Reagents and general proceduresAll chemical reagents and solvents used were reagent grade or higher purity and obtained from commercial sources without further purification. Probe ABP1 and ABP2 were obtained from the Bio-organic Synthesis Group at Leiden University. Analytical thin-layer chromatography (TLC) was performed on silica gel 60 F254 plates (Merk, Supelco, 105554.0001) and analyzed by fluorescence detection with UV light (λ = 254 nm). Staining reagent containing 10% H2SO4 in ethanol was used, followed by heat treatment for visualization.Animal work and stool sample preparationAll animal procedures were performed in accordance with protocols approved by The Institutional Animal Care and Use Committee at Shenzhen Institute of Advanced Technology (Protocol 2015-14). Two sets of 4 female C57BL/6J littermate mice aged 6–8 weeks were obtained from Guangdong Medical Laboratory Animal Center (GDMLAC) and maintained in a 12 h light/12 h dark cycle with ad libitum access to food (GDMLAC) and water. Stool samples were collected twice daily from each mouse, immediately frozen at −20 °C, and stored until microbial extraction. The samples were thawed to room temperature, diluted in cold PBS (pH 7.4), and vortexed to yield slurries, which were then centrifuged at 100 × g for 1 min. The upper layer was removed, and the pellet was rinsed twice with PBS before being resuspended in HEPES extraction buffer (100 mg pellet per 0.5 mL buffer) and subjected to ultrasonic fragmentation at 0 °C (12% power, 3s-on, 5s-off, Scientz-IID) for 30 min. The mixtures were centrifuged at 12,000 rpm for 5 min at 4 °C, and the supernatant was collected for BCA assay to determine concentration.Pull downPull down experiments were performed following a previously reported protocol68. Stool microbiome protein samples (1.0 mg for each sample in 200 μL) was incubated with DMSO/ABP1 for 1 h at 37 °C and extracted by chloroform/methanol precipitation. The protein pellet was air-dried and samples were treated with urea buffer, dithiothreitol (DTT) and Tris-2-carboxyethylphosphine hydrochloride (TCEP) for reduction and alkylation. The samples then underwent another precipitation assay as described above. Next, after pull down buffer washing, the treated proteins were performed with magnetic bead separation and digestion: The magnetic bead samples were divided into two parts: 1/3 for on-bead digestion and 2/3 for in-gel digestion. Finally, the obtained peptides were extracted and dissolved in 0.1% formic acid for LC-MS/MS analysis.LC-MS/MS analysisAll peptides were reconstituted in 0.1% formic acid (vol/vol) and separated on reversed-phase columns. The trapping column had a particle size of 3 μm, C18, and a length of 20 mm (Thermo Fisher Scientific, P/N 164535), while the analytical column had a particle size of 2 μm, C18, and a length of 150 mm (Thermo Fisher Scientific, P/N 164534). The separation was performed using an Ultimate™ 3000 RSLCnano system (Thermo Fisher Scientific, San Jose, CA, USA) with a 60-min gradient (buffer A: 0.1% FA in water, buffer B: 0.1% FA in 80% MeCN) at a flow rate of 300 nL/min. The peptides were then analyzed by Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) in a data-dependent mode. Orbitrap Fusion Lumos mass spectrometer (ThermoFisher Scientific) mass spectrometer was operated in positive ion mode with an ion transfer tube temperature of 275 °C and a positive ion spray voltage of 2.1 kV. One full scan in the Orbitrap (at 60,000 resolution in profile mode with an AGC target of 4 × 105 and maximum injection time of 50 ms) was followed by as many MS/MS scans as could be acquired on the N most abundant ions in 1 s in the ion trap (rapid scan type, HCD collision energy of 30%, AGC target of 5 × 104, maximum injection time of 50 ms). Singly and unassigned charge states were rejected. Dynamic exclusion was enabled, an exclusion duration of 60 s, and an exclusion mass width of ±10 ppm.Metaproteomics data processingMetaLab software (version 1.1.1) was used to generate a sample-specific protein database from original large databases69. The mouse-specific microbiome database comprised 44,478 proteins, which were constructed using a MS/MS clustering strategy based on a non-redundant microbial protein database containing 1,761,33 protein sequences combined with the mouse proteome from Uniprot. Non-redundant microbial protein database was generated by clustering protein sequences of genomes assembled at the chromosomal level (retrieved in August 2018) from NCBI70 using CD-hit71. The MS/MS data were searched against the mouse-specific database using MaxQuant 2.0 with a precursor mass tolerance of 20 ppm and a fragment mass tolerance of 0.5 Da72. The “match between runs” option was enabled to match identifications across different replicates. Enzyme specificity was used, and only tryptic peptides with up to 2 mis-cleavages were allowed in the final data sets. Cysteine carboxamidomethylation was specified as a static modification, and oxidation of methionine residue and acetylation (protein-N) were allowed as variable modifications. Reverse decoy databases were included for all searches to estimate false-discovery rates. Peptide and protein identifications were quantified and filtered for less than 1% false-discovery rate (FDR). In order to facilitate further analysis, the protein group originating from the host mouse was excluded.Quantitative taxonomic analyses were performed by assigning identified proteins to their corresponding taxonomic lineage in Genebank and summing up their intensities. To establish the phylogenetic relationship between different α-galactosidases, multiple alignment of protein sequences was performed using MAFFT v.773. For inferring the maximum-likelihood phylogenetic tree, the best-fitting substitution model (LG + G4) was selected using the Bayesian information criterion in IQ-TREE web server74.Venn diagrams were generated using the online tool available at (http://bioinformatics.psb.ugent.be/) webtools/Venn/. The resulting diagrams were exported as PDF files and further customized by adjusting the color scheme and transparency. Data processing procedures primarily relied on the R programming language, with the ggplot2 package being the main tool for data visualization. Protein sequence similarity was calculated using the Biostrings package. Z-scores for the data were obtained utilizing the scale function. The scatter plot included the calculation of the Pearson correlation coefficient and the fitting curve, which were performed and added using the ggpubr package. The Pearson correlation coefficient between samples was analyzed using the PerformanceAnalytics package. The Games-Howell test, applied to compare groups, was conducted and visualized using the ggstatsplot package. Animal intestinal microbiome data were retrieved from the Animal Microbiome Database (AMDB) at http://leb.snu.ac.kr/amdb. Subsequently, the data were normalized. The fold change (FC) was determined by comparing the intensity of the protein group in the experimental group (probe group) to that of the control group (DMSO group), and any missing values were imputed with a value of 0.CloningThe genes encoding AGAL1 to AGAL5 (Gene names A4V02_01335, A4V02_08955, AT726_08105, AT726_08965, A4V09_19485) (Supplementary Data 2) were optimized and synthesized by GenScript, and were cloned using CloneEZ into a pET-30a (+) expression vector by NdeI/HindIII, respectively, so that a fusion gene with C-terminal His6-tag was obtained. AGAL6 (Gene name N134_04935) was additionally cloned into a pET-30a (+) vector by BamHI/XhoI with a C-terminal His6-tag, respectively. All cloning strains are TOP10. The antibiotic in the plasmid is kanamycin.Expression trialsTo determine ideal expression conditions, the recombinant plasmid was transformed into E. coli BL21 DE3, C41, or Rosetta using the heat-shock method and plated on LB-Agar plates containing 50 μg/ml kanamycin. After overnight growth at 37 °C, a colony was selected and placed in 5 ml of LB medium containing 50 μg/ml kanamycin and grown overnight at 37 °C with shaking at 200 rpm as a starter culture. The following day, the starter culture was added to 100 mL of LB medium containing 50 μg/ml kanamycin in a 250 ml flask and grown at 37 °C with shaking at 200 rpm until the OD600 nm reached 0.6. To test different expression conditions, flasks were induced with either 0.1 mM, 0.5 mM, or 1.0 mM IPTG and placed at 20 °C, 24 °C, 30 °C, or 37 °C. Aliquots of 10 ml were removed at 2, 4, 6, or 16 h, and the bacteria were pelleted by centrifugation at 4000 × g for 10 min. The pellets were then frozen at −20 °C for further analysis.Large-scale expression and purificationA single colony of C41 transformed with the recombinant plasmid was grown in 5 mL of LB medium containing 50 μg/ml kanamycin at 37 °C with shaking at 200 rpm overnight as a starter culture. The next day, the starter culture was added to 500 ml of LB medium containing 50 μg/ml kanamycin in a 1 L flask and grown at 37 °C with shaking at 200 rpm until the OD600 nm reached 0.6. Once the OD600 nm reached 0.6, the bacteria were induced with 1 mM IPTG and grown at 24 °C with shaking at 200 rpm for 16 h. The cells were harvested by centrifugation at 4000 × g for 10 min, and the pellets were resuspended in a lysis buffer (20 mM HEPES pH 7.5, 5% glycerol, 250 mM NaCl) containing 1 mg/ml of lysozyme. The cells were sonicated, pulsing for 3 s with 5 s off for 10 min at 70% maximum power, and then passed through a French press twice. The cells were centrifuged at 30,000 × g at 4 °C for 30 min to remove insoluble components and membrane fractions. The proteins were then concentrated and filtered using Amicon filters with a 30,000 Dalton molecular weight cut off. The samples were filtered through a 0.2 μm filter and loaded onto a HisTrap HP 5 ml Ni-NTA column with wash buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 25 mM imidazole). The column was washed with 10 column volumes of wash buffer and then eluted with elution buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 300 mM imidazole) using a gradient. Peak fractions were collected and dialyzed overnight with 20 mM HEPES pH 7.5, 150 mM NaCl, 1 mM DTT. The dialyzed protein was concentrated again using Amicon filters with a 30,000 Dalton molecular weight cut off. The sample was then loaded onto a Superdex size exclusion column (GE) pre-equilibrated with 20 mM HEPES pH 7.5, 150 mM NaCl, 1 mM DTT. The peaks were pooled, and the purification was confirmed using SDS-PAGE. The protein samples were concentrated using an Amicon filter with a 30,000 Dalton molecular weight cut off until reaching 1 mg/ml. Finally, the proteins were flash-frozen in 50 μl aliquots in liquid nitrogen for further use.Enzyme activity assaysGalactosidase activity was assayed using a colorimetric assay with pNP-α-galactopyranoside (Aladdin, Cat. N100668) as the substrate to measure released PNP. Each assay contained 2.5 mM α-galactopyranoside and 70 mM potassium phosphate buffer pH 6.5, and to ensure values within the linear range, either 100 μg, 10 μg, or 1.0 μg of enzyme or no enzyme as a negative control, in a 96-well plate. The plate was incubated at 37 °C for 5 min with gentle agitation. The reaction was stopped by adding 140 mM sodium borate buffer pH 9.8, and the absorbance at 407 nm was measured. The molar extinction coefficient of PNP, 18.6 mM−1 cm−1, was used to calculate the amount of released pNP.pH dependence and different additives effect on enzyme activityTo determine the optimal pH for AGALs activity, a colorimetric assay was performed with α-galactopyranoside as the substrate using different buffers. HEPES buffer was used for pH 7–8, phosphate buffer was used for pH 6–8, McIlvaine buffer (Na2HPO4 and citric acid) was used for pH 4–8, and Tris-HCl buffer was used for pH 7–9. To assess the effect of different sugars, metal components, or other modifiers on AGALs activity, a colorimetric assay was performed with PNP-α-galactopyranoside as the substrate. The additives included 5 mM glucose, galactose, maltose, melibiose, stachyose, lactose, raffinose, EDTA, MgCl2, KCl, CaCl2, or 1 mM DTT, or 1% Triton X-100, Tween-20, or SDS.Enzyme kineticsTo determine the kinetic constants (Km, Vmax) of the four enzymes, an enzyme assay was performed with increasing concentrations of substrate pNP-α-galactopyranoside ranging from 0 mM to 3.0 mM at pH 6.5 in 70 mM potassium phosphate buffer as described in Enzyme Activity Assays above. The enzyme activity was then plotted against the substrate concentration on GraphPad Prism 8.0. Non-linear regression analysis for Michaelis-Menten enzyme kinetics was performed using the software to calculate Vmax and Km values for each enzyme.Negative-stain electron microscopyNegative-staining electron microscopy was used to investigate AGAL5, CaCl2 treated AGAL5, and AGAL5 bound with an inhibitor. A 5 µL protein solution was applied to a glow-discharged grid with a continuous carbon film, followed by staining with 0.75% w/v uranyl acetate for 10 s and 60 s, respectively. After each staining step, 4 µL of uranyl acetate was added and removed. The processed sample was then examined using a 120 kV electron microscope from FEI Company. The images obtained from the electron microscope were used to analyze the structure and interactions of AGAL5 with CaCl2 and an inhibitor.Cryo-EM grid preparationThree frozen samples were prepared for cryo-EM single-particle analysis. The first sample contained AGAL5 at a concentration of 1.4 mg/ml. The second sample contained AGAL5 incubated with 5 mM CaCl2 for 1 h on ice at a concentration of 1.3 mg/ml. The third sample contained AGAL5 incubated with ABP2 at a ratio of 1:2 for 30 min at a concentration of 1.3 mg/ml. All samples were prepared using a Vitrobot operated at 4 °C and 100% humidity. They were applied to a glow-discharged 300 mesh Quantifoil Cu R1.2/1.3 grid, blotted for 2–3 s with a nominal blot force 1 of the Vitrobot, and immediately plunged into pre-cooled liquid ethane for vitrification.Cryo-EM data processingWe collected 3277 micrographs of AGAL5 wild type, 3199 micrographs of CaCl2 treated AGAL5, and 8959 micrographs of AGAL5 bound with an inhibitor, respectively. The micrographs were motion-corrected using MotionCor275, and CTF parameters were estimated by CTFFIND4.176. For wild-type AGAL5 and CaCl2 treated AGAL5, micrographs were denoised using the TOPAZ program77. High-quality training data of protein particle were filtered from a small manually picked dataset through 2D classification and used as input for TOPAZ program. The trained model then picked a large number of protein particles from all the micrographs. For AGAL5 bound with ABP2, particles were picked manually from selected micrographs, and a subset of these particles was selected through 2D classification. This subset was used as a reference to pick all the particles from all the micrographs using the auto-pick method of Laplacian-of-Gaussian blob in Relion program78. After several rounds of 2D classification, high SNR (signal to noise) 2D classes were retained for further process. In total, 261,006 particles of wild-type AGAL5, 7,720,604 particles of CaCl2 treated AGAL5, and 240,999 particles of AGAL5 bound with ABP2 were selected. 3D initial models generated from good SNR 2D classes served as references in the 3D refinement and classification procedure. The best class of 3D classification was used as an updated reference for 3D auto-refinement. After several rounds of 3D classification and 3D refinement, we obtained consensus particles that were refined in the final reconstruction. As a result, we obtained a 3.28 Å wild-type AGAL5 map, a 3.37 Å map for CaCl2 treated AGAL5 map, and a 3.17 Å map for AGAL5 bound with ABP2 map through CTF refinement and post-processing. Local resolution maps showed that the inner resolution was higher than the surrounding resolution, respectively, and the density map of CaCl2 treated AGAL5 was incomplete. The reported resolution based on the gold-standard Fourier shell correlation at 0.143 criteria (see Supplementary Fig. 7a–c, Supplementary Table 2). The atomic models of AGAL5 were built using COOT79 and refined using the Phenix program. The crystal reference model (PDB: 4FNQ) was used as a starting point for the model building of AGAL5 atomic model. For CaCl2 treated AGAL5 and AGAL5 bound with ABP2, the same procedure was followed, except that the initial model was AGAL5 instead of the crystal model. Refinement was carried out using phenix.real_space_refine80 (Supplementary Table 2).TLC to show hydrolysis and transglycosylationTo demonstrate RFOs hydrolysis, a 50 μL enzyme reaction mixture containing 1.14 μM of enzyme, 50 mM acetate buffer (pH 6.5), and 85 mM of the substrate (melibiose, raffinose, or stachyose) was incubated at 37 °C for 3 h in duplicate with a control containing no enzyme. At every 10-min interval within the 1st hour, 2nd hour, and 3rd hour time points, 10 μL of the reaction mixture was removed and boiled for 10 min to stop the reaction. A silica gel 60 F 254 plate (Sigma) was used for thin-layer chromatography (TLC) analysis, with 2.5–5 μL of the reaction mixture or standards spotted 1 cm from the bottom. The mobile phase used was butanol:ethanol:water (5:3:2), and a mixture of glucose + galactose, melibiose, sucrose, galactose, and raffinose standards were used for comparison. Once the mobile phase reached near the top of the plate, it was dried and developed with orcinol-sulfuric acid reagent (0.1% orcinol m/v and 70% sulfuric acid v/v), and the sugars were visualized by charring using a hotplate (Fig. 7a, Supplementary Fig. 9). The duplicate hydrolysis experiment was repeated three times.To demonstrate blood group antigen trisaccharide hydrolysis, a 15 μL enzyme reaction mixture containing 1.5 μM of enzyme, 1.5 μL substrate (10 mg/ml) blood group antigen A or B, and 12 μL McIvaine buffer (pH 7.0) was incubated at 37 °C for 5 h in duplicate with a control containing no enzyme. After 1 h and 3 h time points, 5 μL reaction mixture was removed and boiled for 10 min to stop the reaction. A silica gel 60 F 254 plate (Sigma) was used for thin-layer chromatography (TLC) analysis, with 2.5–5 μL of the reaction mixture spotted 1 cm from the bottom. The mobile phase used was butanol:ethanol:water (5:3:2). Once the mobile phase reached near the top of the plate, it was dried and developed with orcinol-sulfuric acid reagent (0.1% orcinol m/v and 70% sulfuric acid v/v), and the sugars were visualized by charring using a hotplate (Fig. 7b, Supplementary Fig. 10). The duplicate hydrolysis experiment was repeated three times.To test transglycosylation, a 50 μL enzyme reaction mixture containing 1.14 μM of enzyme, 50 mM acetate (pH 6.5), 15 mM of donor substrate (pNP-α-galactopyranoside), and 85 mM of acceptor substrate (stachyose, raffinose, or melibiose) was incubated at 37 °C for 24 h in duplicate. Reactions without acceptor substrate or donor substrate were used as controls. After each hour, 10 μL of the assay mixture was removed and boiled for 5 min to stop the reaction. A TLC analysis was performed using a silica gel 60 F 254 plate (Merk, Cat. 1.05554.0001), with 2.5–5 μL of the reaction mixture or standards spotted 1 cm from the bottom. The mobile phase used was t-butanol:ethanol:water (5:3:2), and the sugars were visualized by charring using a heat gun after developing the plate with orcinol-sulfuric acid reagent (Fig. 7d, Supplementary Fig. 11). The whole experiment was repeated three times.LC-MS for blood group B antigen trisaccharide analysisAn HPLC-MS system equipped with Electron Spray Ionization (1260-ultivo, Agilent Technologies, Santa Clara, CA, USA) was used for analysis. A Waters Xbridge Amide analytical column (2.1 × 100 mm, 1.7 μm) was used for the separation. Samples were eluted with solvent A [30/70 ACN/H2O with 0.1% NH4OH] and solvent B [80/20 ACN/H2O with 0.1% NH4OH] using the following gradient program at a flow rate of 0.4 mL/min; 0–10 min, 100%–80% solvent B; 10–15 min, 60% solvent B; The injected volume was 1 μL and the column temperature was set at 30 °C. The gas flow and temperature were set to 7 mL/min and 300 °C, respectively. The pressure of the nebulizer was 40 psi, the capillary voltage was set to 4000 V for the positive ionization mode. Selected reaction monitoring (SRM) was chosen as the scan mode, detecting precursor to product ion transitions. Thus, the m/z transitions were 511 → 365 (CE: 33) for the B antigen trisaccharide.100 μL sample, 100 μL PMP methanolic solution (0.5 mol/L) and 100 μL ammonia solution were mixed in a centrifuge tube. The mixture was heated to 70 °C for 30 min. Subsequently, 10% acetic acid aqueous solution and CHCl3 (1.0 mL each) were added. The upper aqueous phase was filtered through a 0.22 um membrane for further analysis. A HPLC-MS system equipped with an Electron Spray Ionization (1260-ultivo, Agilent Technologies, Santa Clara, CA, USA) was used for analysis. Separation was performed on an Agilent Poroshell 120 EC-C18 column (3 × 100 mm, 2.7 μm). Samples were eluted with solvent A [Water with 5 mM ammonium acetate] and solvent B [Acetonitrile with 5 mM ammonium acetate] using the following gradient program at a flow rate of 0.3 mL/min; start with 5% solvent B, change from 5 to 50% solvent B in 15 min and constant 100% solvent B for 5 min. A post-run of 5 min was programmed to equilibrate the column between analyses. The injected volume was 5 μL and column temperature was set at 30 °C. The gas flow and temperature were set at 10 mL/min and 350 °C, respectively. The pressure of the nebulizer was 40 psi, the capillary voltage was set at 4000 V for the positive ionization mode. Selected reaction monitoring (SRM) was chosen as scan mode, detecting precursor to product ion transitions. Thus, the m/z transitions were 657.3 → 373.4 (Fragmentor: 140 V; CE: 40) and 511 → 217.2 (Fragmentor: 150 V; CE: 29) for the hydrolysis product (Fig. 7c).Reporting summaryFurther information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
