Isolation of a gene that is involved in Campylobacter jejuni 81116 cytotoxin activation


  • Editor: George Mendz

Correspondence: Peter J. Coloe, School of Applied Sciences, Building 223 Level 1, Plenty Road, Bundoora, Victoria, Australia. 3083. Tel.: +61 3 99257104; fax: +61 3 99257110; e-mail:


Cytotoxin fractions were isolated from Campylobacter jejuni 81116 and semi-purified by size-exclusion liquid chromatography. The fraction showing the strongest toxicity was injected into mice to produce antiserum. The antiserum was used to screen a C. jejuni 81116 cosmid library. Nine genes were identified in overlapping cosmid inserts that induced reactivity with the antiserum. One of these genes showed high similarity to a periplasmic protein of unknown function and its isogenic mutant showed decreased toxicity compared to the C. jejuni 81116 wild type. This gene contains a Gram-negative bacterial RTX toxin-activating protein C signature, which suggests it may play a role in C. jejuni 81116 cytotoxin activation.


Campylobacter jejuni is now recognized as the most common bacterial cause of diarrhoea throughout the world and its infection is responsible for a major public health and economic burden (Blaser et al., 1983; Skirrow & Blaser, 1992). The pathophysiology of Campylobacter infections is not fully understood; consequently this hinders the development of control mechanisms for Campylobacter-induced disease. The potential virulence determinants of C. jejuni are recognized to include: adherence, motility, invasion and toxin production (Fauchere et al., 1992; Wassenaar, 1997). One important factor in adherence is lipooligosaccharide (LOS), a major component of the Gram-negative outer membrane, which also plays a role in serum resistance and endotoxicity (McSweegan & Walker, 1986; van Vliet & Ketley, 2001). Infection by C. jejuni leads to enterocolitis involving intestinal tissue damage, which indicates that host cell invasion and perhaps cytotoxin production with subsequent tissue destruction are likely to be key elements in pathogenesis (Ketley, 1997).

Toxin production by C. jejuni has been reported by many research groups. Wassenaar (1997) has proposed six different cytotoxins based on cell specificity. Hanel et al. (1998) suggested two different kinds of cytotoxins, ‘cytolethal distending toxin’ (CDT) and ‘cytolethal rounding toxin’ (CRT), based on morphological changes of tissue cell lines. However, the characterisation of cytotoxin production by Campylobacter spp. has been a slow progress. So far there is no evidence on how many different cytotoxins C. jejuni actually produces (Hanel et al., 1998). Except for CDT, no other Campylobacter cytotoxins have been identified and sequenced. As with other bacterial pathogens, it is assumed that the toxin formation by C. jejuni is complex, involving a range of different toxins expressed under various, as yet unknown conditions (Misawa et al., 1994; Wassenaar, 1997). This may be a consequence of different research groups using different tissue culture cell lines for cytotoxin detection, as well as different growth conditions for cytotoxin production. In addition, surface toxins, such as LOS, may also be implicated in cell cytotoxicity (McSweegan & Walker, 1986; van Vliet & Ketley, 2001). It has been postulated that some strains of C. jejuni produce cytotoxin (Lee et al., 2000), and so, to help in understanding if there is a role for a discrete toxin in C. jejuni pathogenesis, the isolation and characterisation of specific toxin-encoding genes is essential.

In mutagenesis studies on the C. jejuni 81116 cdt gene, we have found that this strain can produce toxic factors other than CDT (unpublished data), and so a C. jejuni 81116 cosmid library was screened in a search for cytotoxin(s) encoding genes using specific antiserum raised against a C. jejuni semi-purified cytotoxic fraction prepared in this study.

Materials and methods

Bacterial strains and culture conditions

The bacterial strains and plasmids used in this work are listed in Table 1. Campylobacter jejuni was routinely grown under microaerophilic conditions (10% CO2, 5% O2, in N2) on Columbia agar supplemented with 5% (v/v) defibrinated horse blood and Campylobacter selective Skirrow Supplement for 48 h, or in Brucella Broth for 96 h at 37°C. Escherichia coli strains were grown in Luria–Bertani (LB) broth or agar at 37°C overnight. When necessary to maintain plasmid or mutant selection, ampicillin was added to growth media at a final concentration of 100 mg L−1, tetracycline was added at a final concentration of 20 mg L−1 and kanamycin was added at a final concentration of 30 mg L−1. To minimize the in vitro passage, C. jejuni was stored at −70°C in milk-freezing medium before it was resuscitated on Skirrow medium and then subcultured in Brucella Broth.

Table 1.   Bacterial strains and plasmids used in this study
Bacterial strains and plasmidsRelevant genotype or phenotypeSource or reference
E. coli strains
 DH5αF′/endA1, hsdR17(rKm+K), subE44, thi-1, recA1, gyrA (NalR), relA1, Δ(lacZYA-argF)U169 (m80lacZΔM15)Hanahan (1983)
 HB101Δ(gpt- proA)62, leuB6, thi-1, lacY1, hsdSB20, recA, rpsL20 (StrR), ara-14, galK2, xyl-5, mtl-1, subE44, mcrBBBoyer & Roulland-Dussoix (1969)
C. jejuni
 81116 (Human isolate)N/APalmer & Gully (1983)
 pBluescriptSKIIAmpR, lacZStratagene
 pMW2kanamycin resistance cassette in pBlWösten, unpublished
 pLA2917 Ditta et al. (1980)

Preparation of crude cytotoxin fraction

For general toxicity testing, C. jejuni strains were inoculated in Brucella Broth at a starting OD600 nm of 0.001 and were incubated at 37°C for 96 h. A culture supernate fraction was prepared by centrifuging 10 mL of bacterial culture at 14 000 g for 30 min at 4°C. The supernate was collected and passed through a 0.2-μm filter. A sonicated fraction was prepared by resuspending the cell pellet from 10 mL of culture in 2 mL 10 mM Tris-Cl (pH 8.0) buffer and sonicating at 70 W for 30 s using a Branson digital sonifier 250 (Danbury). Cell debris was removed by centrifugation. The supernate was collected and passed through a 0.2-μm filter.

Semi-purification of C. jejuni 81116 cytotoxins

Two litres of C. jejuni culture were centrifuged at 14 000 g for 30 min at 4°C and the supernatant was collected, passed through a 0.2-μm filter and concentrated 200-fold using an ultrafilter equipped with a 50 kDa molecular weight cut-off membrane (Amicon® series 8000, Millipore). The cytotoxic protein was purified with a self-packed 1.6 × 40 cm Pharmacia C column with Sephacryl S-200 size exclusion resin (Amersham Pharmacia, Sweden), equipped with a Monitor UV-1 (Amersham Pharmacia, Sweden). One millilitre of concentrated sample was added to the column and eluted with 10 mM Tris-Cl (pH 8.0) at a flow rate of 1 mL min−1. Fractions were collected at 1 min intervals and pooled together based on the elution profile. The pooled fractions were then concentrated 10-fold by ultrafiltration and tested for cytotoxicity using the cytotoxin and MTT assay.

Cytotoxin and MTT assay

Protein content was determined using the Bradford method (Bradford, 1976). Tissue culture cells were cultured in 25 cm2 tissue culture flasks (Greiner, Germany) in Dulbecoo's modification of Eagle's Medium (DMEM) supplementing with 10% (v/v) newborn bovine serum (NBS) and 5 μg mL−1 penicillin/streptomycin at 37°C in air with 5% CO2 until confluent. Chinese hamster ovary (CHO) cells and human tumour epithelial (HeLa) cells were purchased from CSL Limited (Melbourne, Australia). Human intestinal epithelial (Int407) cells were supplied by Dr Peter Ward (Royal Children's Hospital, Melbourne, Australia). Human larynx epidermoid carcinoma (HEp-2) cells were stock of the Biotechnology laboratory (RMIT University, Melbourne, Australia). Ninety-six well plates were seeded with 104 tissue culture cells in 100 μL media and incubated overnight before cytotoxicity determination. Cytotoxicity was quantified as previously described (Coote & Arain, 1996), except that the plates were incubated at 37°C for 72 h. Morphological changes were examined at 24 h intervals by phase-contrast microscopy. Uninoculated Brucella Broth and sterile 10 mM Tris-Cl (pH 8.0) were used as the negative controls. After 72 h of incubation, the percentage of cell death was determined using a MTT assay (Coote & Arain, 1996). The plates were read at a wavelength of 600 nm using a Dynatech MR700 microplate reader (Chantilly, VA) to determine the absorbance. 100 microliters of 0.04 N HCl in dimethyl sulfoxide (DMSO) was used as a blank. The percentage of cell death was calculated using the following equation.


The cytotoxicity was expressed as tissue culture dose 50 (TCD50). A TCD50 was defined as the amount of the toxic fraction that caused death in 50% of the cells at the chosen time-point (Thelestam & Florin, 1994; Bacon et al., 1999).

Neutralization and Western blot analysis

A selected fraction eluted from a size exclusion column with a protein concentration of 0.1 mg mL−1 was detoxified by mixing with 0.5% (v/v) formaldehyde solution (pH 7.5) and incubating at 37°C overnight. Female Balb/c mice were used to raise polyclonal antiserum against this fraction. Mice were administered 0.4 mL of pristane intraperitoneally and were stimulated intraperitoneally with 10 μg of detoxified cytotoxic fraction together with 100 μL of Incomplete Freund's adjuvant after 7 and 21 days, followed by a third injection with 10 μg of the cytotoxic fraction and 100 μL of Incomplete Freund's adjuvant after 31 days. At Day 41, mice were injected with 107 Sp/2 cells. Antiserum was collected from Day 51 to 53.

For neutralization studies, the selected fraction at a protein concentration of 1 mg mL−1 was mixed with the specific antiserum at a 1 : 1 ratio (v/v) and incubated at 37°C for 2 h before cytotoxicity determination. The fraction mixed with sterilised 10 mM Tris-Cl (pH 8.0) at a 1 : 1 ratio was incubated the same way and used as the control.

The antiserum was also analysed using Western blots according to Sambrook et al. (1989). After electrophoretic transfer, the nitrocellulose membrane was blocked by incubation with 5% (w/v) skim milk in Tris saline and tween buffer for 1 h. A one in 50 dilution of the antiserum was added to the membrane at 4°C overnight. Goat antimouse horseradish peroxidase-conjugated (HRP) immunoglobulins were used to probe the membrane. The bound peroxidase was visualised by incubating in 4 Chloro-1-naphthol solution for about 5–15 min in the dark.

Screening of the cosmid library by colony lift

The cosmid library used in this study was constructed by Dr V. Korolik with a cosmid vector pLA2917 ligated with Sau3AI partially cleaved C. jejuni 81116 genome DNA fragments and transformed into E. coli HB101 (Fry et al., 1998). The C. jejuni 81116 cosmid library was screened using the colony lift method with E. coli absorbed antiserum (Sambrook et al., 1989). The E. coli HB101 carrying the vector pLA2917 and the C. jejuni 81116 culture supernatant were used as the negative and positive controls, respectively. The experiment was repeated twice for consistency.

DNA manipulation

All restriction endonucleases and T4 DNA ligase were purchased from Promega, Australia, and used according to the manufacturer's instruction. Plasmid and cosmid DNA was isolated from E. coli using a modification of the alkaline lysis method (Birnboim & Doly, 1979). Chromosomal DNA was isolated from pure cultures using the cetyl-trimelthyl ammonium bromide (CTAB) procedure (Sambrook et al., 1989).

Polymerase chain reaction

The cosmids from selected clones were sequenced using the cosmid pLA2917 sequencing primers COS3 (5′-ACAGCAAGCGAACCGGAATTG-3′) and COS4 (5′-ATCATGCGAAACGATCCTCA-3′). Both primers were kindly provided by J. Chang, RMIT University. Primers CJTP1 (5′-GAAATTCAAGGTGGACGATGTATGGAT-3′) and CJTP2 (5′-TTCTTAGGAAAGCGTGATGTTAAAGC-3′) were designed based on the resulting C. jejuni 81116 DNA sequence. PCRs were performed using C. jejuni 81116 genome DNA (positive control), pLA2917 (negative control), and the cosmids of the selected clones using the Expand Long Template PCR system (Roche). The PCR condition was 10 cycles of 10 s at 93°C, 1 min at 60°C, 7 min at 68°C, 20 cycles of 10 s at 93°C, 1 min at 60°C, 7 min at 68°C with 20 s extra each cycle, and a final extension step of 7 min at 68°C.

Sequencing and sequence analysis

DNA was amplified using an ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction kit (Perkin-Elmer, Australia). The amplified DNA was then precipitated using ethanol and sodium acetate as described in the manufacturer's protocol and was sequenced using the ABI Prism 377 DNA Sequencer with XL Upgrade (Perkin-Elmer, Australia) at Monash University (Clayton Campus), Victoria, Australia. Homology searches for DNA and amino acid sequence was conducted using NCBI basic local alignment search tool (blast) ( The Sci Ed Central clone manager suite program was used for DNA sequences alignment, translation, restriction map construction and primer design. The program ‘Identify an unknown protein sequence’ provided by BioManager, ANGIS ( was used to identify proteins with unknown function.

Creation and analysis of knock-out mutants

The orf6 gene plus the orf7 gene were amplified together with the primers ORF6P6 (5′-GCGCGGATCC-CATAAAGGAGTTATTATGAAA-3′) and ORF7P7 (5′-CAGTAAGCTT-GAAAATACCCGCTAATATACC-3′) (restriction sites are indicated in bold). The resulting PCR products were digested with HindIII and BamHI, ligated with pBluescriptSKII digested by HindIII and BamHI, and transformed into E. coli DH5α. The plasmid was named Orf6-7pB. Inverse PCR was performed to delete part of orf6 using Orf6-7pB plasmid DNA and primer pairs ORF6IN1 (5′-CAGTAGATCT-CAGTAGGTAAGCCTAGTTGT-3′) and ORF6IN2 (5′-CAGTAGATCT-CCACCACTTTCTGAAATTCC-3′). The PCR conditions were 35 cycles of 30 s at 94°C, 30 s at 58°C, 9 min at 72°C, and a final extension step of 10 min at 72°C, with Pfu DNA polymerase (Promega, Australia). After inverse PCR, digestion with BglII and self-ligation, plasmid Orf6-7pBIN was obtained containing a unique BglII site in orf6. This site was then used for inserting a kanamycin resistance cassette originating from plasmid pMW2 containing the BamHI end to create plasmid Orf6pBKF. The orientation of the kanamycin resistance cassette was determined by digesting the mutation constructs with restriction enzyme ClaI. Natural transformation was used to introduce the constructs in C. jejuni 81116 as described previously (Wassenaar et al., 1993). The resulting mutants were confirmed using PCR and restriction digestions.

Results and Discussion

Semi-purification of C. jejuni cytotoxins

Size-exclusion low-pressure chromatography was used to separate the C. jejuni 81116 culture supernatant based on molecular weight. The column, of a 30-mL void volume, allowed the semipurified toxin preparation to be separated into 7 × 10 mL fractions. All fractions were tested in a cytotoxin assay and the fraction (Fraction A) collected between retention time 50 and 60 min caused a strong rounding effect on CHO cells (Fig. 1b). Fraction A was used as a semi-purified sample to inject into mice for antiserum production.

Figure 1.

 (a) Size exclusion liquid chromatography elution profile of the C. jejuni 81116 concentrated culture supernatant. ‘a’ indicates the fraction collected according to the elution time and caused a strong rounding effect on CHO cells (b); (c) Normal CHO cells.

Western blot and neutralisation study

To test the antiserum raised against Fraction A, the protein fraction was separated by SDS-PAGE and subjected to Western blot analysis. After incubating with the antiserum, four bands with sizes of 45, 33, 24 and 20 kDa were recognized by the antiserum (Fig. 2b). After incubating Fraction A with the antiserum at 37°C for 2 h, its cytotoxicity decreased 92% (Fig. 2a). This indicated that this antiserum could react with the toxic factor(s) in Fraction A.

Figure 2.

 (a) Specific cytotoxicity of C. jejuni 81116 culture supernatant Fraction A before and after neutralisation by the antiserum raised against Fraction A. (b) SDS-PAGE analysis of the C. jejuni 81116 culture supernatant fractions visualized with Western blot probed with antiserum raised against Fraction A Lane 1, SeeBlue® Plus2 PreStained Protein Standard (Invitrogen, Australia); Lane 2, Fraction A. 30 μg (protein content) of Fraction A was loaded on the gel.

Screening of the cosmid library by colony lift

The antiserum was then used to screen the C. jejuni 81116 cosmid library using the colony lift method with a positive and a negative control. The following eight clones produced a positive reaction: 1C12, 1F12, 3C5, 6A8, 6A9, 6A10, 6A11 and 6A12.

Genetic analysis of the positive clones

Restriction enzyme digestion

The positive cosmids and pLA2917 were isolated and digested with HindIII, EcoRV and BglII. The patterns of these restriction digestion maps revealed by agarose gel electrophoresis showed that there were only four discrete clones (Fig. 3). The clones 6A8, 6A9, 6A10, 6A11 and 6A12 were the same, as was shown in all three patterns. These three digestion patterns also showed that there were differences between 1C12 and 3C5, although they shared a large overlapping fragment.

Figure 3.

 Cosmids of the selected clones digested by BglII (a), EcoRV (b) and HindIII (c). Lane 1, λDNA × PstI; Lane 2-10, pLA2917, cosmids of 1C12, 1F12, 3C5, 6A8, 6A9, 6A10, 6A11 and 6A12.

DNA sequencing and analysis

The cosmid DNA of the clones 1F12, 3C5 and 6A9 were subjected to sequencing. About 1000 bps from both ends of the inserted fragments were sequenced and aligned with the C. jejuni NCTC 11168 genome DNA sequence (accession no. AL139074). The most likely overlap was identified to range from the position 14 383 to the position 22 801 on the C. jejuni NCTC 11168 genome DNA sequence.

PCR, sequencing and ORF analysis

Using primers CTP1 and CTP2, PCR amplicons obtained from the C. jejuni 81116 genome DNA and the clones 1C12, 1F12, 3C5, 6A8, 6A9, 6A10, 6A11 and 6A12 had a similar size. No amplicon was obtained from pLA2917. This confirmed that all the clones had the overlapping sequence. The DNA sequence of the entire overlapping fragment was obtained (GenBank accession no. DQ202384) and showed 97% identity to the C. jejuni NCTC 11168 DNA sequence. The overlapping fragment contained eight complete ORFs and one partial ORF. Their predicted functions are outlined in Table 2.

Table 2.   Predicted functions of the ORFs
ORFAmino AcidsMWpIFunction
  1. Adapted from the C. jejuni NCTC 11168 protein list:

Orf148153 082.308.96Glutamate synthase (NADPH) small subunit (gltD)
Orf215917 944.249.55Ribonuclease HII (rnhB)
Orf3798852.489.44Putative nonspecific DNA binding protein (Cj0011c)
Orf421524 511.855.49Nonhaem iron protein (Cj0012c)
Orf549550 480.738.82Dihydroxy-acid dehydratase (ilvD)
Orf617420 416.236.52Putative integral membrane protein (Cj0014c)
Orf740847 406.615.48Hypothetical protein (Cj0015c)
Orf822425 237.075.13Putative transcriptional regulatory protein (Cj0016)
Orf913915 415.589.47Putative ATP/GTP binding protein (Incomplete)

A conserved domain search showed that Orf1 and Orf5 had putative functions that were responsible for amino acid or carbohydrate transport and metabolism. Orf2 and Orf3 might function in DNA replication, recombination and repair. Orf4 showed high similarity to rubrerythrin, which is responsible for energy production and conversion, and was suggested as an oxidative stress-responsive protein in C. jejuni. Orf8 was predicted to code for a transcriptional regulatory protein. It also showed 99.5% similarity to the PP-loop superfamily ATPases. Apart from the orf6 and orf7 genes, the other six genes were involved in the amino acid and carbohydrate transport and metabolism, DNA replication, recombination, repair, energy production and conversion. No correlation between these functions and cytotoxin production has been found to-date. Orf7 seems to be a bifunctional protein. Its N-terminal region showed high similarity to a competence protein responsible for intracellular trafficking and secretion, and the C-terminal region showed high similarity to a predicted hydrolase of a HD superfamily. These seven genes are considered to be housekeeping genes and hence most likely do not play a role in C. jejuni pathogenesis.

Similarity searches could not predict a function for Orf 6, which was only suggested as an inner membrane protein. A conserved domain search indicated that Orf6 was highly similar to a periplasmic or secreted protein with unknown function. Although these proteins are conserved and exist in a wide range of bacteria, such as E. coli, Staphylococcus aureus, Actinobacillus pleuropneumoniae and Pseudomonas syringae, their function has not been elucidated.

When using the ‘Identify an unknown protein’ program, Orf6 was found to contain a Block similar to PR01489 (RTXTOXINC), which is a four-element fingerprint that provides a signature for the Gram-negative bacterial RTX toxin-activating protein C family. Blocks are multiply aligned, ungapped segments corresponding to the most highly conserved regions of proteins (Henikoff & Henikoff, 1994). The fingerprint, PR01489, was derived from an initial alignment of four sequences: the motifs were drawn from conserved regions spanning virtually the full alignment length (∼150 amino acids). Motif 1 contains the putative active site histidine (Coote, 1992; Trent et al., 1999). Orf6 was found to contain one block similar to this motif at the C-terminal end.

The RTX (repeats in toxin) toxin family includes structurally and functionally related pore-forming cytolysins produced by Gram-negative bacteria. They require Ca2+ for activity and are synthesised as inactive proteins (Coote, 1992; Czuprynski & Welch, 1995). A typical RTX-toxin operon contains four structural genes, which exist in the order rtxCABD. The RtxA protein is the structural component of the exotoxin, whereas RtxB and D are required for extracellular secretion of the toxin. RtxC is an acyl-carrier-protein-dependent acyl-modification enzyme, which is responsible for posttranslational activation of RtxA to form an active toxin (Ludwig, 1996; Lally et al., 1999). The most extensively studied RTX-toxin is the α-haemolysin from E. coli. Studies showed that Protein C of the haemolysin is an internal protein acyltransferase, which catalyses the transfer of a fatty-acyl group from acyl-ACP to R-amino groups of lysine residues 564 and 690 of pro-haemolysin Protein A and converts it to an active toxin (Goebel & Hedgpeth, 1982; Ludwig et al., 1996; Trent et al., 1998). The Blocksearcher results also showed that Orf6 had two blocks similar to UbiA prenyltransferase, but with a high E value. These results might indicate that the Orf6 is a transferase and is involved in toxin activation. However, the similarity between the blocks of Orf6 and RtxC is very low, and Orf6 does not seem to locate in an operon that contains structural toxin genes. To determine whether orf6 is involved in C. jejuni cytotoxin production, this gene was chosen for mutagenesis studies

Cytotoxicity of knock-out mutant

The C. jejuni 81116 wild type and the orf6 mutant were cultured for 96 h at 37°C in Brucella Broth. Their culture supernatant and sonicated culture supernatant fractions were filter sterilized and their cytotoxicity was determined. The Orf6 isogenic mutant showed significant loss in specific toxicity on CHO, HeLa, HEp-2 and Int407 cells compared to the wild type (Fig. 4). Inactivation of Orf6 had a greater effect on HeLa cell cytotoxicity in comparison to the other cell lines used in this study. The loss of activity was mainly in the sonicated culture supernatant, which showed about 70% decrease in toxicity. It indicated that either the protein Orf6 is a toxin and binds very strongly to the cell pellets, or C. jejuni 81116 has a toxic factor that binds very strongly to the cell pellet and needs Orf6 to express activity.

Figure 4.

 Cytotoxin production of the C. jejuni 81116 wild type and the orf6 isogenic mutant. Cytotoxicity was determined using different cell lines. Specific cytotoxicity was calculated by combining the culture supernatant and sonicated culture supernatant fractions. Orf6KF – the C. jejuni 81116 orf6 mutant with the kanamycin resistance cassette in the forward orientation. Cytotoxicity values shown in the figure are average values from three individual samples. Error bars indicate SD.

The results of the mutagenesis experiments described in this paper showed that the orf6 gene was involved in C. jejuni 81116 toxin production. As it contains a Gram-negative RTX toxin-activating protein C signature, it is possible that the Orf6 protein only contributes in C. jejuni cytotoxin activation rather than being an active toxin itself. Further studies need to be conducted to find out its role in C. jejuni cytotoxicity.