The bacterial strains and plasmids used in this study are listed in Supplementary Table S1. Bifidobacterium longum BBMN68 cells (GenBank accession no. NC_014656.1) were grown anaerobically at 37°C in de Man-Rogosa-Sharpe (MRS) broth supplemented with 1% (v/v) L-cysteine (MRSc). Lactococcus lactis NZ9000 was grown at 30°C in M17 medium (Oxoid, Unipath, Basingstoke, United Kingdom) containing 0.5% (w/v) glucose (GM17). Escherichia coli strains were grown aerobically at 37°C in Luria-Bertani (LB) medium with shaking (220 rpm). When required, media were supplemented with the relevant antibiotics at the following concentrations: 100 μg ml⁻¹ ampicillin and 10 μg ml⁻¹ chloramphenicol for E. coli, and 5 μg ml⁻¹ chloramphenicol for L. lactis.

Genomic DNA was extracted from B. longum BBMN68 using a TIANamp Bacteria DNA Kit according to the manufacturer’s instructions (TianGen, Beijing, China). Nine genes encoding predicted surface adhesion proteins were amplified from genomic DNA using the primer pairs listed in Supplementary Table S2. The PCR products digested with KpnI/HindIII or KpnI/XbaI (NEB, Beijing, China) were inserted into the corresponding sites in pNZ81481. The ligated plasmids were then transformed into L. lactis NZ9000 by electroporation using Bio-Rad Gene Pulser Xcell (Bio-Rad, Richmond, CA, United States) as previously described (DeRuyter et al., 1996). The resulting recombinant plasmids were isolated using the E.Z.N.A. Plasmid Mini Kit I (Omega Bio-tek Inc., Doraville, GA, United States). The recombinant strains were confirmed by plasmid sequencing and further analyzed with the DNAMAN software package. Meanwhile, L. lactis NZCK harboring the empty pNZ81481 vector was used as the control strain. Overnight cultures of the recombinant strains were inoculated (1% inocula) into 10 ml of fresh GM17 medium containing 5 μg ml⁻¹ chloramphenicol. To induce gene expression, when the cell density had reached an OD₆₀₀ nm of 0.2~0.3, the cultures were supplemented with 10 ng ml⁻¹ nisin (Sigma-Aldrich, Milwaukee, WI, United States) and incubated for an additional 2 h at 30°C. The cell pellets were then collected after centrifugation at 6,000 × g for 5 min at 4°C for subsequent SDS–polyacrylamide gel electrophoresis (SDS–PAGE) analysis and adhesion assays.

The human colon adenocarcinoma cell line HT-29 and the goblet cell-derived cell line LS174T were obtained from the China Infrastructure of Cell Line Resource. The cells were cultured at 37°C in a humidified atmosphere with 5% CO₂. The HT-29 cells were grown in Dulbecco’s high-glucose modified Eagle’s medium (DMEM, Thermo Fisher Scientific, Rockford, IL, United States) supplemented with 10% (v/v) fetal bovine serum (FBS, Invitrogen, New York, NY, United States) and 100 U ml⁻¹ penicillin/streptomycin (Gibco, Waltham, MD, United States); the mucus-producing LS174T cells (Hews et al., 2017) were grown in RPMI 1640 medium (Gibco) supplemented with 2 mM L-glutamine (Gibco), 10% FBS, and 100 U ml⁻¹ penicillin/streptomycin. The cells were subcultured every 2–3 days. For adhesion analysis, cells were seeded at a density of 1 × 10⁵ cells per well into 24-well plates and grown to ~90% confluence. Epithelial cell monolayers were carefully washed twice with phosphate-buffered saline (PBS, pH 7.4) before the addition of bacterial cells. Recombinant L. lactis strains were washed twice with PBS and resuspended in DMEM or RPMI 1640 medium without antibiotics at a concentration of ~1 × 10⁷ colony-forming units (CFUs) ml⁻¹. Aliquots (1 ml) of L. lactis suspension were added to the wells. The plates were incubated for 1 h at 37°C in 5% CO₂, following which the wells were gently washed five times with PBS to remove unattached bacteria. The epithelial cells with adherent bacteria were detached using 0.25% trypsin (Sigma-Aldrich, St. Louis, MO, United States) treatment. Bacterial counts were determined by plating 10-fold serial dilutions on GM17 plates. Adhesion ratios were calculated as a percentage using the following formula: 100 × the number of adherent bacteria/the number of bacteria inoculated. The results were representative of three independent experiments, each performed in triplicate.

The extracellular matrix proteins (fibronectin, laminin, collagen I, collagen IV, fibrinogen, and plasminogen; Sigma-Aldrich, St. Louis, MO, United States; product codes: 10838039001, L4544, C7624, C5533, F3879, and P7999, respectively), were all of human origin. Mucin (Sigma-Aldrich, St. Louis, MO, United States; product code: M2378) was isolated from the porcine stomach. The ECM proteins and mucin were immobilized in wells of a Nunc Maxisorp 96-well microplate (Maxisorp; Nunc, Roskilde, Denmark) by overnight incubation at 4°C at a concentration of 2.5 pmol well⁻¹. The wells were then washed twice with PBS and incubated with blocking buffer [2% (w/v) BSA in PBS] for 2 h at 37°C. The wells were subsequently washed three times with PBS before adding the bacteria. The bacteria were washed twice with PBS and resuspended in PBS to a final OD₆₀₀ equivalent to ~1 × 10⁸ CFUs ml⁻¹. Aliquots (200 μl) of L. lactis or B. longum suspension were added to coated 96-well plates and incubated at 30°C (L. lactis) or 37°C (B. longum) for 1 h. The unattached bacteria were removed by washing the wells three times with PBS. Bacteria bound to ECM proteins were detached by treatment with PBS containing 0.01% (v/v) Triton X-100 followed by incubation at 30 or 37°C for 30 min with shaking (200 rpm). The bacterial counts were determined on GM17 (L. lactis) or MRSc (B. longum) plates. The adhesion ratios, expressed as percentages, were calculated by comparing the bacterial counts after adhesion to the number of cells in the bacterial suspension added originally to the plate wells. All the results were representative of three independent experiments, each performed in triplicate. For the inhibition assays, L. lactis and B. longum BBMN68 were preincubated with anti-FimM antibody (diluted 1:10,000 in PBS) for 1 h, after which the protocol was continued as outlined above.

The fimM fragment, without the regions encoding the signal sequence and the LPxTG motif, was amplified from B. longum BBMN68 using the primer pair fimM-GST-F and fimM-GST-R (Supplementary Table S2). The predicted product was cloned into the pGEX-4T-1 vector (GE Healthcare, Madison, WI, United Kingdom) for expression as an N-terminal GST-fusion protein in E. coli BL21 (DE3). When the OD₆₀₀ of the recombinant cells had reached 0.4, isopropyl-β-D-thiogalactopyranoside (IPTG) was added to induce protein expression. Soluble proteins were then purified by glutathione-sepharose 4B affinity chromatography (Solarbio, Beijing, China) after a 12 h incubation at 16°C. Subsequently, the GST tag of the purified protein was removed by thrombin cleavage (Solarbio). The purified proteins were detected by SDS-PAGE and the concentration was measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, United States). Polyclonal antibodies against FimM were raised in rabbits by BIOSS CO., LTD (Beijing, China) as previously described (Johnston et al., 1991).

Lactococcus lactis NZCK, L. lactis NZfimM, and the overnight-cultured B. longum BBMN68 were washed three times with PBS and then diluted (OD₆₀₀ of 2.0) in the same buffer. Formvar/carbon-coated copper grids were floated for 10 min on droplets of the diluted cells in PBS, washed several times with the same buffer, and then treated with a blocking solution containing 1% bovine serum albumin (BSA) at 37°C for 30 min. The grids were then floated for 1 h on droplets of blocking solution containing anti-FimM serum (diluted 1:100), washed five times with 0.1% BSA in PBS to remove unbound antibodies, and incubated at 37°C for 1 h with 10-nm diameter gold particle-conjugated protein A diluted 1:55 in blocking solution. After several washes in PBS, the grids were negatively stained with a mixture of 1.8% methylcellulose-0.4% uranyl acetate. The samples were examined under a JEM-1400 transmission electron microscope (JEOL, Ltd., Tokyo, Japan).

Data were analyzed using GraphPad Prism 6 for Windows (GraphPad Software, Inc., La Jolla, CA, United States). When two groups were compared, an unpaired Student’s t-test was used to calculate p values.

Using PSORTdb, a subcellular localization database for bacteria and archaea, 560 extracellular proteins were predicted to be present in the B. longum BBMN68 genome (Peabody et al., 2016). These included 507 proteins that localized to the cytoplasmic membrane and 25 cell wall-anchored proteins (Supplementary Table S3). For adhesion protein prediction, 21 proteins with possible adhesive functions were identified using NCBI’s conserved domain database (Marchler-Bauer et al., 2017; Table 1), including eight that contained a Gram-positive pilin subunit domain, two with a laminin G domain, seven that belonged to the glycosyl hydrolase family, two permeases belonging to the ABC-type transport system, one containing a cadherin-like beta-sandwich domain, and one S-layer protein.

Although 21 genes in the B. longum BBMN68 genome were predicted to encode surface adhesion proteins, only 9 (lspA, fimM, slpA, dppB3, potB, aprE, tadE, tadF, and fimA; Table 1) were successfully cloned into the pNZ81481 expression vector (Supplementary Table S1). The remaining 12 genes could not be ligated into the vector possibly because these genes were too large for PCR amplification or ligation. After sequencing, the correct plasmids (respectively designated as pNZlspA to pNZfimA) were then transformed into L. lactis NZ9000 to generate recombinant strains (designated as L. lactis NZlspA to L. lactis NZfimA; Supplementary Table S1). SDS-PAGE analysis of total protein identified the overproduction of 64-, 41-, 211-, and 56-kDa proteins (Figure 1), which corresponded to the expected sizes of FimM, SlpA, AprE, and FimA, respectively. These results demonstrated that these four proteins could be successfully expressed in L. lactis NZ9000.

To investigate whether the expression of FimM, SlpA, AprE, and FimA could influence the adhesion of the host strain to intestinal epithelial cells, we performed adhesion assays between HT-29 and mucus-secreting LS174T cells (Walsham et al., 2016). The results showed that the adhesion ratios of strains L. lactis NZfimM and L. lactis NZslpA to HT-29 cells were 0.24 and 0.19%, respectively, which were 2.2- and 1.7-fold higher than that of the control L. lactis NZCK strain (Figure 2A). In addition, the adhesion ratios of strains L. lactis NZfimM, L. lactis NZaprE, and L. lactis NZfimA to LS174T cells were 12.87, 11.63, and 3.25%, values that were 5.4-, 4.9-, and 1.4-fold higher than that for L. lactis NZCK, respectively (Figure 2B). Notably, only the overexpression of FimM in L. lactis NZ9000 led to a significant increase in adhesion to both cell types. Consequently, we subsequently focused on investigating the adhesion mechanisms associated with FimM.

To detect the localization of FimM at the surface of L. lactis NZfimM, cells of this strain were visualized by immunogold transmission electron microscopy. First, FimM was expressed as a GST-fusion in E. coli and purified by glutathione-sepharose 4B affinity chromatography. The GST tag was subsequently removed by thrombin cleavage. The purified protein showed an expected molecular mass of 58 kDa (Supplementary Figure S1), and was used to raise anti-FimM antiserum in rabbits. The specificity of the anti-FimM polyclonal antiserum was tested by western blotting. The results showed that the antiserum was specific for FimM, and showed no recognition for the total protein of NZfimM (Supplementary Figure S2). Subsequently, L. lactis NZfimM and L. lactis NZCK (negative control) cells were visualized by immunogold transmission electron microscopy using anti-FimM antiserum in combination with 10-nm gold particle-labeled protein A. A high number of gold particles were observed on the cell surface of L. lactis NZfimM (Figures 3C,D), whereas none were found on the negative control (Figures 3A,B), indicating that FimM localized to the surface of L. lactis NZfimM cells.

To identify the adhesion receptors involved in FimM recognition, we assessed the capacity of L. lactis NZfimM to adhere to mucin and the ECM substrates fibronectin, laminin, collagen type I, collagen type IV, fibrinogen, and plasminogen. The adhesion ratios of the control strain L. lactis NZCK for fibronectin, laminin, collagen type I, collagen type IV, fibrinogen, plasminogen, and mucin were 1.58, 0.93, 2.46, 1.81, 2.08, 1.72, and 1.45%, respectively. The corresponding adhesion ratios for L. lactis NZfimM were 5.94, 1.31, 2.43, 1.65, 4.34, 1.73, and 4.55%. These results showed that the adhesive affinity of L. lactis NZfimM for fibronectin, fibrinogen, and mucin was 3.8-, 2.1-, and 3.1-fold that of the control. No significant differences were observed between L. lactis NZfimM and the control strain for laminin, collagen I, collagen IV, and plasminogen binding (Figure 4). These results indicated that fibronectin, fibrinogen, and mucin are the adhesion receptors for the B. longum BBMN68 FimM protein.

To further confirm the specific binding of FimM to its recognition factors, we performed antibody-mediated inhibition experiments. Lactococcus lactis NZfimM cells were incubated with an anti-FimM antibody before being added to immobilized mucin, fibronectin, and fibrinogen. The results showed that the binding of NZfimM to mucin, fibronectin, and fibrinogen was reduced by 51.6, 52.6, and 68.3%, respectively (Figure 5A). Subsequently, B. longum BBMN68 cells were pretreated with an anti-FimM antibody. Notably, pretreated B. longum BBMN68 cells exhibited average reductions in their adhesion to mucin, fibronectin, and fibrinogen of 54.8, 46.1, and 87.8%, respectively (Figure 5B). These results provided further evidence that mucin, fibronectin, and fibrinogen function as adhesion receptors for FimM.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.