Measuring Kinetic Dissociation/Association Constants Between Lactococcus lactis Bacteria and Mucins Using Living Cell Probes

Abstract

In this work we focused on quantifying adhesion between Lactococcus lactis, the model for lactic acid bacteria (LAB) and mucins. Interactions between two strains of L. lactis (IBB477 and MG1820 as control) and pig gastric mucin–based coating were measured and compared with the use of atomic force microscopy. Analysis of retraction force-distance curves shed light on the differential contributions of nonspecific and specific forces. An increased proportion of specific adhesive events was obtained for IBB477 (20% vs. 5% for the control). Blocking assays with free pig gastric mucin and its O-glycan moiety showed that oligosaccharides play a major (but not exclusive) role in L. lactis-mucins interactions. Specific interactions were analyzed in terms of kinetic constants. An increase in the loading rate of atomic force microscope tip led to a higher force between interacting biological entities, which was directly linked to the kinetic dissociation constant (K_off). Enhancing the contact time between the tip and the sample allowed an increase in the interaction probability, which can be related to the kinetic association constant (K_on). Variations in the loading rate and contact time enabled us to determine K_on (3.3 × 10² M⁻¹·s⁻¹) and K_off (0.46 s⁻¹), and the latter was consistent with values given in the literature for sugar-protein interactions.

Introduction

Understanding the adhesion processes that occur between a microorganism and a surface is definitely a sticky problem (1). Microbial adhesion occurs in both industrial and health domains. On the one hand, it causes harmful side effects, including food spoilage, spread of foodborne diseases, biofouling of materials, and hospital-acquired infections. On the other hand, in parallel with increasing efforts to eradicate uncontrolled biocontamination, it may be useful to promote the adhesion of friendly microorganisms such as probiotic lactic acid bacteria (LAB). Probiotics are living, nonpathogenic microorganisms that when administered in adequate amounts confer a health benefit to the host (2–4). To display such properties in vivo, probiotics need to survive at sufficiently high levels and colonize the gastrointestinal tract. A prerequisite for colonization is adhesion to epithelial mucosa, which is thus generally considered as one of the major criteria for probiotic selection.

Another useful target involving bacterial adhesion is the development of live vaccines. Mucosal routes for vaccine delivery offer several advantages over systemic inoculation, such as reduction of secondary effects, easy administration, and the possibility to modulate both systemic and mucosal immune responses (5). Moreover, it is important for medical molecules that exert their effects at mucosal surfaces to be directly delivered to the appropriate site. Unfortunately, the immunogenicity of soluble proteins is low when administered at the mucosal level. When used as a carrier system, living bacteria can substantially help to overcome such drawbacks, provided that they are able to colonize and multiply in the host without causing disease. The use of living bacterial vectors derived from pathogenic microorganisms such as Mycobacterium, Salmonella, and Bordetella spp. is not totally safe for older people, children, and immunosuppressed patients (6). Therefore, numerous LAB species that meet the Generally Recognized as Safe criteria of the Food and Drug Administration may be suitable for active delivery of therapeutic proteins by genetically modified microorganisms (7).

As part of ongoing efforts to obtain quantitative information about LAB-mucosa interactions in the gastrointestinal tract, we sought to characterize at the nanoscale the adhesion of Lactococcus lactis to intestinal mucus, with a special focus on mucins. L. lactis is widely used as a starter in manufacturing cheese and other fermented dairy products. It is not usually considered a normal element of the human intestinal microbiota, even though it has been sporadically isolated from the feces of many different groups of humans (8). Indeed, an in silico analysis by Boekhorst et al. (9) highlighted the presence of mucus-binding (MUB) domain–containing proteins in lactococci. Furthermore, the biodiversity of a large panel of natural L. lactis strains in terms of surface physicochemical properties (10) and potentially mucoadhesive behavior may provide an important pool of functionalities for food- and health-related applications.

Mucins are heavily O-glycosylated proteins that are found in the mucus layer at the surface of many epithelia (11). They are responsible for the physical properties of mucus gels and are involved in numerous interactions between cells and their environment. Alterations in mucin expression or glycosylation are observed during the development of cancers and influence cellular growth, differentiation, transformation, adhesion, invasion, and immune surveillance (12). Mucins (e.g., MUC1, MUC4, and MUC16) have also been identified as markers of adverse prognosis and thus constitute attractive therapeutic targets (13). In this framework, there is a great interest in analyzing bacterial mucoadhesive properties to aid in the selection of probiotic candidates (14) and improve drug-delivery strategies (15).

The surface determinants that mediate adhesion of LAB to mucus are MUB proteins (9), S-layer (16–18), and other specific proteins (16,19,20). One of the subunits of a proteinaceous surface-exposed polymeric structure known as pili (SpaC) was recently identified in Lactobacillus rhamnosus GG and shown to mediate adhesion to human intestinal mucus (21). Of note, this mechanism was previously established in many Gram-positive pathogens (22). Furthermore, nonproteinaceous compounds have been suggested to play a role in mucosal adhesion. For instance, teichoic or lipoteichoic acids were shown to be involved in adhesion of L. johnsonii (23) or L. reuteri (24) to Caco-2 cells. The role of exopolysaccharides produced by L. rhamnosus GG was also pointed out (25).

In contrast to pathogens, few studies have addressed the attachment of beneficial microorganisms such as L. lactis to mucus-like models, especially at the molecular level. In this context, atomic force microscopy (AFM) and optical or magnetic tweezers (26,27) have emerged as powerful tools for probing single-scale molecular events. They can be used to measure single-molecule interactions between receptors and ligands (28,29), such as antibodies and antigens (30), enzymes and substrates (31), biotin and streptavidin (32), and lectins and carbohydrates (33). Particular attention has also been paid to interactions between a virus and its host cell (34), and between vancomycin and the bacterial cell wall (peptidoglycan) (35). To this end, the target biomolecule was bound to the atomic force microscope tip and the single-molecule-based interactions between the modified tip and the biomimetic surface were measured. For a review of studies conducted on living cells, see Dufrêne (36). In this work, building on our previous results regarding the relevance of the lacto-probe concept for probing L. lactis/pig gastric mucin (PGM) interactions (37), we attached living lactococci to the atomic force microscope tip to obtain a better understanding of bacterial mucoadhesion from a mechanistic point of view. To that end, we measured and compared interactions between two different strains of L. lactis (IBB477 and MG1820 as control) and PGM-based coating at the single-molecule level. We performed blocking assays with free PGM and its constituting O-glycans to elucidate the respective roles of the protein and the oligosaccharide moieties of mucins. Moreover, by varying the loading rate and contact time during dynamic force spectroscopy experiments, we were able to determine, with living cell probes, the kinetic constants (association and dissociation) of the biological interactions between L. lactis and mucins.

Materials and Methods

Bacterial strain, growth conditions, and preparation of suspensions

Throughout this study, we used L. lactis subsp. cremoris strains IBB477 and MG1820. IBB477 was originally isolated from samples of Polish artisanal dairy products (38). Bacterial stock cultures were kept at −80°C in M17 broth (Oxoid) containing 2% (w/v) lactose and 20% (v/v) glycerol. Bacteria were first subcultured at 30°C in M17-lactose (2% (w/v)) medium. This preculture was then used to inoculate a 500 mL flask containing 100 mL of M17-lactose (2% (w/v)) broth, which was incubated overnight at 30°C until the early stationary phase was reached (optical density of 5.0 at a wavelength of 580 nm). Cells were harvested by centrifugation (4000 rpm, 10 min, and room temperature) and washed twice with MilliQ-grade water. The optical density of the suspension at 580 nm was then adjusted to 1.0 in MilliQ-grade water, which approximately corresponded to 5.10⁸ CFU/mL (determined by plating). Suspensions of both IBB477 and MG1820 strains were used to prepare the cell probe.

PGM layer preparation and characterization

The mucin starting material was commercially available as a lyophilized powder (M1778; Sigma, St. Louis, MO). Partially purified type III mucin from porcine stomach (PGM) was dissolved in phosphate-buffered saline (PBS) at pH 7.5 at a final concentration of 10 mg/mL. Coupons of polystyrene were prepared for PGM adsorption as described in our previous work (37), with slight modifications. In brief, polystyrene substrates made from sterile petri dishes (Greiner Bio-One SAS) were incubated overnight with PGM solution at 4°C under gentle agitation. After incubation, the surfaces were rinsed to remove loosely bound material with (in sequence) PBS and MilliQ-grade water, and then dried with N₂. PGM-coated polystyrene was subsequently characterized by an AFM scratch test. An area (3 × 3 μm²) was first recorded at high forces (10 nN), and a larger image (10 × 10 μm²) was then imaged under normal load (1 nN).

Cell probe design and characterization

OTR4 (Si3N4) probes, purchased from Veeco Instruments SAS, were used to prepare the cell probes as described elsewhere (37). Cantilevers and tips were first cleaned for 15 min via UV/O₃ treatment. They were precoated with polyethylenimine (PEI) by immersion for 5 h in a PEI solution (0.2% (w/v) in Milli-Q-grade water), rinsed with a copious amount of Milli-Q-grade water, and stored under light vacuum. L. lactis cells suspended in MilliQ-grade water (OD580 nm = 1) were then attached to the positively charged, PEI-precoated probes during a 20-min contact time. The cell probes were finally rinsed with MilliQ-grade water and were then ready for further experiments. After fabricating the cell probes and performing AFM measurements, we checked for the presence of immobilized bacteria on the AFM tip by scanning electron microscopy (SEM; Hitachi S-3700N). In addition, we evaluated the viability of bacteria after immobilization on the PEI-coated surface. To that end, we first observed the lacto-probe under bright-field microscopy to visualize the total amount of immobilized cells. Then, the lacto-probe was incubated with carboxyfluorescein diacetate (5(6)-CFDA: 5-(and-6)-carboxyfluorescein diacetate-mixed isomers, C195, 492–517 nm; Molecular Probes) for 45 min at 30°C, and reexamined under epifluorescence microscopy. In this method, cells that exhibit an esterase activity are seen in green. Control experiments were performed with dead cells after heat inactivation, and, as expected, no green fluorescence was detected (data not shown).

Force spectroscopy using AFM

We performed AFM experiments at room temperature using the Nanowizard II from JPK Instruments (Berlin, Germany) and the Bioscope II from Veeco Instruments (Santa Barbara, CA). All measurements involving L. lactis were conducted in milliQ-grade water to avoid any modification of the cell wall due to desiccation. Single force-distance curves and a matrix of 32 × 32 force-distance curves on 5 × 5 μm² squares were recorded, giving 1024 force curves to be analyzed for each experiment. We probed the PGM-coated surface with the lacto-probe to compare the adhesion force between the two strains of L. lactis (IBB477 and MG1820). The spring constants of the tips were measured for each probe by the thermal-tune method and were in the range of 0.01–0.02 N.m⁻¹. Unless otherwise specified, all force measurements were recorded with a loading rate of 96,000 pN.s⁻¹ and a contact time of 0 s. We calculated the loading rate by multiplying the spring constant of the cantilever by its retraction velocity. We then adjusted it for related experiments by controlling the cantilever velocity. For measurements dedicated to the determination of kinetic constants (K_on and K_off), the loading rate and contact time were varied in the range of 8000–160,000 pN.s⁻¹ and 0–10 s, respectively. To calculate K_on, we took only force curves that presented biological interactions into account (i.e., considering multiple adhesion events and excluding any nonspecific events). The adhesion rate calculated corresponds to the number of force curves that presented at least one interaction event at several nanometers after the contact point (specific events) over the total number of force curves recorded. Blocking assays were performed with free PGM (10 mg/mL in PBS) and O-glycans (10 mg/mL in PBS). The O-glycans were prepared as follows: PGM was submitted to reductive β-elimination for 72 h at 37°C in 100 mM NaOH containing 1 M NaBH₄. The reaction was stopped by the addition of Dowex (Sigma-Aldrich) 50 × 8 (25–50 mesh, H⁺ form) at 4°C until pH 6.5. After filtration on glass wool and evaporation to dryness, boric acid was eliminated by repetitive distillation as its methyl ester in the presence of methanol. The material was submitted to a cationic exchange chromatography on Dowex (Sigma-Aldrich) 50 × 2 (200–400 mesh, H⁺ form) to remove residual peptides. Sugar-containing fractions were purified on a Bio-Gel P2 column (150 × 2.5 cm; Bio-Rad).

AFM experiments were performed in triplicate with independent PGM-coated surfaces and lacto-probes. We quantified the adhesion forces, as deduced from the force-distance curves, using SPM Image Processing v.03 software from JPK Instruments AG, and Research Nanoscope 7.30R1 software from Veeco Instruments SAS.

Results and Discussion

In this work we used AFM force spectroscopy to quantify at the molecular scale interactions between PGM model mucin, adsorbed on bare polystyrene, and two strains of L. lactis (the control strain MG1820, depicted in terms of in vitro mucoadhesion in our previous work (37), and the IBB477 strain, which exhibited efficient persistence in the rat gastrointestinal tract when tested in vivo (38)). We applied the lacto-probe concept and associated AFM measurements (37) to probe, discriminate, and elucidate the mucoadhesive properties of both strains, and to explore the dynamics of L. lactis-PGM biological interactions.

Characterizing the lacto-probe and the PGM-coated surface

Fig. 1 shows SEM images obtained at each step of the lacto-probe creation process. The Si₃N₄ tip (Fig. 1 A) was first covered with the polycation PEI. The positively charged tip (Fig. 1 B) was then incubated with L. lactis cells, which were shown to be negatively charged (data not shown). The same procedure was applied to create lacto-probes with MG1820 and IBB477 strains (Fig. 1, C and D, respectively). Regardless of the strain involved, the sensitivity of a single cell on the AFM tip could not be achieved, and clusters of cells were rather typically attached, as previously depicted (37). However, the immobilization efficiency was identical for both strains, probably due to the same surface charge (data not shown). We carried out additional experiments to check the viability of the bacteria once they were immobilized on the PEI-coated surface. To that end, we first observed the lacto-probe under bright-field microscopy to visualize the total amount of cells (Fig. 1 E). The lacto-probe was then incubated with CFDA and reexamined under epifluorescence microscopy. Cells that exhibit an esterase activity appear bright and can be considered as physiologically active. Fig. 1 F clearly shows that the L. lactis bacteria on the probe were viable, or at least esterase active. Furthermore, after each AFM experiment, we checked the persistence of bacteria on the AFM tip by SEM (data not shown).

Figure 1.

Representative SEM and epifluorescence images of the biologically functionalized AFM probe, for L. lactis cells immobilized onto AFM tip and cantilever. (A) Si₃N₄ AFM probe. (B) AFM probe after preadsorption of PEI. (C and D) Lacto-probe of MG1820 and IBB477, respectively. (E) Bright-field microscopy image of the MG1820 lacto-probe (total bacteria). (F) Epifluorescence image of the MG1820 lacto-probe after CFDA staining.

We previously depicted the PGM layer formed on bare polystyrene using different multiscale methods (e.g., AFM, x-ray photoelectron spectroscopy (XPS), and the sessile drop method) (37). XPS-based modeling enabled us to estimate a layer thickness of 3.4 nm. To experimentally validate this value, we carried out AFM scratch tests with a bare atomic force microscope tip (Fig. 2). Using this procedure, we removed the PGM layer and revealed the underlying polystyrene surface (arrow in Fig. 2 C). The thickness of the PGM layer, as deduced from such measurements, was 3.15 nm, which is nearly identical to the predicted value of 3.4 nm.

Figure 2.

AFM height (A) and deflection (C) images of the PGM layer formed on polystyrene. (B) Scratch test on a cross section.

Measuring the interaction force of L. lactis-PGM: comparison of the IBB477 and MG1820 strains

Force-distance curves were recorded at a loading rate of 96,000 pN.s⁻¹ between the lacto-probe and the PGM-based coating. Fig. 3 shows a comparison of the AFM results for the MG1820 and IBB477 strains. Adhesion maps are illustrated in Fig. 3, A and B, respectively. Histograms representing the different interaction force distributions (i.e., MG1820 lacto-probe/PGM coating, and IBB477 lacto-probe/PGM coating) are displayed in Fig. 3, C and D, respectively. Illustrative schemes for each configuration tested are also included. Fig. 3, E and F, highlight typical recorded retraction force-distance curves. As previously reported (37), adhesion of L. lactis to PGM was characterized by the combination of three representative curve shapes: 1), no adhesive event detected upon retraction of the tip from the PGM-coated surface; 2), nonspecific adhesive events (showing no extension before rupture); and 3), one, two, or three specific adhesive events occurring at several nanometers after the contact point. We define these specific interactions as biological ones involving ligand-receptor bonding (e.g., antibody-antigen interactions), which are different from nonspecific physicochemical interactions (e.g., hydrophobic, electrostatic, or Lifshitz-van der Waals interactions). Here, the particular shape of group-3 retraction force curves could be due to multiple contact sites of bacteria and/or stretching of the PGM molecules. Therefore, we performed a thorough analysis of such force-distance curves for each strain by focusing on the respective roles of nonadhesive events, and nonspecific and specific adhesive events. For the MG1820 strain, 45% of the force curves corresponded to nonadhesive events, and 50% of the force curves were characteristic of nonspecific physicochemical interactions (including hydrophobic, electrostatic, and van der Waals forces). Only 5% of the force curves were assigned to specific adhesive events. Such force repartition was consistent with previously published data (37). In contrast, for the IBB477 strain, almost all force curves (99%) corresponded to adhesive events and 20% of the force curves could be assigned to specific biological interactions. As a result, the adhesion force of the IBB477 strain to PGM was significantly higher than the value obtained for the MG1820 control (0.22 ± 0.05 nN vs. 0.12 ± 0.06 nN), which is consistent with the in vivo persistence of such a strain in the rat gastrointestinal tract (38).

Figure 3.

(A and B) Adhesion maps. (C and D) Histograms representing the adhesion forces. (E and F) Typical force-distance curves presented for (A, C, and E) the MG1820 lacto-probe after contact with the PGM-coated surface, and (B, D, and F) the IBB477 lacto-probe after contact with the PGM-coated surface.

Elucidating the respective roles of protein and oligosaccharide moieties of PGM in the interaction with L. lactis

To gain a more thorough understanding of the specificity of the interaction events observed for IBB477 and to a lesser extent MG1820, we carried out additional blocking experiments with free PGM and its sole O-glycans (S-PGM) to elucidate the respective roles of protein and oligosaccharide moieties of PGM. To that end, we incubated the lacto-probe in PGM or S-PGM solutions before putting it in contact with the PGM-based coating. We speculated that potential binding sites on the bacteria surface were saturated, with an efficiency correlated to their affinity toward PGM or S-PGM, and thus were no longer able to react with the PGM layer (see Fig. 4 A). For the MG1820 and IBB477 strains, preincubation of the lacto-probe with PGM dramatically reduced the number of curves that showed adhesive events, as well as the measured binding forces, indicating that the adhesion forces measured with the lacto-probe were specific to the L. lactis-PGM interaction. Indeed, the percentage of nonadhesive events drastically increased, from 45% to 81% for MG1820 (Fig. 4 B) and from 1% to 81% for IBB477 (Fig. 4 C). Inhibition of adhesion was lower, albeit significant, with S-PGM for both strains. For MG1820, the nonadhesive event percentage increased from 45% to 71% (Fig. 4 B), whereas for IBB477 it only reached 48% (Fig. 4 C). On the basis of these results, we postulate that O-glycans in PGM plays a major (but not exclusive) role in mucin interactions with L. lactis. Some additional interactions probably occurred between the PGM protein core and other components of the cell surface, such as the polysaccharide pellicle, as recently described for lactococci (39).

Figure 4.

(A) Schematic representation of the AFM blocking test with free PGM or S-PGM. (B and C) Histograms of interaction forces before and after incubation of the lacto-probe with S-PGM and PGM solutions for (B) MG1820 and (C) IBB477.

Exploring the dynamics of L. lactis-PGM interactions

Finally, we sought to elucidate the dynamics of the biological interactions between L. lactis and PGM by determining the kinetic dissociation/association constants (K_off and K_on) by varying the AFM parameters (e.g., loading rate and contact time) during dynamic force spectroscopy experiments. Because of its specificity to PGM (20% specific adhesive events versus 5% for the control strain), we performed these experiments on the IBB477 strain (for determination of both kinetic constants). However, the association kinetic constant (K_on) was also evaluated for MG1820 (see below).

The unbinding force for a specific biological bond is expected to be a function of the loading rate (40;also see (44)). In other words, the faster the tip is retracted, the higher will be the interaction force. The loading-rate dependence comes from the covalent link that exists in what we call biological interactions. Electrostatic, hydrophobic, van der Waals interactions are noncovalent and therefore are not expected to be influenced by the loading rate (41). Fig. 5 displays, for the IBB477 strain, the variation of the adhesion force with the loading rate ranging from 8000 to 160,000 pN.s⁻¹, taking into account both specific and nonspecific contributions (Fig. 5, A and B, respectively). For specific biological interactions, the direct relation between the adhesion force and the applied loading rate is clearly highlighted, with an increase from 85 ± 16 pN at 8000 pN.s⁻¹ to 116 ± 17 pN at 160,000 pN.s⁻¹ (Fig. 5 A). In contrast, as expected, when nonspecific interactions were considered, the adhesion force did not substantially depend on the loading rate applied during retraction, over the experimentally accessible range (Fig. 5 B).

Figure 5.

Loading-rate dependence of specific (A) and nonspecific (B) interaction forces between L. lactis and PGM, for IBB477.

Based on these results, we estimated the kinetic dissociation constant (K_off) of the L. lactis-PGM bond, for the IBB477 strain, according to the following equation (40):

$$F=f_{\beta }\times \ln \left(\frac{r}{f_{\beta }{K}_{off}}\right),$$ (1)

where F is the measured adhesion force; f_β is defined as the ratio between the thermal energy scale (k_BT, where k_B is Boltzmann's constant and T is temperature) and x, which is the length of the bond at the transition state; and r corresponds to the loading rate.

Equation 1 can be rewritten as follows:

$$K_{off}=\frac{r_0}{f_{\beta }},$$ (2)

where r₀ corresponds to the loading rate at zero force.

The values f_β (slope) and r₀ (loading rate at zero force) were deduced from data presented in Fig. 5 A and reached 10.9 J.m⁻¹ and 5.0 pN.s⁻¹, respectively. The K_off parameter, estimated using Eq. 2, was equal to 0.46 s⁻¹.

In parallel, we studied the variation of the adhesion probability (derived from specific interaction forces) with the contact time between the lacto-probe and the PGM layer for IBB477 while keeping the loading rate constant (96,000 pN.s⁻¹). Data for the control MG1820 are also reported. As expected, the adhesion probability increased with contact time (Fig. 6). For MG1820, it rose from 6% to 17% for a contact time in the range of 0–4 s and then reached a plateau. A similar profile was observed for IBB477 (Fig. 6). The adhesion probability first increased from 17% to 32% and tended to level off after a contact time of 4 s. The dependence of the adhesion probability on the contact time indicates that the L. lactis/PGM complex formed via multiple intermolecular bonds. Indeed, the increase in contact time favors the formation of multiple bonds (42) and promotes larger contact surfaces through the viscoelastic adaptation of the cell shape to the interacting surface and the strengthening of active bonds by the cell (43). Moreover, regardless of the contact time, the percentage values remained higher than observed for MG1820, reinforcing our conclusions about the improved mucoadhesive ability of the IBB477 strain. However, we should note that it was not possible to reach a 100% rate of adhesive events.

Figure 6.

Contact-time dependence of the adhesion probability for MG1820 (♦) and IBB477 (▪).

On the basis of these results, we roughly estimated the kinetic association constant (K_on) between L. lactis and PGM for both MG1820 and IBB477 strains according to the following equation (44):

$$K_{on}=t_{0.5}^{-1}\times C_{eff}^{-1},$$ (3)

where t_0.5 is the contact time needed to reach 50% of adhesive events, and C_eff corresponds to the effective concentration (M).

As depicted above, it was not possible to reach a 100% rate of adhesive events. As a consequence, the contact time needed to obtain 50% of adhesive events (t_0.5) could not be determined. Instead, we considered the duration required for the half-maximal probability of binding (1.37 s for both strains). In parallel, we deduced the effective concentration from the following equation (45):

$$C_{eff}=n\times N_A^{-1}\times V_{eff}^{-1},$$ (4)

where n is defined as the number of binding partners within the effective volume V_eff accessible for free equilibrium interaction, and N_A is the Avogadro number.

To evaluate the effective volume V_eff, we defined a semi-ellipsoid (disk shape) assuming a polar radius corresponding to the thickness of the PGM layer (3.15 nm; see above) and an equatorial radius equal to the AFM tip radius (15 nm). Because two peaks were generally observed on force curves corresponding to specific adhesive events (Fig. 3), a number of two binding partners was assumed (n = 2). Under these conditions, the K_on parameter was estimated at 3.3 × 10² M⁻¹.s⁻¹ for both the IBB477 and MG1820 strains. Nevertheless, it should be noted that numerous assumptions had to be formulated, leading to some speculation about the predicted values, which should be considered with caution.

We used dynamic force spectroscopy to determine, for the first time to our knowledge, the kinetic constants (K_off and K_on) of the biological interactions between L. lactis and PGM by varying the AFM parameters (i.e., the loading rate and contact time). Table 1 shows values of K_off and K_on reported in the literature, derived from AFM force spectroscopy, surface plasmon resonance (SPR), and parallel plate flow-chamber experiments. The large dispersion of both K_off and K_on values is impressive, probably due to the different methods used to probe the interactions (see, for example, differences in K_off for the same interaction as revealed by AFM and SPR (see (50)). First, few K_on values are available, and they range from 5.6 × 10⁻⁴ for concanavalin A/carboxypeptidase Y (45) to 4.4 × 10⁶ for P-selectin/P-selectin glycoprotein ligand (47). Our K_on parameter (3.3 × 10² M⁻¹s⁻¹), which was identical for the MG1820 and IBB477 strains, is thus hardly comparable with the literature values. A large variability was also observed for the kinetic dissociation constant (K_off). The lowest K_off values were in the range of 10⁻⁶–10⁻⁵ s⁻¹, corresponding to interactions between E-cadherin/E-cadherin (48) or streptavidin/biotin (49). Our K_off parameter (0.46 s⁻¹) was of the same order of magnitude as values obtained by Dettmann et al. (50) and Sletmoen et al. (51) using AFM on sugar-protein interactions. This reinforces our conclusions about L. lactis-PGM interactions, mainly mediated by O-glycans of PGM and proteinaceous-surface–exposed structures of L. lactis.

Table 1.

Dissociation (K_off) and association (K_on) kinetic constants of different ligand-receptor complexes as found in the literature

Molecular partners	Koff (s−1)	Kon (M−1.s−1)	Methods
Human serum albumin (HSA)/anti-HAS	6.7 × 10−4	5 × 104	AFM (30)
Streptavidin/biotin	1.67 × 10−5	-	AFM (49)
Streptavidin mutant/biotin	10−2	-	AFM (54)
	30		AFM (49)
	6.7 × 10−3
	1.05
Concanavalin A (conA)/carboxypeptidase Y	0.17	5.6 × 10−4	AFM (46)
VE-cadherin/VE-cadherin	1.8	104	AFM (44)
E-cadherin/E-cadherin	0.01	-	AFM (48)
	10−6-10−5
Lysin motif (LysM)/peptidoglycan	0.15	2.4 × 103	AFM (55)
Lactose/bovine heart lectin (BHL)	0.09/0.5 × 10−3	-	AFM/SPR (50)
Lactose/lactose-binding immunoglobulin G (IgG)	0.9/1.3 × 10−3
Lactose/Viscum album agglutinin (VAA)	0.09/1.1 × 10−3
Lactose/Ricinus communis agglutinin (RCA)	0.8/1.1 × 10−3
N-glycan chains of Asialofetuin (ASF)/BHL	1.3/1.1 × 10−3
ASF/VAA	0.9/1.3 × 10−3
ASF/IgG	1.6/0.5 × 10−3
ASF/RCA	0.4/1.2 × 10−3
P-selectin/P-selectin glycoprotein ligand-1 (PSGL-1)	1.4	4.4 × 106	SPR (47)
Mucin-like sialoglycoprotein termed peripheral node addressin (PNAd) /		-	Parallel plate flow chamber (56)
P-selectin	0.93
L-selectin	0.7
E-selectin	6.8
O-glycosylation domain of porcine submaxillary mucin possessing only α-GalNAc residues (Tn-PSM)/soybean agglutinin (SBA)	0.76	-	AFM (51)

Concluding Remarks

Extracellular proteins with domains predicted to be involved in adhesion to mucus have been described for LAB (9). These MUB domains, which are found exclusively in LAB, are variable in size (100–200 residues per domain) and number (one to 15 for a given protein). Among different LAB species (lactobacilli/lactococci), MUB domain-containing proteins are most abundant in lactobacilli, which are the natural inhabitants of the gastrointestinal tract (9). The MUB domain is undoubtedly a specific functional unit that plays an important role in host-bacteria interactions in the gastrointestinal tract (e.g., persistence, competitive exclusion of pathogens, and other health-stimulatory interactions). In addition, a subunit of the proteinaceous surface-exposed polymeric structure known as pili (SpaC) was identified in L. rhamnosus GG and shown to mediate adhesion to human intestinal mucus (21). Hypothetical pili formation and function(s) in L. lactis were recently addressed (identification of a putative class C sortase through in silico analysis) (52).

To explain the low percentage of specific adhesive events for the control strain MG1820 (5%), we should mention that in traditional, domesticated L. lactis strains, which live in a more restricted habitat than lactobacilli of the gastrointestinal tract, the presence of only one single MUB-domain-containing protein has been reported (9). However, Giaouris et al. (10) highlighted the diversity of surface physicochemical properties among 50 natural L. lactis strains isolated from different origins (dairy, vegetal, and animal). In the same way, Passerini et al. (53) recently observed genetic and genomic diversity within a collection of 36 strains (L. lactis subsp. lactis) isolated from different ecological sources and geographical areas. The authors proposed a new classification based on ecological separation between domesticated and environmental strains, the latter being the main contributors to genetic diversity within the subspecies. Such diversity may provide an important pool of phenotypic functionalities, probably including mucoadhesion. In this framework, the identification of molecular determinants involved in the in vitro/in vivo mucoadhesive ability of the IBB477 strain (e.g., MUB domain-containing protein(s) and hypothetical pili) would be highly valuable.

In conclusion, by using AFM force spectroscopy on two different strains of L. lactis (IBB477, which was previously shown to exhibit in vivo persistence in the rat gastrointestinal tract, and the control strain MG1820), we were able to gain new (to our knowledge) insights into the interaction mechanisms between L. lactis and PGM. We directly quantified at the nanoscale interaction forces between an AFM tip functionalized with living L. lactis cells (lacto-probe) and PGM-coated polystyrene. Both nonspecific (showing no extension before rupture) and specific forces (ligand-receptor bonding) were shown to be involved in L. lactis adhesion to PGM. A higher percentage of specific adhesive events was observed for IBB477 (20%) compared with the control strain (5%), probably in line with its in vivo persistence. Blocking assays with free PGM and O-glycans showed that oligosaccharides play a major (but not exclusive) role in interactions between L. lactis and PGM. Moreover, we characterized the biological interactions, for the first time (to our knowledge) with living cell probes on mucins, in terms of both kinetic dissociation and association constants (K_off and K_on). Varying the loading rate and the contact time allowed us to estimate K_off and K_on, respectively. For both strains, the K_on value was identical (3.3 × 10² M⁻¹.s⁻¹). The K_off parameter, which was only evaluated for IBB477, was equal to 0.46 s⁻¹, which is consistent with literature data reported for sugar-protein interactions. This reinforces our conclusions about L. lactis-PGM interactions, mainly mediated by O-glycans of PGM and proteinaceous-surface-exposed structures of L. lactis. Our findings regarding bacterial mucoadhesion may have broad implications for medical and food-related applications, notably by aiding in the selection of probiotic candidates and defining improved drug-delivery strategies.