Human gut microbes express functionally distinct endoglycosidases to metabolize the same N-glycan substrate

Abstract

Bacteroidales (syn. Bacteroidetes) are prominent members of the human gastrointestinal ecosystem mainly due to their efficient glycan-degrading machinery, organized into gene clusters known as polysaccharide utilization loci (PULs). A single PUL was reported for catabolism of high-mannose (HM) N-glycan glyco-polypeptides in the gut symbiont Bacteroides thetaiotaomicron, encoding a surface endo-β-N-acetylglucosaminidase (ENGase), BT3987. Here, we discover an ENGase from the GH18 family in B. thetaiotaomicron, BT1285, encoded in a distinct PUL with its own repertoire of proteins for catabolism of the same HM N-glycan substrate as that of BT3987. We employ X-ray crystallography, electron microscopy, mass spectrometry-based activity measurements, alanine scanning mutagenesis and a broad range of biophysical methods to comprehensively define the molecular mechanism by which BT1285 recognizes and hydrolyzes HM N-glycans, revealing that the stabilities and activities of BT1285 and BT3987 were optimal in markedly different conditions. BT1285 exhibits significantly higher affinity and faster hydrolysis of poorly accessible HM N-glycans than does BT3987. We also find that two HM-processing endoglycosidases from the human gut-resident Alistipes finegoldii display condition-specific functional properties. Altogether, our data suggest that human gut microbes employ evolutionary strategies to express distinct ENGases in order to optimally metabolize the same N-glycan substrate in the gastroinstestinal tract.

Fig. 1. Bacteroides thetaiotaomicron encodes two high-mannose glycan-specific endoglycosidases.

a Sequence similarity network of GH18 family from CAZY database at at a alignment score threshold of 50, representing 10,984 protein nodes. b Schematic representation of PUL16 and PUL72 of B. thetaiotaomicron. ECF σ, extracytoplasmic σ factor; SGBP, surface glycan binding protein; unk, protein with unknown function. Percentage of protein sequence identity between the syntenic genes is indicated. c Competitive kinetic analysis of BT3987 (50 nM) and BT1285 (50 nM) in the presence of four different glycoprotein substrates (2 µM each). Assays were run in technical triplicates (n = 3). Curves represent the complete deglycosylated substrate proportion. d Schematic representation of HM specific PULs in B. thetaiotaomicron indicating predicted subcellular localization of protein components of PUL 16 and 72. Cartoon representation of protein structures were obtained from AF models, with the exception of BT3984 (pdb: 8UWV), BT3984 (pdb: 6TCV), BT3990 (pdb: 2WVX) and BT1281 (pdb: 4MRU). Sugar symbols: GlcNAc, Man, Fuc, Gal, Neu5A. Source data are provided as a Source Data file.

Fig. 2. Structural basis for the HM N-glycan substrate specificity of BT1285.

a Surface representation of the crystal structure of BT1285_D161A-E163A (BT1285i) in complex with Man₉GlcNAc₂ (pdb code: 8U48). The different GH loops around the HM N-glycan are distinctly colored. b Two views of cartoon representation of superimposed crystal structure of BT1285i (yellow) and BT3987_D312A-E314L (BT3987i) (purple) (pdb code:6TCV, both in complex with Man₉GlcNAc₂. c Two views of the electron density map of Man9GlcNAc2 substrate from BT1285_D161A-E163A -Man₉GlcNac₂ crystal structure, shown at 1.0 σ r.m.s deviation. In left part, schematic representation of Man9GlcNAc2 substrate. Sugar symbols: GlcNAc, Man. β-1-4 bond cleaved by ENGases is indicated with scissors. d Close inspection of substrate binding site of BT1285 in complex with Man₉GlcNAc₂. Residues used in the alanine scanning assays are represented as sticks. e Hydrolytic activity of BT1285 and mutants against HM-IgG1, as determined by LC-MS analysis, normalized to BT1285 wt. Data are presented as mean values±SD. Assays were run in technical triplicates (n = 3). f Close inspection of key residues involved in the Mannoses recognition and binding in BT1285. Source data are provided as a Source Data file.

Fig. 3. BT1285 and BT3987 enzymes are differentially affected by temperature and pH changes.

a Thermal stability of BT1285 (blue line) and BT3987 (orange line). Assays were run in technical triplicates (n = 3). Data are presented as mean values ± SD. b Relative activity mesearuments of BT1285 and BT3987 wt enzymes using RNAseB as a substrate at different pH in the range of 2 to 9 at 23 °C. (n = 3 technical triplicates) Data are presented as mean values ± SD. c Close inspection of cartoon representation of crystal structure of BT1285 in complex with Man₉GlcNAc₂ (pdb code: 8U48) showing the differencial aromatic residues with BT3987 and possible residues involved in pH dependence. Active site is composed by D161 and E163 residues. d, e Relative activity of BT1285 mutants measurements using RNAseB as a substrate at different pH in the range of 2 to 9. Endpoint assays were performed by techinical triplicates (n = 3). Data are presented as mean values ± SD. Source data are provided as a Source Data file.

Fig. 4. Binding affinities of BT1285i and BT3987i to different N-glycan substrates by SPR.

In the upper panel are represented the binding kinetics of 1285i (analyte) vs different N-glycan substrates immobilized to a protein A (Man₅, Man₉, CT) and CM5 sensor chip (RNaseB) using amine coupling to a surface density of 100−150 response units (RU). In bottom panel shows binding analysis of BT3987i (analyte). Concentration range of BT1285i analyte plotted (serial dilution 1:2): 37.5 nM to 600 nM for IgG-Man₅/Man₅; 9 nM to 300 nM for IgG Man₉/Man₉; 156 nM to 5 µM for IgG CT, and 3.75 to 240 nM for RNAseB. For BT3987i analyte concentration range plotted (serial dilution 1:2): 625 nM to 20 µM for IgG Man₅/Man₅; 156 nM to 5 µM for Man₉/Man₉ and CT; 1 µM to 32 µM for RNAseB. Black dashes line represent the fitting on kinetic models. ND=not determined K_D. Assays were run in independent duplicates (n = 2). Sugar symbols: GlcNAc, Man, Fuc, Gal. Source data are provided as a Source Data file.

Fig. 5. Kinetics of hydrolysis and binding of BT1285 to both glycans on the IgG homodimer.

a Kinetic analysis of BT3987 and BT1285 vs IgG HM. Asssay is representative of triplicates at 37 °C in HBS buffer. b SPR assays of BT1285i (analyte) with different diglycosylated, monoglycosylated and aglycosylated (N297Q) IgG variants. Concentration range of analyte was 4.7 nM to 300 nM for monoglycosylated IgGs, and 156 nM to 5 µM for IgG deglycosylated. Assays are representative of two independent experiments. Sugar symbols: GlcNAc, Man. Source data are provided as a Source Data file.

Fig. 6. Analysis of BT1285i-Fc HM complex formation.

a SEC analysis of BT1285i-Fc complex run on Superdex S200. Reducing and non-reducing free-stain SDS-PAGE (4–20%) analysis of eluted fraction from gel filtration. Indicated with letter L (ladder), C (complex, fraction injected into the SEC). Experiment were repeated independently three times with similar results. b AUC analysis of BT1285i, Fc HM and BT1285-Fc complex (fraction correspond to fraction eluted in volume 12.5 ml in SEC in Fig. 6a. Samples in HBS Buffer were centrifuged at 45,400 g. Curves represent fits to the continuous c(s) distribution model in SEDFIT. Res.=residuals. c, d 3D reconstruction of SAXS c and NS-TEM d data, respectively, revealed 2:1 complex low-resolution structures. We fiited high-resolution structure of Fc (pdb code:5JII) and BT1285i-HM (pdb code:8URA) in SAXS and NS-TEM maps. e Close inspection of BT1285i-Man₉-Fc interface region showing main resdiudes of both proteins as sticks. f Relative ENGase activity measurements of selected mutants in the interface region of Fc (red) and BT1285 (green). Assays were run in technical triplicates (n = 3). Data are presented as mean values ± SD. Source data are provided as a Source Data file.

Fig. 7. Two different HM-ENGases conservation in Alistipes finegoldii.

a Schematic representation of putative High-Mannose PULs in A. finegoldii in comparison with High-Mannose PULs in B. thetaiotaomicron. SGBP=surface glycan binding protein, PGBP=peptidoglycan binding protein, unk=unknown function, ECF=extracellular factor. Left pannel shows sequence identity percentages of B.thetaiotaoimicron VPI−5482 and A. finegoldii HM-ENGases. b Competitive kinetic analysis of Alfi_0894 and Alfi_0882. Curves represent the complete deglycosylated substrate proportion. c Kinetic analysis of Alfi_0894 and Alfi_0882. Representative of technical triplicates (n = 3). d Tm vs pH curves of both HM-ENGases from A. finegoldii obtained by DSF analysis. e Relative ENGase activity at different pHs, of both HM-ENGases from A. finegoldii obtained by LC-MS analysis using RNAseB as a substrate at 20 °C. Assays were run in technical triplicates (n = 3). Data are presented as mean values ± SD. Sugar symbols: GlcNAc, Man, Fuc, Gal, Neu5A. Source data are provided as a Source Data file.