FlaC, a protein of Campylobacter jejuni TGH9011 (ATCC43431) secreted through the flagellar apparatus, binds epithelial cells and influences cell invasion

Authors


Summary

Type III secretion systems identified in bacterial pathogens of animals and plants transpose effectors and toxins directly into the cytosol of host cells or into the extracellular milieu. Proteins of the type III secretion apparatus are conserved among diverse and distantly related bacteria. Many type III apparatus proteins have homologues in the flagellar export apparatus, supporting the notion that type III secretion systems evolved from the flagellar export apparatus. No type III secretion apparatus genes have been found in the complete genomic sequence of Campylobacter jejuni NCTC11168. In this study, we report the characterization of a protein designated FlaC of C. jejuni TGH9011. FlaC is homologous to the N- and C-terminus of the C. jejuni flagellin proteins, FlaA and FlaB, but lacks the central portion of these proteins. flaC null mutants form a morphologically normal flagellum and are highly motile. In wild-type C. jejuni cultures, FlaC is found predominantly in the extracellular milieu as a secreted protein. Null mutants of the flagellar basal rod gene (flgF) and hook gene (flgE) do not secrete FlaC, suggesting that a functional flagellar export apparatus is required for FlaC secretion. During C. jejuni infection in vitro, secreted FlaC and purified recombinant FlaC bind to HEp-2 cells. Invasion of HEp-2 cells by flaC null mutants was reduced to a level of 14% compared with wild type, suggesting that FlaC plays an important role in cell invasion.

Introduction

Campylobacter jejuni is a leading causative agent of bacterial enteritis worldwide (Tauxe, 1992; Altekruse et al., 1999). C. jejuni is a Gram-negative spiral bacterium with a single unsheathed flagellum at one or both poles of the body. The flagellum is responsible for the high motility of the organism as aflagellate mutants are non-motile (Caldwell et al., 1985; Morooka et al., 1985; Black et al., 1988; Wassenaar et al., 1991; Guerry et al., 1992; Grant et al., 1993; Yao et al., 1994). A number of studies have indicated that the polar flagellum has an important role in the colonization of the viscous mucous lining of the gastric intestinal tract and is an important virulence determinant (Caldwell et al., 1985; Morooka et al., 1985; Newell et al., 1985; Black et al., 1988; Pavlovskis et al., 1991). The basic structure of a bacterial flagellum consists of a filament connected via a universal joint (hook) to a basal body (consisting of a basal rod and three rings) embedded in the cell envelope (Macnab, 1996). The flagellar filament consists of several thousands of self-assembling protein (flagellin) monomers arranged in a helix. These form a hollow tube of a relatively constant diameter and variable length with an overall corkscrew morphology. The flagellum is assembled in a sequential fashion, with the basal body constructed first, followed by the hook and then the filament. The flagellar subunits (hook cap protein, hook protein, hook-associated proteins, flagellin and filament cap protein) are synthesized in the cytoplasm and then discharged into the tubular structure. These subunits are then transported to the tip of the growing axial structure where they are polymerized and assembled. A number of the Escherichia coli and Salmonella flagellar proteins (FlhA, FlhB, FliF, FliH, FliI, FliN, FliP, FliQ and FliR) that are required for flagella assembly are homologous to the conserved components of type III secretion systems (Hueck, 1998; Cornelis and Gijsegem, 2000). Eight flagellar proteins (FlhA, FlhB, FliH, FliI, FliO, FliP, FliQ and FliR) have recently been established as components of the flagellum-specific protein export apparatus (Minamino and Macnab, 1999; Macnab, 2003).

Type III secretion systems have been identified in bacterial pathogens of animals (enteropathogenic and enterohaemorrhagic Escherichia coli, Pseudomonas aeruginosa, Salmonella typhimurium and Shigella spp., Yersinia spp.) and plants (Erwinia spp., P. syringae, Rhizobium spp. Ralstonia solanacearum and Xanthomonas spp.) (Hueck, 1998; Galan and Collmer, 1999). The type III system transposes effectors and toxins directly into the cytosol of the host cells or into the extracellular milieu (Hueck, 1998; Galan and Collmer, 1999; Lee and Schneewind, 1999; Plano et al., 2001). These secreted proteins, unlike proteins of the type III secretion apparatus, are not well conserved among bacteria (Hueck, 1998). The type III secretion system is likely to have evolved from the flagellar export apparatus (Stephens and Shapiro, 1996; Galan and Collmer, 1999; Lee and Schneewind, 1999).

The first non-structural flagellar protein shown to be secreted by the flagellar secretion apparatus is the antisigma factor, FlgM (Hughes et al., 1993). FlgM negatively regulates the transcription of class III flagellar genes and is secreted to the extracellular milieu immediately after the hook basal body has been completed and is ready for flagellar filament assembly (Hughes et al., 1993). A Salmonella non-flagellar structural protein, FlgJ, a flagellum-specific muramidase essential for rod formation, is secreted to the periplasmic space via the flagellar secretion apparatus during the formation of the flagellum (Nambu et al., 1999). FliK, a soluble Salmonella protein involved in hook length control, like FlgM is secreted into the extracellular milieu via the flagellar export machinery (Minamino et al., 1999). FliK, in participation with a membrane protein (FlhB), mediates the switch from exporting hook-type to flagellin-type proteins (Minamino et al., 1999).

Phospholipase YplA, also a non-flagellar protein, has recently been shown to be secreted by the flagellar type III secretion system of Yersinia enterocolitica (Young et al., 1999; Young and Young, 2002). Similarly, the flagellar secretion system of Bacillus thuringiensis is involved in the secretion of non-flagellar proteins, haemolysin BL and phosphatidylcholine-preferring phospholipase C (Ghelardi et al., 2002). Konkel et al. (1999) reported the detection of eight secreted proteins when C. jejuni F3800 were co-cultured with host cells or in INT 407 cell-conditioned medium. One of the eight proteins, CiaB, encoded by ciaB, was shown to be transposed into the host cell cytoplasm. Mutants of ciaB were defective in secretion of the eight proteins and in invasion of INT 407 cells. The authors proposed the involvement of a type III secretion system and suggested that one or more of the secreted proteins affects invasion of the host cells. However, no type III secretion apparatus genes have been found in the complete genomic sequence of C. jejuni NCTC11168 (Parkhill et al., 2000).

In this study, we report the characterization of a 26 kDa protein (designated FlaC) of C. jejuni TGH9011. FlaC has sequence homology to the N- and C-terminal regions of flagellins (FlaA and FlaB) but lacks the central domain. It is not required for the formation of the flagellum or motility. FlaC is secreted through the flagellar export apparatus. Secreted FlaC during C. jejuni infection of HEp-2 cells, and purified recombinant FlaC, binds to epithelial cells and significantly affects cell invasion.

Results

flaC gene

FlaC, the gene product of flaC, has homology to various eubacterial flagellins particularly in the N- and C-terminal of the flagellins. A highly conserved flaC gene is present in the sequenced strain, C. jejuni NCTC 11168, encoding a 249-amino-acid residue polypeptide (Cj0720c) with 99% identity and 100% similarity to FlaC of TGH9011 (http://www.sanger.ac.uk/Projects/C_jejuni) (Parkhill et al., 2000). The flaC gene sequence of TGH9011 was submitted to GenBank in 1997 (accession number U85622).

Flagellum structure and cell morphology

Scanning electron microscopy was used to determine whether a morphologically normal flagellum was formed in the null flaC mutant or basal rod flgF mutant. Polar flagella of wild-type TGH9011 and the null flaC mutant cells have similar structure and length. flgF mutant cells (Chan et al., 1998) are aflagellate (Fig. 1). In MH media, wild-type TGH9011 and flaC mutant cells were both spiral in shape. Thus, the absence of the FlaC protein has no obvious effect on the structure of the flagellum or the shape of the cell.

Figure 1.

Electron micrographs of wild-type and C. jejuni TGH9011 mutant cells. Cells were grown in MH (A–C). The samples were negatively stained with 2% phosphotungstic acid (pH 6.5). Wild-type (A), flaC mutant (B) and flgF mutant (C). Bar = 1 µm.

Motility of flaC, flaA and flgF mutants

The motility of the flaC mutant was compared with wild type, flaA and flgF mutants using semi-solid MH agar plates. flaC mutant and wild-type cells have similar motility (Fig. 2), whereas the flgF (basal rod) mutant was non-motile. The flaA mutant exhibited a low level of motility as observed by Guerry et al. (1992).

Figure 2.

Motility of wild-type, flaC, flaA and flgF mutants. Motility assay was conducted using swarm MH medium containing 0.4% agar. Figures shown are representative of at least 10 separate determinations performed on different days.

FlaC and FlaA localization

Campylobacter jejuni TGH9011 wild-type and flaC mutant whole-cell lysates and flagella preparations were analysed by Western blotting with affinity-purified anti-FlaA antibodies (Khawaja et al., 1992) (Fig. 3A) and anti-FlaC antibodies (Fig. 3B). Using the anti-FlaA antibody, a strong band of about 60 kDa, corresponding to the FlaA flagellin, was observed in the wild-type flagella fraction (Fig. 3A, lane 3) and in the flaC mutant flagella fraction (Fig. 3A, lane 4). The 60 kDa FlaA band was observed in wild-type and flaC mutant whole-cell lysates with similar intensity (Fig. 3A, lanes 1 and 2).

Figure 3.

FlaC was not detected in flagella preparations.
A. Proteins were separated by SDS-10% PAGE, transferred to nitrocellulose membrane and probed with anti-FlaA antibodies. Wild-type whole-cell lysate (5 µl), lane 1; flaC mutant whole-cell lysate (5 µl),lane 2; wild-type flagella sample (10 µg), lane 3; flaC mutant flagella sample (10 µg), lane 4. Flagellin protein (FlaA), indicated by an arrow, was seen in all samples.
B. Proteins were separated by SDS-12% PAGE, transferred to nitrocellulose membrane and probed with anti-FlaC antibodies. Wild-type whole-cell lysate (5 µl), lane 1; flaC mutant whole-cell lysate (5 µl), lane 2; wild-type flagella sample (10 µg), lane 3; flaC mutant flagella sample (10 µg), lane 4. Wild-type whole-cell lysate gives a FlaC signal indicated by an arrow. A weak FlaC signal with a slightly higher molecular mass indicated by an arrow with asterisk is also visible in the wild-type flagella sample.

Western blot analysis using anti-FlaC antiserum identified a major band of ≈ 26 kDa (indicated by an arrow with no asterisk) in the wild-type whole-cell lysate fraction, which was the predicted molecular mass of FlaC (Fig. 3B, lane 1). The 26 kDa band was not detected by the anti-FlaC antiserum in the wild-type flagella fraction. Instead, a weak band with an approximate molecular mass of 27 kDa was observed (Fig. 3B, lane 3). The 27 kDa band is presumably the FlaC protein with an increase in size possibly resulting from post-translational modification. The 26 or 27 kDa band was also not observed in the flaC mutant whole-cell lysate (Fig. 3b, lane 2) or in the flagella fraction of the flaC mutant (Fig. 3B, lane 4). The above results confirm that the 26 kDa band is the product of the flaC gene. The above findings show that FlaA synthesized by wild-type C. jejuni cells was found in high levels in the flagella fraction. In contrast, the FlaC protein was detected only in trace amounts in the flagella fraction. The above findings show that FlaC is unlikely to be a significant structural protein of the flagellum but probably exists primarily in another cellular fraction.

FlaC proteins are found primarily in the extracellular fraction

Further cell fractionation and Western blot analysis was used to determine where FlaC is localized and whether it is secreted to the extracellular milieu. For each cellular fraction, equal amounts of protein were loaded in each lane for wild-type and flaC mutant cells. FlaC protein was detected in the wild-type whole-cell lysate, extracellular protein fraction and cytoplasmic/periplasmic protein fraction (Fig. 4, lanes 1, 3 and 5), but was not detected in the membrane/flagella fraction (Fig. 4, lane 7). In wild-type whole-cell lysates, FlaC has a molecular mass of 26 kDa (Fig. 4, lane 1). Two FlaC bands, one ≈ 27 kDa and the other ≈ 22 kDa were seen in the wild-type extracellular protein fraction (Fig. 4, lane 3). The 22 kDa band is probably a degraded product of FlaC. In the wild-type cytoplasmic/periplasmic fraction, the same 22 kDa protein band can be seen (Fig. 4, lane 5). FlaC, as expected, was not observed in any of the cell fractions of the flaC mutant (Fig. 4, lanes 2, 4, 6 and 8).

Figure 4.

FlaC proteins were found primarily in the extracellular fraction. Western blot analysis of different cell fractions was performed. Proteins were separated by SDS-10% PAGE, transferred to nitrocellulose membrane and probed with anti-FlaC antibodies. Whole-cell lysates were quantified by OD600. Equivalent amounts of wild-type and flaC mutant whole-cell lysates were loaded. Wild-type whole-cell lysate (5 µl), lane 1; flaC mutant whole-cell lysate (7 µl), lane 2; wild-type extracellular proteins (0.6 µg), lane 3; flaC mutant extracellular proteins (0.6 µg), lane 4; wild-type cytoplasmic/periplasmic proteins (2 µg), lane 5; flaC mutant cytoplasmic/periplasmic proteins (2 µg), lane 6; wild-type membrane/flagella proteins (4 µg), lane 7; flaC mutant membrane/flagella proteins (4 µg), lane 8.

The levels of FlaC in the secreted and whole-cell fractions of wild-type cells were compared by loading an equal number of cells into the appropriate lanes (Fig. 5). Similarly, the levels of FlaC in the cytoplasmic/periplasmic and membrane/flagella fractions were compared. FlaC was detected in whole-cell lysates (Fig. 5A, lane 1), cytoplasmic/periplasmic (Fig. 5A, lane 5) and extracellular (Fig. 5A, lane 3) fractions. FlaC synthesized by wild-type cells are found primarily in the extracellular fraction (Fig. 5A, compare lanes 1 and 3), indicating a high level of secretion of FlaC into the extracellular milieu. Non-secreted FlaC molecules were found predominantly in the cytoplasmic/periplasmic fraction but were not detected in the membrane/flagella fraction (Fig. 5A, lanes 5 and 7). As noted earlier, no FlaC cross-reacting band can be seen in the flaC mutant whole-cell lysate (Fig. 5A, lane 9).

Figure 5.

FlaC secretion requires a functional basal body. FlaC and FlaA Western blot analyses were used to compare different cell fractions of wild-type and flgF mutant.
A. Proteins were separated on SDS-10% PAGE, transferred on to nitrocellulose membrane and probed with anti-FlaC antibodies. Equal volumes of proteins in different cell fractions of wild-type and flgF mutant were loaded. Wild-type whole-cell lysate (5 µl), lane 1; flgF mutant whole-cell lysate (5 µl), lane 2; wild-type extracellular proteins (5 µl), lane 3; flgF mutant extracellular proteins (5 µl), lane 4; wild-type cytoplasmic/periplasmic proteins (5 µl), lane 5; flgF mutant cytoplasmic/periplasmic proteins (5 µl), lane 6; wild-type membrane/flagella proteins (5 µl), lane 7; flgF mutant membrane/flagella proteins (5 µl), lane 8; flaC mutant whole-cell lysate (8 µl), lane 9. Equivalent amounts of proteins from whole-cell lysate of wild type (lane 1) and flaC mutant (lane 9) were loaded. FlaC signals are indicated by arrows with and without an asterisk. The probable degraded FlaC products are indicated by arrows with two asterisks.
B. Proteins separated on SDS-10% PAGE, transferred to nitrocellulose membrane were probed with anti-FlaA antibodies. The sample lanes and the amount loaded in each lane are the same as in (A).

FlaC is secreted via the flagellar secretion apparatus

The flgF gene (Chan et al., 1998) codes for a proximal basal rod protein, FlgF. An flgF mutant would not be expected to form a functional hook–basal body structure. In C. jejuni flgF mutant cells, FlaC was detected in whole-cell lysates (Fig. 5A, lane 2) and cytoplasmic/periplasmic (Fig. 5A, lane 6) fractions but was undetectable in the extracellular fraction (Fig. 5A, lane 4) or membrane fraction (Fig. 5A, lane 8). Thus, FlaC is produced in the flgF mutant but is not secreted into the extracellular milieu. A null flagellar hook (flgE) mutant was also constructed, and mutation was verified by polymerase chain reaction (PCR). The flgE mutant, like the flgF mutant, was unable to secrete FlaC (data not shown). The C. jejuni flgE gene used in this study was first cloned and characterized by Lüneberg et al. (1998), corresponds to Cj 1729c of NCTC11168 and was annotated as flgE2 by Parkhill et al. (2000). In this paper, we refer to it as flgE as in Lüneberg et al. (1998).

The same cellular fractions of wild-type and flgF mutant cells were also analysed with an anti-FlaA antibody. FlaA, the major flagellin, was detected at high levels in wild-type membrane/flagella fractions (Fig. 5B, lane 7) and in wild-type whole-cell lysates (Fig. 5B, lane 1). Low levels of FlaA were detected in the wild-type secreted fraction (Fig. 5B, lane 3). FlaA was detected at high levels in flaC mutant whole-cell lysates (Fig. 5B, lane 9). In the flgF mutant, FlaA was detected at low levels in the whole-cell fraction (Fig. 5B, lane 2) but was not detectable in other cell fractions (Fig. 5B, lanes 4, 6 and 8). The synthesis of FlaA is presumably repressed in the flgF (a class II gene) mutant. Mutations that affect the expression of class II flagellar genes are known to reduce the expression of a flagellin gene (class III) in S. typhimurium (Macnab, 1996). A similar mechanism may occur in C. jejuni, resulting in lower levels of FlaA.

To ensure that the presence of high levels of FlaC in the medium was not the result of extensive cell lysis, we monitored the presence of a cytoplasmic protein, Fur (Chan et al., 1995), by Western blotting using anti-Fur antibodies. A Fur (18 kDa) band was detected in the cytoplasmic/periplasmic fraction of wild-type and flgF mutant cells (Fig. 6A). This 18 kDa band was not observed in the extracellular fraction of wild-type and flgF mutant cells, indicating an insignificant amount of cytoplasmic protein contaminants in the extracellular fraction. Similarly, Western blot analysis was performed using anti-CipA antibodies raised against another cytoplasmic protein, CipA (Joe et al., 1998). CipA was detected at high levels in the cytoplasmic fraction, and only a trace amount was observed in the extracellular fraction of wild-type and flgF mutant cells (Fig. 6B), again indicating that there was little cytoplasmic protein contamination of the extracellular fraction resulting from cellular lysis (Fig. 6).

Figure 6.

Absence of major cell lysis. Proteins in different cell fractions were separated on SDS-10% PAGE, transferred to nitrocellulose membrane and probed with anti-Fur antibodies (A) or anti-CipA antibodies (B). Wild-type cytoplasmic proteins (1 µg), lane 1; wild-type extracellular proteins (1 µg), lane 2; flgF mutant extracellular proteins (1 µg), lane 3.

Detection of secreted FlaC by immunofluorescence microscopy

Secretion of FlaC by wild-type and flaC mutant cells was analysed using immunofluorescence microscopy. HEp-2 cells were infected with wild-type C. jejuni or flaC mutant [multiplicity of infection (MOI) = 50] for 3 h. Infected cells were fixed with paraformaldehyde, permeabilized using 0.1% Triton X-100 and blocked with 1% BSA–PBS. Cells were then incubated with rabbit anti-FlaC antibody. FlaC was observed on the wild-type C. jejuni-infected HEp-2 cell surface. No FlaC was detected in flaC mutant-infected HEp-2 cells (Fig. 7).

Figure 7.

Secretion of FlaC in C. jejuni infection of HEp-2 cells. HEp-2 cells were infected with wild-type C. jejuni TGH9011, flaC or flgF mutant. C. jejuni cells were labelled by FITC (shown in green). HEp-2 actin filaments were stained by Texas red–X-phalloidin. FlaC (in blue) was detected by primary rabbit anti-FlaC antibody and cy5-coupled goat anti-rabbit IgG (secondary antibody).

GST-FlaC binds to HEp-2 cell surface

To characterize further the localization of FlaC on the cell surface of HEp-2, we examined the binding of purified fluorescein isothiocyanate (FITC)-labelled GST-FlaC and GST to HEp-2 cells. Bound GST-FlaC or GST alone was probed with polyclonal anti-GST-FlaC antibodies from rabbits followed by FITC-conjugated secondary antibody. HEp-2 cells were then permeabilized with 0.1% Triton X-100, and F-actin was visualized by staining with Texas red-conjugated phalloidin. Bright fluorescence was observed on the cell surface of HEp-2 cells treated with FITC-labelled GST-FlaC, whereas only a background level of fluorescence was observed on FITC-labelled GST-treated HEp-2 cells (Fig. 8). Binding of FITC-labelled GST-FlaC on to HEp-2 cells was blocked with 10-fold excess of unlabelled GST-FlaC (Fig. 8, bottom).

Figure 8.

Immunofluorescence microscopy shows that FlaC binds to the surface of HEp-2 cells. HEp-2 cells were treated with 0.05 µg µl−1 FITC-labelled GST (centre) or FITC-labelled GST-FlaC (top). Fixed HEp-2 cells were incubated with GST-FlaC (0.5 µg µl−1) for 1 h before incubating with 0.05 µg µl−1 FITC-labelled GST-FlaC. Actin filaments were stained with Texas Red–X phalloidin.

flaC null mutant is defective in invasion of epithelial cells

The binding of secreted FlaC to HEp-2 cells led us to ask whether binding of FlaC has an effect on adherence and/or invasion of the epithelial cells. Table 1 summarizes the findings of nine independent adherence and invasion assays using HEp-2 cells that compared wild-type C. jejuni and flaC null mutant cells. The flaC null mutants adhered equally well to HEp-2 cells compared with wild type, but invasion levels were reduced to 14% that of wild-type cells. Pairwise comparisons of strain invasion were done using two-sample t-tests. The difference between the mean invasion level of TGH9011 and that of TGH9011flaC is highly significant (P < 0.001). A null mutant (orf1::kan) of orf1, an open reading frame (designated as Cj0719c in the sequenced strain NCTC11168) located immediately downstream of the flaC gene, was also constructed by gene replacement, and this mutant, compared with wild-type cells, had similar levels of adherence to and invasion of HEp-2 cells (data not shown). Thus, the defect in invasion of the flaC mutant was not likely to be attributed to a polar effect on the orf1 (Cj0719c) gene by the insertion mutation at the flaC gene. Furthermore, Northern analysis using a flaC-specific probe identified a strong hybridizing RNA band of 1.0 kb (data not shown). The size of the transcript suggests that the flaC gene is transcribed as a monocistronic mRNA. Cj0719c is homologous to jhp0986 of the Helicobacter pylori J99 genome (Alm et al., 1999) and encodes a hypothetical protein of unknown function.

Table 1. Null mutant of flaC is defective in invasion of HEp2 cells.
StrainAdherenceaInvasionb
  • a. 

    Adherence levels were calculated as a percentage of input bacteria adhered after extensive washing without gentamicin treatment. The means and standard errors for adherence were calculated from nine independent experiments for TGH9011 and TGH9011flaC and four for TGH9011TC and TGH9011hipO. Pairwise comparisons of strains show no difference in adherence at the 0.05 significance level.

  • b. 

    Invasion levels were expressed as a percentage of adhered bacteria surviving after gentamicin treatment. Means and standard errors for invasion were calculated from nine independent experiments for TGH9011 and TGH9011flaC and four for TGH9011TC and TGH9011hipO. Pairwise comparisons of strain invasion were done using two-sample t-tests. The difference between the mean invasion level of TGH9011 and that of TGH9011flaC is highly significant (P < 0.001). The difference between that of TGH9011flaC and TGH9011TC is also highly significant (P < 0.001). The difference between that of TGH9011 and TGH9011TC is not significant (P = 0.16). Similarly, the difference between that of TGH9011TC and TGH9011hipO is not significant (P = 0.69).

  • c. 

    The percentages shown within the brackets were calculated relative to the value of the parental strain TGH9011.

TGH90110.95 ± 0.07 (100%)c1.32 ± 0.14 (100%)
TGH9011flaC0.91 ± 0.13 (96%)0.18 ± 0.02 (14%)
TGH9011TC1.11 ± 0.16 (117%)0.98 ± 0.13 (74%)
TGH9011hipO0.96 ± 0.18 (101%)1.08 ± 0.18 (82%)

If the defect in invasion observed results from the insertion mutation in the flaC gene, the defect should be corrected by a wild-type flaC gene inserted in trans. TGH9011TC (flaC hipO::CmflaC) was constructed to test this hypothesis. In this TGH9011 derivative, its chromosome contains a single copy of the flaC mutation and a copy of the wild-type flaC gene inserted into a non-essential gene, hipO (Hani and Chan, 1995). Insertion of the chloramphenicol and wild-type flaC gene (CmflaC) was confirmed by PCR, and expression of FlaC in TGH9011TC was similar to that of TGH9011 as shown by Western blot analysis (data not shown). TGH9011 hipO mutant was constructed previously (Hani, 1997). Like the flaC mutant, TGH9011 hipO and TGH9011TC have similar levels of adherence to HEp-2 compared with wild-type TGH9011. Invasion levels of TGH9011TC are 74% that of wild-type TGH9011 and similar to that of TGH9011 hipO (82%) (see Table 1). C. jejuni hipO mutants have been observed to have normal levels of adhesion to and invasion of epithelial cells (Steel, 2002). These results showed that the defect in invasion observed for the TGH9011 flaC mutant results from the flaC insertion mutation, which can be complemented by a single copy of the wild-type gene.

Discussion

FlaC is a conserved protein in C. jejuni. A homologous sequence of flaC is present in Campylobacter coli, Campylobacter lari and Campylobacter upsaliensis as demonstrated by Southern blot analysis (data not shown). C. coli strains tested also secrete a FlaC-like protein as demonstrated by Western blot analysis (data not shown). The H. pylori genome does not have a FlaC orthologue (Parkhill et al., 2000). Thus, the flaC gene was evolved early in the speciation of Campylobacter but after the branching of Helicobacter. The observation that the flaC null mutant has a normal flagellum structure and is highly motile indicates that FlaC is not required for the formation of a functional flagellum. FlaC is a secreted protein, and secretion requires a functional hook basal body as the basal rod protein mutant (flgF) and the hook protein mutant (flgE) were unable to secrete FlaC. Presumably, a functional flagellar type III secretion apparatus cannot be formed in the flgF or flgE mutant.

The secreted form of FlaC is 27 kDa while the cytoplasmic form is 26 kDa, suggesting post-translational modification. C. jejuni has N- and O-linked protein glycosylation systems (Szymanski et al., 2003). Over 30 secreted proteins of C. jejuni are subjected to N-linked glycosylation, and the glycans are linked to proteins through asparagine in the motif Asn-Xaa-Ser/Thr (Young et al., 2002). TGH9011 FlaC has five such motifs. The major C. jejuni flagellin protein, FlaA, is extensively glycosylated with 19 glycosylation sites either serine or threonine residue, all confined to the central variable domain of the flagellin (Thibault et al., 2001). FlaC is also subjected to degradation or proteolytic processing as a 22 kDa form was found in both the extracellular fraction and the membrane/periplasmic fraction. We have not yet determined whether the 22 kDa form binds to mammalian cells or has any biological function. Our future plans include determining the molecular nature of the post-translational modification of FlaC and the structure and function of the 22 kDa product.

YplA, a phospholipase virulence factor in Y. enterocolitica, was shown to be secreted by the flagellar secretion apparatus and two other type III secretion systems (Young et al., 1999; 2002). Presumably, Yersinia uses the flagellar type III system to secrete proteins before the acquisition of the type III secretion systems by horizontal gene transfer. As no non-flagellar type III secreted apparatus genes were identified in the C. jejuni NCTC11168 genome (Parkhill et al., 2000), we propose that Campylobacter uses the flagellar export apparatus to secrete flagellar and non-flagellar proteins. C. jejuni F38011 secrete eight proteins, ranging in size from 12.8 to 108 kDa (Konkel et al., 1999). One of these proteins is CiaB, a 73 kDa protein, and the encoding gene ciaB was cloned and characterized (Konkel et al., 1999). FlaC, described in this study, has a similar size to one of the eight secreted proteins of C. jejuni F38011. Null ciaB mutants of C. jejuni F38011 are defective in secretion of all eight proteins and in invasion of host cells, suggesting that one or more of the secreted proteins may be involved in stimulating uptake of bacteria (Konkel et al., 1999). We constructed a TGH9011 ciaB null mutant, and the mutant secretes normal levels of FlaC (data not shown), suggesting that FlaC is not one of the eight secreted proteins observed in C. jejuni F38011. Regulation of expression of FlaC in TGH9011 is clearly different from that of the eight Cia secreted proteins observed by Konkel et al. (1999). Unlike the Cia proteins, synthesis of FlaC in TGH9011 does not require contact with epithelial cells or culturing in conditioned medium. Bleumink-Pluym and van Putten (2003) reported at the 12th International Workshop on Campylobacter, Helicobacter and Related Organisms that the flagellar export apparatus of C. jejuni 81116 is involved in protein secretion. A ciaB mutant constructed using this strain showed no defect in secretion of proteins through the flagellar secretion apparatus. C. jejuni genomes from different strains are very diverse (Dorrell et al., 2001) and have very different capability to invade mammalian cells (Kopecko et al., 2001); thus, different strains are likely to have differences in the complement of secreted proteins. For instance, type IV secreted proteins present in 81-176 are absent in most C. jejuni isolates including the sequenced strain NCTC11168 (Bacon et al., 2000).

The flaC null mutant of TGH9011 is normal in adherence to HEp-2 cells but is defective in cellular invasion. The mechanism of C. jejuni cell invasion is not well understood. Hu and Kopecko (1999) proposed that a successful Campylobacter ligand–host cell receptor interaction activates dynein bound in caveolae leading to invagination of the dynein-bound membrane, resulting in engulfment of the adjacent bound external bacterium. It is possible that binding of FlaC on to the cell surface of HEp-2 triggers the invagination of the C. jejuni-bound membrane. Further work is required to identify the host cell protein(s) that interact(s) with FlaC during the C. jejuni invasion process.

Experimental procedures

Bacterial strains and plasmids

Campylobacter jejuni TGH9011 (ATCC 43431) is a reference strain for heat-stable Penner serotype O:3 and was obtained from J. L. Penner (University of Toronto). The pBluescript library of C. jejuni contains inserts ranging between 4 and 9 kb. Escherichia coli JM101 was used as the host for most plasmid isolations, and plasmid DNA was isolated by the alkaline lysis method (Sambrook et al., 1989). Restriction endonucleases and other enzymes were purchased from either Pharmacia or Boehringer Mannheim Biochemicals. All required radioisotopes were from ICN Biomedicals.

Media and growth conditions

Campylobacter jejuni cultures were grown in either Mueller–Hinton (MH) broth or agar at 37°C under microaerobic conditions. E. coli cells were grown on LB agar or LB broth. Where appropriate, ampicillin and kanamycin were added to a final concentration of 100 mg ml−1 and 50 mg ml−1 respectively. Motility of C. jejuni mutant and wild-type cells was determined on semi-solid (0.4%) MH agar plates as described previously (Guerry et al., 1991; 1992).

Electron microscopy

Samples were prepared for electron microscopy using the direct application negative staining method. A small drop of undiluted sample was placed directly on to a 400 mesh Formvar carbon-coated grid that had been glow discharged to render the carbon surface negatively charged. The sample was allowed to sit on the grid for 2 min without drying. The sample was then negatively stained by adding a drop of aqueous 2% phosphotungstic acid (pH 6.5). After 30 s, excess fluid was drawn off the grid with filter paper, and a thin film of stained sample was left to dry on to the grid surface. The samples were examined and photographed in a Hitachi H600 transmission electron microscope at an accelerating voltage of 75 kV using Kodak electron microscope cut film.

Null mutant of flaC , flgF and flaA

A null flaC mutant of TGH9011 was constructed by gene replacement strategy. Briefly, a 1.4 kb BamHI fragment of pKAN-HINDA (Hani, 1997) containing a kanamycin resistance cassette (KmR) derived from pILL550 (Labigne-Roussel et al., 1987) was inserted at the BglII site within the encoding sequence of flaC, 148 nucleotides (nt) from the start codon (Fig. 1). The mutant gene was transformed into C. jejuni TGH9011 as described previously (Wang and Taylor, 1990). The gene replacement insertion mutant was isolated as a KmR colony and confirmed by Southern blot analysis. A null flgF mutant of TGH9011 was constructed by insertion of the KmR cassette into the NsiI site, 187 nt from the start codon of flgF, the proximal basal rod protein gene (Chan et al., 1998), as described above. Similarly, a null flaA mutant of TGH9011 was constructed by the gene replacement method. A 1.4 kb SmaI fragment of pKAN-HINDA (Hani, 1997) containing a kanamycin resistance cassette (KmR) was inserted at the EcoRV site, 343 nt from the start codon of the flaA gene. The FlaA coding fragment of TGH9011 was isolated by PCR amplification using primers flaAF (5′-CGCTGATTTAAATAGTAAAAG-3′) and flaAR (5′-ACTTTGAGCTAAGATATTTGC-3′) and cloned into the EcoRV site of pBluescript KS-II. The mutant flaA::Kan gene was transformed into C. jejuni TGH9011 as described previously (Wang and Taylor, 1990). The gene replacement insertion mutant was isolated as a KmR colony and confirmed by PCR.

Construction of C. jejuni TGH9011 flaC hipO::CmflaC (designated TGH9011TC)

A 0.9 kb Cm resistance gene cassette was isolated from pAV35 (Van Vliet et al., 1998) by PvuII digestion, cloned into the SmaI site of pBluescript KS, and the new construct was designated pBSCm. The entire flaC gene including the putative regulator binding sequences upstream of the promoter region was amplified by PCR using forward primer flaCF (5′-ATTGCGATTTTGTTCAGTGTTAATT-3′) and reverse primer B8 (5′-GTCTTTGTCCATAGC-3′). The 1.4 kb PCR product was cloned into the EcoRV site of pBSCm forming recombinant pBSCmflaC. The Cm-flaC cassette was isolated from pBSCmflaC using BamHI and SalI digestion, blunt-ended and then ligated into the hipO gene in pHK1.The hipO gene of C. jejuni TGH9011 in pHK1 (Hani and Chan, 1995) was first digested with SphI that is located in the middle of the hipO gene. The SphI digests were then blunt-ended before ligation. The resulting pHipO::CmflaC was transformed into C. jejuni TGH9011-flaC mutant to obtain C. jejuni TGH9011 flaC hipO::CmflaC (designated TGH9011TC) in which the mutant flaC::Km was complemented by a wild-type flaC gene inserted at the hipO gene. A Cm-resistant gene was used as a selective marker. The construct was verified by PCR, and complementation of the flaC mutation in C. jejuni TGH9011 flaC was confirmed by Western blotting using anti-FlaC antibodies.

Construction of ciaB mutant

Two specific primers, ciaBF (5′-TATGGTACCTTAGACAAA GAAGATGTGGG-3′) and ciaBR (5′-ACAGGATCCATCAAGT CATCTTGTTCATG-3′), were designed based on ciaB gene sequences of C. jejuni NCTC11168 (Parkhill et al., 2000) and C. jejuni F38011 (Konkel et al., 1999). A 1.37 kb ciaB internal sequence was amplified by PCR and cloned into KpnI and BamHI sites in pBluescript KS, resulting pBSciaB. Restriction enzyme analysis showed that there are two HindIII sites about 100 bp apart in the middle of the ciaB gene. As Km resistance gene cassette was isolated from pSK5 (Hani, 1997) by digestion with HindIII, and the fragment was then inserted into the HindIII site of ciaB in pBSciaB forming pBSciaB::Km in which the ciaB gene was disrupted by a Km cassette. pBSciaB::Km was used to transform C. jejuniTGH9011 to obtain a C. jejuni TGH9011 ciaB mutant strain. The disruption of the ciaB gene by a Km cassette in the mutant strain was confirmed by PCR.

Construction of pGEX-flaC and generation of rabbit antiserum against FlaC

The flaC gene of C. jejuni TGH9011 was amplified by PCR using pD2-2 as a template and Vent DNA polymerase (New England Biolabs). The PCR fragment produced by oligonucleotides FC1–B4 was inserted into the SmaI site of pGEX-2T (Pharmacia) to generate pGEX-flaC. The construct was transformed in E. coli JM101. Expression and purification of GST-FlaC was performed according to the manufacturer's instructions (Pharmacia). Briefly, an overnight culture of E. coli JM101/pGEX-flaC was diluted 1:10 in fresh LB broth containing 100 µg ml−1 ampicillin and grown at 37°C to an OD600 of 0.5. The expression of the GST-FlaC protein was induced with 0.1 mM IPTG at 37°C for 4 h. Cells recovered by centrifugation were resuspended in 1/20th of the original volume in phosphate-buffered saline (PBS). Cells were sonicated on ice to disrupt the bacteria, and the lysate fraction was separated from residual intact cells and cellular debris by centrifugation. The clarified lysate was incubated with 1% Triton X-100 for 30 min to aid solubilization of the GST-FlaC protein. After centrifugation, detergent-soluble proteins in the supernatant fraction were incubated with glutathione-Sepharose 4B beads for 30 min at room temperature to purify GST-FlaC by affinity binding. The beads were washed five times with PBS and then incubated with bovine thrombin overnight at room temperature to cleave FlaC from the GST carrier protein. The thrombin reaction mixture was centrifuged to recover purified FlaC in the supernatant fraction, whereas the pellet contained GST bound to glutathione-Sepharose 4B beads. Purified recombinant FlaC (about 500 µg) with Freund's complete adjuvant were used to immunize each rabbit. The rabbit was boosted with purified FlaC every 3 weeks until the desired antibody titre was obtained.

Flagella purification

Ten plates of cells grown on MH agar plates for 3 days were obtained. The cells were suspended in 10 ml of 10 mM Tris-HCl (pH 7.4) and homogenized for 3 min (Power et al., 1994). Whole cells and cell debris were removed by centrifuging the samples twice at 8000 r.p.m. in a Beckman JA-20 rotor for 20 min. The supernatant was then ultracentrifuged in a Beckman SW41Ti rotor at 25 000 r.p.m. at 4°C for 1 h. The pellet was resuspended in 5 ml of 1% SDS and ultracentrifuged as before. The resulting pellet containing flagella was resuspended in 100 µl of 10 mM Tris-HCl and stored at −20°C.

Isolation of extracellular proteins

Overnight cell cultures grown in 40 ml of MH broth were obtained. Extracellular proteins in the medium were separated from whole cells by centrifugation with a Beckman JA-20 rotor at 6000 r.p.m. for 10 min. The cells were washed twice by resuspending in 10 ml of PBS (pH 7.4) and centrifuging at 6000 r.p.m. for 10 min. The washed cells were suspended in 50 µl of PBS and stored at −20°C. Extracellular proteins in the medium were passed through a 0.22 µm filter (Millipore). The proteins were precipitated with 10% trichloroacetic acid (TCA) overnight at 4°C. Precipitated proteins were pelleted by centrifugation with a Beckman JA-20 rotor at 9000 r.p.m. for 20 min. The pellet was resuspended in 1.5 ml of 90% acetone (stored at −20°C), centrifuged for 5 min at 14 000 r.p.m. and air dried. The resulting pellet was resuspended in 50 µl of PBS (same volume as whole-cell sample) and stored at −20°C.

Isolation of membrane/flagella and cytoplasmic/periplasmic proteins

Five plates of cells grown on agar plates for 2 days were obtained. The cells were washed twice with 20 ml of PBS (pH 7.4) and centrifuged with a Beckman JA-20 rotor at 6000 r.p.m. for 10 min. The cells were resuspended in 10 ml of PBS and sonicated with a Branson sonifier 450 microtip for 5 × 1 min on ice with duty cycle setting at 50% and output control at 6. Whole cells were removed by centrifugation at 6000 r.p.m. for 10 min. The supernatant was subjected to ultracentrifugation in a Beckman SW41Ti rotor at 25 000 r.p.m. at 4°C for 30 min twice. The pellet containing cell membranes and flagella was suspended in 200 µl of PBS and stored at −20°C. The cytoplasmic/periplasmic proteins contained in the supernatant fraction were precipitated with 10% TCA on ice for 1 h 30 min. Precipitated proteins were obtained by centrifugation with a Beckman JA-20 rotor at 9000 r.p.m. for 15 min. The pellet was resuspended in 1.5 ml of 90% acetone (stored at −20°C), centrifuged for 5 min at 14 000 r.p.m. and air dried. The resulting pellet was resuspended in 200 µl of PBS (same volume as membrane/flagella sample) and stored at −20°C.

Electrophoresis and immunoblotting analysis

Whole-cell and protein samples were diluted 1:1 with 2× SDS sample buffer and heated to 100°C for 10 min. Proteins were separated by SDS-PAGE. Samples were loaded on to 4% polyacrylamide stacking gel, pH 6.8. The proteins were separated in 10% or 12% polyacrylamide separating gel, pH 8.8, by running at 135 V for 45 min. Kaleidoscope prestained standard (10 µl; Bio-Rad) was used as a protein marker. Separated proteins were transferred to a 0.45 µm pore size nitrocellulose membrane (Protran) with 1× transfer buffer (25 mM Tris, 192 mM glycine, 0.02% SDS, 20% methanol) at 80 V for 3 h on ice. The membrane was then blocked with 10% blocking reagent [10% skim milk powder in 1× TBS (Tris-buffered saline), pH 7.6, containing 0.3% Tween-20 (Bio-Rad)] for 15 min with shaking. The membrane was incubated for 30 min with primary antibodies raised in rabbits. Dilutions of antibodies with 1× TBS, 0.3% Tween-20 were as follows: 1:300 dilution of anti-FlaC crude antiserum; 1:250 dilution of anti-Fur crude antiserum; 1:500 dilution of anti-CipA crude antiserum; 1.5 µg ml−1 (1:160 dilution) FlaA flagellin (O:1 serogroup) affinity-purified anti-FlaA antibody. Anti-rabbit Ig horseradish peroxidase-conjugated antibody from donkey (Amersham) in a 1:2000 dilution was used as secondary antibody and incubated with the membrane for 15 min. The membrane was washed with 1× TBS (pH 7.6), 0.3% Tween-20 according to the ECL Western blotting analysis system rapid method (Amersham) throughout the hybridization process. Reactive proteins were detected using 2 ml each of ECL Western blotting detection reagents 1 and 2 (Amersham) and exposed to film (Kodak).

Quantification of whole cells and proteins

Whole-cell samples were diluted 20-fold with PBS and quantified by spectrophotometry, measuring at OD600. Extracellular, membrane/flagella, cytoplasmic/periplasmic and purified flagella protein fractions were diluted in PBS. The proteins were quantified by the addition of 0.1 ml of dye reagent concentrate (Bio-Rad) to 0.4 ml of diluted protein samples and read at OD595 as described by the manufacturer (Bio-Rad). Protein standards of 1–25 µg ml−1 were prepared using bovine albumin (Sigma) suspended in PBS.

Immunofluorescence detection of secreted FlaC

Wild-type C. jejuni TGH9011 and flaC mutant were grown for 2 days on MH plates. The bacteria were washed three times with PBS. HEp-2 cells were grown overnight on sterile glass coverslips in MEM media with 15% fetal bovine serum. HEp-2 cells washed in serum-free MEM media were infected with wild-type C. jejuni TGH9011 or flaC mutant (MOI = 50) for 3 h. The coverslips were then washed three times in PBS and fixed with 3.7% paraformaldehyde in PBS for 30 min at room temperature. Fixed cells were washed three times in PBS, permeabilized using 0.1% Triton X-100 in PBS for 15 min at room temperature and washed three times in PBS before preblocking using 1% BSA–PBS with 10 min incubation. Cells were then incubated with rabbit anti-FlaC antibody in PBS–BSA for 1 h at room temperature, washed five times in PBS, then incubated with anti-JlpA antibody (for bacterial staining) for 1 h at room temperature followed by five washes with PBS. The cells were incubated with cy5-coupled goat anti-rabbit IgG, FITC-coupled goat anti-rabbit IgG and Texas red-coupled phalloidin for 1 h, followed by five washes with PBS. The coverslips were then mounted on to glass slides using Mowiol and viewed using a digital fluorescence microscope.

Preparation of FITC-labelled GST-FlaC and GST

About 1 mg of GST-FlaC and GST was used to coat 200 µl of glutathione Sepharose 4B beads (Amersham Pharmacia Biotech). After washing with PBS, the beads were resuspended in 500 µl of 50 mM carbonate buffer (pH 9.2) containing 5 µg ml−1 FITC and incubated at room temperature for 30 min with gentle shaking. Beads were then washed seven times with PBS buffer containing 1% BSA, followed by two washes with PBS alone. FITC-labelled GST-FlaC and GST proteins were eluted from beads with elution buffer, aliquoted and stored at −20°C.

GST-FlaC binding to HEp-2 cell surface by immunofluorescence microscope

For determination of FITC-labelled GST and GST-FlaC binding on cell surfaces, HEp-2 cells were seeded and grown on glass coverslips for 24 h. Cells were washed with PBS and fixed in 3.7% paraformaldehyde in PBS at room temperature for 30 min. The fixed cells were blocked with 1% BSA in PBS for 10 min, followed by treatment with 0.01 µg of FITC-labelled GST-FlaC and GST for 1 h. The cells were washed five times with PBS and permeabilized with 0.1% Triton X-100 in PBS for 15 min, then washed three times with PBS. HEp-2 cells were blocked with 1% BSA in PBS again, and then incubated with Texas red–X phalloidin for 1 h, followed by washing with PBS at room temperature. The coverslips were mounted on to glass slides using Mowiol (Sigma-Aldrich) and viewed using a digital immunofluorescence microscope.

For unlabelled GST-FlaC blocking assays, HEp-2 cells grown on coverslips were fixed in 3.7% paraformaldehyde in PBS for 30 min and blocked with 1% BSA for 10 min. The cells were treated with 0.1 µg of GST-FlaC for 1 h and washed five times with PBS, followed by treatment with 0.01 µg of FITC-labelled GST-FlaC or GST for 1 h. The cells were washed five times with PBS and permeabilized with 0.1% Triton X-100 in PBS for 15 min, then washed three times with PBS. HEp-2 cells were blocked with 1% BSA in PBS and incubated with Texas red–X phalloidin for 1 h, followed by washing with PBS at room temperature. The coverslips were mounted on to glass slides using Mowiol and viewed using a digital immunofluorescence microscope.

Adherence and invasion assays

HEp-2 cells (ATCC CCL23) were grown at 37°C in Eagle minimal essential medium (EMEM) supplemented with 10% fetal bovine serum (FBS), 100 µg ml−1 streptomycin and 100 units ml−1 penicillin G (Sigma) in a humidified 5% CO2 incubator. Confluent HEp-2 monolayers were trypsinized, seeded into 24-well tissue culture plates at about 1 × 105 cells per well in EMEM−10% FBS without antibiotics, incubated at 37°C for 18 h and then washed twice with EMEM. The adherence and invasion assays were performed by co-incubating flaC mutant or wild-type C. jejuni TGH9011 with HEp-2 cells at a cell–bacteria ratio of about 1:100. The medium was removed after 2 h incubation, and monolayers were washed five times with EMEM then lysed with 0.1% (w/v) Triton X-100 at 37°C for 15 min. For the invasion assay, the monolayers were washed twice with EMEM after 2 h incubation and then incubated for an additional 2 h with EMEM containing 100 µg ml−1 gentamicin. The monolayers were washed twice with EMEM and lysed with 0.1% (w/v) Triton X-100 for 15 min. The released bacteria were enumerated by plate counting on MH agar plates.

Acknowledgements

This work was supported by a grant from the Crohn's and Colitis Foundation of Canada. We thank Drs Scott Gray-Owen, Eric Hani and Angela Joe for valuable comments on the manuscript, Mr Thomas Wong and Dr S. Yuwen Hong and Mingshan Qi for early contributions to the project, Steven Doyle (Microscopy Imaging Laboratory) for the electron microscopy work, Dr Susan F. Koval for discussion of the electron micrographs, Drs John Penner and Dawn Mills for the affinity-purified anti-FlaA antibody, Chee Chan for help with the figures and manuscript preparation, and Soo Chan Carusone for performing the t-test analysis.

Ancillary