Gastrointestinal MUC proteins contain characteristic tandem repeats of threonine, proline, and serine residues, where O-glycosidic linkages occur between the protein core and N-acetylgalactosamine (GalNAc) termini of oligosaccharides [46,47]. Neither the amino nor the carboxy termini of secretory mucins are generally glycosylated [48], but contain cysteine rich regions that promote intermolecular disulfide bonding (as shown in Figure 1). The dense glycosylation of the protein core and intermolecular bonding of the terminal regions effectively protects the mucin polymers from protease activity, preserving the protective structural matrix [49].

The predominant genes expressing membrane-bound mucins in human colonic goblet cells are MUC1, MUC3A/B, MUC4, and MUC12. Membrane-bound mucins could play a role in immunomodulatory effects of bacterial interactions with the epithelial membrane when the secretory mucin matrix is bypassed [50], however bacteria more frequently come in contact with secretory mucins considering the majority of bacteria only inhabit the outer portions of the mucus layers [51]. MUC2 is the principal secretory mucin gene expressed in the colon, comprising the majority of the mucous gel protecting the underlying tissue [52]. The role and mechanisms of mucin in innate immunity is reviewed more thoroughly by Dharmani et al. [53] and for a more detailed structural analysis of MUC2, see Allenet al. [15].

Oligosaccharide chains are affixed to MUC proteins by membrane-bound transferases in the Golgi apparatus and endoplasmic reticulum of goblet cells. GalNAc is affixed to the mucin protein from a sugar-nucleotide donor and a collection of specific glycosyltransferases continues to add residues, resulting in an oligosaccharide with a particular structure and terminus [54,55]. Glycosylation biosynthesis pathways are highly complex; glycosyltransferase gene expression levels, variability in spatiotemporal concentrations of enzymes, cofactors, and substrates, as well as the number of branching configurations possible all contribute to the wide range of potential protein-modifications [55]. This leads to glycoproteins forming from the same mucin gene product that will vary in glycan modification with location or tissue.

The oligosaccharide modifications can comprise up to 80% of the weight of a mucin and vary in length and structure. Secreted colonic mucins commonly contain side-chains of 4-15 monosaccharides with galactose and GalNAc backbones and branched chains terminating with GalNAc, fucose, or sialic (neuraminic) acid to varying degrees [56,57].

The predominance of acidic mucin subtypes, those with side-chains containing terminal ester sulfates and sialic acid groups, varies by location in the GI tract from species to species, as does the type of acidic modification most heavily expressed [58]. The presence of acidic side-chains can result in greater inhibition of bacterial growth in vitro [59] and reduced enzymatic degradation [60,61], but what causes the prevalence of these modifications in different parts of GI tissues is likely due to the tissue-dependence of specific collections of glycosyltransferase enzymes [62].

intestinal mucin polymers are considered nutritive glycans for commensal bacteria in the promotion of their residence and associated benefits [63,64]. Host glycosylation patterns in the gut may have coevolved with intestinal microbiota to accommodate the filling of niches beneficial to the host [65,66]. Host provision of mucin oligosaccharides specific to particular bacterial enzymes could provide a nutritional advantage to bacteria with those enzymes and differential expression of mucin oligosaccharides by tissue could hypothetically regulate host-microbe interactions to direct certain microbial populations to fill particular host-niches. So-called host “legislation” of glycosylation to promote particular microbial populations is evaluated in greater depth in a review by Patsos and Corfield [67]. Whether this plays a crucial role in Lactobacillus adhesion or is primarily a mechanism of promoting maintenance of other commensal microbiota is currently unclear.

One broad example of host legislation comes from the analysis of mucin oligosaccharide composition along the human intestinal tract, which showed that certain glycosylation patterns were conserved regionally despite inter-individual variation [68]. A gradient of sialylated mucin concentration was observed, decreasing from the ileum to the colon, running against an increasing gradient of more heavily fucosylated mucin.

A more specific example of microbial legislation by hosts lies in the presence of O-glycans on mucin that exhibit Lewis type or blood group ABO antigens. The secretor genes that determine host blood type also control the specificity of the ABO blood group type terminal glycosides of certain mucin oligosaccharide chains [69]. The glycosyltransferase responsible for blood group antigen precursors has been identified in secretory tissues producing mucins and glycoproteins [70]. There is evidence that populations of bacteria that produce specific blood type antigen-degrading glycosidases are present at levels 50,000 times greater in individuals with that particular blood type [71].

While evidence of mucin oligosaccharide degradation by bacteria is fairly well established [64,72,73,74,75], the dramatic impact of blood type on the composition of enteric microbial populations could imply that there is some degree of binding preference at play with host glycan legislation as well. This is supported by evidence of bacteria binding with human milk oligosaccharides, which can exhibit structural similarities with mucin oligosaccharides and blood type antigens [76].

Glycosidases of lactic acid bacteria have been fairly well characterized in terms of oligosaccharide breakdown and metabolism [77,78], but knowledge of glycoconjugate adhesion remains poorly described. Figure 2 displays a model of molecular binding mechanisms that may play a role in host-bacteria interactions. Elucidation of these binding mechanisms may be the key to understanding adhesion of lactobacilli in the gut.

For a more detailed characterization of mucin production, structure, and host function see the review by Lindén et al. [79]. Further information on the study of mucin glycomics, including identified O-linked glycan modifications characteristic of GI mucin and their biosynthetic pathways, can be obtained from glycobiology resources such as the Kyoto Encyclopedia of Genes and Genomes [80] and the Consortium for Functional Glycomics [81].

Several lactobacilli proteins have been shown to promote mucus adhesion (Table 2). The most studied example of mucus-targeting bacterial adhesins is the mucus-binding protein, MUB, produced by L. reuteri [82,94]. The MUB protein contains repeated functional domains, referred to by the authors as Mub domains, which are responsible for the protein’s adhesive properties. The Mub domain has since been designated a member of the MucBP domain family (Pfam PF06458). Numerous MUB homologues and MucBP domain containing proteins have been found, but almost exclusively in lactic acid bacteria and predominantly in lactobacilli found naturally in intestinal niches (Table 3). This suggests that MucBP domain containing proteins play an important role in establishing host-microbial interactions in the gut and promoted the evolution of the species as primarily GI organisms [93,95,96,97].

MUB and most MucBP domain containing proteins exhibit characteristics typical of Gram-positive cell surface proteins; a C-terminal sortase recognition motif (LPXTG) for anchoring the protein to peptidoglycan, repeated functional domains and an N-terminal region signaling the protein for secretion [94,95].

Roos and Jonsson’s competitive adhesion study showed that the binding of MUB to mucus was inhibited by the glycoproteins fetuin and asialofetuin as well as fucose, suggesting that MUB interacts with specific muco-oligosaccharides [94]. The study also demonstrated equivalent adhesion to mucus from different hosts indicating that MUB binding has little to no host specificity regarding mucus components. The recent resolution of the crystal structure of a MucBP domain in MUB, dubbed Mub2 [92], and subsequent discovery of immunoglobulin binding, provides further evidence of a broad binding specificity. This suggests that binding specificities of MucBP domain containing proteins are dictated by multiple factors, not solely resulting from the presence of MucBP domains.

Fimbrial genes have been reported in L. johnsonii NCC533 [99], but the direct visualization of pili on Lactobacillus cells has only been shown for L. rhamnosus GG [100]. Fimbriae, also referred to as pili, are thin proteinaceous extensions from bacterial cells, predominantly in Gram-negative bacteria, that promote adhesion. In many pathogens, pili are virulence factors that promote attachment to the host [101]. Kankainen et al. [100] isolated a pilin subunit, SpaC, located within the pili structure and found at the pilus tip, which was concluded to be essential to the interaction of L. rhamnosus GG with host mucus. A mutant strain lacking spaC expression showed significantly reduced binding. While these genes are uncommon among lactobacilli, this study has shown for the first time that fimbrial interaction with mucus can mediate host adhesion in lactobacilli.

SlpA, an S-layer protein in L. acidophilus,has been implicated in promoting adhesion directly to the GI surface, because slpA knockouts showed decreased adhesion capability [103]. However, this could possibly be due to disruption of other surface proteins. S-layer proteins and glycoproteins can form a latticed monolayer coating the surface of bacterial cells [114]. S-layer components can vary widely by species, but function to protect the cell from enzymatic damage, low pH, bacteriophages and phagocytosis. While S-layers are present in only some Lactobacillus species, they are beginning to be studied for their adhesive functions. A number of studies have begun associating S-layer proteins in probiotic bacteria with competitive exclusion of pathogens and pathogen adhesion to mucus [115,116,117].

Certain other surface proteins are implicated in contributing to adhesive properties of lactobacilli but are otherwise not well characterized or their importance to adhesive mechanisms is poorly defined. For instance, a 32 kDa protein associated with adhesion to porcine mucus in L. fermentum, named 32-Mmubp, was identified as a homologue of the substrate binding domains of the OpuAC ABC-transport protein family [102]. A mannose-specific adhesin protein (Msa, a MucBP domain containing protein) is responsible for the binding of mannose by L. plantarum [104]. While this was initially discovered as a protein responsible for agglutinating Saccharomyces cerevisiae,the presence of L. plantarum in many intestinal niches suggests that the MucBP domains of Msa may also play a role in adhesion to non-mannosylated muco-oligosaccharides as well. Elongation Factor Tu is a guanosine binding protein that is important in protein synthesis in the cytoplasm, but has been identified as a membrane associated protein as well [107,108]. More recently it has been found on the cell surfaces of many lactobacilli [109,110] and the demonstration of its upregulation in the presence of mucus suggests it may play a role in adhesion to the GI tract [111]. Mucus adhesion-promoting protein (MapA) is reported to mediate the binding of L. reuteri and L. fermentum to mucus [105,106]. Interestingly, it is also degraded into an antimicrobial peptide, which lends the host anti-pathogenic properties and provides an example of how large surface proteins may exhibit evolutionarily beneficial pleiotropic effects [118].

Numerous factors have been shown to influence binding of lactobacilli to mucus in vitro. Certain aspects of experimental design in particular should be reviewed when choosing or comparing methods to study adhesion in vitro because of the direct effects they have on adhesion. Time allotted for incubation of bacteria on immobilized mucus can have a significant influence on observed adhesion if microbial sedimentation occurs and the substrate is saturated at an artificial level [119].

Ramiah et al. [111] showed that growth conditions mimicking the GI environment have significant effects on the expression of several mucus adhesins in vitro in L. plantarum. MapA was upregulated 6-8-fold when incubated in the presence of mucin and up to 25-fold when exposed to physiological concentrations of pancreatin and bile compared to MRS grown controls. It was also found that mapA was significantly downregulated in the presence of cysteine, and suggested that cysteine is an effector molecule that represses transcription of mapA. Mub was expressed 80-140-fold more in the presence of mucin, but was suppressed 7-30-fold under normal gut physiological conditions containing bile and pancreatin. EF-Tu was expressed 33-100 times greater in media containing mucus, but was not affected by bile or pancreatin concentrations. This may connote interplay between different mechanisms regulating adhesin expression to adapt to particular environments.

The possible mechanisms whereby food components affect the adhesion of probiotic organisms in vivo have not been investigated thoroughly. Exposure to milk and milk fatty acids has been observed to reduce the adhesive properties of some probiotic lactobacilli [120,121] to human intestinal mucus in vitro, which may also be relevant in vivo. It is also hypothesized that entrapment in food matrices in vivo, resulting in binding to or steric hindrance of adhesins, can decrease adhesion of bacteria to intestinal surfaces [119].

All bacterial adhesion in the gut is also likely inhibited to some degree by competitive exclusion of access to binding sites by commensal organisms, but quantification of these effects have yet to be studied thoroughly.

The simplest method to measure the adhesion of bacterial strains to mucus is by immobilizing commercially available mucin. Mucin is bound to microtitre well plates, bacterial culture is bound to the mucin and strains are compared thereafter in any number of methods, qualitatively or quantitatively [91,132,133]. The use of mucin alone in adhesion assays allows for the targeting of interactions between bacteria and particular mucins known to be expressed in a given host locations. It also isolates microbe-mucus interactions from other interactions, such as host cell-microbe interactions, which may or may not be desirable. One drawback of this model is the complication of microbial hydrophobic properties with mucus-binding properties. The comparison of hydrophobic binding interactions of bacteria to untreated polycarbonate wells with mucus binding interactions in treated wells in one study [131] showed that hydrophobic binding interactions are not easily separable from mucus binding interactions.

Cultures of human intestinal cell lines are often presumed to better represent the environment in vivo because of the presence of actual tissue. The availability of a simple in vitro intestinal tissue model, as in the Caco-2 cell line, has provided valuable insight into cellular interaction mechanisms that would have been much more difficult to obtain with more complex in vitro techniques or in vivo. Caco-2 and HT29 cells, the two most commonly used intestinal cell lines, can be grown in culture to form a homogeneous polar monolayer of mature enterocytes resembling the tissue of the small intestine [134]. These models were developed primarily for the study of absorption and permeability in the small intestine and are derived from intestinal carcinomas, so they may or may not be accurate models for adhesion to healthy colonic tissue [135]. Several studies have compared the extent of Caco-2 cell binding by potential probiotic bacteria to adhesionin vivo with mixed results [136,137]. Regardless, these cell lines do not take into account the omnipresent mucus layer found in the healthy intestinal tract. The HT29 cell line, however, can be treated with methotrexate (MTX) to differentiate the cells into mucin-secreting goblet cells [138,139]. The production of mucus by HT29-MTX cells increase adhesion of bacterial cells relative to Caco-2 or HT29 cells alone [140,141], further supporting the importance of the presence of mucus to bacterial adhesion.

The HT29-MTX line is composed primarily of goblet cells, which incorporates a mucus layer into the model, but it still does not accurately represent the enterocyte/goblet cell ratio of the gut epithelial layer. In response to this drawback, Caco-2/HT29-MTX co-cultures have been developed [145,146,147]. Unfortunately, HT29-MTX differentiated goblet cells express MUC5AC and MUC5B at a much greater rate than MUC2 [139], which could be a significant drawback when studying microbial adhesion to the colon, where MUC2 is prevalent. HT29 cells also differentiate in the presence of 5-fluorouracil (FU) to secrete MUC2 [142], and while this would emulate the colonic environment more closely, the HT29-FU model only seems to have been used in the study of pathogens thus far [139,140].

The disadvantage of many models is that they don’t take into account the presence of normal GI microbiota and the competitive exclusion that would take place in vivo between established commensal populations and exogenous microbes. For a more complete model of intestinal tissue in vitro, encompassing the mucus layer and epithelial tissue accurately, but also accounting for the presence of commensal microbiota, whole intestinal tissue fragments can be used [148,149]. Resected fragments of healthy colonic tissue may be difficult to obtain, but likely display characteristics closer to those probiotic bacteria would be expected to encounter in vivo than other models. Similarly, organ culture can be employed to maintain the viability and architecture of the tissue and has been used to assess adhesive properties of pathogens [150,151]. As of yet, it does not seem that organ culture has been used in the study of probiotic organisms; the expense of using organ culture is more easily justifiable with pathogenic organisms that could not otherwise be used in human models safely, unlike probiotic organisms.