Sialidases from gut bacteria: a mini-review

1In the gastrointestinal (GI) tract, sialic acid [N-acetylneuraminic acid (Neu5Ac)] is commonly found in terminal location of mucins. Mucins are large glycoproteins, which can be broadly grouped as membrane-bound or secreted. Membrane-bound mucins are essential contributors of the glycocalyx of mucosal surfaces where they play important biological roles in cell interactions and signalling. Secreted mucins are the main structural components of the mucus gel covering the epithelium and essential to the maintenance of a homoeostatic relationship with our gut microbiota. Mucins are characterized by a proline–threonine–serine (PTS) domain which is the site of extensive O-glycosylation with carbohydrates accounting for up to 80% of the total mucin mass. The synthesis of mucin oligosaccharides starts with the transfer of N-acetyl-galactosamine (GalNAc) to serine and threonine residues of the mucin backbone to form mucin O-glycan core structures. These core structures can be further elongated with galactose (Gal), N-acetyl-glucosamine (GlcNAc), GalNAc and frequently modified by terminal fucose or sialic acid residues via α1-2/3/4 and α2-3/6 linkages, respectively (Figure 1). The proportion of the major mucin glycan epitopes, sialic acid and fucose, varies along the GI tract with a decreasing gradient of fucose and an increasing gradient of sialic acid from the ileum to the rectum in humans and a reverse gradient in mice. The Neu5Ac α2-6 N-acetylgalactosaminitol epitopes and Sda/Cad antigens found in humans are absent or rare in mice where as the Neu5Ac–GlcNAc epitope and disialylated epitopes are more common along the murine GI tract.

2The GI tract is heavily colonized with bacteria. Most species belong to the phyla Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria and Verrucomicrobia. The microbiota composition varies longitudinally along the GI tract but also transversally from the mucosa to the lumen. The terminal mucin O-glycans have been proposed to serve as metabolic substrates, providing a nutritional advantage to bacteria which have adapted to the GI mucosal environment. The release of sialic acid from non-reducing ends by sialidases is an initial step in the sequential degradation of mucins, since the terminal location of sialic acid residues in the mucin oligosaccharide chains may prevent the action of other glycoside hydrolases (GHs). In bacteria, the genes involved in sialic acid metabolism are usually found clustered together forming what is denominated as a Nan cluster. The canonical nanA/K/E cluster was first described in Escherichia coli and an alternative pathway defined by the nanLET cluster was later discovered in Bacteroides fragilis. The majority of the bacteria that harbour a Nan cluster colonize mucus regions of the human body, such as the gut where sialic acid is highly abundant and can serve as a source of energy, carbon and nitrogen. As described below, a number of gut bacteria employ sialidases in the release of host sialic acids, including multiple species of Clostridia, Bacteroides, certain subspecies/serovars/strains of Bifidobacterium longum, Vibrio cholerae, Ruminococcus gnavus and Akkermansia muciniphila. However, some bacteria appear to have incomplete packages of enzymes for utilizing host sialic acids. For example, Bacteroides thetaiotaomicron VPI-5482 encodes a sialidase and can release free sialic acid, but lacks the Nan operon required to consume the liberated monosaccharide and does not appear capable of consuming free sialic acid. On the other hand Clostridium difficile strain 630 encodes the Nan operon but lacks the sialidase and thus relies on other sialidase-producing organisms to acquire this potential nutrient source from the mucosal environment. In contrast, some bacteria appear to possess the complete pathway of sialic acid catabolism including a predicted sialidase gene e.g. B. fragilis strains. A recent study reported that mice monoassociated with B. thetaiotaomicron exhibited a significantly higher concentration of free Neu5Ac compared with germ-free mice, consistent with the ability of B. thetaiotaomicron to liberate but not consume the monosaccharide, whereas colonization of mice with B. fragilis, which is able to catabolize Neu5Ac, did not result in increased free sialic acid.

3In contrast with gut commensals, which appear to use sialidases primarily for nutrient acquisition, some pathogens of the GI tract such as V. cholerae or Clostridium perfringens strains also use sialidases to decrypt adhesin or toxin-binding sites. All toxigenic strains of V. cholerae have a sialidase encoded within a pathogenicity island in their genomes. However it is also worth noting that, within particular pathogenic or commensal species, the presence or absence of sialidase-encoding genes in bacterial genomes is often strain-specific. For example R. gnavus ATCC 29149 but not E1 expresses a sialidase, several strains of E. coli, e.g. enteropathogenic E. coli O127 strain (EPEC) or probiotic strain Nissle 1917 possess a sialidase-encoding gene whereas commensal E. coli strains such as E. coli strain EHV2 lack a sialidase. However E. coli sialidases remain to be biochemically characterized. Similarly, not all Salmonella enterica strains encode a putative sialidase and only one sialidase has been functionally-characterized from S. enterica serovar typhimurium although it appears to have been acquired by horizontal transfer.

4Sialic acid catabolism in the gut is important as increased free sialic acid levels in the intestinal mucosal compartment, e.g. post-antibiotic treatment, will favour outgrowth of some bacterial pathogenic strains of S. Typhimurium and C. difficile or the outgrowth of E. coli during inflammation. Such cross-feeding activity has also been reported between commensal bacteria, e.g. Bifidobacterium breve UCC2003 (containing a functional Nan cluster for sialic utilization) can utilize sialic acid released by the sialidase activity of Bifidobacterium bifidum PRL201048. The gut symbiont, R. gnavus ATCC 29149, is different from the above as it possesses the complete Nan cluster and an intramolecular trans-sialidase (IT-sialidase), thus producing (2,7-anhydro-Neu5Ac) 2,7-anydro-N-acetylneuraminic acid instead of free Neu5Ac, suggesting a novel mechanism of adaptation to the mucosal environment. The biological role of bacterial sialidases produced by human gut commensal and pathogenic bacteria has been reviewed previously. Here we focus on the structural and biochemical properties of characterized sialidases involved in the interaction between gut bacteria and epithelial surfaces.

5Sialidases (also commonly referred to as neuraminidases) are a large group of enzymes, the majority of which are exo-sialidases catalysing the cleavage of terminal sialic acids from complex carbohydrates on glycoproteins or glycolipids. Based on amino acid sequence similarities, bacterial exo-sialidases are classified in the GH family 33 (GH33) of the CAZy classification (www.cazy.org). Hydrolysis occurs via an acid/base-catalysed double-displacement mechanism involving a covalent sialyl–enzyme intermediate, resulting in overall retention of configuration at the anomeric centre (Figure 2). Unusually the catalytic nucleophile is a tyrosine residue activated by a proximal glutamic acid, due to the charge on sialic acid itself as shown biochemically and confirmed structurally (Figure 2). Sialidases from the retaining sialidase families GH34 and GH83 are both restricted to viruses and examples of GH58 inverting endo-sialidases are found in some E. coli strains (www.cazy.org).

6Based on their substrate specificity and catalytic mechanism, exo-sialidases can be separated into three classes: hydrolytic, trans-sialidases and IT-sialidase. Hydrolytic-sialidases cleave the glycosidic bond of terminal sialic acids and release free sialic acid, whereas trans-sialidases transfer the cleaved sialic acid to other glycoconjugates; according to the Enzyme Commission both classes belong to exo-α-sialidases (EC 3.2.1.18). Hydrolytic-sialidases usually have wide substrate specificity and cleave α2-3-, α2-6- and α2-8-linked terminal sialic acids. Trans-sialidase activity with specificity for α2-3-linked substrates was first discovered for the Trypanosoma cruzi sialidase TcTS. Trans-sialidases with activity against α2-6- and α2-8-linked sialic acid substrates have been discovered in the intervening years. The third class is the IT-sialidase (EC 4.2.2.15). Currently, the discovered and characterized IT-sialidases are strictly α2-3-linkage specific and produce 2,7-anhydro-Neu5Ac. However, the substrate and linkage specificity of sialidases is often unknown due to reliance on artificial substrates such as 4-methylumbelliferyl-Neu5Ac (4MU-Neu5Ac) or 2-O-(p-Nitrophenyl)-α-D-Neu5Ac (PNP-Neu5Ac; Table 1).

7The GH33 catalytic domains adopt a six-bladed β-propeller fold (Figures 3A and 3B). GH33 catalytic domains are often associated with additional domains including membrane-binding domains and carbohydrate-binding modules (CBMs) such as sialic acid-specific CBM40 and broadly specific CBM32 as classified in CAZy (www.cazy.org). CBMs are believed to mediate adherence of the enzyme to cognate carbohydrate substrates and enhance the hydrolase activity of the catalytic domains by increasing enzyme substrate proximity. Both exo- and trans-sialidases share a set of active site residues and cleave the terminal α-linked sialic acid residue by the same catalytic mechanism. This conserved active site includes a glutamic acid–tyrosine charge relay with the tyrosine acting as the catalytic nucleophile and an aspartate residue as the general acid/base (Figure 2). The incoming sialic acid residue is orientated in the active site via a trio of arginines, which co-ordinates the sialic acid carboxylate moiety and a hydrophobic pocket which accommodates the ligand N-acetyl group (Figure 3C). Aspartic acid-boxes are motifs commonly found at the termini of sialidase β-propeller blades, they may stabilize the protein fold by providing inter-blade contacts.

8Among GI commensals, Bacteroidetes species are found at high abundance and many of them express sialidases in culture. B. thetaiotaomicron spp. encode the sialidases required to cleave and release terminal sialic acid from the mucosal glycoconjugates, but do not encode the Nan cluster required to consume the liberated monosaccharide. The purified sialidase from B. thetaiotaomicron VPI-5482 has been shown to hydrolyse sialylglycoconjugates including fetuin and transferrin (Table 1). Presumably, the release of sialic acids allows B. thetaiotaomicron to access highly coveted underlying carbohydrates in the mucus. Recently, a sialidase from Bacteroides vulgatus BVU 4143 has been shown to be active against 4MU-Neu5Ac and PNP-Neu5Ac and inhibited by N-acetyl-2,3-didehydro-2-deoxyneuraminic acid (Neu5Ac2en) inhibitor (Table 1). B. fragilis strains are among those bacteria that have been shown to possess the complete pathway of sialic acid catabolism including the sialidase. Several sialidases from different B. fragilis strains have been characterized, showing a broad specificity with some preference for the α2-8 linkage (Table 1). Sialidases have also been identified in the genomes of infant-derived Bifidobacteria, including two intracellular sialidases from B. longum subsp. infantis ATCC 15697, two predicted extracellular exo-α-sialidases of B. bifidum PRL 2010 and a putative sialidase from B. breve UCC2003. However, the only sialidase from this group of infant-associated bacteria to be functionally characterized is SiaBb2 from B. bifidum JCM 1254, a strain for which the genome sequence is not yet publicly available. SiaBb2 has a strong preference for α2-6 linkages and was shown to be sufficient to confer B. longum 105-A with the ability to degrade human milk oligosaccharides (HMOs). This sialidase can also transfer Neu5Ac to 1-alkanols at high acceptor concentrations (Table 1). All four putative sialidases annotated in the genome of the mucin-degrading bacteria A. muciniphila ATCC BAA-835 have recently been characterized. The enzymes are active against a range of sialylated substrates with either α2-3 or α2-6 linkages (Table 1).

9Among gut pathogens, NanH/STSA from S. typhimurium TA262/LT2 strain has been biochemically (Table 1) and structurally characterized (Figure 3A), revealing conservation of key catalytic residues with the GH34 viral sialidases, including the nucleophilic charge relay, the aspartic acid acid/base and the arginine triad. This enzyme shows kinetic preference for sialyl α2-3 linkages over sialyl α2-6 linkages and preferentially cleaves Neu5Ac residues rather than N-glycolylneuraminic acid (Neu5Gc) residues (Table 1). Some strains of C. perfringens encode multiple sialidases (Table 1). The evolutionary rationale for this is unclear but may be because the enzymes differ in their cellular location, properties and sensitivities to inhibitors. NanI from C. perfringens is unusual in that it is a hydrolytic enzyme which can also hydrate the inhibitor Neu5Ac2en to Neu5Ac in vitro. These enzymes differ from V. cholerae sialidases which can hydrolyse both α-2,3- and α-2,6-linked sialic acid substrates and produce the Neu5Ac2en inhibitor (Table 1). The active site has many features in common with other viral and bacterial sialidases but, uniquely, has an essential Ca2+ ion which plays a crucial structural role.

10Most trans-sialidases have been characterized from trypanosome species. Trans-sialidases have not been reported in the gut microbiota. However a few examples of exo-sialidases from gut bacteria have been reported to perform trans-glycosylation reactions under certain experimental conditions. These include the aforementioned SiaBb2 from B. bifidum and NanI from C. perfringens (Table 1).

11IT-sialidases are unique in that they catalyse an intramolecular reaction in which the O7-hydroxy group of the bound sialic acid glycerol group attacks the positively charged C2 atom of the oxocarbenium intermediate. The altered reaction pathway leads to release of 2,7-anhydro-Neu5Ac instead of Neu5Ac (Figure 2). The first example of this enzyme class was described in NanL, which is purported to be from the leech Macrobdella decora, but may be from a bacterial source in the leech GI tract, as previously suggested. Three IT-sialidases have been biochemically and structurally described: NanL, NanB from Streptococcus pneumonia and RgNanH from R. gnavus (Figure 3B). The active site of IT-sialidases is characterized by a conserved threonine residue which sterically hinders the substrate glycerol group, forcing it into an axial position whence it can attack at the anomeric C2 carbon and form the intramolecular linkage (Figure 3C). An additional characteristic feature of these enzymes is a hydrophobic rim close to the arginine triad, formed by tryptophan–tyrosine stack. This feature provides strict specificity for α2-3-linked substrates and may also provide an important contribution to the reaction mechanism by providing a desolvated, hydrophobic environment. This allows the intramolecular reaction to proceed, as the O7 hydroxy of the glycerol group must outrun any incoming water molecules that would otherwise attack the C2 carbon and produce Neu5Ac (Figure 3C).

12RgNanH from R. gnavus ATCC 29149 is the first example of an IT-sialidase functionally characterized in gut bacteria. The enzyme produces 2,7-anhydro-Neu5Ac with strict specificity towards α2-3 glycosidic substrate linkages. RgNanH is a three-domain modular protein with an N-terminal lectin-like domain (L-domain) classified as a CBM40, a GH33 catalytic domain (N-domain) and a domain inserted into the catalytic domain (I-domain). The crystal structure of the RgNanH catalytic domain has been solved and demonstrates the six-bladed β-propeller fold characteristic of sialidases. A domain of unknown function protrudes from between two blades of the β-propeller (I-domain). Crystal structures in complex with 2,7-anhydro-Neu5Ac and known inhibitors of hydrolytic sialidases, allowed interrogation of the active site. Of particular importance is the conservation of the active site threonine (Thr557), which is proposed to sterically force the substrate glycerol group into a position from where it can attack the C2 atom (Figure 3C). This residue also impacts on the response to sialidase inhibitors, as shown by poor inhibition by Neu5Ac2en and micromolar inhibition by siastatin B.

13Bioinformatics analyses revealed that the presence of IT-sialidases is shared by other members of the gut microbiota, in particular Blautia hansenii, Ruminococcus torques, all 10 strains of Clostridium perfringens with available genome data, C. sp. 7 2 43 FAA, C. celatum, C. nexile, C. spiroforme, three unclassified Lachnospiraceae, more than 100 strains of Streptococcus agalactiae and three of the genome-sequenced publicly available Lactobacillus salivarius strains. The detection of IT-sialidase homologues in at least 11% of gut metagenomes of a population of diseased and healthy humans confirmed that this enzyme is widespread across gut bacteria, especially in Firmicutes. This analysis also revealed a greater abundance of IT-sialidase encoding species in patients with inflammatory bowel diseases (IBD) as compared with healthy individuals. The specific niche colonization of these bacteria may reflect their adaptation to particular mucin glycosylation profiles associated with intestinal inflammation and/or infection.

14Bacterial sialidases and their sialoglycan targets contribute to host–microbe interactions at the mucosal surface. An imbalance in the proportion of gut commensals able to modulate mucosal sialic acid levels or a change in host mucin sialylation is often associated with enteric infection or intestinal inflammation. Maintaining a balance in the ability of gut commensals to produce and/or consume sialic acid in the mucosal compartment is therefore essential to gut homoeostasis.

15Further investigations of bacterial sialidases should clarify the type of sialylated structures that are accessible to the gut bacteria and the specificity of sialidases towards sialic acids with different modifications and in different linkages. These include gaining structural insights into the diversity of sialic acid derivatives that can be produced and/or taken up by commensal and pathogenic bacteria. Thus, for therapeutic purposes, modulation of sialidase expression might be effectively achieved by appropriate use of specific inhibitors or pro/prebiotic approaches targeting specific bacterial strains.