Restoration of Flagellar Biosynthesis by Varied Mutational Events in Campylobacter jejuni

doi:10.1111/j.1365-2958.2008.06428.x

Mol Microbiol. Author manuscript; available in PMC 2009 October 1.

Published in final edited form as:

Mol Microbiol. 2008 October; 70(2): 519–536.

Published online 2008 August 29. doi: 10.1111/j.1365-2958.2008.06428.x

PMCID: PMC2576521

NIHMSID: NIHMS73266

Restoration of Flagellar Biosynthesis by Varied Mutational Events in Campylobacter jejuni

David R. Hendrixson^*

David R. Hendrixson, Department of Microbiology, The University of Texas Southwestern Medical Center, Dallas, TX USA;

^* Corresponding author: David R. Hendrixson, University of Texas Southwestern Medical School, Department of Microbiology, 5323 Harry Hines Boulevard, Dallas, TX 75390-9048, Phone: (214) 648-5949, Fax: (214) 648-5907, e-mail: david.hendrixson/at/utsouthwestern.edu

The publisher's final edited version of this article is available at Mol Microbiol

See other articles in PMC that cite the published article.

Abstract

Both a complex regulatory cascade involving the FlgSR two-component system and phase variation control expression of σ⁵⁴-dependent flagellar genes in Campylobacter jejuni. In this study, mutational mechanisms influencing production of the FlgS histidine kinase were discovered. Random non-motile, non-flagellated flgS variants were impaired for growth in the chick intestinal tract. Spontaneous revertants restored for flagellar biosynthesis, gene expression, and motility identified by in vivo and in vitro studies had undergone diverse intragenic and extragenic mutational events relative to flgS. Restorative intragenic events included true phase variation, second-site intragenic reversion, and insertion and deletion of short DNA segments within flgS. In vivo-isolated motile revertants possessed an identical, single extragenic mutation to create a partially constitutively-active FlgR protein in the absence of FlgS. Considering that FlgR production is also influenced by phase variation, these new findings suggest that the FlgSR two-component system is a unique in that each protein is controlled by phase variation and phosphorylation. In addition, this study highlights the mutational activities of C. jejuni and suggests that the bacterium may possess a repertoire of mutational mechanisms to overcome genetic lesions that impair production of virulence and colonization determinants while lacking a normal mismatch repair system.

Keywords: motility, flgS, phase variation

Introduction

Flagellar motility of Campylobacter jejuni is required for optimal interactions with both human and animal hosts. Infection of the human intestinal tract by C. jejuni can lead to gastroenteritis with symptoms ranging from a mild, watery diarrhea to a severe, bloody diarrheal syndrome (Blaser and Reller, 1981; Skirrow and Blaser, 2000). Infection of the gastrointestinal tracts of many animals other than non-human primates by C. jejuni, however, results in a harmless, commensal colonization (Beery et al., 1988; Lindblom et al., 1986; Pokamunski et al., 1986). One proven virulence and colonization factor necessary for interaction with both humans and animals is flagellar motility. In a human volunteer study, non-motile, non-flagellated phase variants of C. jejuni were unable to compete with a wild-type, motile strain during in vivo growth (Black et al., 1988). Similarly, flagellar motility of C. jejuni is required for wild-type levels of colonization of the chick cecum in a natural poultry model of infection (Hendrixson and DiRita, 2004; Hendrixson, 2006; Nachamkin et al., 1993; Wassenaar et al., 1993; Wosten et al., 2004). Thus, production of flagella and motility of C. jejuni are required to promote associations with the intestinal mucosal surfaces of hosts.

Biosynthesis of a single flagellum at one or both poles of C. jejuni requires coordinating expression of more than 40 flagellar genes and the proper ordering of sequential interactions of the encoded proteins (Hendrixson, 2008). Previous studies have determined that the alternative σ factors, σ²⁸ and σ⁵⁴, are required for expression of many flagellar genes (Carrillo et al., 2004; Hendrixson et al., 2001; Hendrixson and DiRita, 2003; Jagannathan et al., 2001; Wosten et al., 2004). σ²⁸ is involved in expression of flaA, encoding the major flagellin that comprises a large proportion of the flagellar filament. σ⁵⁴ is hypothesized to be required for expression of almost all genes that encode components of the flagellar basal body, rod, and hook. Indeed, expression of flgDE2, encoding hook-associated proteins, and flaB, encoding a minor flagellin, have been experimentally shown to be σ⁵⁴-dependent, whereas other proposed σ⁵⁴-dependent promoters remain to be investigated (Hendrixson et al., 2001; Hendrixson and DiRita, 2003; Wosten et al., 2004).

To activate expression of members of its regulon in C. jejuni, σ⁵⁴ requires the FlgSR two-component system, the putative GTP-binding protein FlhF, and components of the flagellar export apparatus, a secretion system that exports many flagellar proteins located beyond the cytoplasmic membrane (Hendrixson and DiRita, 2003). The FlgS histidine kinase is hypothesized to sense an as yet uncharacterized signal to begin a phosphorelay event terminating in activation of the FlgR σ⁵⁴-dependent response regulator. In support of this hypothesis, in vitro analysis has shown that FlgR is phosphorylated upon the addition of FlgS (Joslin and Hendrixson, 2008; Wosten et al., 2004). Genetic expression analyses indicate that phosphorylation of FlgR activates the transcriptional regulator, presumably allowing for a productive interaction with the σ⁵⁴-RNA polymerase holoenzyme complex at target promoters for specific flagellar genes (Joslin and Hendrixson, 2008; Wosten et al., 2004). FlgR is kept inactive in the absence of phosphorelay mediated by FlgS by the N-terminal input and C-terminal domains of the protein. Deletion of either domain of FlgR results in a partially constitutively-active protein capable of activating expression of σ⁵⁴-dependent flagellar genes in the absence of FlgS (Joslin and Hendrixson, 2008). How factors such as FlhF and components of the flagellar export apparatus (including FlhA, FlhB, FliP, and FliR) contribute to influencing the FlgSR system or σ⁵⁴ to stimulate expression of flagellar genes is currently unclear.

Flagellar biosynthesis and gene expression in C. jejuni are also controlled by phase-variable mechanisms (Caldwell et al., 1985; Hendrixson, 2006; Nuijten et al., 1989). Phase variation occurs through random, reversible mutations that switch on and off expression of specific genes or translation of encoded proteins (van der Woude and Baumler, 2004). Causative mutations usually occur within a homopolymeric nucleotide tract or a repetitive heteropolymeric unit in the promoter or coding sequence of a gene. A recent study revealed that production of FlgR can undergo phase variation to affect expression of σ⁵⁴-dependent flagellar genes (Hendrixson, 2006). Specific mutations occur within two different homopolymeric poly A or poly T tracts of flgR. Spontaneous loss of a nucleotide within the tracts shifts the coding sequence of the gene out-of-frame to prematurely terminate translation of FlgR. Reversion of the non-motile, non-flagellated flgR phase ‘off’ state to the motile, flagellated flgR phase ‘on’ state occurs in vivo during commensal colonization of the intestinal tracts of chicks, a natural host for C. jejuni, to provide a colonization advantage for the phase ‘on’ revertants. Mutational mechanisms to restore FlgR production involve either correct insertion of the proper nucleotide in the homopolymeric tract originally affected in the phase ‘off’ variants or by a second-site intragenic reversion via an insertion of a nucleotide near the site of the original lesion that restored the flgR coding sequence with minimal, harmless amino acid substitutions in FlgR (Hendrixson, 2006). Even though this study identified FlgR production as a major phase-variable phenomenon controlling expression of σ⁵⁴-dependent flagellar genes, other variants with wild-type flgR sequences were identified. Thus, other genes of C. jejuni required for flagellar gene expression may be controlled by phase variation.

In the present study, additional spontaneous mutants defective for expression of σ⁵⁴-dependent flagellar genes were identified and analyzed to determine alternative targets for phase variation or other random mutational mechanisms that ultimately influence flagellar motility. This work revealed that production of the FlgS histidine kinase is also subject to random mutation due to either loss of a single nucleotide within one of two poly A tracts or the loss of a single heteropolymeric repeat within the coding sequence of flgS. These variants were able to revert during in vivo growth in chicks and in in vitro motility assays to restore flagellar motility and expression of σ⁵⁴-dependent flagellar genes. The mechanisms of reversion uncovered in these revertants revealed diverse intragenic and extragenic mutational events relative to flgS that restored expression of σ⁵⁴-dependent flagellar genes, flagellar biosynthesis and motility. Intragenic mutational events included a typical phase-variable mechanism in addition to second-site intragenic reversion, and insertion or deletion of DNA segments to restore production of FlgS. Extragenic mutational events were found to occur at a specific nucleotide within flgR to translate a protein able to activate flagellar gene expression in a FlgS-independent manner. This study demonstrates that multiple mutational strategies allow C. jejuni to overcome genetic lesions that impair expression of an important virulence and colonization factor such as flagellar motility. Considering that C. jejuni naturally lacks a mismatch repair system, this study provides an example of the challenges of C. jejuni encounters to maintain and restore production of virulence factors in the presence of a hypermutator phenotype.

Results

Isolation and analysis of non-motile variants unaffected for FlgR production

Production of the FlgR response regulator required for activation of expression of σ⁵⁴-dependent flagellar genes is controlled by a phase-variable mechanism targeting two homopolymeric nucleotide tracts within flgR (Hendrixson, 2006). In the course of this work, other non-motile variants that failed to express σ⁵⁴-dependent flagellar genes, but still produced wild-type FlgR were identified. Discovery of these variants suggested that other genes required for expression of σ⁵⁴-dependent flagellar genes may also be subject to phase variation or other random mutational mechanisms.

To expand the repertoire of variants of C. jejuni defective for flagellar gene expression, spontaneous mutants of DRH665 (81–176 Sm^R ΔastA flaB::astA) were identified. DRH665 is a streptomycin-resistant derivative of C. jejuni 81–176 with an in-frame deletion of the native copy of astA, encoding arylsulfatase, and a transcriptional fusion of a promoterless copy of astA to flaB on the chromosome (Hendrixson and DiRita, 2003). Expression of astA is solely driven from the σ⁵⁴-dependent flaB promoter. flaB encodes a non-essential minor flagellin; mutants lacking flaB produce flagella and are motile, similar to wild-type strains (Guerry et al., 1991; Wassenaar et al., 1991). Thus, DRH665 is a flagellated, motile strain that produces arylsulfatase from the flaB promoter. On Mueller-Hinton (MH) agar containing 5-bromo-4-chloro-3-indolyl sulfate (X-S; a chromogenic substrate for arylsulfatase), DRH665 produces a blue-colony phenotype in contrast to a white-colony phenotype of an isogenic flgR mutant (Hendrixson and DiRita, 2003). To identify spontaneous variants of DRH665 lacking expression of σ⁵⁴-dependent flagellar genes, dilutions of freshly-grown DRH665 were spread on MH agar containing X-S in two separate experiments. The spontaneous white colony phenotype was found at a rate of 1.97 x 10⁻⁵ to 1.92 x 10⁻⁶ mutants per cfu of the total population with three different variants identified (DRH665 5W2, DRH665 5W6, and DRH665 10W1). In comparison to the wild-type parental DRH665 strain, the variants were reduced more than 650-fold for production of arylsulfatase from the flaB::astA transcriptional fusion (Table 1). The variants were also non-motile and did not produce flagella in contrast to wild-type DRH665 (Figures 1A and 1E). Further analysis revealed that all variants produced wild-type levels of FlgR and possessed wild-type flgR sequences (Figure 2 and data not shown). Thus, these mutants acquired at least one mutation in a gene other than flgR to negatively affect expression of σ⁵⁴-dependent flagellar genes and flagellar motility.

Table 1

Arylsulfatase activity due to expression of the flaB::astA transcriptional fusion in C. jejuni DRH665 non-motile variants and associated in vitro-isolated revertants.^a

Figure 1

Motility and flagellar biosynthesis phenotypes of C. jejuni wild-type, non-motile variants, and in vitro-isolated revertants

Figure 2

FlgR production in DRH665 and DRH665 non-motile variants

Experiments were then employed to determine if these spontaneous variants could randomly and reversibly switch back to the flagellated, motile state. In vitro reversion assays were performed by stabbing individual colonies of non-motile variants in motility agar and then isolating motile revertants over time. After three to seven days, 46 – 83% of stabs gave rise to spontaneous motile flares originating from the point of inoculation (data not shown). Bacteria from the flares were recovered and examined for flagellar gene expression, motility, and biosynthesis. Seven different revertants were isolated from DRH665 5W2. Each of these revertants, represented by DRH665 5W2 rev1 and rev5, were restored to wild-type levels of expression of the flaB::astA transcriptional fusion (Table 1), motility (Figure 1B), and flagellar biosynthesis (Figure 1E). Similarly, six revertants were isolated for DRH665 5W6 (rev3, rev4, rev5, rev7A, rev7B, and rev8) and two revertants were isolated for DRH665 10W1 (DRH665 10W1 rev3 and rev4). These revertants were also restored for expression of flaB::astA (Table 1), motility (Figures 1C–1D), and flagellar biosynthesis (Figure 1E). Thus, these variants were able to reversibly switch to restore expression of σ⁵⁴-dependent flagellar genes, production of flagella, and motility.

Identification of impaired FlgS production as the cause of the non-motile phenotype in the variants

To identify genes with mutations that resulted in lack of expression of σ⁵⁴-dependent flagellar genes in the variants, a complementation procedure using chromosomal DNA from a darkhelmet transposon mutant library of wild-type C. jejuni 81–176 was employed (see Experimental Procedures for details). After uptake and recombination of exogenous DNA, two transformants of the DRH665 5W2 non-motile variant were isolated that had acquired a transposon-linked DNA fragment that restored expression of the flaB::astA transcriptional fusion (data not shown). Both transformants were found to have a transposon insertion at a different location within cjj81176_0815 (Figure 3A). This gene is immediately downstream of flgS which encodes the cognate sensor histidine kinase for FlgR (Fouts et al., 2005; Hendrixson and DiRita, 2003; Joslin and Hendrixson, 2008; Parkhill et al., 2000; Wosten et al., 2004). Thus, these findings suggest that the DRH665 5W2 variant may have lacked FlgS and recovered expression of flaB::astA by transformation with a wild-type flgS allele.

Figure 3

Identification of FlgS production as the variable factor in non-motile variants and associated in vitro-isolated revertants

Immunoblotting analysis of proteins from whole-cell lysates was performed to determine if FlgS production was impaired in the original DRH665 5W2 non-motile variant. FlgS appears as a 38-kDa protein in wild-type DRH665 (Figure 3B). No FlgS production was observed in DRH665 5W2, similar to DRH939 (an isogenic mutant of DRH665 lacking flgS) (Hendrixson and DiRita, 2003). Furthermore, FlgS was also absent in the DRH665 5W6 and DRH665 10W1 non-motile variants, indicating that these variants were also defective for production of FlgS (Figure 3B). In all the in vitro-isolated revertants associated with each non-motile variant, production of FlgS was restored, which correlated with their flagellar motility and gene expression phenotypes (Figure 3B). Other than in DRH665 5W6 rev5, the size of FlgS in the revertants was equivalent to wild-type FlgS.

To determine the genetic basis for the lack of FlgS production, the flgS sequences from the non-motile variants and their associated revertants were analyzed. Three different mutations were found to occur in the variants. The first mutation represented by the flgS allele of DRH665 5W2 occurs within the first 203 bases of the flgS coding sequence (Figure 3A). This region of flgS contains a poly A tract of seven nucleotides beginning at base 197. In DRH665 5W2, a six-member poly A tract was present that shifted the coding sequence of the gene out-of-frame to create a premature stop codon 29 bases downstream of the homopolymeric tract (Figure 4A).

Figure 4

Sequence of the flgS variable regions in C. jejuni variants and associated revertants

The second type of mutation occurring within flgS was found in the variant DRH665 10W1. A four-member poly A tract within the coding sequence of wild-type flgS begins at base 268 (Figure 3A). In DRH665 10W1, a single nucleotide within this homopolymeric tract was lost to result in a shift of the coding sequence to create a premature stop codon 40 bases downstream of poly A tract (Figure 4B). The FlgS products of DRH665 5W2 and DRH665 10W1 are likely unstable as no truncated FlgS protein could be detected in these variants (Figure 3B).

The type of mutation occurring in DRH665 5W6 was different from those of DRH665 5W2 and DRH665 10W1. In this non-motile variant, the flgS allele had lost one unit of a heteropolymeric nucleotide repeat at the 3’ end of the flgS coding sequence. Fifteen bases before the stop codon of flgS, two heteropolymeric repeat units consisting of the sequence ACCTT are present (Figure 3A). In DRH665 5W6, one repeat unit was missing to result in a frame shift that extends the coding sequence by 68 nucleotides after the original flgS stop codon (Figure 4C). This 21-amino acid extension to the C-terminus of the translated FlgS protein appears to make the protein unstable as no FlgS in DRH665 5W6 was present by immunoblotting analysis (Figure 3B).

To ensure that these mutated flgS alleles were solely responsible for the non-flagellated, non-motile phenotypes observed in the variants, wild-type flgS in C. jejuni 81–176 Sm^R and 81–176 Sm^R ΔastA flaB::astA were replaced with the flgS5W2, flgS5W6, or flgS10W1 alleles (see Experimental Procedures for details). The resulting mutants were defective for flagellar gene expression, biosynthesis and motility, identical to the phenotypes of the original variants (data not shown). Thus, the phenotypes of the original variants can be attributed to the mutated flgS alleles.

Multiple mutational events restored FlgS production in in vitro-isolated revertants

Analysis of the flgS sequences of the in vitro-isolated revertants associated with each non-motile variant revealed multiple mutational strategies to restore FlgS production. These revertants can be categorized into two different classes based on these mutational strategies.

Class I revertants had undergone one of four different intragenic mutational mechanisms in flgS to restore FlgS production. One mechanism typical of phase variation was observed in the seven in vitro-isolated revertants of DRH665 5W2. Six of these phase ‘on’ revertants (represented by DRH665 5W2 rev1) correctly repaired flgS by inserting an adenine into the affected poly A tract (extending it from six to seven nucleotides) so that the wild-type flgS sequence was restored (Figure 4A). The other phase ‘on’ revertant, DRH665 5W2 rev5, deleted two adenine residues of the poly A tract thereby creating a homopolymeric tract of four nucleotides to result in a FlgS amino acid sequence lacking one residue, K67.

A second intragenic mutational strategy was represented by DRH665 10W1 rev4. In this revertant, a second-site intragenic reversion had occurred by insertion of a single thymine residue after the poly A tract in flgS to restore FlgS production with a single amino acid change of the isoleucine residue at position 90 to a phenylalanine (Figure 4C).

The third type of mutational event of Class I revertants represented by DRH665 10W1 rev3 and DRH665 5W6 rev7B involved insertion of a DNA repeat within flgS. In DRH665 10W1 rev3, a 17-base sequence immediately after the poly A tract was repeated to result in an insertion of seven amino acids within FlgS (Figure 4B). Similarly, the reversion of flgS in DRH665 5W6 rev7B was due to insertion of a 26-base sequence at the 3’ end of the flgS5W6 allele (Figure 4C).

The last type of mutational event of the Class I revertants represented by DRH665 5W6 rev3, rev5, and rev7A involved deletion of DNA after the last remaining heteropolymeric nucleotide repeat unit of the flgS5W6 allele. These deletions removed 167 to 368 bases downstream of the repeat unit to cause a replacement of amino acid sequences at the C-terminus of FlgS. These altered C-termini appeared to stabilize the respective FlgS proteins since the proteins were visible at wild-type levels in immunoblots (Figure 3B). In DRH665 5W6 rev5, the deletion caused an in-frame fusion to the downstream gene, cjj81176_0815, resulting in the translation of a chimeric FlgS-0815 protein with predicted size of 77 kDa. The larger form of FlgS observed in this revertant appeared to be of similar size, indicating that the chimeric protein is produced and stable (Figure 3B).

All FlgS proteins resulting from the reversion mechanisms described above, other than those of phase variation, had changes in the amino acid sequence of the protein. Despite these relatively minor changes (Figure 4B–4C), the FlgS proteins were able to promote wild-type levels of expression of σ⁵⁴-dependent flagellar genes, motility, and flagellar biosynthesis in the revertants (Table 1 and Figures 1B–1E).

Class II revertants, represented by DRH665 5W6 rev4 and DRH665 5W6 rev8, likely had extragenic mutations in the chromosome. These revertants did not change the flgS allele present in the parental DRH665 5W6 non-motile variant but FlgS was produced at similar levels to wild-type DRH665 (Figure 3B). However, the strains were wild-type for expression of σ⁵⁴-dependent flagellar genes, motility and flagellar biosynthesis (Table 1, Figure 1C and 1E). To determine if expression of flaB::astA was dependent on FlgS and FlgR in the revertants, the chromosomal copies of flgS and flgR were insertionally inactivated in the revertants. The resulting flgS or flgR mutants were each reduced over 300-fold for production of arylsulfatase from the flaB::astA transcriptional fusion compared to the isogenic parental revertants, demonstrating that the regulatory cascade involving the FlgSR two-component system was intact and required for expression of σ⁵⁴-dependent flagellar genes (data not shown). Thus, these findings suggest that one or more extragenic mutations have occurred within other genes encoding proteins that allow for the production of FlgS with an altered C-terminus encoded by the flgS5W6 allele. These mutations remain to be identified and characterized.

A specific point mutation in flgR was the result of an extragenic mutational event in in vivo-isolated revertants

A previous study showed that phase variation of FlgR production from the non-motile, non-flagellated phase ‘off’ state to the motile, flagellated phase ‘on’ state in C. jejuni occurs in vivo during commensal colonization of the gastrointestinal tract of chicks (Hendrixson, 2006). A similar study was performed to determine if restoration of FlgS production could also occur in the non-motile variants during infection of chicks. One-day-old chicks were orally gavaged with wild-type DRH665 or the non-motile variants DRH665 5W2 and DRH665 5W6 and the C. jejuni from the chick ceca were isolated at days 1, 7, and 14 post-infection. Bacteria were recovered on selective agar containing X-S to enumerate C. jejuni colonizing the ceca and determine the ratio of C. jejuni expressing σ⁵⁴-dependent flagellar genes (blue colony phenotype, AstA+) versus not expressing these genes (white colony phenotype, AstA-).

At day 1 post-infection, wild-type DRH665 and the non-motile variants displayed a similar colonization capacity despite the different motility phenotypes (Figure 5). However, by day 7 post-infection, both variants demonstrated 300- to 5,000-fold reduced colonization capacities compared to the wild-type strain. These results support previous findings that motility is not absolutely necessary for initial colonization events but is required for maintaining high levels of colonization of C. jejuni in the ceca (Hendrixson, 2006). By day 14 of infection, the variants continued to exhibit large colonization defects compared to the wild-type strain (Figure 5). Furthermore, in three chicks infected with the DRH665 5W6 non-motile variant, no C. jejuni could be recovered from the ceca of chicks. The mean colonization capacity of wild-type DRH665 at days 7 and 14 were significantly different from those of the variants (P value < 0.05).

Figure 5

Analysis of in vivo phase variation and the colonization capacity of C. jejuni non-motile variants in chicks

All wild-type DRH665 isolates recovered from the ceca of chicks at all time points were positive for σ⁵⁴-dependent flagellar gene expression by producing arylsulfatase from the flaB::astA transcriptional fusion and the blue-colony phenotype. All DRH665 5W2 and DRH665 5W6 isolates recovered from the chicks at days 1 and 7 post-infection remained negative for expression of σ⁵⁴-dependent flagellar genes. However, at day 14 post-infection, reversion of the variants had occurred in some of the chicks. In three chicks infected with DRH665 5W2, 0.3%, 87.3% and 87.5% (with DRH665 5W2 rev8, rev9, and rev10 as representative isolates, respectively) of the recovered C. jejuni displayed the blue-colony phenotype and thus, expressed σ⁵⁴-dependent flagellar genes (Figure 5). In one of the chicks infected with DRH665 5W6 at day 14 post-infection, 98.1% of the bacteria recovered (with DRH665 5W6 rev10 as a representative isolate) displayed the blue-colony phenotype.

The in vivo-isolated revertants were then analyzed for flagellar gene expression, motility and biosynthesis. Even though the revertants had a level of motility comparable to wild-type DRH665 in agar plates in vitro (Figure 6A), dark-field microscopy revealed differences in the number of motile bacteria between the strains. Approximately 85% of the population of wild-type bacteria showed rapid, darting motility but only 15% of the in vivo-isolated revertants showed similar motility with the remainder being non-motile (data not shown). Further analysis of the revertants revealed that the strains were only restored for expression of flaB::astA to a level approximately 25 – 30% of that of wild-type DRH665 (Table 2). Electron microscopy was then employed to analyze any defects in flagellar biosynthesis in the in vivo-isolated revertants. Unlike wild-type DRH665, which produced a single flagellum at one or both poles of the bacterium, these in vivo-isolated revertants rarely produced two flagella per bacterium. Instead, the bacteria most often produced a single flagellum or no flagella (Figure 6C; data not shown). For a more accurate measurement of flagellar biosynthesis, over 120 individual bacteria of wild-type DRH665 or the in vivo-isolated revertants were examined by electron microscopy to determine the number of flagella produced by each bacterium. For wild-type DRH665, over 48% of individual bacteria produced two flagella, 41% produced one flagellum and fewer than 10% produced no flagella (Table 3). However, a large majority (>81% of the population) of each in vivo-isolated revertant produced no flagella, with approximately 11–17% of the remaining population producing one flagellum. Fewer than 2% of each revertant produced two flagella. Thus, the reduced restoration of σ⁵⁴-dependent flagellar gene expression in the in vivo-isolated revertants may have hindered the ability of these bacteria to produce normal numbers of flagella, thereby reducing the number of motile bacteria.

Figure 6

Motility and flagellar biosynthesis phenotypes of wild-type C. jejuni and in vivo-isolated revertants

Table 2

Arylsulfatase activity due to expression from the flaB::astA transcriptional fusion in C. jejuni DRH665 non-motile variants and associated in vivo-isolated revertants.^a

Table 3

Analysis of flagellar number produced by C. jejuni strains.

The DRH665 5W2 rev8, DRH665 5W2 rev9, DRH665 5W2 rev10, and DRH665 5W6 rev10 revertants were then analyzed for production of FlgS. Unlike the in vitro-isolated revertants, immunoblotting analysis was unable to detect FlgS in these strains (data not shown). Furthermore, the flgS alleles of these revertants had not changed and were identical to the flgS alleles of the parental DRH665 5W2 or DRH665 5W6 non-motile variants (data not shown). Thus, these strains appeared to have undergone extragenic mutations that allow for expression of σ⁵⁴-dependent flagellar genes in the absence of FlgS.

The in vivo-isolated revertants were then examined for FlgS- or FlgR-dependency of expression of the flaB::astA transcriptional fusion. To this end, the chromosomal copies of flgS or flgR in these bacteria were replaced with respective insertionally-inactivated flgS or flgR alleles. As shown in Table 4, insertional inactivation of flgR caused over 25- to 600-fold reductions in production of arylsulfatase from the flaB::astA transcriptional fusion compared to the parental revertants. However, insertional inactivation of flgS in the revertants did not cause similar decreases. These data demonstrate that expression of σ⁵⁴-dependent flagellar genes in the in vivo-isolated revertants occurs by a FlgR-dependent, FlgS-independent mechanism.

The sequences of the flgR alleles of the in vivo-isolated revertants were then examined. Each flgR allele in these revertants was found to have undergone an identical mutation at nucleotide 1145 within flgR. This mutation resulted in a G to A switch resulting in substitution of a lysine for an arginine at position 382 of FlgR. The flgRR382K mutation appeared to create a FlgR mutant protein that was stable and produced at similar levels as wild-type FlgR in whole-cell lysates of C. jejuni as observed by immunoblot analysis (data not shown).

R382 of FlgR is the first amino acid of the C-terminal domain of FlgR. This domain has been hypothesized to prevent phosphorylation of FlgR by non-FlgS sensor kinases or phosphodonors present in C. jejuni so that FlgR is only specifically modified by FlgS (Joslin and Hendrixson, 2008). Deletion of this domain results in a FlgR protein that can partially activate expression of σ⁵⁴-dependent flagellar genes in the absence of FlgS. Thus, the production of FlgR R382K in these in vivo-isolated revertants may mimic FlgR lacking the C-terminal domain and cause partial constitutive activation of the protein in the absence of FlgS.

To test this hypothesis, flgS mutants of C. jejuni 81–176 Sm^R were manipulated to replace wild-type flgR with the flgR5W6rev10 allele that encodes the FlgR R382K protein of DRH665 5W6 rev10. DRH460 (81–176 Sm^R ΔflgS) contains an in-frame deletion of the flgS coding sequence from the chromosome whereas DRH2724 (81–176 Sm^R ΔastA flgS5W6 flaB::astA) has had wild-type flgS replaced with the flgS allele from the non-motile variant DRH665 5W6. DRH2724 also lacks native astA so that expression of the σ⁵⁴-dependent flaB::astA transcriptional fusion can be analyzed in this strain. Both the deletion of flgS in DRH460 and the expression of the flgS5W6 allele in DRH2724 caused the lack of production of FlgS resulting in a non-motile, non-flagellated phenotype (Figure 6B; data not shown). Even though wild-type FlgR is produced in these mutant strains, it is inactive presumably due the absence of phosphotransfer from FlgS (data not shown) (Joslin and Hendrixson, 2008). Replacement of flgR with the flgR5W6rev10 allele allowed for restoration of motility in both types of flgS mutants (Figure 6B). When individual bacteria were examined for the number of flagella produced, no flagella were produced in strains lacking FlgS with production of wild-type FlgR, as expected (Table 3). However, replacing wild-type flgR with flgR5W6rev10 in a ΔflgS or flgS5W6 mutant resulted in about 1 to 2% of bacteria producing two flagella and 9 to 11% of bacteria producing one flagellum (Table 3). This partial restoration of flagellar motility and biosynthesis in these mutants containing flgR5W6rev10 in the absence of FlgS is similar to what was observed in the in vivo-isolated revertants.

The expression of the flaB::astA transcriptional fusion was then analyzed in these strains. Only baseline levels of expression of the flaB::astA transcriptional fusion was observed in 81–176 Sm^R ΔastA flgS5W6 flaB::astA (DRH2737), which lacks production of FlgS to activate wild-type FlgR (Table 5). However, co-expression of flgS5W6 with flgR5W6rev10 resulted in a greater than 400-fold increased level of expression of the flaB::astA transcriptional fusion (Table 5). This fold-increase in the level of expression of flaB::astA was comparable to the partial restoration of expression of the transcriptional fusion observed in the in vivo-isolated revertants compared to the original non-motile variants (Table 2). Thus, the extragenic mutational event in the in vivo-isolated revertants that resulted in production of FlgR R382K resulted in a FlgR mutant protein that is partially constitutively active in the absence of wild-type flgS. Therefore, these data suggest that FlgR R382K can partially suppress the flgS mutation of the original variants.

Table 5

Arylsulfatase activity due to expression of the flaB::astA transcriptional fusion in flgS5W6 mutants containing wild-type flgR or flgR5W6rev10.^a

Discussion

Flagellar motility is one of the most important virulence and colonization factors of C. jejuni, required for promoting colonization of the intestinal tract of human and avian hosts (Black et al., 1988; Hendrixson and DiRita, 2004; Hendrixson, 2006; Nachamkin et al., 1993; Wassenaar et al., 1993; Wosten et al., 2004). C. jejuni has developed a complex regulatory cascade involving the FlgSR two-component system, the flagellar export apparatus, and the FlhF accessory protein to control expression of a subset of flagellar genes encoding components of the basal body and hook (Hendrixson and DiRita, 2003; Hendrixson, 2008). The regulatory processes governing transcriptional control of these genes most likely ensures correct temporal expression so the encoded proteins are secreted and interact in a defined order for proper biogenesis of the flagellar organelle.

Like other bacteria, C. jejuni must ensure that detrimental mutations within the coding sequence of genes for virulence and colonization factors do not occur when these proteins are needed to interact with a host. When mutations do occur, the bacterium must be able to repair these sequences to initiate and maintain infection. The importance of this feat is heightened in C. jejuni which evidently lacks a typical mismatch repair system usually present in bacteria to lessen the frequency of spontaneous mutations (Fouts et al., 2005; Parkhill et al., 2000). A minimally functioning mismatch repair system to correct DNA replication errors consists of the MutS, MutH, and MutL proteins (Kunkel and Erie, 2005). Similar to Helicobacter pylori, the genome sequences of C. jejuni strains lack mutH and mutL (Alm et al., 1999; Tomb et al., 1997). The MutS proteins of C. jejuni and H. pylori are homologous to the proteins of MutS2 family (Eisen, 1998). MutS in H. pylori has been shown to function in repairing DNA after oxidative damage rather than in mismatch repair (Wang et al., 2005). Since C. jejuni most likely lacks a typical mismatch repair system, spontaneous mutations may occur with increased frequency, contributing to a hypermutator phenotype. Experimental evidence from a collection of Campylobacter isolates suggest that a majority of strains possess an appreciable hypermutator phenotype (Hanninen and Hannula, 2007). Hypermutation may be important for the bacterium in generating diversity which can contribute to differential production or structural variability of virulence and colonization factors so that the population as a whole has an advantage in maintaining prolonged colonization of a host. Much diversity in glycosylation of LOS, capsular polysaccharide, and protein glycosylation has already been shown for C. jejuni strains (Bacon et al., 2001; Godschalk et al., 2007; Guerry et al., 2002; Karlyshev et al., 2005a; Karlyshev et al., 2005b; Logan et al., 2002; Szymanski et al., 2003).

However, having a natural hypermutator phenotype can lessen the fitness of a bacterium if the mutation rate is too high to maintain correct production of proteins so that the bacterium can maintain a niche in a host or environment. Shown both previously and in this study, naturally occurring spontaneous mutations in genes such as flgS and flgR abolish flagellar biosynthesis and motility in C. jejuni, which reduce the colonization capacity of the bacterium for the natural avian host (Hendrixson, 2006). Thus, C. jejuni must be able to overcome genetic lesions that impede production of these elements.

Phase variation is one genetic mechanism frequently employed by C. jejuni to randomly and reversibly control expression and production of important virulence and colonization factors such as LOS glycosylation, motility, and capsular polysaccharide production (Bacon et al., 2001; Caldwell et al., 1985; Guerry et al., 2002; Hendrixson, 2006; Karlyshev et al., 2005a; Linton et al., 2000; Nuijten et al., 1989; Prendergast et al., 2004; Szymanski et al., 2003). This work revealed that in addition to FlgR, production of the cognate sensor kinase FlgS is also controlled by random and reversible mutational mechanisms. Mutations were found in repeating homopolymeric or heteropolymeric tracts within the flgS coding sequence that affected translation of FlgS and, consequently, downstream regulatory events required for σ⁵⁴-dependent flagellar gene expression and flagellar motility. In some cases, repair of flgS occurred by a mechanism common to phase variation, where the mutated homopolymeric tract that had originally lost a nucleotide to shift the coding sequence out-of-frame regained a nucleotide to set the coding sequence back to its wild-type state, ultimately restoring flagellar motility. However, this phase-variable mechanism was only one of many different mechanisms identified that restored expression of flagellar genes and production of flagella. Additional mechanisms included intragenic and extragenic mutations relative to flgS. Intragenic mutational events included second-site reversions and insertion and deletion of sequences to restore the flgS coding sequence so that a functional protein is produced. Extragenic mutations included one that specifically affected flgR to produce a partially constitutively-active response regulator that functions in the absence of FlgS. Thus, a vast array of mutational and recombinational events can repair genetic lesions in flgS to restore motility to the organism.

The FlgSR system continues to represent an unusual regulatory network in bacteria. Having a two-component regulatory system which is controlled by phosphorelay to govern expression of flagellar genes is typical of many different motile bacteria (Brahmachary et al., 2004; Correa et al., 2000; Dasgupta et al., 2003; Klose and Mekalanos, 1998; Niehus et al., 2004; Prouty et al., 2001; Ritchings et al., 1995; Spohn and Scarlato, 1999). However, C. jejuni has placed another independent level of control involving phase variation to influence production of both FlgS and FlgR. In the literature, only the bacterial sensor histidine kinase BvgS of Bordetella pertussis has been shown to undergo phase-variable production (Stibitz et al., 1989). Evidence also suggests that AgrA of Staphylococcus aureus may undergo phase variation to affect its activity as a response regulator (Traber and Novick, 2006). However, having both the sensor kinase and response regulator components of a single two-component system be controlled by phase variation in addition to phosphorelay makes the FlgSR system truly unique amongst all currently studied systems. While the reasons for this additional control are not fully defined, one hypothesis for this feature suggests that having FlgS or FlgR under control of phase variation may make the bacterium more diverse as a population when expression of both the non-motile, non-flagellated and the motile, flagellated phenotypes are advantageous in different host or environmental settings.

One mutational strategy to restore flagellar gene expression in a flgS variant was specific for in vivo-isolated revertants. This mechanism involved an extragenic mutation at a specific nucleotide in flgR to result in the production of the FlgR R382K mutant protein. This specific mutation in four separate in vivo-isolated revertants is curious because it occurs at the first residue of a domain that has been implicated in controlling FlgR activation (Joslin and Hendrixson, 2008). As reported in this work, the FlgR response regulator can be divided into three different domains based on homology to other NtrC-like transcriptional regulators. The C-terminal domain of FlgR is unusual in that it appears to not contain a typical DNA-binding domain but rather functions to prevent phosphorylation of FlgR by non-FlgS sensor kinases or small phosphodonors in C. jejuni. Removal of this domain allowed the mutated FlgR protein (FlgR ΔCTD) to partially stimulate expression of σ⁵⁴-dependent flagellar genes in a FlgS-independent manner (Joslin and Hendrixson, 2008). However, activity of FlgR ΔCTD was found to still depend on phosphorylation. Thus, FlgR ΔCTD appears to be phosphorylated by other kinases or phosphodonors in the absence of FlgS, whereas the full-length wild-type FlgR protein remains inactive without FlgS. The FlgR R382K protein in the in vivo-isolated revertants appears to function similarly as this FlgR ΔCTD protein (Joslin and Hendrixson, 2008). FlgR R382K is able to partially activate expression of σ⁵⁴-dependent flagellar genes and restore flagellar biosynthesis and motility to a flgS variant, a ΔflgS mutant, or a reconstructed flgS mutant expressing the flgS5W6 allele. Thus, the extragenic mutational mechanism to generate this version of FlgR in vivo naturally created a FlgR protein with a presumably inactive C-terminal domain so that it functions without wild-type FlgS. Further biochemical analyses are required to determine if this mutation does indeed inactivate the C-terminal domain to cause the apparent partial constitutive activation of the protein. Identification of this flgR allele in the in vivo-isolated revertants suggests that alteration of the C-terminal domain of FlgR can occur in in vivo settings to ultimately impact the behavior of the bacterium significantly in its flagellar gene expression, flagellar biosynthesis, and colonization capacity. The kinase or phosphodonor that may be used to modify FlgR R382K through phosphorylation remains to be identified.

This study has also revealed that additional unidentified components that influence the FlgSR system exist. Previous to this study, it was hypothesized that the flagellar export apparatus and FlhF may influence signaling events to result in autophosphorylation of FlgS to begin the phosphorelay that terminates in activation of FlgR for expression of σ⁵⁴-dependent flagellar genes (Hendrixson and DiRita, 2003). In the present study, two different classes of revertants were identified that contained extragenic mutations to influence activation of the FlgSR system. As discussed above, extragenic mutations that resulted in formation of FlgR R382K that can function in the absence of FlgS suggest another unidentified kinase or a small phosphodonor can be used to phosphorylate and activate FlgR. In addition, two in vitro-isolated revertants, DRH665 5W6 rev4 and DRH665 5W6 rev8, were found to produce FlgS with the flgS5W6 allele. This allele fails to produce FlgS in the parental DRH665 5W6 strain. Thus, the isolated revertants of DRH665 5W6 may have undergone an extragenic mutation in a gene that allows FlgS produced from the flgS5W6 allele to be stable or in a gene whose inactivation may eliminate a component such as a protease that normally degrades improperly formed FlgS. Identification of this type of factor will be pursued in future research activities using transposon mutagenesis screens and other genetic approaches.

One finding uncovered in the in vivo-isolated revertants is that the FlgR R382K mutant protein in the absence of FlgS only partially restored flagellar biosynthesis (as measured by counting the number of flagella on individual bacteria) to a small portion of the population. One reason for the decreased flagellar numbers may be due to the observation that expression of one σ⁵⁴-dependent flagellar gene, flaB, in the revertants was only about 25% of the level seen in the wild-type strain. Mutations that reduce but not abolish the number of flagellar organelles have not been appreciated before in C. jejuni. Identification of this type of mutation may have implications for understanding how the level of expression of σ⁵⁴-dependent flagellar genes directly contributes to the efficiency of flagellar biosynthesis and flagellar numbers.

One of the sites of mutation within flgS that contributed to variation was a homopolymeric tract consisting of adenine residues. Until recently, phase variation in C. jejuni strains was largely assumed to occur in homopolymeric G tracts due to their infrequent occurrence in the bacterial genomes, which has an average A+T content of over 69% (Fouts et al., 2005; Parkhill et al., 2000). Indeed, poly G tracts have been found in genes in some phase-variable loci, and at least one of these tracts has been shown to contribute to phase variation of LOS (Linton et al., 2000). This work, along with the previous work analyzing phase variation of FlgR production, has found that poly A and poly T tracts are responsible for phase variation of flagellar biosynthesis and motility (Hendrixson, 2006). In addition, one poly A tract in flgS that also showed spontaneous mutation and reversion by intragenic mutational events other than phase variation consisted of only four residues. Since the genome of C. jejuni is rich in adenine and thymine residues, this work demonstrates that many more sites in the bacterial genome may be subject to phase variation and random mutation, thereby potentially creating much more diversity in C. jejuni than previously appreciated. Furthermore, this work emphasizes the care that may need to be performed with the passage and manipulation of C. jejuni strains in vitro to prevent spontaneous deleterious mutations from occurring in genomes to affect other phenotypes of strains being analyzed.

Genetic analysis of production of virulence and colonization factors in C. jejuni continues to reveal multiple and complex mechanisms that converge to control production of these proteins. This work sheds light into the types of mutational mechanisms of C. jejuni that repair and maintain production of one such virulence and colonization factor, flagellar motility. The lack of a mismatch repair system in C. jejuni would seem to increase the chances of detrimental mutations occurring in factors required by the bacterium for growth in certain environments or hosts to ultimately decrease the fitness of the bacterium for these niches. However, C. jejuni is a successful pathogen of humans and is rampant in agriculture where it is able to colonize the intestinal tract of different types of livestock. Thus, C. jejuni appears to balance a potential hypermutator phenotype that may increase mutations to affect production of important virulence and colonization factors with varied mutational strategies to overcome such detrimental mutations. Further investigation of other known variably-produced virulence and colonization factors of C. jejuni will determine if a variety of mutational events are also in place to restore production of these proteins and contribute to the understanding of how this potential hypermutator maintains genome stability while promoting optimal fitness for various niches.

Experimental procedures

Bacterial strains and growth conditions

All C. jejuni strains used in this study are derived from strain 81–176, a clinical isolate that has been shown to cause disease in humans and promote commensal colonization of the gastrointestinal tract of chickens (Bingham-Ramos and Hendrixson, 2008; Black et al., 1988; Hendrixson and DiRita, 2004; Hendrixson, 2006; Korlath et al., 1985). DRH212 (81–176 Sm^R), DRH460 (81–176 Sm^R ΔastA ΔflgS), DRH461 (81–176 Sm^R ΔastA), DRH665 (81–176 Sm^R ΔastA flaB::astA), and DRH737 (81–176 Sm^R ΔflgR) are previously described (Hendrixson et al., 2001; Hendrixson and DiRita, 2003). DRH939 is a derivative of 81–176 Sm^R ΔastA ΔflgS (DRH911) which contains the flaB::astA transcriptional fusion (Hendrixson and DiRita, 2003). C. jejuni was routinely grown on Mueller-Hinton (MH) agar containing 10 μg/ml trimethoprim (TMP) at 37 ºC in microaerobic conditions (10% CO₂, 5% O₂, and 85% N₂). When necessary, antibiotics or other compounds were used during growth at the following concentrations: kanamycin, 100 μg/ml; chloramphenicol, 15 μg/ml; cefoperazone, 30 μg/ml; streptomycin, 0.5, 1, 2, or 5 mg/ml; and 5-bromo-4-chloro-3-indolyl sulfate (X-S), 35 μg/ml. All C. jejuni strains were stored at −80 ºC in a solution of 85% MH broth and 15% glycerol. E. coli DH5α was cultured in Luria Bertani (LB) agar or broth with one or more antibiotics added at the following concentrations as necessary: ampicillin 100 μg/ml; kanamycin 100 μg/ml; and chloramphenicol 15 μg/ml. All E. coli strains were stored at −80 ºC in a solution of 80% LB and 20% glycerol.

Isolation of spontaneous, non-motile variants lacking expression of σ⁵⁴-dependent flagellar genes

DRH665 was grown from frozen stocks on MH agar containing TMP under microaerobic conditions for 48 hours at 37 ºC. Colonies were restreaked and grown for 16 hours in microaerobic conditions at 37 ºC. Bacteria were suspended from agar plates and diluted to OD₆₀₀ 0.4, diluted 1,500-fold, and 67.5 μl were spread on MH agar containing kanamycin and XS in 150 mm petri dishes. After incubation for three to five days under microaerobic conditions at 37 ºC, colonies displaying the white-colony phenotype (AstA-) were identified, restreaked and grown for storage as frozen stocks.

Immunoblot analysis

Whole-cell lysates (WCL) of C. jejuni strains were prepared by growing strains from frozen stocks on MH agar with TMP in microaerobic conditions for 48 hours at 37 ºC. Bacteria were restreaked, grown in microaerobic conditions at 37 ºC for 16 hours, and then resuspended from agar plates in MH broth. Samples were diluted to OD₆₀₀ 0.9 and 1 ml aliquots of each sample were centrifuged at 13,000 rpm for 3 minutes. Samples were washed once with PBS and then suspended in 50 μl of 1X Laemmli buffer. To detect FlgR by immunoblotting, 4 μl of each sample were separated by 10% SDS-PAGE. Similar electrophoresis with 7.5 μl of each sample was used to detect FlgS. Immunoblot analyses were performed using a 1:5,000 dilution of α-FlgR Rab13 or 1:3500 dilution of α-FlgS Rab11 (Hendrixson, 2006), followed by a 1:5,000 dilution of horseradish peroxidase-conjugated goat α-rabbit secondary antibody (Bio-Rad).

Motility assays and transmission electron microscopy

C. jejuni strains were grown from frozen stocks on MH agar with TMP in microaerobic conditions at 37 ºC for 48 hours. Strains were restreaked and grown for an additional 16 hours in microaerobic conditions at 37 ºC. Strains were washed from plates in MH broth and diluted to OD₆₀₀ 1.0. For motility assays, strains were stabbed into semisolid MH agar plates containing 0.4% agar as previously described (Hendrixson et al., 2001). Motility phenotypes were observed 25 to 30 hours after incubation in microaerobic conditions at 37 ºC. For direct observation of motile bacteria, bacterial cultures were grown as described above but diluted to OD₆₀₀ 0.75. A ten-fold dilution of each culture was then made in MH broth. Ten microliters of each culture was placed on a glass slide and topped with a glass coverslip. Bacteria were then visualized under 400X magnification using an Olympus BH2 dark-field microscope. For transmission electron microscopy, 1 ml samples of bacteria were centrifuged at 13,000 rpm for 3 minutes and resuspended in 2% glutaraldehyde. After incubation on ice for 1 hour, samples were stained with 1% uranyl acetate and visualized with a FEI Technai G2 Spirit BioTWIN transmission electron microscope.

Arylsulfatase assays

Arylsulfatase assays were performed as previously described (Hendrixson and DiRita, 2003), which was based on previously established methods (Henderson and Milazzo, 1979; Yao and Guerry, 1996). C. jejuni strains were grown from frozen stocks on MH agar in microaerobic conditions at 37 ºC for 48 hours. Strains were restreaked on MH agar and grown at 37ºC for 16 hours in microaerobic conditions. Strains were washed from plates with PBS and diluted to an absorbance at OD₆₀₀ 0.6 to 1.0, washed once in assay buffer, and then incubated with 10 mM nitrophenylsulfate and 1 mM tyramine at 37 ºC for 1 hour. After termination of the reactions with NaOH, the amount of nitrophenol released in each sample was determined by measuring the absorbance at OD₄₁₀ of each sample. Values were compared to a standard curve of OD₄₁₀ readings of known nitrophenol concentrations to determine the number of arylsulfatase units produced by each strain. One arylsulfatase unit is defined as the amount of enzyme catalyzing the release of 1 nmol of nitrophenol per hour per OD₆₀₀ unit. Each strain was tested in triplicate and each assay was performed three times.

Construction of the darkhelmet transposon and the transposon mutant library of C. jejuni 81–176

The ori from pACYC184 (Genbank XO6403; bases 581 to 1493) was amplified by PCR with oligonucleotides that added 5’ PstI and 3’ KpnI restriction sites to the amplified DNA (Chang and Cohen, 1978; Rose, 1988). After digestion with PstI and KpnI, the ori was cloned into PstI- and KpnI-digested pEnterprise2 containing the picard transposon (Hendrixson and DiRita, 2003). Successful cloning of the DNA allowed for insertion of the ori upstream of the chloramphenicol acetyltransferase (cat) gene within the transposon. The resulting plasmid and transposon were named pSpaceball1 and darkhelmet, respectively.

Construction of a random transposon library of C. jejuni 81–176 with the darkhelmet transposon followed previously published protocols (Hendrixson et al., 2001; Hendrixson and DiRita, 2003, 2004). Briefly, twelve in vitro transposon mutagenesis reactions were performed with each reaction consisting of 2 μg of chromosomal DNA from C. jejuni 81–176, 1 μg of pSpaceball1, and 250 ng of Himar1 C9 transposase purified from DH5α/pMalC9 (Akerley and Lampe, 2002). After transposition, the transposed DNA was repaired and transformed into C. jejuni 81–176 as previously described (Hendrixson et al., 2001). Transposon mutants were recovered after growth on MH agar containing chloramphenicol. After combining the transformants, a library containing approximately 24,000 transposon mutants was obtained.

DNA transformation for identification of the gene mutated in non-motile variants

Chromosomal DNA from the C. jejuni 81–176 darkhelmet transposon library was purified and 1 μg of the DNA was digested with 25 units of BglII in a 50 μl reaction for 16 hours at 37 ºC. Ten additional units of BglII were added and the reaction was incubated an additional 3 hours at 37ºC. Since BglII restriction sites are present frequently in the C. jejuni genome but absent from the darkhelment transposon, use of BglII to digest chromosomal DNA from the transposonmutant library allowed for DNA fragments of shorter lengths containing both the transposon and a wild-type copy of the gene of interest to be generated. Therefore, digestion with BglII increased the ease of finding the gene potentially affected by variation. For transformation, DRH665 5W2 was grown for 48 hours from frozen stocks on MH agar containing TMP at 37 ºC under microaerobic conditions. The bacteria were restreaked and grown for an additional 16 hours on MH agar with TMP under microaerobic conditions at 37 ºC. Bacteria were washed from plates and diluted in MH broth to an OD₆₀₀ of 0.5. One ml of molten MH agar was added to each of five tubes. After solidification, the agar plugs were overlayed with 0.5 ml of bacterial suspension and the cultures were then incubated for 3 hours at 5% CO₂ at 37 ºC. Ten microliters of the BglII-digested DNA (~0.2 μg) were added to each tube and the tubes were incubated for an additional four hours at 5% CO₂ at 37 ºC. Transformants were recovered on MH agar containing chloramphenicol and X-S after growth for four days under microaerobic conditions at 37 ºC. Approximately 10,000 colonies were recovered in all with two colonies displaying the blue-colony phenotype. Chromosomal DNA from these transposon mutants, designated DRH665 5W2 darkhelmet6111 and DRH665 5W2 darkhelmet6122, was purified.

To identify the location of the transposon insertion, 2 μg of chromosomal DNA from DRH665 5W2 darkhelment6111 and DRH665 5W2 darkhelment6122 were digested with 100 units of BglII at 37 ºC overnight in a total volume of 100 μl. To ensure complete digestion, 20 units of BglII were added to each reaction and the reactions were incubated for another 2 hours. Digestion with BglII allows for relatively short DNA fragments containing the entire transposon and flanking DNA sequences to be recovered by this plasmid rescue technique. After phenol:chloroform extraction and ethanol precipitation, digested DNA fragments were circularized with 400 units of T4 DNA ligase overnight at 16 ºC. The ligation reactions were ethanol precipitated and the DNA was then electroporated into DH5α. Transformants were recovered on LB agar containing chloramphenicol. Plasmid DNA was prepared from individual transformants and then sequenced with a primer annealing to the 3’ end and reading outward from the transposon. DNA sequences obtained were compared to the genomic sequence of C. jejuni 81–176 to determine the location of the transposon-chromosomal junction (Fouts et al., 2005).

In vitro isolation of spontaneous motile revertants

DRH665 5W2, DRH665 5W6, and DRH665 10W1 non-motile variants were grown from frozen stocks on MH agar in microaerobic conditions at 37 ºC for 72 hours. Ten to sixteen individual colonies were stabbed into semisolid MH agar and incubated in microaerobic conditions at 37 ºC for up to 7 days. For stabs that gave rise to spontaneous motile flares originating from the point of inoculation, an agar plug was obtained and vortexed with 1 ml of MH broth. Dilutions of the sample were plated on MH agar with TMP. After growth for 48 hours in microaerobic conditions at 37 ºC, individual motile revertants were recovered from each flare. The flgS and flgR alleles from each revertant were amplified by PCR and sequenced.

Chick colonization assays

Oral infection with 1-day old white leghorn chicks strain Δ was performed as previously described (Hendrixson and DiRita, 2003; Hendrixson, 2006). Fertilized eggs (SPAFAS) were incubated in an egg incubator (Sportsman Incubator Model 1202; Georgia Quail Farms) for 21 days at 37.8 ºC with appropriate humidity and rotation of eggs until hatch. Approximately 12–24 hours after hatch, chicks were orally infected with C. jejuni strains.

C. jejuni strains for infections were grown from frozen stocks on MH agar in microaerobic conditions at 37 ºC for 48 hours. Bacteria were restreaked and grown on MH agar under microaerobic conditions at 37 ºC for 16 hours. Strains were resuspended from plates in MH broth and diluted to OD₆₀₀ 0.4. Samples were diluted 1:5,000 in PBS and 100 μl of the diluted sample (containing approximately 2 x 10⁴ cfu) were used to orally infect the 1-day old chicks. Dilutions of each inoculum were plated on MH agar containing TMP to determine the actual number of bacteria used to infect chicks. Inocula were also grown on MH agar containing TMP and X-S to verify that all strains were in the same phase for FlgS production.

At days 1, 7, and 14 post-infection, five to ten chicks were sacrificed and the cecal contents were collected, weighed, and suspended to a concentration of 0.1 g per ml PBS. Tenfold serial dilutions of the cecal contents were spread on MH agar containing TMP and cefoperazone and on MH agar containing TMP, cefoperazone, and X-S to enumerate C. jejuni in the ceca and determine the percentage of bacteria positive for expression of flaB::astA (displaying the AstA⁺, blue-colony phenotype). Statistical analysis to determine the significance of differences in the colonization abilities of the wild-type and phase ‘off’ variants was determined by using the Mann-Whitney U test.

Construction of mutants with flgS alleles from in vitro-isolated, non-motile variants

The flgS alleles of the in vivo-isolated variants DRH665 5W2, DRH665 5W6, and DRH665 10W1 were amplified with oligonucleotides containing BamHI restriction sites at their 5’ ends. The amplified DNA contained the flgS coding sequence along with approximately 500 bases of upstream and downstream sequence. After cloning of the alleles into BamHI-digested pUC19, the plasmids were sequenced to ensure that the correct sequence was amplified and cloned. The correctly-constructed plasmids included pDRH2672 (containing flgS5W2), pDRH2708 (containing flgS5W6), and pDRH2762 (containing flgS10W1).

For construction of C. jejuni 81–176 Sm^R ΔastA strains containing alleles from the flgS variants, a strain containing flgS::cat-rpsL first had to be created. For this step, DRH461 (81–176 Sm^R ΔastA) was electroporated with pDRH426 (containing flgS::cat-rpsL; (Hendrixson and DiRita, 2003)) to create DRH2704 (81–176 Sm^R ΔastA flgS::cat-rpsL), which replaced wild-type flgS with flgS::cat-rpsL.

To replace flgS::cat-rpsL with the flgS alleles from the variants, DRH441 (81–176 Sm^R flgS::cat-rpsL; (Hendrixson and DiRita, 2003)) was electroporated with pDRH2672 and pDRH2708 and DRH2704 (81–176 Sm^R ΔastA flgS::cat-rpsL) was electroporated with pDRH2672, pDRH2708, and pDRH2762. Transformants were recovered on MH agar containing streptomycin and colony PCR was used to identify correct transformants. The flgS alleles of putative transformants were sequenced to verify correct replacement of flgS::cat-rpsL with the flgS alleles from the variants. Recovered transformants included DRH2714 (81–176 Sm^R flgS5W2), DRH2715 (81–176 Sm^R flgS5W6), DRH2717 (81–176 ΔastA Sm^R flgS5W2), DRH2724 (81–176 ΔastA Sm^R flgS5W6), and DRH2777 (81–176 ΔastA Sm^R flgS10W1).

Introduction of the flaB::astA transcriptional fusion into DRH2717, DRH2724, and DRH2777 was achieved by electroporation of DRH610 into the strains (Hendrixson and DiRita, 2003). Transformants were selected on MH agar containing kanamycin and colony PCR was performed to ensure correct construction of strains. The resulting transformants included DRH2731 (81–176 ΔastA Sm^R flgS5W2 flaB::astA), DRH2737 (81–176 ΔastA Sm^R flgS5W6 flaB::astA), and DRH2781 (81–176 ΔastA Sm^R flgS10W1 flaB::astA).

Construction of flgR and flgS mutants in in vivo- and in vitro-isolated revertants

DRH665 5W2 rev8, DRH665 5W2 rev9, DRH665 5W2 rev10, DRH665 5W6 rev4, DRH665 5W6 rev8, and DRH665 5W6 rev10 were electroporated with pDRH426 (containing flgS::cat-rpsL; (Hendrixson and DiRita, 2003)) and pDRH431 (containing flgR::cat-rpsL). pDRH431 was constructed by first digesting pDRH265 with SmaI to release a DNA cassette containing cat linked to a wild-type rpsL allele (Hendrixson et al., 2001). This SmaI fragment was then inserted into the PmeI site of flgR in pDRH428 to create pDRH431 (Hendrixson and DiRita, 2003). C. jejuni transformants were recovered on MH agar with chloramphenicol after growth in microaerobic conditions at 37 ºC and screened by colony PCR to verify correct replacement of flgS with flgS::cat-rpsL or flgR with flgR::cat-rpsL on the chromosome.

Replacement of flgR with the flgR5W6 rev10 allele in ΔflgS and flgS5W6 mutants

To clone the flgR allele from DRH665 5W6 rev10, chromosomal DNA was purified and used in PCR with primers to amplify a 3.2-kb DNA fragment with 5’ and 3’ BamHI restriction sites. After digestion with BamHI, the DNA fragment was ligated into BamHI-digested pUC19 to create pDRH2836.

To create the C. jejuni 81–176 ΔastA Sm^R flgS5W6 flgR::cat-rpsL strain necessary for replacement of flgR with the flgR5W6rev10 allele, DRH2724 (81–176 ΔastA Sm^R flgS5W6) was electroporated with pDRH443 (containing flgR::kan-rpsL) (Hendrixson and DiRita, 2003). A correct transformant, DRH2851, was identified after growth on MH agar containing kanamycin. For construction of the final mutant, SNJ767 (81–176 Sm^R ΔflgS flgR::kan-rpsL; (Joslin and Hendrixson, 2008)) and DRH2851 (81–176 Sm^R ΔastA flgS5W6 flgR::kan-rpsL) were electroporated with pDRH2836. Tranformants were recovered on MH agar containing streptomycin and colony PCR was used to identify correct transformants. The flgR locus was sequenced from putative transformants to verify correct replacement of flgR::kan-rpsL with flgR5W6rev10. This approached allowed the recovery of DRH2853 (81–176 Sm^R ΔflgS flgR5W6rev10) and DRH2861 (81–176 Sm^R ΔastA flgS5W6 flgR5W6rev10). DRH2861 was then electroporated with pDRH610 to replace flaB with flaB::astA to create DRH2874.

Table 4

Arylsulfatase activity due to expression of the flaB::astA transcriptional fusion in flgS and flgR mutants of in vivo-isolated revertants of C. jejuni DRH665 5W2 and DRH665 5W6.^a

Acknowledgments

I thank Eric Hansen and Deborah Ribardo for helpful discussions critical reading of this manuscript. This work was supported by the National Institutes of Health grant R01 AI065539 to D.R.H. and the United States Department of Agriculture National Research Initiative Grants no. 2006-35201-17382 from the USDA Cooperative State Research, Education, and Extension Service Food Safety (32.0) program.

References

Akerley BJ, Lampe DJ. Analysis of gene function in bacterial pathogens by GAMBIT. Methods Enzymol. 2002;358:100–108. [PubMed]
Alm RA, Ling LS, Moir DT, King BL, Brown ED, Doig PC, et al. Genomic-sequence comparison of two unrelated isolates of the human gastric pathogen Helicobacter pylori. Nature. 1999;397:176–180. [PubMed]
Bacon DJ, Szymanski CM, Burr DH, Silver RP, Alm RA, Guerry P. A phase-variable capsule is involved in virulence of Campylobacter jejuni 81–176. Mol Microbiol. 2001;40:769–777. [PubMed]
Beery JT, Hugdahl MB, Doyle MP. Colonization of gastrointestinal tracts of chicks by Campylobacter jejuni. Appl Environ Microbiol. 1988;54:2365–2370. [PMC free article] [PubMed]
Bingham-Ramos LK, Hendrixson DR. Characterization of two putative cytochrome c peroxidases of Campylobacter jejuni involved in promoting commensal colonization of poultry. Infect Immun. 2008;76:1105–1114. [PMC free article] [PubMed]
Black RE, Levine MM, Clements ML, Hughes TP, Blaser MJ. Experimental Campylobacter jejuni infection in humans. J Infect Dis. 1988;157:472–479. [PubMed]
Blaser MJ, Reller LB. Campylobacter enteritis. N Engl J Med. 1981;305:1444–1452. [PubMed]
Brahmachary P, Dashti MG, Olson JW, Hoover TR. Helicobacter pylori FlgR is an enhancer-independent activator of σ⁵⁴-RNA polymerase holoenzyme. J Bacteriol. 2004;186:4535–4542. [PMC free article] [PubMed]
Caldwell MB, Guerry P, Lee EC, Burans JP, Walker RI. Reversible expression of flagella in Campylobacter jejuni. Infect Immun. 1985;50:941–943. [PMC free article] [PubMed]
Carrillo CD, Taboada E, Nash JH, Lanthier P, Kelly J, Lau PC, et al. Genome-wide expression analyses of Campylobacter jejuni NCTC11168 reveals coordinate regulation of motility and virulence by flhA. J Biol Chem. 2004;279:20327–20338. [PubMed]
Chang AC, Cohen SN. Construction and characterization of amplifiable multicopy DNA cloning vehicles derived from the P15A cryptic miniplasmid. J Bacteriol. 1978;134:1141–1156. [PMC free article] [PubMed]
Correa NE, Lauriano CM, McGee R, Klose KE. Phosphorylation of the flagellar regulatory protein FlrC is necessary for Vibrio cholerae motility and enhanced colonization. Mol Microbiol. 2000;35:743–755. [PubMed]
Dasgupta N, Wolfgang MC, Goodman AL, Arora SK, Jyot J, Lory S, Ramphal R. A four-tiered transcriptional regulatory circuit controls flagellar biogenesis in Pseudomonas aeruginosa. Mol Microbiol. 2003;50:809–824. [PubMed]
Eisen JA. A phylogenomic study of the MutS family of proteins. Nucleic Acids Res. 1998;26:4291–4300. [PMC free article] [PubMed]
Fouts DE, Mongodin EF, Mandrell RE, Miller WG, Rasko DA, Ravel J, et al. Major structural differences and novel potential virulence mechanisms from the genomes of multiple campylobacter species. PLoS Biol. 2005;3:e15. [PMC free article] [PubMed]
Godschalk PC, Kuijf ML, Li J, St Michael F, Ang CW, Jacobs BC, et al. Structural characterization of Campylobacter jejuni lipooligosaccharide outer cores associated with Guillain-Barre and Miller Fisher syndromes. Infect Immun. 2007;75:1245–1254. [PMC free article] [PubMed]
Guerry P, Alm RA, Power ME, Logan SM, Trust TJ. Role of two flagellin genes in Campylobacter motility. J Bacteriol. 1991;173:4757–4764. [PMC free article] [PubMed]
Guerry P, Szymanski CM, Prendergast MM, Hickey TE, Ewing CP, Pattarini DL, Moran AP. Phase variation of Campylobacter jejuni 81–176 lipooligosaccharide affects ganglioside mimicry and invasiveness in vitro. Infect Immun. 2002;70:787–793. [PMC free article] [PubMed]
Hanninen ML, Hannula M. Spontaneous mutation frequency and emergence of ciprofloxacin resistance in Campylobacter jejuni and Campylobacter coli. J Antimicrob Chemother. 2007;60:1251–1257. [PubMed]
Henderson MJ, Milazzo FH. Arylsulfatase in Salmonella typhimurium: detection and influence of carbon source and tyramine on its synthesis. Journal of Bacteriology. 1979;139:80–87. [PMC free article] [PubMed]
Hendrixson DR, Akerley BJ, DiRita VJ. Transposon mutagenesis of Campylobacter jejuni identifies a bipartite energy taxis system required for motility. Mol Microbiol. 2001;40:214–224. [PubMed]
Hendrixson DR, DiRita VJ. Transcription of σ⁵⁴-dependent but not σ²⁸-dependent flagellar genes in Campylobacter jejuni is associated with formation of the flagellar secretory apparatus. Mol Microbiol. 2003;50:687–702. [PubMed]
Hendrixson DR, DiRita VJ. Identification of Campylobacter jejuni genes involved in commensal colonization of the chick gastrointestinal tract. Mol Microbiol. 2004;52:471–484. [PubMed]
Hendrixson DR. A phase-variable mechanism controlling the Campylobacter jejuni FlgR response regulator influences commensalism. Mol Microbiol. 2006;61:1646–1659. [PubMed]
Hendrixson DR. Regulation of Flagellar Gene Expression. In: Nachamkin I, Szymanski CM, Blaser MJ, editors. Campylobacter. Washington, DC: ASM Press; 2008. pp. 545–558.
Jagannathan A, Constantinidou C, Penn CW. Roles of rpoN, fliA, and flgR in expression of flagella in Campylobacter jejuni. J Bacteriol. 2001;183:2937–2942. [PMC free article] [PubMed]
Joslin SN, Hendrixson DR. Analysis of the Campylobacter jejuni FlgR response regulator suggests integration of diverse mechanisms to activate an NtrC-like protein. J Bacteriol. 2008;190:2422–2433. [PMC free article] [PubMed]
Karlyshev AV, Champion OL, Churcher C, Brisson JR, Jarrell HC, Gilbert M, et al. Analysis of Campylobacter jejuni capsular loci reveals multiple mechanisms for the generation of structural diversity and the ability to form complex heptoses. Mol Microbiol. 2005a;55:90–103. [PubMed]
Karlyshev AV, Ketley JM, Wren BW. The Campylobacter jejuni glycome. FEMS Microbiol Rev. 2005b;29:377–390. [PubMed]
Klose KE, Mekalanos JJ. Distinct roles of an alternative sigma factor during both free-swimming and colonizing phases of the Vibrio cholerae pathogenic cycle. Mol Microbiol. 1998;28:501–520. [PubMed]
Korlath JA, Osterholm MT, Judy LA, Forfang JC, Robinson RA. A point-source outbreak of campylobacteriosis associated with consumption of raw milk. J Infect Dis. 1985;152:592–596. [PubMed]
Kunkel TA, Erie DA. DNA mismatch repair. Annu Rev Biochem. 2005;74:681–710. [PubMed]
Lindblom GB, Sjorgren E, Kaijser B. Natural campylobacter colonization in chickens raised under different environmental conditions. J Hyg (Lond) 1986;96:385–391. [PMC free article] [PubMed]
Linton D, Gilbert M, Hitchen PG, Dell A, Morris HR, Wakarchuk WW, Gregson NA, Wren BW. Phase variation of a beta-1,3 galactosyltransferase involved in generation of the ganglioside GM1-like lipo-oligosaccharide of Campylobacter jejuni. Mol Microbiol. 2000;37:501–514. [PubMed]
Logan SM, Kelly JF, Thibault P, Ewing CP, Guerry P. Structural heterogeneity of carbohydrate modifications affects serospecificity of Campylobacter flagellins. Mol Microbiol. 2002;46:587–597. [PubMed]
Nachamkin I, Yang XH, Stern NJ. Role of Campylobacter jejuni flagella as colonization factors for three-day-old chicks: analysis with flagellar mutants. Appl Environ Microbiol. 1993;59:1269–1273. [PMC free article] [PubMed]
Niehus E, Gressmann H, Ye F, Schlapbach R, Dehio M, Dehio C, et al. Genome-wide analysis of transcriptional hierarchy and feedback regulation in the flagellar system of Helicobacter pylori. Mol Microbiol. 2004;52:947–961. [PubMed]
Nuijten PJM, Bleumink-Pluym NMC, Gaastra W, van der Zeijst BAM. Flagellin expression in Campylobacter jejuni is regulated at the transcriptional level. Infect Immun. 1989;57:1084–1088. [PMC free article] [PubMed]
Parkhill J, Wren BW, Mungall K, Ketley JM, Churcher C, Basham D, et al. The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences. Nature. 2000;403:665–668. [PubMed]
Pokamunski S, Kass N, Borochovich E, Marantz B, Rogol M. Incidence of Campylobacter spp. in broiler flocks monitored from hatching to slaughter. Avian Pathology. 1986;15:83–92. [PubMed]
Prendergast MM, Tribble DR, Baqar S, Scott DA, Ferris JA, Walker RI, Moran AP. In vivo phase variation and serologic response to lipooligosaccharide of Campylobacter jejuni in experimental human infection. Infect Immun. 2004;72:916–922. [PMC free article] [PubMed]
Prouty MG, Correa NE, Klose KE. The novel σ⁵⁴- and σ²⁸-dependent flagellar gene transcription hierarchy of Vibrio cholerae. Mol Microbiol. 2001;39:1595–1609. [PubMed]
Ritchings BW, Almira EC, Lory S, Ramphal R. Cloning and phenotypic characterization of fleS and fleR, new response regulators of Pseudomonas aeruginosa which regulate motility and adhesion to mucin. Infect Immun. 1995;63:4868–4876. [PMC free article] [PubMed]
Rose RE. The nucleotide sequence of pACYC184. Nucleic Acids Res. 1988;16:355. [PMC free article] [PubMed]
Skirrow MB, Blaser MJ. Clinical aspects of Campylobacter infection. In: Nachamkin I, Blaser MJ, editors. Campylobacter. Washington, DC: ASM Press; 2000. pp. 69–88.
Spohn G, Scarlato V. Motility of Helicobacter pylori is coordinately regulated by the transcriptional activator FlgR, an NtrC homolog. J Bacteriol. 1999;181:593–599. [PMC free article] [PubMed]
Stibitz S, Aaronson W, Monack D, Falkow S. Phase variation in Bordetella pertussis by frameshift mutation in a gene for a novel two-component system. Nature. 1989;338:266–269. [PubMed]
Szymanski CM, Michael FS, Jarrell HC, Li J, Gilbert M, Larocque S, Vinogradov E, Brisson JR. Detection of conserved N-linked glycans and phase-variable lipooligosaccharides and capsules from campylobacter cells by mass spectrometry and high resolution magic angle spinning NMR spectroscopy. J Biol Chem. 2003;278:24509–24520. [PubMed]
Tomb JF, White O, Kerlavage AR, Clayton RA, Sutton GG, Fleischmann RD, et al. The complete genome sequence of the gastric pathogen Helicobacter pylori. Nature. 1997;388:539–547. [PubMed]
Traber K, Novick R. A slipped-mispairing mutation in AgrA of laboratory strains and clinical isolates results in delayed activation of agr and failure to translate delta- and alpha-haemolysins. Mol Microbiol. 2006;59:1519–1530. [PubMed]
van der Woude MW, Baumler AJ. Phase and antigenic variation in bacteria. Clin Microbiol Rev. 2004;17:581–611. table of contents. [PMC free article] [PubMed]
Wang G, Alamuri P, Humayun MZ, Taylor DE, Maier RJ. The Helicobacter pylori MutS protein confers protection from oxidative DNA damage. Mol Microbiol. 2005;58:166–176. [PubMed]
Wassenaar TM, Bleumink-Pluym NMC, van der Zeijst BAM. Inactivation of Campylobacter jejuni flagellin genes by homologous recombination demonstrates that flaA but not flaB is required for invasion. Embo J. 1991;10:2055–2061. [PubMed]
Wassenaar TM, van der Zeijst BAM, Ayling R, Newell DG. Colonization of chicks by motility mutants of Campylobacter jejuni demonstrates the importance of flagellin A expression. J Gen Microbiol. 1993;139(Pt 6):1171–1175. [PubMed]
Wosten MMSM, Wagenaar JA, van Putten JPM. The FlgS/FlgR two-component signal transduction system regulates the fla regulon in Campylobacter jejuni. J Biol Chem. 2004;279:16214–16222. [PubMed]
Yao R, Guerry P. Molecular cloning and site-specific mutagenesis of a gene involved in arylsulfatase production in Campylobacter jejuni. J Bacteriol. 1996;178:3335–3338. [PMC free article] [PubMed]