Identification and molecular characterisation of CmeB, a Campylobacter jejuni multidrug efflux pump


  • Lilian Pumbwe,

    1. Antimicrobial Agents Research Group, Division of Immunity and Infection, The Medical School, University of Birmingham, Birmingham B15 2TT, UK
    Search for more papers by this author
  • Laura J.V. Piddock

    Corresponding author
    1. Antimicrobial Agents Research Group, Division of Immunity and Infection, The Medical School, University of Birmingham, Birmingham B15 2TT, UK
    Search for more papers by this author

*Corresponding author. Tel.: +44 (121) 414 7915; Fax: +44 (121) 414 3457, E-mail:


A multidrug efflux pump gene (cmeB) was identified from the published Campylobacter jejuni genome sequence. Secondary structural analysis showed that the gene encoded a protein belonging to the resistance nodulation cell division (RND) family of efflux transporters. The gene was inactivated by insertional mutagenesis. Compared with the wild-type strain (NCTC 11168), the resultant knockout strain (NCTC 11168-cmeB::kanr) displayed increased susceptibility to a range of antibiotics including β-lactams, fluoroquinolones, macrolides, chloramphenicol, tetracycline, ethidium bromide, the dye acridine orange and the detergent sodium dodecyl sulfate. Accumulation of ciprofloxacin was increased in the knockout mutant, but carbonyl cyanide m-chlorophenyl hydrazone, a proton motive force inhibitor, had less effect upon ciprofloxacin accumulation in the knockout mutant compared with NCTC 11168. These data show that the identified gene encodes an RND-type multi-substrate efflux transporter, which contributes to intrinsic resistance to a range of structurally unrelated compounds in C. jejuni. This efflux pump has been named CmeB (for Campylobactermultidrug efflux).


Campylobacter is the leading cause of gastrointestinal infection in the Western world, causing more disease than Salmonella and Shigella combined [1,2]. Most human illness is caused by Campylobacter jejuni. Infections are self-limiting with symptoms resolving in about 3–5 days [1,2]. Antibiotic therapy is used in immunocompromised individuals, in cases of bacteraemia or extraintestinal infection and if symptoms worsen or persist [1–3]. Due to resistance to macrolide antibiotics, fluoroquinolones, particularly ciprofloxacin, are often the first choice of therapy [1,3]. However, fluoroquinolone resistance, mainly due to mutations in target topoisomerases (gyrA and/or parC) is increasing [3–7]. In our laboratory, some isolates from both man and animals have shown multiple resistance to different classes of antibiotics [7]. There have also been reports, from other investigators, of cross-resistance to both macrolide and fluoroquinolone antibiotics and of multiply resistant clinical isolates of C. jejuni[8]. However, this resistance has not been explained by the mechanisms of resistance already identified in Campylobacter and is of major concern, especially for the immunocompromised patients, in whom it may increase the number of deaths. In other Gram-negative bacteria, secondary active transporters (efflux pumps), particularly of the resistance nodulation cell division (RND) family, have been shown to contribute to intrinsic antibiotic resistance as well as high level multiple resistance because they recognise a broad variety of structurally unrelated substrates [9–13]. These pumps are proton (H+) driven antiporters which extrude compounds from the bacterial cytoplasm against a proton concentration gradient (PMF) and hence are inhibited by carbonyl cyanide m-chlorophenyl hydrazone (CCCP) [14]. RND-type efflux transporters are large and have a primary structure composed of approximately 1000 amino acid residues [15]. Their secondary structure is made up of 12 transmembrane helical domains (TMs) with two large extracytoplasmic loops between TMs 1 and 2, and 7 and 8 [14–17]. Several specifically localised residues within the TMs are charged and highly conserved and may play a role in substrate binding [14]. The efflux pumps also contain four highly conserved regions or motifs, A, B, C and D, distributed throughout their structure [14]. Motif A is located in the external loop between TMs 1 and 2. Motif D covers the whole of TM 4 and the cytoplasmic loop between TMs 4 and 5. Motif B covers TM 6, the external loop between TMs 5 and 6, and the internal loop between TMS 6 and 7. Motif C covers TM 11 and the external loop between TMs 11 and 12 [14]. In Gram-negative bacteria, RND efflux pumps occur as part of a three-component system, where they are linked to an outer membrane channel protein by a membrane fusion protein (MFP), possibly via the two large cytoplasmic loops. This tripartite protein complex allows bacteria to extrude substrates directly into the external medium, bypassing the periplasm [14]. The three proteins are usually encoded as an operon with a common promoter [15]. An upstream regulator, usually of the tetracycline resistance (TetR) family, with a separate promoter regulates transcription of the efflux operons [18,19]. Charvolos et al. implicated efflux as the mechanism of multidrug resistance in a laboratory isolated mutant of C. jejuni[20], but no genes or proteins conferring the phenotype were identified. In the present study a multi-substrate RND-type efflux pump was identified and characterised in C. jejuni.

2Materials and methods

2.1Bacterial strains and plasmids

The bacterial strains used in this study were Escherichia coli JM109 (Promega) for general transformation, E. coli XL1-Blue (Stratagene) for recombinant plasmid transformation, and C. jejuni NCTC 11168, which is amenable to genetic manipulation and is the genome sequence strain [21]. NCTC 11168-cmeB::kanr was derived in this study (see below). The cloning vectors used were pGEMT-Easy (Promega) for PCR product cloning and as a suicide vector for C. jejuni NCTC 11168 mutagenesis, and pJMK30, a Campylobacter compatible plasmid containing a gene encoding resistance to kanamycin (alph3′-III) for insertional mutagenesis [22].

2.2Bioinformatic DNA and protein sequence analysis, and promoter prediction

BLASTN (NCBI) searches were performed on the C. jejuni genome sequence [21] for homologues of E. coli acrB, Pseudomonas aeruginosa mexB and Salmonella typhi acrB, which are characterised RND efflux pump genes. Sequences were compared using GCG (Genetics Computer Group; University of Birmingham) and imported into GeneDoc ( for colour scheme alignment. The promoter of the identified gene was predicted using the Neural Network Promoter Prediction software (Berkeley). The secondary structure of the deduced protein was then determined by hydrophobicity profiling using TMHMM version 2.0 ( which predicts the transmembrane arrangement of proteins.

2.3Recombinant DNA methods

Chromosomal DNA was isolated from C. jejuni NCTC 11168 using the DNAce spin kit (Bioline). Primers cmeBF (5′-GACGTAATGAAGGAGAGCCA-3′) and cmeBR (5′-CTGATCCACTCCAAGCTATG-3′) were used to amplify a 1-kb fragment within the coding region of the putative efflux pump gene. The PCR product was cloned into pGEMT-Easy (to give construct pI) and restriction digested at a unique BclI and ligated with a 1.4-kb BamHI restriction fragment from pJMK30 containing the kanamycin resistance cassette to give construct pKI. To ensure that the kanamycin resistance cassette had inserted in the correct position, plasmid was analysed by restriction digestion with EcoRI, and by standard PCR with original template primers (cmeBF and cmeBR) and a set of primers designed for amplification of the kanamycin resistance cassette; kancasF (5′-AAGCTACCAAGAGCAAGAGG-3′) and kancasR ((5′-GCTCGACATACTGTTCTTCC-3′). To confirm that the kanamycin resistance cassette had inserted in the same orientation as the insert and that there were no polar effects, pKI was amplified with combinations of template and kanamycin cassette primers, cmeBF and kancasR and, cmeBR and kancasF. Plasmids containing the correct insertional mutation were used to transform C. jejuni NCTC 11168 [23]. PGEMT-Easy was used for mutagenesis in C. jejuni NCTC 11168 because it has an E. coli origin of replication. All known vectors with E. coli origin of replication cannot replicate in Campylobacter, and are therefore used as suicide vectors [23]. When mutant constructs of Campylobacter DNA on E. coli vector are introduced into Campylobacter, a double crossover event occurs, leading to the elimination of vector sequences and replacement of the wild-type gene with the disrupted copy [23]. Therefore, transformation of C. jejuni NCTC 11168 with pKI resulted in incorporation of the disrupted insert gene and the loss of pGEMT-Easy. Transformants were cultured on Muller Hinton agar plates (Oxoid), supplemented with 50 μg ml−1 kanamycin, for 3–5 days. Since pGEMT-Easy could not replicate in C. jejuni, mutants with the incorporated chromosomal mutation were selected. To ensure that the observed kanamycin resistance was on the chromosome, the transformants were checked for plasmid, and chromosomal DNA isolated and analysed by PCR with cmeBF and cmeBR.

2.4RNA isolation, reverse transcription (RT)-PCR and PCR of cDNA

Total cellular RNA was extracted using the RNAce spin kit (Bioline). To ensure that the RNA samples were free of any contaminating DNA, they were treated with RNase-free DNase I (BDH) for 30 min at 37°C. RNA was reverse transcribed with Superscript II RNase H-reverse transcriptase (Gibco BRL) using the cmeBF primer. Expression was measured on a 1/10 dilution of cDNA by standard PCR with primers cmeBF and cmeBR. RT-PCR and PCR were performed on three separate occasions, using three separate sets of primers, to cover three separate regions of the entire gene. Region 2 (above) covered the coding region in the middle of the gene. Region 1 covered the upstream region of the gene and used primer sets uscmeBF (5′-GCTGGAGCTATAGGTCTT-3′) and uscmeBR (5′-GGAGAGCTAACTTCTTGC-3′). Region 3 covered the upstream region of the gene and used primer sets dscmeBF (5′-GATCGCAATGCTTCAAGT-3′) and dscmeBR (5′-GATGCTGCGATCATTCCA-3′).

2.5Minimum inhibitory concentrations (MICs) and efflux assays for NCTC 11168 and NCTC 11168-cmeB::kanr mutants

All agents used in this study were supplied and used according to the manufacturers’ instructions: the antibiotics, ampicillin, chloramphenicol, kanamycin, tetracycline, ethidium bromide, acridine orange and N-lauryl sulfate (SDS) (Sigma); ciprofloxacin (Bayer); and erythromycin (Abbot). The MIC of each agent for parent and mutant strains was determined in triplicate using a microtitre tray doubling dilution method and confirmed by the agar dilution method [24,25]. The MIC of each agent was defined as the lowest concentration at which no growth was observed. Accumulation of ciprofloxacin was measured by fluorescent spectrophotometry [25]. Cells were grown in Muller Hinton broth to an OD600 of 0.5 and preincubated for 15 min at 37°C. Ciprofloxacin was added to a final concentration of 10 μg ml−1 and accumulation kinetics measured on 0.5 ml aliquots between t0 min and t20 min at 5 min intervals. Accumulation was measured with or without CCCP, added to a final concentration of 150 μM after 5 min. The assays were repeated three times and the data expressed as mean±standard deviation.

3Results and discussion

3.1cmeB gene sequence identity

BLASTN searches for efflux genes on the C. jejuni genome identified gene Cj0366c [19]. This gene, which we designated cmeB (Campylobactermultidrug efflux gene B) was 3120 bp in size, which was comparable to the sizes of E. coli acrB (3150 bp), P. aeruginosa mexB (3141 bp) and S. typhi acrB (3180 bp), and showed greater than 40% homology to these gene sequences. Amino acid sequence alignment of CmeB with the other efflux pump proteins yielded: E. coli AcrB (52.6% similarity and 42.3% identity), P. aeruginosa MexB (53.2% similarity and 42.5% identity), S. typhi AcrB (52.4% similarity and 47.2% identity). Most of the consensus was in motifs A–D. Sequence analysis and promoter prediction showed that CmeB is expressed from an operon (Cj0365c-Cj0366c-Cj0367c) with a separate upstream tetR-like regulator gene (Cj0368c). All these genes are transcribed in the same orientation on the complementary strand, with cmeB sharing the same promoter as the two other genes in the operon, Cj0365 (putative outer membrane channel protein) and Cj0367 (putative membrane fusion protein).

3.2Disruption of cmeB gene and inactivation of expression

The PCR of chromosomal DNA from NCTC 11168 with primers cmeBF and cmeBR gave an amplimer size equal to that of the original template. However, NCTC 11168-cmeB::kanr gave an amplimer equal to the combined size of the template and the kanamycin cassette, showing that the gene was disrupted (Fig. 1A). RT-PCR gave no product for the mutant for all three regions of the gene (Fig. 1B), confirming the lack of CmeB mRNA expression.

Figure 1.

Construction and characterisation of cmeB mutants of NCTC 11168. A: PCR analysis of gene disruption with cmeB specific primers. B: RT-PCR, with cmeBF and cmeBR, for cmeB mRNA expression to confirm gene inactivation. Lanes: 1 and 7, hyperladder I DNA marker; 2, no template negative control; 3 and 4, wild-type chromosomal DNA and cDNA, respectively; 5 and 6, mutant chromosomal DNA and cDNA, respectively.

3.3CmeB secondary structure and biological function

CmeB has a predicted 12 transmembrane helical domain structure typical of either major facilitator superfamily (MFS) or RND efflux transporters. However, it also has two large loops between domains 1–2 and 7–8, which are only characteristic of RND efflux pumps [14]. Absence of CmeB in NCTC 11168-cmeB::kanr resulted in a 2–4-fold increase in susceptibility to ampicillin, ciprofloxacin, chloramphenicol, erythromycin, tetracycline, ethidium bromide, acridine orange and SDS (Table 1). Accumulation of ciprofloxacin increased from a steady state concentration (SSC) of 13.80±0.07 ng mg−1 in wild-type to 24.20±0.19 ng mg−1 in mutant, suggesting a 1.8-fold increase. CCCP increased the concentration of ciprofloxacin accumulated by NCTC 11168 to 28.95±0.23 ng mg−1, and the cmeB knockout mutant to 28.10±0.21 ng mg−1. These accumulation data suggested that CmeB was responsible for two-thirds of the efflux of ciprofloxacin (Fig. 2).

Table 1.  Susceptibility of C. jejuni NCTC 11168 (wild-type) and its derivative knockout mutant NCTC 11168-cmeB::kanr to antibiotics, ethidium bromide, acridine orange and SDS (μg ml−1)
 NCTC 11168NCTC 11168-cmeB::kanr
Ethidium bromide42
Acridine orange168
Figure 2.

Ciprofloxacin accumulation by NCTC 11168 (wild-type) and NCTC 11168-cmeB::kanr (mutant) cells, and effect of CCCP on this accumulation. —?—•—?—, accumulation by wild-type; –•–, accumulation by wild-type+CCCP; —?—▴—?—, accumulation by mutant; –▴–, accumulation by mutant+CCCP. Data shown are the mean of three experiments. Error bars show the standard deviation.

Efflux is a well documented mechanism of both intrinsic and selected antibiotic resistance in several bacteria including E. coli and P. aeruginosa[9–13]. However, to date no efflux systems have been identified in C. jejuni. These data present the first characterised multiple efflux pump which contributes to intrinsic antibiotic resistance in C. jejuni. We have designated this pump CmeB (for Campylobactermultidrug efflux). We have demonstrated that CmeB contributes to intrinsic resistance in C. jejuni. However, the finding that the effect of CCCP was not completely abolished in the knockout mutant suggests that CmeB is not the only mechanism of resistance in C. jejuni.


This project is funded by MAFF (project ref. MAFF/VLA CTA9903). We are thankful to Prof. C.W. Penn and A. Jagannathan for providing pJMK30 and their advice on the knockout technique, Dr D.G. Griggs for the initial BLASTN searches and Mrs M. Johnson for help with the agar dilution MICs.