is shown in Fig. 7, which reveals they are very similar. The glycerol in EfMan-I that coordinates the calcium ion superimposes more on the −1 subsite saccharide while the second glycerol occupies a position in between the two saccharides of the thio-α2-mannobioside substrate analog in Bt3990. The only main difference in the active sites is the movement of Asn601 in Bt3990. The equivalent residue (Asn562) of EfMan-I coordinates the calcium ion, while in Bt3990, it shifts down and a water molecule occupies the vacated coordination space. However, in another Bt3990 structure complexed with the inhibitor swainsonine (PDBID 2ww0), Asn601 does indeed coordinate the calcium ion similar to Asn562 of EfMan-I.

Fig. 7.

Superposition of EfMan-I (salmon color) and Bt3990 (green) active sites. The side chains of residues that coordinate the calcium ion are shown along with ordered glycerol molecules found in EfMan-I structure (pink-colored carbon atoms) and the substrate analog thio-α2-mannobioside in Bt3990 (pale-green carbon atoms). Location of the calcium ion is shown by spheres with colors corresponding to their respective active sites.

In conclusion, a new α1–2-mannosidase from Enterococcus faecalis (EfMan-I) has been identified and characterized. It has been found to catalyze the cleavage of α1–2-linked mannose residues highly efficiently from not only oligomannosides but also glycoproteins containing high-mannose-type N-glycans using ribonuclease B as a model glycoprotein. It is an inverting glycosidase similar to two characterized GH92 α1–2-mannosidases (Bt3990 and Bt2199) from Bacteroides thetaiotaomicron VPI-5482. The crystal structure of EfMan-I reveals a similar fold to other GH92 α-mannosidases and identifies residues in the active site that ligate the essential calcium ion and are involved in substrate binding and catalysis. Ordered glycerol molecules, used for cryoprotection, bind in the active site with one molecule ligating to the active-site calcium ion. An unexpected potassium ion was observed bridging two domains near the active site. With a good expression level in E. coli, high activity and high selectivity on cleaving α1–2-linked mannose residues in oligomannose structures including those attached to glycoproteins, the newly characterized EfMan-I is a powerful tool for in vitro enzymatic modification of high-mannose-type N-glycans on glycoproteins. Its characterization is also helpful for better understanding the use of host or dietary glycoproteins as nutrients by Enterococcus faecalis, a Gram-positive commensal bacterium and an opportunistic pathogen that can cause healthcare-associated infection (HAI).