Prominent members of the human gut microbiota express endo-acting O-glycanases to initiate mucin breakdown

Abstract

The thick mucus layer of the gut provides a barrier to infiltration of the underlying epithelia by both the normal microbiota and enteric pathogens. Some members of the microbiota utilise mucin glycoproteins as a nutrient source, but a detailed understanding of the mechanisms used to breakdown these complex macromolecules is lacking. Here we describe the discovery and characterisation of endo-acting enzymes from prominent mucin-degrading bacteria that target the polyLacNAc structures within oligosaccharide side chains of both animal and human mucins. These O-glycanases are part of the large and diverse glycoside hydrolase 16 (GH16) family and are often lipoproteins, indicating that they are surface located and thus likely involved in the initial step in mucin breakdown. These data provide a significant advance in our knowledge of the mechanism of mucin breakdown by the normal microbiota. Furthermore, we also demonstrate the potential use of these enzymes as tools to explore changes in O-glycan structure in a number of intestinal disease states.

Fig. 1. Mucin structure and genomic context of the loci encoding the mucin-associated GH16 enzymes.

a Left: the main structural features of a model mucin O-glycan chain. All mucin oligosaccharides are linked via an α-GalNAc to serine and threonine residues in the peptide backbone. A number of different core structures are then attached, with core 3 (shown) being the most common in the large intestinal MUC2. The cores are then often extended with polyLacNAc repeats of varying lengths which are decorated along their length by sulfation and fucosylation and capped at the non-reducing end by a variety of α-linked monosaccharides. Right: a model of an intestinal mucin glycoprotein showing complexity and variability of glycan chains attached to peptide backbone. b Genetic context of the GH16 encoding genes identified as being upregulated in the four species shown during growth on mucin (see Supplementary Figs. 1–3). In Bacteroides spp. the GH16 genes (highlighted red) are part of discrete polysaccharide utilisation loci (PULs), cluster of co-regulated genes encoding glycan degradation and uptake apparatus (SusC-like and SusD-like outer membrane proteins, additional CAZymes and putative surface glycan binding proteins (pSGBPs), often adjacent to a hybrid two component system (HTCS) sensor-regulators that likely control expression of the associated PUL. Glycan utilisation genes are not organised into PULs in the A. muciniphila genome.

Fig. 2. Domain structure of the mucin-associated GH16 enzymes characterised in this study.

GH16 catalytic domain (green), Type I signal peptide (light blue), Type II signal peptide (orange). BT2824 also has an N-terminal DUF4971 domain (PF16341) of unknown function. Sequence identity is between 22 and 38% for most of the enzymes, but two pairs of the GH16 enzymes are close homologues - BF4058 and BACCAC_02679 (red asterisk) display 87% identity, while BF4060 and BACCAC_02680 (black asterisk) display 79% identity.

Fig. 3. Activity of the GH16 O-glycanases against porcine small intestinal mucin.

SI mucin was incubated with the GH16 enzymes alongside a broad-acting sialidase (BT0455^GH33). The control is sialidase-only. Products of small intestinal mucin digestion were labelled with procainamide at the reducing end and analysed by LC-FLD-ESI-MS. The results show a variety of oligosaccharides are released by the GH16 enzymes and majority are similar between the samples. The oligosaccharides have a variety of fucose and sulphate decorations. Species capped with α-GalNAc were determined using exo-acting enzymes specific to that sugar and linkage (see Supplementary Fig. 8).

Fig. 4. Heat map showing the activity of the GH16 O-glycanases against different oligosaccharides.

The data summarises the specificity of the GH16 O-glycanases described in this report. From left to right the glycans are TetraLacNAc, TriLacNAc, paraLacto-N-neohexaose, Lacto-N-neotetraose, Lacto-N-tetraose, Lacto-N-triose, Galβ1,3GalNAc β1,3Galβ1,4Glc tetrasaccharide Blood group A hexasaccharide, Blood group B hexasaccharide, Blood group H pentasaccharide, LacNAc, Blood group H tetrasaccharide II, 3-sialyllactose and P1 antigen. The linkages are β unless otherwise labelled and the bonds cleaved are indicated by the black arrows. Partial and trace activity are the estimation of greater than or less than 50% degradation, respectively, under the assay conditions used. A more detailed summary can be found in Supplementary Table 4. The predicted cellular locations of each enzyme is indicated on the far right of each row.

Fig. 5. Surface activity of mucin-grown A. muciniphila against TriLacNAc.

aA. muciniphila cells were grown on PGM III, harvested and TriLacNAc added to assess surface enzyme activity. Samples were analysed at different time points following addition of TriLacNAc. Equivalent data for the Bacteroides species are shown in Supplementary Fig. 13. b Samples from two different time points of the A. muciniphila whole cell assay were incubated with two different exo-acting enzymes of known specificity to identify the products. The asterisks indicate the time points analysed. The top and bottom of the glycan structures shown is the non-reducing and reducing end, respectively. A: original sample, B: +β1,4-galactosidase BT0461^GH2, C: +broad-acting β-GlcNAc’ase BT0459^GH20. For example, in both time points, the two different types of trisaccharides can be seen to be hydrolysed in the separate digests: (1) Galβ1,4GlcNAcβ1,3Gal disappears in ‘B’ and products Gal and GlcNAcβ1,3Gal are now present, (2) GlcNAcβ1,3Galβ1,4GlcNAc disappears in ‘C’ and products GlcNAc and Galβ1,4GlcNAc are now present. These experiments were carried out once, but multiple pilot experiments were run and were consistent with the data shown. The source data are provided in the source data file.

Fig. 6. Structures of four of the O-glycan active GH16 family members characterised in this study.

a Crystal structures of BACCAC_02680^E143Q, BF4060, BACCAC_03717, and Amuc_0724. The loops extending from the active site that are proposed to be involved in substrate specificity in GH16 enzymes are termed ‘fingers’ and are colour coded. b Subsites −1 to −3 of BACCAC_02680^E143Q and BF4060 have the product of TriLacNAc cleavage bound (Galβ1,4GlcNAcβ1,3 Gal; shown in symbol form next to the structures with the sugar in each subsite labelled). The residues interacting directly with sugar are shown as sticks. The aromatic residues shared with β-glucanase GH16 family members that drive specificity for a β1,3 between the −1 and −2 sugars are shown (W129, W138 and W131, W140; See Supplementary Fig. 20 for active sites of BACCAC_03717, and Amuc_0724). c A surface representation of the regions surrounding the −1 subsite showing the selection for the axial O4 of Gal in the three Bacteroides enzymes, while Amuc_0724 has a more open ‘tunnel’ like space that appears to also allow accommodation of the equatorial O4 of Glc. The product from BACCAC_02680^E143Q was overlaid in the BACCAC_03717, and Amuc_0724 structures. Colours represent the different ‘fingers’. d A view of the predicted +1 subsite of BF4060 and BACCAC_03717 overlaid with the glucose from the +1 subsite of a laminarinase from Phaenerochaete chrysosporium. The +1 subsites are much more closed for BF4060 and BACCAC_02680^E143Q compared to BACCAC_03717, and Amuc_0724. e, An overview of the monosaccharides occupying the different subsites in GH16 family members with different activities. Linkages also shown. It should be noted that the sulfation will be variable along the O-glycan chain and there will also be fucose decorations. This situation is similar to porphyran, where the polysaccharide can have a variable composition, but the subsite occupancy shown here reflects the observations of structures of enzyme-glycan complexes currently available for porphyranases.

Fig. 7. Examples of the O-glycan profiles that can be produced from a variety different human mucins by the Amuc_0724 O-glycanase.

Mucins from different samples were pre-treated with the broad acting sialidase BT0455^GH33 and then digested with the GH16 and products analysed by LC-FLD-ESI-MS. a Inflamed colonic tissue removed during a laproascopic panproctocolectomy from a patient with UC. b Bowel tissue from neonates with necrotising enterocolitis. c Colorectal cancer cell lines. Small amounts of Neu5Gc are seen in some of the UC samples (e.g., panel a) suggesting either the presence of contaminating dietary animal O-glycans remaining in the mucus layer or that this xenobiotic sugar has been incorporated into human mucins from dietary sources. The O-glycan profiles of all the different samples analysed are shown in Supplementary Fig. 21.

Fig. 8. Model for the role of GH16 O-glycanases in mucin breakdown.

A model of the initial steps of mucin degradation on the surface of Bacteroides spp. and A. muciniphila. Sialic acid is removed by surface-localised sialidases and the GH16 enzymes remove oligosaccharides for import into the periplasm for further degradation by other CAZymes, including periplasmic GH16 O-glycanases. In Bacteroides species, glycan import is via energy dependent SusCD-like complexes, but in A. muciniphila the mechanism of glycan import across the outer membrane is unknown. The remaining mucin glycoprotein is likely further degraded by other extracellular CAZymes and glycopeptidases^,,.