Carbohydrate-active enzymes from Akkermansia muciniphila break down mucin O-glycans to completion

Abstract

Akkermansia muciniphila is a human microbial symbiont residing in the mucosal layer of the large intestine. Its main carbon source is the highly heterogeneous mucin glycoprotein, and it uses an array of carbohydrate-active enzymes and sulfatases to access this complex energy source. Here we describe the biochemical characterization of 54 glycoside hydrolases, 11 sulfatases and 1 polysaccharide lyase from A. muciniphila to provide a holistic understanding of their carbohydrate-degrading activities. This was achieved using a variety of liquid chromatography techniques, mass spectrometry, enzyme kinetics and thin-layer chromatography. These results are supported with A. muciniphila growth and whole-cell assays. We find that these enzymes can act synergistically to degrade the O-glycans on the mucin polypeptide to completion, down to the core N-acetylgalactosaime. In addition, these enzymes can break down human breast milk oligosaccharide, ganglioside and globoside glycan structures, showing their capacity to target a variety of host glycans. These data provide a resource to understand the full degradative capability of the gut microbiome member A. muciniphila.

Fig. 1. Activity of the recombinantly expressed glycoside hydrolases from AM against α-linked monosaccharides.

a, Structural features that are expected in natural secreted mucin glycoproteins with epitopes highlighted. Only α-linkages are labelled apart from the core GalNAc monosaccharides that are also α-linkages. The structure of an O-glycan chain is generally accepted to be categorized into three sections: (1) the core, consisting of an α-linked GalNAc attached to a serine or a threonine, (2) the polyLacNAc extensions linked to the core GalNAc and (3) the terminal ‘capping’ epitopes. The polyLacNAc extensions are generally LacNAc disaccharides linked through β1,3-bonds, but variable sulfation, fucosylation and branching add to the complexity and heterogeneity along these chains. Key ganglioside and globoside structures are also included. b, Volcano plot highlighting the differential gene expression of AM when grown with PGMIII compared with glucose. Genes that did not pass the threshold of significance (FDR (false discovery rate; Benjamini–Hochberg) < 0.05) are coloured grey. Genes that passed the threshold of significance (FDR < 0.05) but had a log₂(fold change (FC)) between −1.5 and 1.5 are coloured blue. Genes that were significantly differentially expressed (FDR < 0.05) and had a log₂(FC) <−1.5 or >1.5 are coloured red. Enzymes from this study that were significantly differentially expressed are labelled. The total number of genes in the analysis was 2,171. c, TLC of the whole-cell assay of PGMIII-grown AM against fresh PGMIII. A smear can be seen increasing in concentration over time. The glycans in these samples were then labelled with procainamide and analysed by LC–FLD–ESI–MS. The chromatogram for the 9 h sample is shown, and the different glycan peaks are labelled. The data for all the time points are in Supplementary Fig. 6. The data allow the reconstruction of the order of monosaccharides in an oligosaccharide but do not provide information about linkages. It is also not always possible to tell where a fucose or sulfate group is along the chain. Example mass spectra and MS–MS fragment data are presented in Extended Data Figs. 2 and 3. d, A cocktail of enzymes from A. muciniphila BAA-835 can completely degrade the O-glycans from PGMIII down to the core GalNAc. The top and bottom panels are a TLC and HPAEC-PAD of the results, respectively. PGMIII was incubated with a cocktail of enzymes (in the pale-blue box), and the reaction terminated by boiling (1). The result of this reaction was a large amount of monosaccharides that can be seen on the TLC and by HPAEC-PAD. This reaction was then dialysed to remove all free monosaccharides and glycans (2) and concentrated (3). Untreated PGMIII was also included as a control (4). The application of Amuc_1008^GH31 is indicated by plus signs, and in samples 3 and 4, GalNAc can be seen to be released. Standards have also been included. Enzyme assays were carried out at pH 7 and 37 °C, overnight, and with 1 μM enzymes. Neu5Ac, N-acetylneuraminic acid; Neu5Gc, N-glycolylneuraminic acid; o/n, overnight. Source data

Fig. 2. Activity of the recombinantly expressed glycoside hydrolases from AM against α-linked monosaccharides capping mucin.

a, Activity of enzymes against PGMIII (pre-treated with Amuc_1835^GH33 and Amuc_1120^GH95) and analysed using HPAEC-PAD. Standards were also run to identify monosaccharide products. Top panel: relevant area of chromatograms. Bottom panel: enlarged image of the area where galactose elutes, showing details (dashed blue box in the top panel). b, Activity of enzymes releasing α-GlcNAc, α-GalNAc and α-fucose from GH16-released O-glycans from PGMIII. Top panel: full chromatograms with the two types of sialic acid highlighted in pink and purple. Bottom panel: enlarged image of the smaller peaks (highlighted by the dashed orange box in the top panel). Black diamonds and black stars indicate glycan number 10 and number 20 peaks, respectively. Highlighted in yellow and blue are the number 11 and number 12 peaks. The red asterisks indicate where a glycan is absent. Source data

Fig. 3. Activity of the glycoside hydrolases from families 2, 35 and 43 from AM against β-linked galactose substrates.

a, A heat map of recombinant enzyme activities against defined oligosaccharides. The dark blue and white indicate full and no activity, respectively, and partial activities are represented by the lighter blues. Partial activity is when all the substrate has not been broken down in an end-point assay. b, The activity of the panel of β-galactosidases against PGMIII that had been sequentially degraded. Control 1, no enzymes added; control 2, Amuc_1835^GH33 and Amuc_1120^GH95; control 3, control 2 plus Amuc_2108^GH16. The top and bottom panels are the TLC and the LC–FLD–ESI–MS results, respectively. These assays were performed in stages, so the reactions were boiled in between steps. Enzyme assays were carried out at pH 7 and 37 °C, overnight and with 1 μM enzymes. Monosaccharide standards are shown in the right lanes. c, Activity of the GH2, GH35 and GH43 enzymes against lactosylceramide. The assays were analysed using HPAEC-PAD, and the different controls confirm the release of galactose. Left panel: the different controls and samples are stacked. Right panel: the assay chromatograms are overlaid for comparison so the galactose peaks can be observed in more detail. The substrate could not be resolved using this method, so whether the reaction had gone to completion could not be determined and TLC of the samples was also inconclusive. Enzyme assays were carried out at pH 7 and 37 °C, overnight and with 1 μM enzymes. LNT, lacto-N-tetraose; LNnH, lacto-N-neohexaose; LNFP II, lacto-N-fucopentaose II; Cer, ceramide. Source data

Fig. 4. Activity of β-HexNAcases from glycoside hydrolase families 20, 84, 123 and 4 from AM.

a, Heat map of recombinant enzyme activities against defined oligosaccharides. The dark blue and white indicate full and no activity, respectively, and partial activities are represented by the lighter blues. Asterisks represent those substrates that were generated by pre-treating with other CAZymes. b, The activity of the panel of β-HexNAcases against PGMIII that had been sequentially degraded. Control 1, no enzymes added; control 2, Amuc_1835^GH33 and Amuc_1120^GH95; control 3, control 2 plus Amuc_2108^GH16; and control 4, control 3 plus Amuc_0771^GH35. The relevant area of the chromatograms of the LC–FLD–ESI–MS data are shown, and the two glycans are highlighted. Red asterisks indicate where a glycan is not present. These assays were performed in stages, so the reactions were boiled in between steps. Enzyme assays were carried out at pH 7 and 37 °C, overnight and with 1 μM enzymes. c, TLC results of sequential assays against ganglioside GD1a. GDa1 was sequentially treated with Amuc_1835^GH33 and then Amuc_0771^GH35 (lane B). The sample was then boiled and a panel of β-HexNAcases added. Source data

Fig. 5. Illustration of the CAZymes characterized in this report against host glycans by AM.

The different types of substrate are grouped where possible. The enzymes listed for each reaction are the ones that will work alone, but for the Lewis B structures, ‘and’ signifies that both enzymes are required to remove the fucose. For the sialylfucosyllacto-N-tetraose (SFLNT), the possible fucosidase activities were not included, but Amuc_0392^GH29 can act on this substrate also. Only α-linkages are labelled apart from the core GalNAc monosaccharides that are also α-linkages.

Fig. 6. A model for the degradation of mucins by A. muciniphila based on current understanding.

The GH enzymes included are only where activity has been observed, and the colour indicates the type of activity (see key). The localization of all the enzymes is based on the SignalP 6.0 prediction, apart from Amuc_2108^GH16 owing to it being observed in outer membrane analysis when AM was grown on mucin, and our whole-cell assays support a GH16 being localized to the outside of the cell. The peptidases included are both those that have been characterized and those that have been highlighted as upregulated on mucin. The signal sequences of a peptidase and sulfatases potentially localize them to the outside of the cell also. The numbers indicate the order that mucin is degraded. 1: There is some processing of mucin on the surface, with both exo- and endo-GH activities. Large sections of mucin are then imported into the cell for further breakdown. 2: Initially, there will be further exo- and endo-GH activities to produce fragments of O-glycans and mucin polypeptide with only the core glycan decoration remaining. 3: The O-glycan fragments will then be broken down to monosaccharides through the alternating action of sulfatases, fucosidases, β-galactosidases and β-HexNAcases. 4: The remaining glycopeptides are known targets for characterized A. muciniphila glycopeptidases, and Amuc_1008^GH31 will remove core GalNAc from the polypeptide. The stars indicate Mul1A association.

Extended Data Fig. 1. Growth of A. muciniphila ATCC BAA-835 on a variety of glycans and polysaccharides.

A. muciniphila was grown anaerobically in minimal media containing a different potential nutrient sources. Growth was monitored continuously at OD600 using a 96-well plate in a plate reader. Concentrations of substrates are listed in Supplementary Table 2. For the dialysed Scmannan, the black curves are original sample and the green is dialysed. For the GAG substrates, this was completed with varying concentrations of mucin:GAG. The black lines are the mucin growth curves and the darkest purple are the GAGs alone. The lighter the purple gets the more mucin in those growths. Each line is one bacterial culture. Source data

Extended Data Fig. 2. Characterisation of surface enzymes activity of AM using whole cell assays.

The glycan products from the different whole cell assay timepoints were labelled with procainamide at the reducing end and analysed by LC-FLD-ESI-MS. The results show a release of a variety of different O-glycan fragments, predominantly with galactose at the reducing end, which is indicative of GH16 endo-O-glycanase activity. The panel on the left emphasizes glycans 3 and 4 and the panel on the right emphasizes glycans eluting between 25-60 minutes. Source data

Extended Data Fig. 3. Characterisation of surface enzymes activity of AM using whole cell assays.

Examples of mass spectra for each glycan from the whole cell assay. The glycan products from the different whole cell assay timepoints were labelled with procainamide at the reducing end and analysed by LC-FLD-ESI-MS. The numbering corresponds to the numbering in the numbering in Fig. 1. Black and green are Y and B fragments, respectively. MS/MS data is shown for some glycans to provide examples of how structures were determined. The grey arrows clarify which peak the MS/MS data belongs to. Source data

Extended Data Fig. 4. Characterisation of surface enzymes activity of AM using whole cell assays continued.

Examples of mass spectra for each glycan from the whole cell assay. The glycan products from the different whole cell assay timepoints were labelled with procainamide at the reducing end and analysed by LC-FLD-ESI-MS. The numbering corresponds to the numbering in the numbering in Fig. 1. Black and green are Y and B fragments, respectively. MS/MS data is shown for some glycans to provide examples of how structures were determined. The grey arrows clarify which peak the MS/MS data belongs to. Source data

Extended Data Fig. 5. Sequential degradation of O-glycans on BSM with CAZymes from A. muciniphila BAA-835.

The sialidases, β-galactosidases, and CAZymes associated with α-linked monosaccharide hydrolysis were added sequentially to understand specificity of the CAZymes. a, Structural features that are expected in bovine submaxilliary mucin (left) and how it is broken down in this experiment. Only alpha linkages are labelled apart from the core GalNAc monosaccharides which are also alpha linkages. b, Thin layer chromatography results of the sequential degradation. Standards have also been included on the left of the TLCs. c, The chromatograms of the assays analysed by HPAEC-PAD. Maltose was used as the internal standard and other standards were run separately. d, The areas of the peaks were quantified and are presented on a scatter dot plot with the mean and SD. The data are from at least three distinct reactions. Source data

Extended Data Fig. 6. Activity of GH enzymes from A. muciniphila BAA-835 against raffinose-type defined oligosaccharides and a summary of activities against substrates with α-capping sugars.

a, melibiose. b, raffinose. c, stachyose. Standards have also been included on the TLCs on the right. Enzyme assays were carried out at a final substrate concentration of 1 mM, pH 7, 37 °C, overnight, and with 1 μM enzyme. d, Heat map of recombinant enzyme activities against defined oligosaccharides. The dark blue and white indicate full and no activity, respectively, and partial activities are represented by the lighter blues. Source data

Extended Data Fig. 7. Activity of the GH33 and GH181 family members from A. muciniphila BAA-835 against defined oligosaccharides, polysaccharides, and mucin substrates.

a, 3’-sialyllactose. b, 6’-sialyllactose. c, Sialylated Lewis A. d, Sialylated Lewis X. e, colominic acid. f, sialylfucosyllacto-N-tetraose. g, biantennary complex N-glycans (α₁acid glycoprotein). h, PGM III. i, GD1a j, and GT1b. k, Exploring if removing the galactose also allows the release of the final sialic acid. l, Testing Amuc_1216^GH177 and Amuc_0623^GHnc against three different sialylated substrates. m, Heat map of recombinant enzyme activities against defined oligosaccharides. The dark blue and white indicate full and no activity, respectively, and partial activities are represented by the lighter blues. Standards have also been included on the TLCs and substrates have been included on the right where possible. For the gangliosides (I and j) the structures of the different products are in the key to the right. Enzyme assays were carried out at a final substrate concentration of 1 mM (except N-glycans at 10 mg/ml and colominic acid at 25 mg/ml), pH 7, 37 °C, overnight, and with 1 μM enzymes. BT0455 from Bacteroides thetaiotaomicron has been used as a positive control a-i. Source data

Extended Data Fig. 8. Activity of the GH29 and GH95 family members from A. muciniphila BAA-835 against defined oligosaccharides, Part 2.

a, Exploring the full degradation of LNDFH I by pre-treating with Amuc_0846^GH29 and then adding the other fucosidases. b, Fucosidase activity against different types of blood group B structures where the α-linked galactose has been removed. c, Amuc_0010^GH29 activity against a range of blood group H structures. For all assays, standards have been included on the TLCs. Enzyme assays were carried out at a final substrate concentration of 1 mM, pH 7, 37 °C, overnight, and with 1 μM enzyme. d, A. heat map of recombinant enzyme activities against defined oligosaccharides. The dark blue and white indicate full and no activity, respectively, and partial activities are represented by the lighter blues. The asterisk indicates that one fucose has been remove by Amuc_0846^GH29. e, Structures of the defined substrates used to characterize the fucosidases. Source data

Extended Data Fig. 9. Activity of fucosidases and β-galactosidases from A. muciniphila BAA-835 against sulfated Lewis Structures.

a, Standards have also been included on the TLCs. Enzyme assays were carried out at a final substrate concentration of 1 mM, pH 7, 37 °C, overnight, and with 1 μM enzyme. b, Heat map of recombinant enzyme activities against defined oligosaccharides. The dark blue and white indicate full and no activity, respectively, and partial activities are represented by the lighter blues. Source data

Extended Data Fig. 10. Activity of the GAG-active enzymes and breakdown of sulfated host glycans by AM.

a, Activity of Amuc_0778^PL38 against known PL38 substrates glucuronan (left) and alginate (right). Enzyme assays were carried out at pH 7, 37 °C, overnight, and with 1 μM enzyme. b, Activities of the Amuc_0771^PL38 and Amuc_0863^GH105. c, Testing Amuc_0771^PL38 and Amuc_0863^GH105 against PGMIII. d, Left: substrate depletion of different GAGs (2 g/L) by Amuc_0771^PL38. This wavelength allows the monitoring of the generation of unsaturated products (double bonds) of the PL38. Middle: Highest initial rates of Amuc_0771^PL38. Right: Initial rates of the Amuc_0863^GH105. The change in absorbance monitors the GH105 activity against the Amuc_0771^PL38 product. e, Quantification of the Amuc_0771^PL38 activity over time from four different GAG substrates by LC-MS. The area under the peaks from the chromatograms are shown and each glycan has been given a different colour. For sulfated glycans, the position of the sulfation is unknown. CSA and DS typically has 4S on GalNAc, whereas CSC has relatively high levels of 6S on GalNAc, but also low levels of 4S. The GlcA in CSC can also have 2S sulfation. HA has no sulfation. Data are presented as mean values +/− standard deviation and experiments were completed in triplicate. Source data