FbpA, a novel multifunctional Listeria monocytogenes virulence factor

Authors


Summary

Listeria monocytogenes is a Gram-positive intracellular bacterium responsible for severe opportunistic infections in humans and animals. Signature-tagged mutagenesis (STM) was used to identify a gene named fbpA, required for efficient liver colonization of mice inoculated intravenously. FbpA was also shown to be required for intestinal and liver colonization after oral infection of transgenic mice expressing human E-cadherin. fbpA encodes a 570-amino-acid polypeptide that has strong homologies to atypical fibronectin-binding proteins. FbpA binds to immobilized human fibronectin in a dose-dependent and saturable manner and increases adherence of wild-type L. monocytogenes to HEp-2 cells in the presence of exogenous fibronectin. Despite the lack of conventional secretion/anchoring signals, FbpA is detected using an antibody generated against the recombinant FbpA protein on the bacterial surface by immunofluorescence, and in the membrane compartment by Western blot analysis of cell extracts. Strikingly, FbpA expression affects the protein levels of two virulence factors, listeriolysin O (LLO) and InlB, but not that of InlA or ActA. FbpA co-immunoprecipitates with LLO and InlB, but not with InlA or ActA. Thus, FbpA, in addition to being a fibronectin-binding protein, behaves as a chaperone or an escort protein for two important virulence factors and appears as a novel multifunctional virulence factor of L. monocytogenes.

Introduction

Listeria monocytogenes is a ubiquitous Gram-positive, food-borne bacillus responsible for life-threatening infections in humans and animals. It is a facultative intracellular pathogen able to enter into a wide variety of host cells including intestinal epithelial cells, hepatocytes, endothelial cells and fibroblasts. The cellular and molecular basis of its intracellular cycle have largely been elucidated (for recent reviews, see Vazquez-Boland et al., 2001; Cossart, 2002). Infection occurs in several successive steps, which are entry of the bacterium into the host, lysis of the phagosomal vacuole, multiplication in the cytosol and direct cell-to-cell spread using actin-based motility. Each step requires expression of specific virulence factors. The major virulence genes are located in a cluster of genes encoding a regulatory protein PrfA, a phosphatidylinositol-specific phospholipase C (PlcA), a haemolysin called listeriolysin O (LLO), a metalloprotease (Mpl), an actin-recruiting protein (ActA) and a lecithinase (PlcB). On a different locus are encoded two proteins involved in invasion, InlA and InlB. Expression of virulence genes is co-ordinately and positively regulated by the transcriptional regulator PrfA (Kreft and Vazquez-Boland, 2001; Milohanic et al., 2003). There is overwhelming evidence that the primary L. monocytogenes virulence determinant is LLO, which is responsible for escape from a vacuole and thus entrance in the cytosol (Dramsi and Cossart, 2002). Mutants lacking LLO are completely avirulent in the mouse model of infection (Cossart et al., 1989).

Adherence of L. monocytogenes to the host cell surface is a critical event during infection. It involves a number of surface proteins, including InlA (internalin), ActA, Ami and p104. InlA is the first member of the internalin multigene family characterized by the presence of a leucine-rich repeat region (Gaillard et al., 1991). InlA promotes adherence and entry into cell lines expressing its receptor, the adhesion protein E-cadherin (Mengaud et al., 1996a). ActA, the surface protein required for actin-based motility, may also promote attachment via host cell proteoglycans (Alvarez-Dominguez et al., 1997; Suarez et al., 2001). Ami is an autolysin, with an amidase activity present on the surface of L. monocytogenes that contributes to bacterial adherence to host cells (Milohanic et al., 2001). Finally, the involvement of a cell surface protein of 104 kDa (p104) in the adhesion of L. monocytogenes to the human intestinal cell line Caco-2 has been proposed (Pandiripally et al., 1999).

Fibronectin is a dimeric glycoprotein (≈ 440 kDa) that is present in a soluble form in plasma and extracellular fluids and in a fibrillar form of higher molecular weight on cell surfaces. Although fibronectin has critical roles in eukaryotic cellular processes, such as adhesion, migration and differentiation, it is also a common substrate for the attachment of bacteria. The binding of pathogenic Streptococcus pyogenes and Staphylococcus aureus to epithelial cells via fibronectin facilitates their internalization (Fowler et al., 2000). The best characterized fibronectin-binding proteins from streptococci and staphylococci share a similar structural organization and mechanism of ligand recognition (Joh et al., 1999). These surface proteins are cell wall-anchored proteins with a signal peptide, an LPXTG motif and a fibronectin-binding domain that consists of 45-amino-acid-long repeats within the C-terminal region. Streptococci also express several atypical fibronectin-binding proteins. These proteins do not contain any conventional secretion signal, anchorage motif or typical fibronectin-binding sequences. Examples include Fbp54 in S. pyogenes (Courtney et al., 1994), PavA in Streptococcus pneumoniae (Holmes et al., 2001) and FbpA in Streptococcus gordonii (Christie et al., 2002).

In this work, signature-tagged transposon mutagenesis (STM) has allowed the identification of a new L. monocytogenes gene, fbpA, required for efficient colonization of host tissues. The fbpA gene encodes a 570-amino-acid polypeptide with strong homologies to the fibronectin-binding proteins PavA of S. pneumoniae, Fbp54 of S. pyogenes and FbpA of S. gordonii. We show here that FbpA is a fibronectin-binding protein present on the listerial surface that can mediate adherence to host cells but also modulates the protein levels of two virulence factors, LLO and InlB. These results point to the multiple contributions of FbpA to L. monocytogenes virulence.

Results

Isolation of an attenuated mutant by STM

Signature-tagged mutagenesis of L. monocytogenes strain EGD was performed to identify new genes important for the development of listeriosis in the mouse model using plasmid pID408t, which takes advantage of the transposition properties of Tn917, as described previously for S. aureus (Mei et al., 1997). Ninety-six tagged plasmids, each carrying a different tag, were introduced individually by electroporation into the wild-type strain EGD and allowed to generate banks of L. monocytogenes insertion mutants. Pools of 96 differently tagged mutants were used to infect 7- to 8-week-old Swiss mice intravenously. We used a dose of 106 bacteria per mouse. Infected mice died at day 4. In our protocol, mutants were recovered from infected spleens and livers (output) 72 h after infection and compared with the inoculated pool (input) by hybridization. Screening of ≈ 2000 mutants resulted in obtaining 12 attenuated mutants, among which one mutant (1D12) was non-haemolytic as a result of transposon insertion in the hly gene, one of the main listerial virulence genes. This mutant validated our approach. One attenuated mutant (3G11) was analysed further.

Virulence analysis of the 3G11 mutant

To quantify the degree of virulence attenuation of the 3G11 mutant, competition assays were performed in the mouse model of infection. Persistence of 3G11 mutant was compared with that of the isogenic wild-type strain in the spleen and liver of mice infected at a 1:1 ratio, using the intravenous route at an infecting dose of 106 bacteria per mouse. Bacteria recovered from spleens and livers at 48 h were enumerated. The in vivo competitive index (CI) was calculated by dividing the ratio of mutant to wild-type bacteria recovered from the organs by the ratio of mutant to wild-type bacteria inoculated into each animal. As shown in Fig. 1, the 3G11 mutant was attenuated at a level comparable to that of an inlB mutant strain in both liver and spleen. As expected, the hly mutant (1D12) is highly attenuated in liver and spleen. The growth rate of the 3G11 mutant strain was similar to that of the wild-type strain in rich medium at 37°C (data not shown).

Figure 1.

Virulence of the 3G11 mutant in the mouse model. Competitive indexes of 3G11 mutant in the liver and spleen of mice inoculated intravenously with equivalent numbers of wild-type and mutant bacteria (106 bacteria per mouse). hly and inlB mutants were used as controls.

As oral infections are more appropriate for the study of listeriosis, we infected orally non-transgenic and transgenic mice expressing the human E-cadherin in enterocytes (iFABP-hEcad mice) (Lecuit et al., 2001) with the 3G11 mutant. We followed bacterial invasion in the small intestine and translocation into the mesenteric lymph nodes, liver and spleen of mice over a 3 day period after oral infection with 1010 wild-type L. monocytogenes or fbpA mutant. In transgenic mice, a significant difference (100-fold) was observed in bacterial counts of the 3G11 mutant in the intestine and liver compared with the isogenic wild-type strain at 72 h after inoculation (Fig. 2A). In the mesenteric lymph nodes, a 10-fold difference was observed in bacterial counts between mutant and wild-type strains whereas, in the spleen, bacterial counts of wild-type EGD strain and 3G11 mutant were not significantly different. In contrast, oral infection of non-transgenic mice did not show any difference in colonization of host organs between wild type and 3G11 mutant (Fig. 2B). Taken together, these results indicate that the gene interrupted in the 3G11 mutant is involved in the intestinal and hepatic phases of listeriosis and hence represents a novel virulence factor.

Figure 2.

Virulence of the 3G11 mutant in transgenic mice.
A. Bacterial counts in the mesenteric intestine, lymph nodes, liver and spleen of iFABP-hEcad transgenic mice at various time points after oral infection with 1010 wild-type or mutant bacteria (mean of five mice per strain at each time point).
B. Bacterial counts in the mesenteric intestine, lymph nodes, liver and spleen of iFABP-hEcad transgenic mice and of non-transgenic mice infected orally with 1010 wild-type or mutant bacteria at 72 h after infection.

Localization of the transposon insertion in the 3G11 mutant and genetic organization of the fbpA locus

Southern hybridization confirmed the presence of a single copy of Tn917 inserted into the chromosome of 3G11 mutant (data not shown). Sequencing of the listerial DNA–Tn917 junction and comparison with the L. monocytogenes EGDe complete genome sequence (Glaser et al., 2001) showed that the Tn917 had inserted into an open reading frame (ORF) encoding a protein exhibiting strong homologies (≈ 60% similarity and 40% identity over ≈ 550 amino acids) to the fibronectin-binding proteins FbpA of S. gordonii, PavA of S. pneumoniae, Fbp54 of S. pyogenes and YloA of Bacillus subtilis (Supplementary material, Fig.S1). This gene, listed as lmo1829 in the EGDe genome (http://genolist.pasteur.fr/ListiList), was designated fbpA for fibronectin-binding protein A. It encodes a protein of 570 amino acids with a predicted molecular mass of 65161 Da and a pI of 6.68. Figure 3 depicts the genetic organization of the fbpA-containing region in L. monocytogenes strain EGDe (Glaser et al., 2001). Flanking fbpA (lmo1829) are two genes (lmo1828 and lmo1830) transcribed in opposite directions and encoding conserved proteins of unknown functions. Lmo 1828 is homologous to a putative stress-induced protein of the YicC family, and Lmo 1830 shows homology to a dehydrogenase. A putative transcription terminator is present downstream from fbpA making unlikely a polar effect of the Tn917 insertion on the downstream lmo1830 gene. The transposon insertion in mutant 3G11 was located between nucleotides 1097 and 1098 of the ORF. Interestingly, the gene encoding fbpA is also present in the closely related non-pathogenic Listeria species, L. innocua (designated lin1943 in L. innocua CLIP 11262 genome). Analysis of the primary and secondary structures of the protein did not reveal any putative signal peptide in the N-terminus or motifs (LPXTG signature, hydrophobic region, GW repeats) characteristic of surface proteins in Gram-positive bacteria.

Figure 3.

Genetic organization of the fbpA locus on the L. monocytogenes chromosome. The transposon (Tn917) insertion in mutant 3G11 is shown by a black arrow. ORFs are indicated according to the L. monocytogenes genome nomenclature (Glaser et al., 2001). The fbpA gene corresponds to lmo1829. The putative transcription terminator is indicated.

FbpA binds to immobilized fibronectin and mediates adherence to HEp-2 cells

A full-length recombinant C-tagged FbpA-6xHis protein was purified from Escherichia coli and shown to bind to human fibronectin immobilized onto microtitre plate wells (Fig. 4A). Binding of FbpA to fibronectin was saturable, and maximal binding levels were dependent upon the amounts of immobilized fibronectin. Note that bovine serum albumin (BSA) used as a control did not bind to fibronectin, nor did an unrelated His-tagged protein InlE-6xHis. FbpA did not bind to fibrinogen (data not shown).

Figure 4.

Binding of recombinant FbpA protein to immobilized human fibronectin and adherence of fbpA mutant to HEp-2 cells.
A. Saturation plots for binding of FbpA protein (1 mg l−1 and 5 mg l−1) to fibronectin (from Sigma) immobilized on to microtitre plate wells and showing no binding of BSA or InlE-6xHis (not shown) to fibronectin. Binding was measured by enzyme-linked immunosorbent assay (ELISA) using anti-FbpA polyclonal antibodies (dilution 1:3000) and horseradish peroxidase (HRP)-conjugated anti-rabbit secondary antibody (dilution 1:4000).
B. Adherence of L. monocytogenes wild-type EGD and isogenic fbpA mutant to HEp-2 cells in the absence (w/o) or presence of exogenous fibronectin (100 nM) was analysed by immunofluorescence as described previously (Milohanic et al., 2001) and expressed as the number of bacteria per cell (black bars) from ≈ 300 cells in two separate experiments. Human purified fibronectin (Sigma) was added to the media at the final concentration of 100 nM 1 h before bacterial infection. Insert. Equivalent amounts of HEp-2 and Caco-2 protein extracts were analysed by immunoblotting with anti-fibronectin antibodies (Sigma).

Most fibronectin-binding proteins are known to promote adherence to eukaryotic cells through their affinity for fibronectin acting as a bridge between bacteria and host cells (Fowler et al., 2000). The role of L. monocytogenes FbpA in bacterial adherence to human epithelial HEp-2 cell line was investigated. HEp-2 cells were chosen because they are deficient in fibronectin synthesis compared with the human intestinal epithelial cell line Caco-2 (see insert in Fig. 4B). As shown in Fig. 4B, adherence was reduced by ≈ twofold in the EGD:fbpA Tn917 mutant compared with the isogenic wild-type strain. Addition of exogenous fibronectin (100 nM) to HEp-2 cells increased the adherence of wild-type strain EGD, but not that of the fbpA mutant strain, resulting in a greater difference in adherence between wild type and the isogenic fbpA mutant strain. These results reinforce the idea that FbpA is a fibronectin-binding protein that contributes to L. monocytogenes adherence to the human epithelial HEp-2 cells.

FbpA is a listerial cell surface protein

To detect expression of FbpA protein by L. monocytogenes, FbpA-specific polyclonal antibodies were generated against the purified recombinant FbpA-6xHis. In addition, complementation of the fbpA mutant was carried out by transformation with the plasmid pFB1.5 carrying fbpA under the control of its own promoter.

Immunofluorescence microscopy with affinity-purified FbpA antibodies demonstrated the surface localization of the FbpA protein on both wild-type and complemented L. monocytogenes strains. As expected, the anti-FbpA immunoreactive species was absent from the surface of the fbpA mutant strain (Fig. 5A).

Figure 5.

Display of FbpA on the listerial surface.
A. The EGD wild-type (wt), fbpA mutant (mt) and the complemented mt(fbpA) strains were analysed by phase-contrast and immunofluorescence staining with affinity-purified anti-FbpA polyclonal antibodies.
B. Bacterial surface extracts prepared from exponentially growing bacteria at 37°C were analysed by immunoblotting with anti-FbpA antibodies.

The surface proteins from wild-type, fbpA mutant and complemented strains were prepared by solubilization with 1% SDS as described previously (Kocks et al., 1992) and subjected to Western blotting analysis using affinity-purified FbpA polyclonal antibodies. The antiserum recognized a band of ≈ 60 kDa in both wild-type and comple-mented strains (Fig. 5B). The 60 kDa band was absent in the fbpA mutant. Altogether, these results indicate that FbpA is expressed as a cell surface protein in L. monocytogenes.

FbpA associates with the bacterial cytoplasmic membrane and regulates the levels of LLO and InlB proteins

As the primary sequence of FbpA lacks the conventional secretion or anchorage signals, we analysed in which bacterial compartment FbpA was located using a previously described fractionation procedure (Jonquieres et al., 1999). Purity of the various fractions, i.e. cytosolic, membrane, cell wall and culture medium compartments, was verified by including in our analysis several proteins with known cellular locations. InlA, which is covalently bound to peptidoglycan, was used as a marker of the cell wall fraction. ActA, an integral surface protein, was used as a marker of the membrane fraction. InlB, an invasion protein, non-covalently associated with lipoteichoic acid, was used as a second marker of the membrane fraction. LLO, a secreted cytolysin, was used as a marker of the supernatant fraction. Finally, GroEL, a general chaperone, was used as an internal control. As expected from previous investigations (Dramsi et al., 1995; Lebrun et al., 1996), InlA was detected in the cell wall fraction and in the supernatant (Fig. 6B). ActA was detected mostly in the cytoplasmic and membrane fractions and, to a lesser extent, in the cell wall fraction. InlB was detected in the membrane and supernatant fractions (Jonquieres et al., 1999). LLO was detected in the supernatant. GroEL was present in the cytoplasmic and membrane fractions. The membrane localization of GroEL was unexpected. It probably results from adsorption of the protein on the inner face of the cytoplasmic membrane. FbpA was mainly located in the membrane fraction (Fig. 6A) and absent from the supernatant fraction. The mobility of the FbpA protein on SDS-PAGE was estimated at ≈ 60 kDa. As expected, the anti-FbpA immunoreactive species was absent in the fbpA mutant strain.

Figure 6.

Subcellular localization of FbpA and effect of fbpA inactivation on LLO and InlB levels. L. monocytogenes wild-type strain EGD (w) and fbpA mutant (m) cultures were fractionated into cytosolic (Cyt), membrane (Mb), cell wall (CW) and supernatant (SN) compartments and immunoblotted with specific anti-FbpA, anti-ActA, anti-InlA, anti-InlB, anti-LLO and anti-GroEL antibodies.
A. FbpA is associated with the membrane compartment.
B. Immunolocalization of known virulence factors (InlA, InlB, ActA and LLO) as controls.
C. The quantity of LLO found in the supernatant is reduced in the fbpA mutant and restored to wild-type level in the complemented strain.

Interestingly, we found that the quantity of InlB and of secreted LLO was reproducibly reduced in the fbpA mutant. To confirm this result, we investigated whether the wild-type level of LLO was restored in the supernatant of the complemented strain and whether the level of InlB was restored in the surface extracts of the complemented strain. As shown in Fig. 6C, the levels of both LLO and InlB were restored in the complemented strain.

Inactivation of fbpA does not affect the transcription of hly and inlB

To investigate whether fbpA inactivation affects hly and inlB transcription, semi-quantitative reverse transcription polymerase chain reaction (RT-PCR) analysis was performed on total RNAs extracted from L. monocytogenes wild-type and fbpA mutant strains grown in brain–heart infusion (BHI) at 37°C (OD600 = 0.3). As shown in Fig. 7, equal levels of inlB and hly transcripts were detected in wild-type and mutant strains. InlA and fbpA transcripts are also shown as controls. As expected, no fbpA transcript was detected in the mutant strain. In the absence of reverse transcriptase, no DNA fragments were amplified by PCR from L. monocytogenes RNA (data not shown), excluding the possibility of DNA contamination in the RNA preparation. These results strongly suggest that FbpA modulates the levels of LLO and InlB at a post-transcriptional stage.

Figure 7.

RT-PCR analysis of fbpA, hly, inlA and inlB transcription in wild-type (w) and fbpA mutant strains (m). Total RNA extracted from wild-type and mutant strains is shown on the left. Two DNA ladders are included, and the PCR products obtained for each gene corresponded to the predicted length. Semi-quantitative RT-PCR using serial dilution of the RNA template (1:250; 1:50; 1:10) and gyrA as a standard gene.

FbpA interacts with LLO and InlB

We investigated whether FbpA could interact physically with LLO and InlB in total extracts of wild-type L. monocytogenes. Immunoadsorptions were performed on total protein extracts of exponentially growing cells using 2 µg of affinity-purified anti-FbpA polyclonal antibodies, anti-LLO monoclonal antibody or affinity-purified anti-InlB polyclonal antibody. Immunocomplexes were captured using protein G-Sepharose beads and analysed by immunoblotting. As shown in Fig. 8, FbpA can interact with both LLO and InlB. No signal was detected with control proteins such as ActA or InlA, thus demonstrating a high degree of specificity. These results indicate that FbpA can interact specifically with at least two virulence factors of L. monocytogenes, LLO and InlB.

Figure 8.

FbpA forms a complex with LLO and InlB in wild-type L. monocytogenes. Immunoadsorptions were performed on total protein extracts of exponentially growing bacteria using 2 µg of affinity-purified anti-FbpA polyclonal antibodies, anti-LLO monoclonal antibody or affinity-purified anti-InlB polyclonal antibody (indicated on the top of each lane). Immunocomplexes were captured using protein G-Sepharose beads and analysed by immunoblotting. Antibodies used for immunoblotting are indicated at the right.

Discussion

We report here the characterization of FbpA, a novel L. monocytogenes surface protein involved in virulence. This protein, although lacking a typical signal peptide, is exposed on the listerial surface, binds to immobilized fibronectin and contributes to cell adherence. In addition, we show that FbpA expression can affect the protein level of two virulence factors, LLO and InlB.

Using STM (Hensel et al., 1995) with transposon Tn917, screening of ≈ 2000 mutants has led to the isolation of 12 attenuated mutants in the mouse model by the intravenous route (H. Fsihi and P. Cossart, unpublished results). One mutant (1D12) was highly attenuated (>3 logs): it was non-haemolytic on blood agar plate. The transposon insertion mapped into the hly gene that encodes LLO, the main listerial virulence factor, validating our screening procedure. We have characterized another attenuated mutant (3G11) in which the transposon had inserted in a gene encoding a protein homologous to fibronectin-binding proteins, therefore called FbpA. Competition assays revealed that the attenuation of the fbpA mutant was similar to that of an inlB mutant in the same experimental conditions, i.e. is affected in liver colonization. Virulence studies in a more relevant animal model for human listeriosis, i.e. oral infection of transgenic mice expressing the human E-cadherin in enterocytes (iFABP-hEcad mice), showed that the fbpA mutant was markedly altered in its capacity to colonize and survive in the intestine and the liver, two organs essential for the development of listeriosis. This result suggests that, during an oral infection, FbpA either synergizes the action of InlA, for example by mediating adherence to the intestinal cells, and/or is necessary downstream from the InlA site of action.

fbpA is maximally expressed in exponentially growing bacteria at 37°C like all the virulence genes identified so far in L. monocytogenes (data not shown). However, unlike the majority of known virulence genes (Milohanic et al., 2003), fbpA expression is not dependent upon the transcriptional activator PrfA (data not shown).

blast sequence analysis shows that FbpA exhibits strong homologies to atypical fibronectin-binding proteins such as PavA of S. pneumoniae (Holmes et al., 2001), Fbp54 of S. pyogenes (Courtney et al. (1994) and FbpA of S. gordonii (Christie et al., 2002). Components of the extracellular matrix, such as fibronectin, collagen, laminin and proteoglycans, are found in all eukaryotic tissues and constitute ideal microbial adherence targets that many intracellular and extracellular pathogens, including L. monocytogenes through its binding to heparan sulphate proteoglycans (Alvarez-Dominguez et al., 1997) and glycosaminoglycans (Jonquieres et al., 2001), have exploited for colonization of tissues and initiation of infection. In addition, five putative fibronectin-binding proteins expressed by L. monocytogenes have been reported using fibronectin overlays on listerial total extracts. Only one with an estimated mobility on SDS-PAGE at 55 kDa has been proved to be present at the bacterial surface (Gilot et al., 1999). It might be FbpA. It was thus predictable that Listeria uses these extracellular matrix compounds during infection.

The primary sequence of FbpA does not show any characteristic of a surface-exposed protein, i.e. no N-terminal signal peptide that might direct the export of the protein via the general secretion pathway, no transmembrane domain, no typical signature sequence that could explain cell surface association in L. monocytogenes (Cossart and Jonquieres, 2000). However, our immunoblotting and immunofluorescence data demonstrate unambiguously that FbpA is localized on the cell surface of L. monocytogenes. Orthologues of FbpA including PavA of S. pneumoniae and Fbp54 of S. pyogenes were also shown to be cell surface proteins (Courtney et al., 1996; Holmes et al., 2001). Moreover, immunizations with Fbp54 induce protective immune responses against S. pyogenes challenge in mice, indicating that Fbp54 is expressed in vivo and, in physiological conditions, behaves as a cell surface protein (Kawabata et al., 2001). Whether it is also the case for FbpA is unknown.

Fractionation studies showed that FbpA is associated with the bacterial cytoplasmic membrane. Moreover, FbpA could be extracted from listerial surface with 1% SDS, reinforcing the finding of its association with the membrane. It is the first time, to our knowledge, that the precise localization of this type of atypical surface protein from Gram-positive bacteria has been determined. To test whether mislocalization of LPXTG-containing proteins in the cell wall of L. monocytogenes could alter the subcellular localization of FbpA, we analysed the localization of FbpA in L. monocytogenes sortase A and B single and double mutants (Bierne et al., 2002; 2004). Expression and membrane localization of FbpA in the sortase mutants were identical to that in isogenic wild-type strain EGDe (data not shown).

Recently, a second secA machinery has been discovered in L. monocytogenes (Lenz et al., 2003). This pathway allows secretion of 17 listerial proteins amongst which 10 do not have a signal peptide. We therefore suspected that FbpA might be exported through this non-essential secretion pathway. It is indeed the case (S. Dramsi and P. Cossart, unpublished). Whether other FbpA-like proteins of Gram-positive bacteria also use this pathway is under current investigation.

The inactivation of fbpA correlated with a significant reduction in the protein levels of two virulence factors, LLO and InlB. Interestingly, the protein levels of InlA and ActA, two other virulence factors, were not affected by the fbpA mutation, suggesting that the effect of FbpA is specific and independent of PrfA, the central regulator of virulence genes in L. monocytogenes. Semi-quantitative RT-PCR analysis indicated that the regulation of hly and inlB by fbpA occurs at a post-transcriptional level. It is worth noting that inactivation of fbpA in S. gordonii led to a significant reduction in the expression of the major cell wall-anchored fibronectin-binding protein, CshA, albeit at the transcriptional level (Christie et al., 2002).

In view of these results, we propose that FbpA, in addition to being a fibronectin-binding protein, also acts as a ‘chaperone’ that can prevent degradation and/or increase the secretion mechanism of specific virulence proteins such as LLO and InlB. These findings raise the possibility of an indirect role for FbpA and the other homologous fibronectin-binding proteins of this family in virulence. Indeed, in S. pneumoniae, pavA mutants were strongly impaired (104-fold) in virulence in the mouse sepsis model, which was somewhat surprising as mutants in adhesins are expected to be moderately attenuated (10- to 100-fold). We hypothesize that, as in L. monocytogenes, PavA can modulate the expression of key virulence factors in S. pneumoniae. It is important to note that, in another STM study of type 3 S. pneumoniae, Lau et al. (2001) also isolated an attenuated mutant in pavA.

Thus, although the role of these FbpA proteins is not fully understood, they are all clearly important for infection.

Experimental procedures

Strains, plasmids and growth conditions

Brain–heart infusion (BHI; Difco Laboratories) and Luria–Bertani (LB; Difco Laboratories) broth and agar were used to grow Listeria and Escherichia coli strains respectively. The wild-type virulent strain of L. monocytogenes EGD (BUG 600) belongs to the serovar 1/2a. The EGD:fbpA Tn917 mutant (3G11) was put in the collection as BUG 1898. Strains harbouring plasmids were grown in the presence of the following antibiotics: pAT28 derivatives, 60 mg l−1 spectinomycin, pET22b+ derivatives, 100 mg l−1 ampicillin. L. monocytogenes strains carrying the Tn917 transposon were grown in the presence of 5 mg l−1 erythromycin.

General DNA techniques

Plasmid DNA from E. coli was prepared by rapid alkaline lysis using the QIAprep spin prep kit (Qiagen), and PCR products were purified using the QIAquick kit (Qiagen) according to the manufacturer's instructions. Genomic DNA from L. monocytogenes was prepared using the Boehringer–Roche high purification kit. Standard techniques were used for DNA fragment isolation, DNA cloning and restriction analysis (Sambrook et al., 1989). Restriction enzymes and ExpandTM High-fidelity Taq polymerase were purchased from Boehringer–Roche. Nucleotide sequencing was carried out by Genome Express.

Trans-complementation with fbpA

A 1987 bp fbpA fragment was amplified by PCR from EGD genomic DNA using the primers starting 256 bp before the start codon (5′-GGGGTACCCCAATTCGGAGTAGCGGTGG-3′) and 20 bp downstream from the stop codon (5′-ACATG CATGCATGTACGGCTCCACGAAGGAAC-3′), which contain KpnI and SphI sites respectively (underlined). After KpnI and SphI digestion, the PCR product was cloned into the shuttle vector pAT28 (Trieu-Cuot et al., 1990). The resulting plasmid, pFB1.5 in E. coli, BUG 2122, was introduced into L. monocytogenes mutant strain 3G11 by conjugation as described previously (Poyart and Trieu-Cuot, 1997) giving rise to BUG 1900. The construction was verified by sequencing the insert on both strands.

RNA isolation and RT-PCR assays

Bacteria (10 ml) were grown in BHI broth until OD600 reached 0.3. Bacteria were centrifuged, washed twice with 1 ml of 1× PBS and resuspended in 1 ml of Trizol (Invitrogen). Bacteria were broken using a FastPROTEIN BLUE tube (Bio101 Systems) and homogenized using a FastPrep instrument at a speed of 6.5 for 30 s. RNA was extracted with 300 µl of chloroform–isoamyl alcohol 24:1 (Sigma). After 10 min of centrifugation at 13 000 g, the aqueous phase was transferred to a tube containing 270 µl of isopropanol. Total RNA was then precipitated overnight at 4°C and washed with 1 ml of a 75% ethanol solution before suspension in RNase-free water. Contaminating DNA was removed by digestion with RNA-free DNase I according to the manufacturer's instructions (Roche), and RNAs were cleaned up using RNeasy protocol (Qiagen). RT-PCR analysis was carried out using the SuperscriptTM one-step RT-PCR system according to the manufacturer's instructions (Life Technologies). Sequences were as follows: inlB primers, forward AGCAGGGACGCG GATAACTGC and reverse CCTTGGTAGACCGATAGCT TATTC; hly primers, forward CCACGGAGATGCAGTGACAA ATG and reverse GACGGCCATACGCCACACTTGAG; inlA primers, forward CAACATTTAGTGGAACCGTGACGC and reverse GGGTCATAAGCGTTCATTGTACTTG; fbpA primers, forward CGCCAACTTGAGTACTCTTCTAG and reverse CGATTGAATCACCACATGCGAAC; gyrA primers, forward GTTGCCCGTGCCCTACCTG and reverse GCGCGACGGA TTCCGCTAC.

Expression and purification of recombinant FbpA-6xHis

A 1730 bp PCR fragment was produced using genomic DNA from EGD as template and the primers 5′-(GGAATTC) CATATGGCGTTTGATGCAATG-3′ and 5′-(CCG)CTCGAGAT TTTTTAACTCTAAAACGAGC-3′, which contain NdeI and XhoI sites respectively (underlined). The resulting fragment was digested with NdeI and XhoI and inserted in frame upstream from the His tag sequence in the expression vector pET22b+ (Novagen). The resulting plasmid, pFB2.6, was verified by sequencing the insert from both junctions. It was used to transform E. coli DH5α giving rise to BUG 1958 and E. coli BL21(DE3) (Novagen) as recommended by the manufacturer. Recombinant FbpA-6xHis was purified using a two-step chromatographic procedure. The first step of purification by metal affinity chromatography (Novagen) has been detailed elsewhere (Braun et al., 1998). The fractions containing FbpA-6xHis were pooled and subjected to anion exchange chromatography on a POROS HQ20 column (PerSeptive Biosystems) using a Biocad Sprint HPLC (PE Biosystem). The loading buffer was 50 mM Tris (pH 8) and 150 mM NaCl. Protein was then eluted by an NaCl gradient up to 1 M in the same buffer, and the UV detector at 280 nm monitored the protein presence. The protein fractions were pooled and concentrated on Centriprep 30 devices (Amicon). Protein concentrations were determined with the BCA system (Pierce).

Cell culture, adherence assays

The human HEp-2 epithelial cell line was cultured in Dulbecco's modified Eagle medium (DMEM with Glutamax, 25 mM glucose; Gibco Laboratories) supplemented with 10% fetal bovine serum (FBS; Bio West) and incubated in a 10% CO2 atmosphere at 37°C. For the assays, cells were seeded at 5 × 104 cells ml−1 on to 22 mm square glass coverslips in 24-well plates (adherence assays). Adherence assays were performed as described previously (Milohanic et al., 2001) with exponentially growing bacteria (0D600≈ 0.8). Human purified fibronectin (Sigma) was added to the media 1 h before bacterial infection.

Microscopy, fractionation and immunoblots

Generation of polyclonal antibodies against recombinant FbpA-6xHis and subsequent affinity purification were performed as described previously (Friederich et al., 1995). Immunofluorescence staining of FbpA was performed according to procedures described elsewhere (Lebrun et al., 1996) using polyclonal antibody (this work) and a fluorescein isothiocyanate (FITC)-conjugated secondary antibody (Biosys). For analysis of FbpA expression, bacterial proteins were separated into cytoplasm, membrane, cell wall and culture medium, according to a previously described protocol (Jonquieres et al., 1999). Bacteria were grown in rich medium (BHI) at 37°C and harvested for protein analysis during the exponential phase of culture. Equivalent amounts of all fractions, representing 100 or 200 µl of bacterial culture, were analysed by immunoblotting. Alternatively, cell surface extracts were prepared as described previously (Kocks et al., 1992). Proteins were boiled in Laemmli sample buffer, resolved on a 10% SDS-PAGE gel and transferred to nitrocellulose membrane. FbpA was detected using polyclonal antibodies and horseradish peroxidase (HRP)-coupled anti-rabbit secondary antibodies (Biosys) and the ECL kit (Amersham). The other primary antibodies used in this study were polyclonal antibodies raised against ActA (Cabanes et al., 2003), InlB (Braun et al., 1997) and GroEL (Sigma), and monoclonal antibodies raised against InlA (L7.7; Mengaud et al., 1996b) and LLO (Nato et al., 1991). Secondary HRP-conjugated goat anti-mouse antibodies were from Biosys.

Immunoblotting to detect fibronectin synthesis in HEp-2 and Caco-2 cell extracts was performed using polyclonal antibodies raised against fibronectin (Sigma).

Binding of FbpA to immobilized fibronectin

Binding of C-terminal 6xHis-tagged FbpA protein to fibronectin immobilized on to microtitre plate wells was measured essentially as described previously (Braun et al., 2000). Briefly, wells were coated overnight at 4°C with fibronectin (or BSA as a control) in the range 0.1–10 µg per well in PBS. After blocking with 1% BSA in PBS, the plates were incubated with 100 µl of various concentrations of FbpA-6xHis, BSA or InlE-6xHis diluted in PBS. Bound FbpA was detected using an anti-FbpA polyclonal antibody, and InlE-6xHis was revealed using the penta-His monoclonal antibody (Qiagen). Antibodies were detected using HRP-conjugated anti-rabbit or anti-mouse antibody and the chromogenic substrate 1,2-phenylenediamine dihydrochloride (Dako). The absorbance of the resulting colour development was measured at 492 nm.

Infection of mice

Competition experiments were performed by co-infecting the same animal (four mice) with equal amounts of wild-type and mutant strains (106 bacteria final) and sampling each strain at day 3 in both liver and spleen by plating on selective plates and counting the bacteria recovered from each organ. Determination of the competitive index is as follows:

image

where wt represents the number of wild-type bacteria and mt represents the number of mutant bacteria.

Oral infections were performed by infecting 8-week-old iFABP-hEcad transgenic mice with 1010 wild-type or mutant bacteria diluted in 0.5 ml of PBS containing 50 mg of CaCO3 (Sigma) and injected intragastrically with a needle (0.9 mm by 38 mm, ref 20GX1; 1/2′′ Popper) as described previously (Lecuit et al., 2001). Four animals for each time point were killed and dissected for determination of bacterial counts in ground organs.

Acknowledgements

We are extremely grateful to Edith Gouin for invaluable help in antibody preparation and purification, Renaud Jonquières for expertise in running the HPLC, Javier Pizzaro and Eliane Milohanic for assistance in the cellular assays, and Pierre Dehoux for bioinformatic analysis. We thank Ji-Min Mei for supplying the Tn917 tags, and David Holden for helpful discussions on STM. This work received financial support from the Pasteur Institute, the Ministère de l’Education Nationale, de la Recherche et de la Technologie (Programme de Microbiologie Fondamentale et Appliquée, Maladies Infectieuses, Environnement et Bioterrorisme) and the European Commission (contract QLG2-CT 1999-00932). P.C. is an international scholar from the Howard Hughes Medical Institute.

Supplementary material

The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/mmi/mmi4138/mmi4138sm.htm

Fig. S1. Multiple sequence alignment of FbpA-like proteins from different species.

Ancillary