Characterization of a fibronectin-binding protein from Lactobacillus casei BL23

Authors

  • D. Muñoz-Provencio,

    1. Laboratorio de Bacterias Lácticas y Probióticos, Instituto de Agroquímica y Tecnología de Alimentos-CSIC, Burjassot, Valencia, Spain
    Search for more papers by this author
  • G. Pérez-Martínez,

    1. Laboratorio de Bacterias Lácticas y Probióticos, Instituto de Agroquímica y Tecnología de Alimentos-CSIC, Burjassot, Valencia, Spain
    Search for more papers by this author
  • V. Monedero

    1. Laboratorio de Bacterias Lácticas y Probióticos, Instituto de Agroquímica y Tecnología de Alimentos-CSIC, Burjassot, Valencia, Spain
    Search for more papers by this author

Vicente Monedero, Instituto de Agroquímica y Tecnología de Alimentos-CSIC, PO Box 73, 46100 Burjassot, Valencia, Spain. E-mail: btcmon@iata.csic.es

Abstract

Aims:  To characterize the functionality of the Lactobacillus casei BL23 fbpA gene encoding a putative fibronectin-binding protein.

Methods and Results:  Adhesion tests showed that L. casei BL23 binds immobilized and soluble fibronectin in a protease-sensitive manner. A mutant with inactivated fbpA showed a decrease in binding to immobilized fibronectin and a strong reduction in the surface hydrophobicity as reflected by microbial adhesion to solvents test. However, minor effects were seen on adhesion to the human Caco-2 or HT-29 cell lines. Purified 6X(His)FbpA bound to immobilized fibronectin in a dose-dependent manner. Western blot experiments with FbpA-specific antibodies showed that FbpA could be extracted from the cell surface by LiCl treatment and that protease digestion of the cells reduced the amount of extracted FbpA. Furthermore, surface exposition of FbpA was detected in other L. casei strains by LiCl extraction and whole-cell ELISA.

Conclusions:  FbpA can be found at the L. casei BL23 surface and participates in cell attachment to immobilized fibronectin. We showed that FbpA is an important, but not the only, factor contributing to fibronectin binding in BL23 strain.

Significance and Impact of the Study:  This is the first report showing the involvement of FbpA in fibronectin binding in L. casei BL23 and represents a new contribution to the study of attachment factors in probiotic bacteria.

Introduction

Lactobacilli have been used for the fermentation of food products, and they have attracted much attention as probiotic bacteria for their beneficial effects on human health. Adhesion of probiotic bacteria to the host intestinal epithelium is an important criterion for strain selection, and several methods (binding to cultured epithelial cells, to immobilized tissue components or to resected tissue) have been employed for characterization and screening of new strains (Ouwehand et al. 2001; Styriak et al. 2003; Tuomola and Salminen 1998; Vesterlund et al. 2006). Adhesion is believed not only to play a role in the persistence of a particular strain in the digestive tract but also to participate in pathogen exclusion by competition and blocking of their binding sites at the mucosa (Collado et al. 2007; Lee et al. 2003; Vesterlund et al. 2006). Also, it may contribute to immunomodulation (Galdeano et al. 2007). However, some authors have hypothesized that attachment factors in lactic acid bacteria are risk factors that might be an indicative of their pathogenic potential (Vesterlund et al. 2007).

Lactobacilli can bind to mucin, a component of the mucus epithelial layer, and to a variety of proteins present in the extracellular matrix (ECM), such as fibronectin, collagen and laminin, which are shed into the mucus or can be exposed to the intestinal lumen in case of trauma, infection or inflammation (Lorca et al. 2002; Styriak et al. 2003). While in most cases protein factors have been identified as responsible for this attachment, with the exception of mucin-binding proteins from lactobacilli, information about specific binding proteins is still scarce for this group of micro-organisms (Vélez et al. 2007).

Fibronectin is a dimeric 454-kDa glycosylated protein, which is present in soluble form in plasma and in immobilized form on the host cells surfaces and in the ECM. It is an important target for bacterial attachment in many pathogens, such as Streptococcus pneumoniae and Streptococcus pyogenes, where fibronectin-binding proteins are important pathogenic factors (Holmes et al. 2001; Jedrzejas 2007; Molinari et al. 1997). There are numerous works describing the attachment of lactic acid bacteria to fibronectin, but information about molecules implicated in the mechanism of binding is limited. The surface layer protein (SlpA) from Lactobacillus brevis ATCC8287 is involved in fibronectin binding (de Leeuw et al. 2006). Moreover, inspection of lactobacilli genome sequences reveals that they carry genes-encoding proteins homologous to fibronectin-binding proteins from streptococci. Lactobacillus casei is a species widely used in the dairy industry. It is also a normal constituent of the intestinal microbiota in humans, and probiotic capacities have been reported for many strains, for which it is included as a probiotic in food products. Some L. casei strains survive the passage through the digestive tract in humans and persist in it for several days (Oozeer et al. 2006). In this work, we sought to analyse factors involved in interaction with host cells and colonization of intestinal mucosa in L. casei BL23. This strain has been widely used for genetic and physiology studies (Acedo-Félix and Pérez-Martínez 2003). It showed anti-inflammatory effects in animal inflammatory bowel disease models (Foligne et al. 2007), and its genome sequence is available. We describe the characterization of a gene encoding a fibronectin-binding protein that was detected during the in silico analysis of putative adhesion factors encoded in the L. casei BL23 genome.

Materials and methods

Strains and growth conditions

Lactobacillus casei strains are listed in Table 1 and were grown in Le Man, Rogosa and Sharpe (MRS) broth (BD Difco, Le Pont de Claix, France) at 37°C under static conditions. Escherichia coli DH5α was used for gene cloning, and E. coli M15[pREP4] was used for protein purification. Both strains were grown in LB medium under agitation (200 rev min−1) at 37°C. Ampicillin and kanamycin were used for E. coli at 100 and 25 μg ml−1, respectively. Erythromycin was used for L. casei at 5 μg ml−1. Solid medium was prepared by adding 1·8% agar. Bacterial growth curves were determined in microtitre plates (200 μl MRS broth per well) at 37°C in a Polarstar Omega plate reader (BMG Labtech, Offenburg, Germany).

Table 1. Lactobacillus casei strains used in this study
StrainOrigin/referenceSelected characteristics
  1. CECT, Colección española de cultivos tipo; ATCC, American type culture collection; CRL, Centro de referencia para lactobacilos; eryr, erythromycin resistant.

BL23CECT5275Laboratory strain; sequenced genome (Acedo-Félix and Pérez-Martínez 2003)
BL32CECT4040Cheese isolate
BL82ATCC25598Sour milk isolate
BL83CECT4043Cheese isolate
BL87ATCC11578Oral cavity isolate
BL90ATCC334Cheese isolate; sequenced genome
BL91ATCC4545Dental caries isolate
BL101Laboratory stockIsolated from commercial probiotic drink
BL106Laboratory stockIsolated from commercial probiotic drink
BL193Laboratory stockIsolated from commercial probiotic drink
BL199CRL 87Exopolysaccharide producer (Mozzi et al. 1996)
BL208Laboratory stockHuman intestinal isolate
BL212CRL686Dry cured sausage isolate (Fadda et al. 1998)
BL227Laboratory stockCommercial probiotic
BL229Laboratory stockCommercial probiotic
BL308BL23 fbpA::pRV300, eryrThis work

Construction of an L. casei BL23 fbpA mutant

A 600-bp internal DNA fragment from the fbpA gene (LCABL_16620) was amplified by PCR using oligonucleotides FBP1 (5′-CTTAAGCTTCGCAGCGTTGTTGC) and FBP2 (5′-TGAGGTACCTGGGCAACGGCATTAC), which introduced HindIII and KpnI restriction sites (underlined), using L. casei BL23 genomic DNA and EcoTaq DNA polymerase (Ecogen, Barcelona, Spain). The fragment was digested with HindIII and KpnI and cloned into the integrative vector pRV300 (Leloup et al. 1997) treated with the same enzymes. The resulting plasmid, pRVfbp, was transformed by electroporation into BL23 strain by using a Gene-Pulser (Biorad) as previously described (Posno et al. 1991) and transformants were selected in solid media by erythromycin resistance. Integration at the correct locus and fbpA disruption was checked by southern blot on HindIII-digested genomic DNA. The probe was the pRVfbp insert labelled with digoxigenin (DIG) with the PCR DIG-labelling mix (Roche). Hybridization and detection was performed in Hybond-N membranes (GE Healthcare) by using alkaline phosphatase-conjugated anti-DIG and the CDP-star chemiluminescent reagent as recommended by the manufacturer (Roche). The insertional mutation was shown to be stable for at least c. 40 generations in the absence of antibiotic (screening of 600 colonies after two consecutive overnight cultures gave a 100% of erythromycin resistants). Therefore, to discard interferences resulting from growth with antibiotics, bacteria used for the binding experiments in Fig. 2a and for growth curves were grown in the absence of erythromycin.

Figure 2.

 Binding of Lactobacillus casei to fibronectin. (a) Effect of protease treatment and fbpA disruption on L. casei binding to immobilized fibronectin. ( inline image ), control; ( inline image ), Proteinase K; (b) Inhibition of L. casei binding to immobilized fibronectin by soluble-added fibronectin. (inline image), BL23; (inline image), BL308; (c) Binding of L. casei to soluble fibronectin. Cells were incubated with fibronectin, washed and bound fibronectin was detected by Western blot. In parallel experiments, cells were treated with protease prior to fibronectin-binding assays. BL23 is wild-type L. casei; BL308 is the fbpA mutant strain. The bars represent standard deviations.

Microtitre plate binding assays

Binding of L. casei to immobilized human fibronectin (Sigma) was performed in 96-well Polysorp plates (Nunc) with bacterial cells grown to late exponential phase (OD550 of 3·5–4; 1·2 × 109 to 1·4 × 109 CFU ml−1). Plates were covered with 50 μg ml−1 of fibronectin in carbonate/bicarbonate buffer 50 mmol l−1 pH 9·6 at 4°C overnight. Wells were washed three times with PBS and blocked for 1 h with PBS plus 1% Tween 20. One hundred microlitres of each strain was added to each well in PBS adjusted to an OD550nm of 1 (7 × 108 CFU ml−1), and plates were incubated overnight at 4°C. After removing nonadhered cells by three washes with 200 μl of PBS plus 0·05% Tween 20 (PBST), the plates were dried, and adhered cells were detected by staining with crystal violet (1 mg ml−1 for 45 min). After washing, the colourant was released with citrate buffer 50 mmol l−1 at pH 4·0 (100 μl per well), and the absorbance at 595 nm was determined in a Multiskan Ascent plate reader (Thermo-Labsystems, Helsinki, Finland). The effect of protease treatment was assayed by incubating bacterial cells at an OD550nm of 1 (7 × 108 CFU ml−1) in PBS with 100 μg ml−1 of proteinase K (Roche) at 37°C for 1 h. After incubation, the protease was inactivated by addition of 1 mmol l−1 phenylmethylsulfonyl fluoride followed by three washes with PBS containing 1 mmol l−1 phenylmethylsulfonyl fluoride. Bacteria were resuspended in PBS to an OD550nm of 1 (7 × 108 CFU ml−1) and used for binding assays. Control bacterial cells were treated exactly as digested bacterial cells but without the addition of proteinase K. Inhibition of binding by soluble fibronectin was assessed by adding different quantities of fibronectin (1–10 μg per well) to the binding assay described earlier. Blank wells without bound fibronectin were run as controls in all experiments, and their absorbance values were subtracted from the values of wells covered with fibronectin. Experiments were carried out in triplicate three times with bacteria coming from independent cultures.

For whole-cell ELISA, L. casei bacterial cells were bound to immunoplates in PBS buffer at an OD550nm of 0·1 overnight at 4°C. The wells were washed and blocked with 2% BSA in PBS, and the content of the wells were reacted with a 1 : 200 dilution of anti-FbpA mouse serum or pre-immune serum followed by a 1 : 1000 dilution of peroxidase-conjugated anti-mouse IgG. Colour was developed with the 1-Step™ Ultra TMB-ELISA substrate (Pierce).

To assay binding to soluble fibronectin, L. casei bacterial cells at an OD550nm of 1 (7 × 108 CFU ml−1) were incubated with 100–500 ng of fibronectin in 1 ml of PBST containing 1% BSA for 1 h at 37°C. After three washes with PBST, bound fibronectin was released by boiling the bacteria in SDS–PAGE buffer and detected by Western blotting with a rabbit anti-fibronectin serum (Sigma).

Adhesion to solvents test

Microbial adhesion to solvent (MATS) test has been used to assess the surface properties of lactobacilli (Vinderola et al. 2004). This test was performed with L. casei essentially as described (Bellon-Fontaine et al. 1996). Five millilitres of overnight cultures of each strain was washed with PBS and resuspended in PBS to a final OD600nm of 0·4 (A0). This suspension was mixed (1 : 3) with different solvents (chloroform, ethyl-acetate or hexadecane) and vortexed for 1 min at full speed. After phase separation, absorbance of the aqueous phase was measured at 600 nm (A1). The percentage of adhesion was calculated from: %Adhesion = 100 × [1−(A1/A0)]. Each experiment was carried out in triplicate with bacteria coming from independent cultures.

Adhesion to Caco-2 and HT-29 cell lines

Epithelial cells were seeded at 4 × 104 cells cm−2 (Caco-2) or 2 × 105 cells cm−2 (HT-29) in 24-well plates in DMEM medium (with Glutamax, glucose 25 mmol l−1; Gibco) supplemented with 1% (v/v) nonessential amino acids solution (Gibco), 1% (v/v) sodium pyruvate solution (Gibco), 1% (v/v) sodium bicarbonate solution (only for HT-29 cells; Gibco), 1% (v/v) of antibiotics (100 U ml−1 penicillin, 100 μg ml−1 streptomycin; Gibco) and 10%(v/v) foetal calf serum and incubated at 37°C in a CO2 incubator. After the cells reached confluence (incubation for 6 and 3 days for Caco-2 and HT-29, respectively), plates were incubated for additional 15 (Caco-2) or 21 (HT-29) days to allow cell differentiation and the medium was changed every 2 days. Log-phase L. casei bacterial cells were added to each well in 0·5 ml of culture medium adjusted to an OD550nm of 0·2 (108 CFU ml−1), and the plates were incubated for 1 h at 37°C with mild agitation. Nonadhered bacteria were removed by washing three times with 1 ml of PBS, and the bacteria were detached by covering the monolayer with 200 μl of a 15% (v/v) solution of trypsin–EDTA (Gibco) in PBS. After addition of 300 μl of culture medium, serial dilutions were plated on MRS agar plates, and the bacterial colonies were counted after 48 h of incubation. The experiments were made in triplicate three times with bacteria coming from independent cultures. Adhesion was expressed as percentage of adhered bacteria with respect to input bacteria.

Purification of 6X(His)FbpA and binding assays

The L. casei BL23 fbpA gene was amplified with oligonucleotides FBP3 (5′-CGGGGATCCATGTCATTTGACGGAATC) and FBP4 (5′-ACGAAGCTTTTACTTGGTAGGCGGGTTGC) that included restriction sites (underlined) and Pfx DNA polymerase (Invitrogene). The amplified fragment was digested with BamHI and HindIII and cloned into the expression vector pQE30 (Qiagen) digested with the same enzymes. The plasmid construct was transferred to E. coli M15[pREP4], and cells of the transformed strain were grown in 500 ml of LB medium at 37°C until OD550 nm reached 0·6. Then, IPTG was added to 1 mmol l−1, and induction was carried out for 3 h at 37°C. Cells were collected by centrifugation, washed and resuspended in 10 ml of Tris–HCl 100 mmol l−1 pH7·4, lysozyme 1 mg ml−1, phenylmethylsulfonyl fluoride 0·5 mmol l−1, dithiothreitol 0·5 mmol l−1 and disrupted by sonication. The cellular debris were eliminated by centrifugation at 6000 g for 30 min at 4°C, the supernatant containing recombinant 6X(His)FbpA was applied to a Ni-NTA column (1 ml bed volume; Qiagen), and the recombinant protein purified according to the manufacturer’s instructions. Fractions containing 6X(His)FbpA were analysed using SDS–PAGE and dialysed overnight at 4°C in Tris–HCl 50 mmol l−1 pH8, EDTA 1 mmol l−1, NaCl 500 mmol l−1, glycerol 15% and stored at −80°C until use. Protein concentrations were determined with the BioRad dye-binding assay. To test in vitro fibronectin binding of FbpA, different protein amounts of 6X(His)FbpA were added in 100 μl of PBS buffer plus 0·1% BSA to microwell plates covered with fibronectin or BSA (50 μg ml−1 in carbonate/bicarbonate buffer 50 mmol l−1 pH 9·6, overnight at 4°C). After 1 h of incubation at 37°C, unbound protein was removed by washing three times with PBST and His-tagged FbpA was detected with the HisProbe™-HRP reagent (Pierce) and 1-Step™ Ultra TMB-ELISA (Pierce) as recommended by the manufacturer.

Preparation of antiserum to 6X(His)FbpA

Fifteen micro-grams of purified 6X(His)FbpA was intraperitoneally administered to 8-week-old female Balb/c mice (kept at the animal facilities of the University of Valencia) in 50 μl of PBS-containing adjuvant. Three doses were applied at 2-week intervals. Ten days after the last administration, mice were bled, and the presence of anti-FbpA antibodies in sera was tested by Western blot technique.

Isolation of cellular fractions and Western blot

Lactobacillus casei bacterial cells were grown in 50 ml of MRS to late exponential phase (OD550 nm of 3·5–4; 1·2 × 109 to 1·4 × 109 CFU ml−1) and washed two times with PBS. The pellet was resuspended in Tris–HCl 10 mmol l−1 pH8, LiCl 1·5 mol l−1 and incubated at 4°C for 1 h. Bacteria were pelleted by centrifugation at 6000 g for 10 min, and proteins in the supernatant were precipitated by adding trichloroacetic acid to 10% and incubation at 4°C for 1 h, followed by centrifugation at 10 000 g 20 min, washing with cold 96% ethanol and resuspension of the pellet in urea 7 mol l−1. The cell pellet was disrupted with glass beads (0·1 mm) in a Mini-Bead Beater (BioSpec Products, Bartlesville, OK, USA) with four cycles of 30 s at maximal speed, and unbroken cells were discarded by centrifuging the supernatant three times at 6000 g for 5 min. The supernatant was then centrifuged at 22 000 g, 20 min at 4°C. The soluble fraction was retained as the cytoplasm fraction, whereas the pellet was washed three times at 22 000 g for 15 min with Tris–HCl 50 mmol l−1 pH8 plus NaCl 0·5 mol l−1 and retained as the cell-envelope fraction (cell-wall/membrane fragments). To assess the effect of protease digestion on LiCl extraction of FbpA, the bacteria were treated before extraction with protease as described earlier. Samples of the different fractions were separated by 10% SDS–PAGE, and the gels were electro-transferred to Hybond-ECL membranes (GE Healthcare). FbpA was detected with a mouse anti-FbpA serum (1 : 5000) and the ECL-advance western blotting detection kit (GE Healthcare).

Statistical analysis

Results are indicated as means ± standard deviation. The significance of the difference of the means in experiments carried out with wild-type L. casei and the fbpA mutant was calculated using the Student’s t-test with the prism 4.0 software (Graph Pad Software, San Diego, CA, USA).

Results

Characterization of an L. casei BL23 strain mutated in fbpA

Inspection of the L. casei BL23 genomic sequence (GenBank FM177140) revealed the presence of a gene (LCABL_16620, designated fbpA from now), which encoded a protein showing homology to FBP54 (46% identity) or PavA (43% identity) from Streptococcus pyogenes and S. pneumoniae, respectively, two proteins which have been reported to mediate fibronectin binding (Courtney et al. 1994; Holmes et al. 2001). The product of fbpA was a 64-kDa protein that contained the typical pfam05833 and pfam05670 domains (Pfam database) present in a variety of bacterial fibronectin-binding proteins. To construct a mutant affected in fbpA, an internal fragment of the gene was cloned into the nonreplicative plasmid pRV300. Electroporation of this construct (pRVfbp) in BL23 yielded erythromycin-resistant clones in which the plasmid was integrated at the fbpA locus leading to a disruption of the gene (Fig. 1). One of such integrants was chosen and designated BL308 (fbpA::pRV300). Compared to the wild type, the fbpA mutant showed a reduced specific growth rate (0·320 ± 0·005 and 0·290 ± 0·01 h−1 for BL23 and BL308 strains, respectively; P = 0·0217). Both strains were able to bind fibronectin immobilized on immunoplates in a protease-sensitive manner, as treatment of the cells with proteinase K drastically reduced the binding (Fig. 2a). Interestingly, the presence of the fbpA mutation produced a 50% reduction in binding (P = 0·004) to immobilized fibronectin compared to the wild type (Fig. 2a). Adding soluble fibronectin to the binding assays resulted in a decrease of bacterial binding in both the wild-type and the fbpA mutants, suggesting that L. casei BL23 was interacting with immobilized as well as soluble fibronectin (Fig. 2b). This latter idea was confirmed by incubating L. casei cells with fibronectin. After several washings, fibronectin attached to the cell surface could be released and detected by immunoblotting (Fig. 2c). In these assays, no differences between the wild-type and the fbpA mutant were observed, and protease treatment reduced the binding in both strains (Fig. 2c). In conclusion, the attachment ability to immobilized as well as soluble fibronectin probably involved surface proteinaceous substances, and FbpA played a role in attachment to the immobilized form.

Figure 1.

 Construction of an Lactobacillus casei BL23 mutant in fbpA. (a) Wild-type strain was transformed with the nonreplicative pRVfbp plasmid, and erythromycin-resistant colonies were selected. (b) Integration at the correct locus was verified by Southern blot on HindIII-digested genomic DNA from wild-type strain (BL23) and the disrupted mutant (BL308).

Two other tests were performed to detect changes in cell surface characteristics induced by the fbpA mutation. First, we measured adhesion of the strains to cultured intestinal epithelial cells lines. A small but significant increase in adhesion to the HT-29 cell line was observed in the fbpA mutant strain with respect to the wild type (% adhesion of 3·03 ± 0·7 and 4·11 ± 1·8 for the wild-type and the fbpA mutant, respectively; P = 0·02), whereas no significant changes were detected in the binding ability to Caco-2 (% adhesion of 0·96 ± 0·6 and 1·3 ± 0·34 for the wild-type and the fbpA mutant, respectively; P = 0·167). As a second approach, we used the microbial adhesion to solvent test (MATS) with three different solvents: chloroform (acidic solvent and electron acceptor) ethyl-acetate (basic solvent and electron donor) and hexadecane (hydrophobic solvent). The results showed that the fbpA mutation did not induce changes in the acid–base characteristics of the cell surface; however, fbpA-disrupted cells showed a clear diminishing in their hydrophobicity, as reflected by a 70% decrease in the affinity for hexadecane (Fig. 3).

Figure 3.

 Cell-surface characteristics of Lactobacillus casei wild-type (BL23) and the fbpA mutant (BL308) measured by the MATS test. The bars represent standard deviations. (inline image, chloroform; inline image hexadecane; inline image, ethyl-acetate).

Lactobacillus casei FbpA binds to fibronectin

The fbpA gene was cloned in E. coli and FbpA was purified after expression as a His-tagged protein. When the purified protein was tested for binding to fibronectin immobilized on immunoplates, it was shown that 6X(His)FbpA bound to fibronectin in a dose-dependent and saturable manner. A low binding was observed when the immunoplates were covered with the control protein BSA (Fig. 4a). In inhibition experiments where soluble fibronectin was added to the binding reaction, a low inhibition (around 20%) in FbpA binding was only found at the highest fibronectin concentration (Fig. 4b). These results were in agreement with the previous characterization of the L. casei fbpA mutant and those reported for the S. pneumoniae protein (Holmes et al. 2001), which showed that FbpA preferentially binds to immobilized fibronectin.

Figure 4.

 Binding of purified FbpA. (a) Binding of 6X(His)FbpA to fibronectin immobilized on immunoplates (inline image, Fibronectin; inline image, BSA). After washing, the bound protein was detected with a His-tag detection reagent. (b) Inhibitory effect of soluble fibronectin on 6X(His)FbpA binding. The bars represent standard deviations (inline image, BSA; inline image, Fibronectin).

Cellular location of FbpA

To address the question whether FbpA was present at the L. casei cell surface several L. casei fractions were tested by Western blot against an anti-FbpA serum. Results showed that a 64-kDa band, the molecular weight of FbpA, was detected in all cellular fractions (surface proteins extracted with LiCl, cell-envelope proteins and cytoplasmic proteins, Fig. 5a). Additional unspecific bands were also shown to react with the antiserum. These bands were not detected in the LiCl fraction, indicating the absence of cross-contamination. The 64-kDa band disappeared in the fbpA-disrupted mutant (Fig. 5b), thus confirming its identity as FbpA. Similar amounts of extracted proteins were loaded onto each lane, which led us to the conclusion that most of FbpA was present intracellularly. Treatment of the cells with protease reduced the amount of FbpA extracted by LiCl treatment, reinforcing the idea that part of FbpA is surface-exposed and accessible to hydrolytic enzymes (Fig. 5c). However, similar to the rest of homologue proteins, no signal peptide responsible for protein secretion was identifiable in the FbpA sequence.

Figure 5.

 Detection of FbpA at the cell surface. (a) Western blot detection of FbpA in different cellular fractions of Lactobacillus casei wild type (BL23). One μg of extracted proteins was loaded onto each lane. LiCl, surface proteins extracted with LiCl; E, surface proteins from the cell-envelope (cell-wall/membrane fraction); C, cytoplasmic proteins; (b) Detection of FbpA in cell extracts (1 μg total protein per lane) of L. casei BL23 and the fbpA mutant (BL308); (c) The effect of protease digestion on the LiCl extraction of FbpA from L. casei BL23.

FbpA in other L. casei strains

We screened a collection of L. casei strains from different origins (food and human isolates, including probiotic strains, Table 1) for the presence of FbpA. As expected from the presence of a gene homologous to fbpA in its genome (LSEI_1439), the BL90 (ATCC334) strain showed a reacting protein band similar to BL23 (Fig. 5). FbpA homologue proteins were also extracted at different levels by LiCl treatment in the rest of L. casei strains and were also present in the corresponding surface fractions (Fig. 6). The cross-reacting bands varied in size, indicating that FbpA from different L. casei strains are not totally identical. Furthermore, Southern blot hybridization with an fbpA probe showed that a single copy of fbpA was present in all strains (data not shown). Whole-cell ELISA, in which the bacterial cells were bound to microtitre plates and probed with the anti-FbpA serum, led also to the detection of FbpA (Fig. 7). These results concluded that the presence of FbpA on the cell surface is a common feature in L. casei.

Figure 6.

 Detection of FbpA in different Lactobacillus casei strains. Equal amount of LiCl-extracted proteins (equivalent to the protein extracted from ten absorbance units (600 nm) of cells) (a) and one μg of cell-envelope proteins (b) were resolved by 10% SDS–PAGE and tested by Western blot with an anti-FbpA serum.

Figure 7.

 Whole-cell ELISA to detect the presence of FbpA on the surface of Lactobacillus casei strains. Bacterial cells were bound to the surface of immunoplates and an ELISA test was carried out with an anti-FbpA serum. Data presented are subtracted from the absorbance values of control wells incubated with a pre-immune serum. The bars represent standard deviations.

Discussion

In this work, we tried to get some insight into the mechanisms that mediate interactions of lactobacilli to host cells. To this end, we have characterized FbpA from L. casei, a protein homologous to fibronectin-binding proteins described in other bacteria. Genome search at the NCBI database revealed that all sequenced lactobacilli genomes encode FbpA homologues with amino acid identities ranging from 41% to 60% compared to L. casei FbpA. In the search for host adhesion factors in L. acidophilus NCFB, Buck et al. (2005) constructed a mutant in fbpA, which displayed a strong reduction in Caco-2 cells attachment. These experiments established that L. acidophilus FbpA participates in adhesion to epithelial cells. Nevertheless, no assays on fibronectin binding were carried out in this study. After all, our results showed that, compared to the wild type, L. casei fbpA strain adhered slightly better to HT-29 cells, whereas no significant changes were detected on Caco-2 adherence. This striking result suggests that changes in the bacterial surface resulting from an fbpA mutation (as evidenced by decreased surface hydrophobicity) lead to a slightly improved capacity to attach to HT-29 cell surfaces. However, the reason for this observation is not known. L. casei probably utilizes other FbpA-independent mechanisms for attachment to the HT-29 and Caco-2 cell lines or the contribution of FbpA to binding in these models is low. It has been reported that FBP54, an FbpA homologue from S. pyogenes, had minor effects on binding to some types of epithelial cells, whereas binding to others was strongly influenced by this adhesion (Courtney et al. 1996).

Similar to fbpA-mutated streptococci, where reduction of bacterial fibronectin binding ranged from 50% to 25% (Christie et al. 2002; Miller-Torbert et al. 2008), disruption of L. casei fbpA led to only a 50% reduction in binding. This strengthens the idea that adhesion is a multifactor process and suggests the presence of additional not yet identified fibronectin-binding molecules, presumably of proteinaceous nature. In any case, inspection of the BL23 genome does not reveal the presence of genes encoding other types of fibronectin-binding proteins.

Previous studies showed that some lactobacilli were able to bind to the immobilized but not to the soluble form of fibronectin (Lorca et al. 2002). We showed that L. casei BL23 can bind both forms of fibronectin, but FbpA is binding more efficiently to only the immobilized form. The FbpA proteins belong to an atypical group of fibronectin-binding proteins that lack the repetitive, secretion and cell-wall anchoring (LPXTG motif) sequences present in other characterized fibronectin-binding proteins (Jedrzejas 2007). Lack of conventional signal for secretion and anchoring is a common feature of numerous proteins that decorate the bacterial surface. In lactobacilli, many of the characterized attachment factors are surface ‘moonlighting’ proteins that are implicated in other processes. These include the elongation factor Tu (EF-Tu) (Granato et al. 2004), the heat shock protein GroEL (Bergonzelli et al. 2006) and glycolytic enzymes (Hurmalainen et al. 2007; Kinoshita et al. 2008; Ramiah et al. 2008). How these proteins are transported and localized at the cell surface is still unknown.

Although L. casei FbpA can be found at the cell surface, the vast majority of the protein was of intracellular location. Similar results were found for the homologue protein Fbp68 from Clostridium difficile (Hennequin et al. 2003). Despite of the fact that all FbpA-homologues characterized to date have a surface location, posses in vitro binding ability to fibronectin and are important virulence factors in pathogens (Dramsi et al. 2004; Holmes et al. 2001), some controversy exists about the exact role of FbpA. In S. gordonii, fbpA is clustered with a gene (cshA) encoding a distinct fibronectin-binding protein whose expression is down-regulated upon fbpA mutation, for which it was postulated that FbpA might play a role in the transcriptional regulation of adhesion factors (Christie et al. 2002). Likewise, a mutation in fbpA of Listeria monocytogenes reduces the amount of two virulence factors (listeriolysin O and InlB) acting at the post-transcriptional level and FbpA co-precipitates with them; therefore, it was postulated that it might function as a chaperone or an escort protein for these factors (Dramsi et al. 2004). The genome context of fbpA in L. casei BL23 does not allow prediction of putative functions for FbpA. The fbpA gene is monocistronic, and no adhesion-related genes can be found adjacent to it.

The pleiotropic effects of fbpA mutations largely differ between species. S. pneumoniae pavA (fbpA) mutants bound less to fibronectin and were attenuated in virulence; however, they showed no changes in cell surface physicochemical properties or in the expression of virulence factors (Holmes et al. 2001). A different situation was found in S. gordonii (Christie et al. 2002) and in L. casei BL23 fbpA mutants, where a clear decrease in the cell surface hydrophobicity was observed. In S. gordonii, this decrease was related to the lower expression of CshA (Christie et al. 2002). Further research is needed to disclose the changes produced by an fbpA mutation on the cell surface of L. casei BL23. Whether FbpA directly interacts with fibronectin in vivo or it modulates the expression and functionality of other interacting proteins, or both, is still unknown.

It has been reported that the adhesive capacity of lactobacilli to ECM proteins is not exclusively found in probiotics or human isolates (Vesterlund et al. 2007). Therefore, no link between attachment and probiotic character appears to exist. FbpA orthologues are encoded in the genomes of all sequenced lactobacilli, and the protein can be extracted from the cell surface of all L. casei strains tested in this work, from both human and food origins. While the exact role of FbpA in pathogen, commensal or probiotic bacteria is not yet understood, the study of this protein in lactobacilli may lead to a better understanding of the relations established between these bacteria and the intestinal epithelium.

Acknowledgements

This work was supported by projects AGL2004-00176/ALI and Consolider Fun-c-Food CSD2007-00063 from the Spanish Ministry of Science and Innovation. Diego Muñoz-Provencio was recipient of a predoctoral fellowship from the Conselleria de Cultura of the Generalitat Valenciana and of a research fellowship from the Instituto Danone. We thank Dr Javier Buesa and Rebeca Montava from the Microbiology Department of the University of Valencia for their help in obtaining anti-FbpA antibodies.

Ancillary