Mucus-binding proteins are bacterial surface adhesins with typical signal peptide and C-terminal LPxTG motif. They contain (multiple) Mub domains (around 200 residues in length) and/or MucBP domains (Pfam PF06458, around 50 residues), which have been shown to bind mucins (Juge, 2012). According to structural studies, the Mub domains consist of two modules B1 and B2 with B2 being a MucBP domain (MacKenzie et al., 2009; Etzold et al., 2014). Contrary to the MucBP domain also present in pathogens (Popowska et al., 2017), the Mub domain is predominantly found in LAB and is highly abundant in lactobacilli of the gut microbiota (Boekhorst et al., 2006). The best characterized MUB protein from Lactobacillus reuteri endowed with 14 Mub domains (Roos and Jonsson, 2002) binds mucus via multiple interactions involving terminal sialylated mucin glycans (Etzold and Juge, 2014; Gunning et al., 2016). In L. lactis plasmid free laboratory strains IL1403 and MG1363, only one MUB protein with four Mub domains and devoid of signal sequence, is encoded in the chromosome sequence (Boekhorst et al., 2006). However, these strains do not exhibit muco-adhesive properties (Radziwill-Bienkowska et al., 2017), suggesting that this MUB protein is not expressed or not functional. In contrast, three other MUB proteins found in dairy or vegetal strains were shown to contribute to bacterial adhesion to mucins as tested by different methods and tools (Table 1). These three proteins have different structures (Table 1), with three MucBP domains in MbpL protein from dairy L. lactis BGKP1 (Kojic et al., 2011), two Mub domains in Muc protein from vegetal L. lactis TIL448 (Meyrand et al., 2013) and one MucBP domain in AJ89_07570 protein from dairy L. lactis IBB477 (Radziwill-Bienkowska et al., 2016).

Pili are elongated protein structures of 1–10 nm in diameter and a few μm in length, protruding outside bacterial cells. They were long considered as characteristic of pathogens (Danne and Dramsi, 2012) but have been later discovered in LAB, in probiotic Lactobacillus rhamnosus GG, where they bind human mucins and could explain the strain persistence in human GIT (Kankainen et al., 2009). Pili were also described in L. lactis. Until now, three different gene clusters specifying sortase-dependent heterotrimeric pili have been identified in several L. lactis strains, one chromosomal present in all L. lactis strains (Oxaran et al., 2012) and two plasmid-encoded others in plant (Meyrand et al., 2013) and dairy (Tarazanova et al., 2016) strains respectively. These gene clusters encode three pilins, the major pilin which is polymerized by the sortase C to form the pilus backbone, a minor anchoring pilin and a tip pilin usually endowed with adhesive properties, as well as sortase C (Mandlik et al., 2008; Hendrickx et al., 2011). After synthesis, pili are covalently anchored to peptidoglycan by sortase A (Dieye et al., 2010). Noteworthy, the three identified L. lactis pili gene clusters exhibit limited sequence identity and have different syntheny. The chromosomal pilus gene cluster studied in the laboratory strain IL1403, as well as the plasmid one in L. lactis NCDO712 dairy strain, are not expressed in standard conditions but overexpression allowed pili synthesis (Oxaran et al., 2012; Tarazanova et al., 2016). In contrast, in the plant isolate L. lactis TIL448, the synthesis of pili at the bacterial surface was revealed without overexpression, by a proteomic analysis with the “shaving” approach and the presence of pili was visualized by electron microscopy and atomic force microscopy (AFM). These pili were further shown to contribute, together with the protein Muc, to mucin specific binding (Table 1). Interestingly, the tip pilin is a large protein with an LPxTG motif and containing a lectin domain (PF00139, Lectin_legB domain) that could recognized mucin glycan (Meyrand et al., 2013). Pili were also visualized by electron microcopy in four other L. lactis plant or clinical natural isolates but the genes involved in their synthesis were not identified (Oxaran et al., 2012).

A plasmid-encoded aggregation factor, AggL, was characterized in L. lactis BGKP1. It is a cell-wall anchored protein with a LPxTG motif and seven collagen-binding protein B domains (CnaB domain) and it is responsible for cell surface hydrophobicity and bacterial aggregation phenotype (Kojic et al., 2011). It was shown to contribute to mucus adhesion (Table 1), probably as a result of non-specific hydrophobic interactions with the hydrophobic mucosal surface (Lukic et al., 2012).

The cell-wall anchored proteinase, PrtP, has a crucial role in milk casein degradation required for efficient growth of L. lactis in milk (Siezen, 1999). The presence of PrtP at the surface of L. lactis was shown to modify the cell surface physico-chemical properties, leading to a greater hydrophobicity and increased adhesion to abiotic surfaces (Habimana et al., 2007). In L. lactis IBB477, the cell surface PrtP was shown to contribute to mucin adhesion (Radziwill-Bienkowska et al., 2017) (Table 1).

In vitro models that are most commonly used for evaluation of L. lactis adhesive properties, involve static microtiter plate assays using immobilized commercially available mucin (pig gastric mucin, PGM) and/or gut-related epithelial cell lines (Figure 1). Adherent bacterial cells are quantified using various approaches such as culturing/microscopical enumeration (Kimoto et al., 1999; Meyrand et al., 2013), radiolabeling (Rintahaka et al., 2014), crystal violet staining (Lukic et al., 2012; Radziwill-Bienkowska et al., 2016), or FISH (Radziwill-Bienkowska et al., 2017). The two most commonly used intestinal epithelial cell lines to study LAB adhesion are Caco-2 and HT29, which are originally derived from a human colorectal adenocarcinoma. Their major disadvantage is the lack of substantial mucus production. However, a mucus-secreting subpopulation of goblet cells from HT29 cell population (HT29-MTX) has been obtained after growth adaptation of HT29 cells to anticancer drug methotrexate (MTX) (Lesuffleur et al., 1990) and this cell line is increasingly used to investigate the muco-adhesive properties of L. lactis (Lukic et al., 2012; Radziwill-Bienkowska et al., 2017).

Biophysics-based tools have provided new in vitro insights into the interaction mechanisms between L. lactis and mucins (Figure 1). Interaction forces were quantified at nanoscale by AFM force spectroscopy using a L. lactis-functionalized tip (“lacto-probe”) and PGM-coated surface (Dague et al., 2010). Both non-specific and specific forces (ligand/receptor bonding) were at play in L. lactis adhesion to PGM. For L. lactis IBB477, the percentage of specific adhesive events was high (20%), in comparison with its low-adhesive counterpart L. lactis MG1820 (about 5%) (Le et al., 2011). Furthermore, mucin oligosaccharides were essential in interactions between L. lactis and PGM, as shown with blocking assays. For the first time with living cell probes and mucin, specific interactions were analyzed through kinetic constants: (i) the kinetic dissociation constant K_off was determined with increasing the tip loading rate, which led to a higher adhesion force. The K_off value (0.46 s^-1) was consistent with values corresponding to sugar/protein interactions (Le et al., 2011); (ii) the kinetic association constant K_on (3.3 × 10² M^-1 s^-1) was deduced from enhancing the tip/sample contact time. Furthermore, the high adhesion of IBB477 vs. MG1820 to PGM was confirmed at the multicellular level and under flow field conditions, using real-time quartz crystal microbalance with dissipation monitoring (QCM-D) (Le et al., 2012) and shear stress flow chamber (Radziwill-Bienkowska et al., 2016). Later studies revealed the combined contribution of chromosomal and plasmid-encoded cell-wall proteins (MucBP-Big_3 containing protein and proteinase PrtP respectively) in the muco-adhesive properties of IBB477 (Radziwill-Bienkowska et al., 2016, 2017).

A further work on the plant isolate L. lactis TIL448 enabled to elucidate the L. lactis muco-adhesive phenotype, based on the respective contribution of pili and MUB protein (Le et al., 2013). With AFM force spectroscopy, a high proportion of specific adhesive events to PGM was detected (60%), consistent with the weak shear-flow induced detachment of bacterial cells from the PGM coating. Rupture events were observed in AFM at short (100–200 nm) and long distances (up to 600–800 nm). With AFM force spectroscopy on pili and MUB protein defective mutants, the equivalent role exerted by these two cell surface determinants was established. However, under shear flow, a more critical contribution of MUB protein than pili was found. The importance of pili was further dissected in relation with their nanomechanical properties as probed with optical tweezer (Castelain et al., 2016a,b). AFM blocking assays also revealed that mucin neutral oligosaccharides were involved in adhesion of L. lactis TIL448 to PGM (Le et al., 2013).

A relatively low number of in vivo/ex vivo studies (i.e., biopsy samples) have been conducted to confirm the muco-adhesive phenotype of L. lactis in “real” environmental conditions (Figure 1). Application of such methods seems to be the next step, as shown for lactobacilli (Da Silva et al., 2015; Walsham et al., 2016). Nevertheless, first experimental evidence has been provided for lactococci. In mice, the GFP-labeled L. lactis WH-C1 strain was found to adhere to the gut mucosa (Wang et al., 2011). In another study in rats, in ex vivo and in vivo experiments, AggL protein was found to confer adhesive properties to L. lactis BGKP1 to colonic tissue through non-specific hydrophobic interactions. In contrast, for this particular strain, the MbpL protein did not contribute to bacterial adhesion to colonic tissues but was rather involved in gastric mucin binding (Lukic et al., 2012). More recently, it was shown that the cell-wall proteinase PrtP, albeit contributing to in vitro muco-adhesion of L. lactis IBB477, probably through non-specific interactions, could not confer a selective advantage to this strain in the gut of conventional C57BL/6 mice (Radziwill-Bienkowska et al., 2017).