Macrolide Resistance in Campylobacter jejuni and Campylobacter coli: Molecular Mechanism and Stability of the Resistance Phenotype

Campylobacter isolates used in this study.

This study was carried out using 23 macrolide-resistant isolates of Campylobacter. All isolates were microaerobic, spiral, gram-negative rods that were isolated and selected at a growth temperature of 42°C. The isolates were identified as Campylobacter as described previously (28). Identification of the isolates as C. jejuni or C. coli was carried out using the following PCR-based assays: hipO PCR (22), mapA PCR (39), Fermér PCR-restriction fragment length polymorphism (11), aspartokinase PCR (22), and ceuE PCR (17). Campylobacter isolates were routinely cultured on brain heart infusion agar (Difco, Becton-Dickinson, Sparks, Md.), supplemented with 0.4% yeast extract (Difco), and incubated at 37°C under microaerobic conditions (5% CO₂, 10% H₂, balance N₂) for 48 h. The bacteria were stored at −70°C in 25% glycerol Luria-Bertani broth (37).

Determination of MICs.

The determination of the MICs of Ery, Cla, and tetracycline was performed using the agar dilution method as previously described (15), whereas the determination of ciprofloxacin MICs was carried out using the Etest (44). To ensure reproducibility, MIC determinations were repeated at least twice. Isolates were considered resistant to Ery, Cla, tetracycline, and ciprofloxacin if they had MICs of ≥8, ≥8, ≥16, and ≥4 μg/ml, respectively.

Preparation of genomic DNA.

All genomic DNA templates were prepared using a Promega Wizard DNA isolation kit (Promega, Madison, Wis.).

Gel electrophoresis and purification of PCR products.

The PCR products were visualized using agarose gel electrophoresis according to a standard method (37). PCR products were purified using either a PCR purification kit or a gel extraction kit (QIAGEN, Mississauga, Ontario, Canada).

Characterization of macrolide resistance in Campylobacter isolates.

In the case of C. coli, an internal part of domain V of the 23S ribosomal DNA (300 bp) was amplified by PCR using primers DP1 and CJ1 (Table 1). A reaction mixture containing the oligonucleotide primers at 0.5 μM each; dATP, dCTP, dGTP, and dTTP at 200 μM each; 1× reaction buffer (50 μM KCl, 10 mM Tris-HCl [pH 8.3], 1.5 mM MgCl₂, 0.01% [wt/vol] gelatin); 1 U of Taq polymerase (Perkin-Elmer Cetus, Norwalk, Conn.); and 50 ng of genomic DNA template was used for PCR amplification. Thirty cycles of amplification were performed. Each cycle consisted of a 0.5-min denaturation at 95°C, a 1.0-min annealing step at 52°C, and a 1.0-min extension at 72°C. The PCR product was sequenced after purification as described above.

To assess whether some isolates had a mutation at position 2611 (E. coli numbering) or at position 2717 (H. pylori numbering) of the 23S rRNA gene, about 300 bp of the gene was amplified by PCR using (2611) mutation-F and (2611) mutation-R primers (Table 1). The amplification was performed using the above-mentioned PCR mixture and PCR conditions except that the annealing step was carried out at 56°C. The PCR product was sequenced after purification.

Genes coding for L4 and L22 ribosomal proteins were amplified by PCR using the PCR mixture mentioned above. Primer pairs CJ18-CJ19 and CJ20-CJ21 were used for the amplification of the genes coding for L4 and L22 proteins, respectively (Table 1). Thirty cycles of amplification were performed. Each cycle consisted of a 1-min denaturation at 94°C, a 1-min annealing step at 60°C in the case of the L4 protein or at 50°C for the L22 protein, and a 1-min extension at 72°C. The PCR products (800 bp and 600 bp in the cases of L4 and L22, respectively) were then sequenced after purification.

Different copies of the 23S rRNA genes of the C. jejuni isolates were amplified by PCR using three oligonucleotide pairs, FI-CJ copy-R, FII-CJ copy-R, or FIII-CJ copy-R (Table 1). Amplification was performed in a total volume of 50 μl containing the PCR primers at 0.1 μM each; 1× AccuPrime PCR buffer II [60 mM Tris-SO₄ (pH 8.9), 18 mM (NH₄)₂SO₄, 0.4 mM MgSO₄, 0.2 mM dGTP, 0.2 mM dATP, 0.2 mM dTTP, 0.2 mM dCTP, thermostable AccuPrime protein, 1% glycerol]; 200 ng of the genomic DNA; and 1 U of AccuPrime Taq DNA polymerase (Invitrogen, Burlington, Ontario, Canada). A hot start at 95°C for 2 min was performed before the AccuPrime Taq DNA polymerase was added to the reaction mixture. Twenty-five cycles were performed, and each cycle consisted of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 68°C for 7 min. The PCR cycles were followed by a final extension period of 15 min at 68°C. The PCR products (5.7 kbp from the first copy of the 23S rRNA gene [FI-CJ copy-R], 5.8 kbp from the second copy [FII-CJ copy-R], and 5.7 kbp from the third copy [FIII-CJ copy-R]) were analyzed by gel electrophoresis, purified as described above, and investigated by nucleotide sequence analysis using the DP1 primer (Table 1).

Detection of erythromycin-modifying activity.

Extracellular erythromycin-modifying enzyme assays were carried out as described by Yan and Taylor (53).

Stability of the erythromycin resistance phenotype.

The stability of Ery^r was tested by repeated subculturing of seven representative isolates on Ery-free Mueller-Hinton (MH) agar plates (BBL, Becton-Dickinson, Cockeysville, Md.). Ery MICs were determined after 15, 35, and 55 passages using the Etest.

Efflux pump inhibitor assay.

The effect of PAβN on the Ery^r of Campylobacter isolates was investigated as described previously (26) except that standard susceptibility disks (Oxoid, Nepean, Ontario, Canada) containing 15 μg of Ery were used. Inhibition zone sizes were interpreted according to the guidelines of the National Committee for Clinical Laboratory Standards (30). A control plate of MH agar containing an equivalent concentration of PAβN was also included to assess the effect of PAβN on the growth of the isolates investigated. PAβN was purchased from Sigma-Aldrich (St. Louis, MO).

Natural transformation and the effect of erythromycin-associated mutation on the growth rate of the transformants.

The PCR fragments containing the 23S rRNA mutations from two isolates (one carried an A-2059→G transition, and the other had an A-2058→C mutation) were introduced into a susceptible isolate of C. jejuni (isolate 81116) by natural transformation using the biphasic system as described previously (50). Ery^r transformants were selected on MH agar (BBL) plates containing 16 μg/ml of Ery. The resulting Ery^r transformants in each case were screened for the existence of erythromycin-associated mutations by PCR amplification using the DP1 and CJ1 primers (Table 1). The Ery MICs were also determined for the resulting transformants. Growth experiments were performed to assess the effect of the erythromycin-associated mutation on the growth patterns of Ery^r transformants compared to the effect on their parental isolate. A 24-h culture of each transformant was inoculated into 40 ml of MH broth (BBL, Becton-Dickinson, Cockeysville, MD) to an optical density of 0.003 at 625 nm, which was determined using a spectrophotometer (Ultrospec 3000; Pharmacia Biotech, Piscataway, NJ). Ery was added to the culture of the transformants to obtain a final concentration of 15 μg/ml, whereas the culture of the parent isolate, C. jejuni 81116, was left free of Ery. All cultures were incubated at 37°C with shaking (140 rpm) under microaerobic conditions for 54 h. Sampling of 0.2 ml from each culture was done at 0, 6, 24, 30, 48, and 54 h. The number of viable cells (CFU) in each sample was estimated by spreading an aliquot of the appropriate dilutions onto MH agar plates. The plates were subsequently incubated at 37°C in a microaerobic atmosphere for 48 h. The growth experiments were carried out in triplicate. Similarly, the growth pattern of Ery^r isolate 001A-168 was compared to that of its susceptible isogenic isolate, resulting from 55 subcultures on a drug-free medium.

PCR screening for the tet(O) gene.

Tetracycline-resistant isolates were screened for the existence of the tet(O) gene by PCR using the primers and method described previously (16).

Analysis of the QRDR.

The genetic basis of ciprofloxacin resistance in Campylobacter isolates was determined by PCR amplification of the quinolone resistance determining region (QRDR) of the gyrA gene as described previously (49). The PCR products were purified and sequenced as mentioned above.

DNA sequencing.

DNA samples were prepared for sequencing as described previously (16).

Characterization of Campylobacter isolates.

Twenty-three Ery^r Campylobacter isolates collected from different geographic areas were included in this study (Tables 2 and 3). They comprised 6 isolates from poultry, 1 from sheep, 1 from cattle, and 15 clinical isolates. All isolates were initially identified as C. jejuni or C. coli based on the results of the hippurate hydrolysis test. The identification of the isolates was further confirmed by the use of other molecular PCR-based assays which target different genes, such as hipO, mapA, ceuE, the aspartokinase gene, and the 23S rRNA gene. In the latter case, species differentiation was accomplished by digestion of the PCR product, representing the highly polymorphic part of the 23S rRNA gene, by two restriction enzymes, AluI and Tsp509I (11). Of the 23 isolates, 9 were classified as C. coli (2 from poultry, 1 from sheep, and 6 clinical isolates) (Table 2), and 14 isolates were classified as C. jejuni (4 from poultry, 1 from cattle, and 9 from humans) (Table 3).

Detection of the resistance patterns of Campylobacter isolates.

The MICs of Ery and Cla for all resistant isolates were determined using the agar dilution method. The MICs for Ery and Cla ranged from 128 to >1,024 and from 64 to 256 μg/ml, respectively (Table 4). All Ery^r isolates showed cross-resistance to Cla with the exception of the clinical isolate 001A-114, which was resistant to Ery only (Table 4). In addition to macrolide resistance, some C. jejuni and C. coli isolates were also resistant to tetracycline (14 isolates) or tetracycline and ciprofloxacin (2 isolates) (Tables 2 and 3). The MIC of the two ciprofloxacin-resistant isolates was 32 μg/ml. None of the isolates collected between 1974 and 1989 were resistant to ciprofloxacin. The MICs of the tetracycline-resistant isolates ranged from 16 to 512 μg/ml, with the highest MICs being demonstrated by the isolates collected from poultry (256 to 512 μg/ml). The rate of incidence of tetracycline resistance among the isolates collected between 1974 and 1989 was lower than that of the isolates collected in Quebec between 2000 and 2001. This supports recent studies showing that the increase in the incidence of tetracycline and ciprofloxacin resistance is a trend occurring not only across Canada (14, 16) but also in other countries (19, 24, 25, 32). In this study, the QRDR of the gyrA gene of the ciprofloxacin-resistant isolates showed the C-to-T transition at nucleotide 256, resulting in the substitution of Ile for Thr at amino acid 86 which was previously observed to mediate high quinolone resistance in Campylobacter isolates (41). Also, all tetracycline-resistant isolates were found to carry the tet(O) gene, the resistance marker commonly detected in tetracycline-resistant Campylobacter isolates (16, 40).

The mechanism of macrolide resistance in Campylobacter coli isolates.

In C. coli isolates, the complete genome sequence is not yet available; therefore, the detection of a potential macrolide-associated mutation in domain V of the 23S rRNA gene was mainly analyzed by amplifying a 300-bp fragment of the target gene (where the majority of point mutations related to macrolide resistance are expected to exist) without specific amplification of the three copies of the target gene. The PCR products were sequenced to verify the specificity of the primers (CJ1-DP1) (Table 1) and the presence of a mutation(s) that could be associated with macrolide resistance. The numbering used throughout the paper is based on the 23S rRNA gene of E. coli unless otherwise indicated. All C. coli isolates exhibited an A→G transition at position 2059 (Table 4). In one case (isolate UA585) (Table 4), repeated amplification and DNA sequencing revealed that different copies of the 23S rRNA gene may have an A→G transition at two variable positions. On one occasion, an A→G transition was detected at position 2058, and in another instance, the same mutation was observed at a neighboring position (position 2059). Confirmation of this hypothesis awaits completion of the genome sequence of C. coli in order to carry out the specific amplification of each copy of the target gene.

The mechanism of macrolide resistance in Campylobacter jejuni isolates.

In C. jejuni isolates, operon-specific PCRs were performed to amplify the three copies of the 23S rRNA gene. In the majority of C. jejuni isolates (13 of 14 isolates), the amplification resulted in PCR fragments of the expected size. However, in one case (UA261) (Tables 3 and 4), repeated attempts yielded no products at all. Amplification of a 300-bp fragment of the target gene in isolate UA261, where the majority of point mutations related to macrolide resistance are expected to occur, was performed using the CJ1-DP1 primer pair (Table 1). DNA sequencing of the resulting PCR products revealed an A→G transition at position 2059. In the other 13 isolates, each copy of the target gene was specifically amplified and analyzed by DNA sequencing. Two of the 13 isolates (001B-22 and UA336) (Table 4) exhibited an A→G transition at position 2059 in only two of the three copies of the target gene. Six of the 13 isolates showed an A→G transition at position 2059 in all three copies of the target gene (Table 4). Three isolates (UA695, UA697, and UA709) (Table 4), however, carried an A→C transversion at position 2058 in all three copies of 23S rRNA gene. Surprisingly, isolate 001A-168 exhibited an A→G transition at position 2058 (Table 4).

In contrast, the molecular mechanism of macrolide resistance remains undetermined in the case of isolate UA710. Sequencing of the three copies of the 23S rRNA gene in UA710 revealed no point mutation at either position 2058 or position 2059. Mutations at other positions associated with macrolide resistance, such as position 2611 (27) or position 2717 (H. pylori numbering) (12), were not detected in any of the copies of the target gene. To rule out the possibility that isolate UA710 is resistant to macrolides due to the production of an extracellular enzyme capable of modifying erythromycin and related macrolides, a biological assay for the detection of erythromycin-modifying activity was performed. E. coli BM2571(pIP1527), which contains the ereA gene encoding erythromycin esterase, was included in the assay as a positive control. The assay indicated that extracellular degradation of erythromycin was not responsible for macrolide resistance in isolate UA710. In addition, no correlation was found between macrolide resistance in UA710 and alteration in the ribosomal protein L4 or L22 (see below).

Investigation of the ribosomal proteins L4 and L22 in some isolates of C. jejuni and C. coli.

Five representative isolates were examined regarding alterations in the ribosomal proteins L4 and L22. The Campylobacter isolates examined included two C. coli isolates (UA585 and 001B-15) (Table 4) and three isolates of C. jejuni (UA710, UA695, and 001B-22) (Table 4). The representative isolates exhibited variable resistance levels and also carried different point mutations in the 23S rRNA gene. Two C. jejuni strains, NCTC 11168 and 81-176, were also included as representatives of macrolide-susceptible strains. The L4 and L22 proteins of C. jejuni NCTC 11168 showed variation in some amino acid residues compared to the corresponding proteins of strain 81-176 (data not shown). Also, the L22 protein of C. jejuni 81-176 has an insertion of six amino acids (₁₁₈APAAKK) compared to the L22 protein of NCTC 11168. The ribosomal proteins L4 and L22 of the five representative isolates, however, demonstrated complete identity with the corresponding proteins of either C. jejuni strain 11168 or C. jejuni strain 81-176 with the exception of the L22 protein in isolate UA695. This isolate exhibited a P13S alteration in the L22 protein. The mutation at this position has not been identified as a macrolide resistance-associated mutation in other bacterial species (21, 51). This might imply that the P13S alteration in isolate UA695 is not related to macrolide resistance.

Effect of efflux pump inhibition on erythromycin resistance in Campylobacter isolates.

The activity of PAβN as an efflux pump inhibitor was investigated for all Ery^r isolates. The patterns of resistance to Ery in 18 of the 23 isolates, including UA710, were not affected by the presence of 15 μg/ml of the PAβN inhibitor, indicating that the efflux pump plays no role in mediating macrolide resistance in these isolates. One isolate of C. jejuni (001A-15), however, demonstrated an increase in the inhibition zone diameter for 15 μg of Ery from 0 to 45 mm in the presence of 15 μg/ml PAβN, indicating complete susceptibility to Ery. In C. coli isolates UA749 and 001B-15 (Table 2), the presence of PAβN resulted in a slight increase in the diameter of the inhibition zone from 0 to 22 mm and from 0 to 24 mm, respectively, although this was not sufficient to completely restore susceptibility to Ery. In two other isolates of C. coli (001B-17 and 001A-18) (Table 2), the presence of 15 μg/ml PAβN during susceptibility testing resulted in a marked inhibition of their growth interfering with the assay. The effect of two lower concentrations of PAβN inhibitor (5 and 10 μg/ml) on the growth and Ery^r of C. coli isolates 001B-17 and 001A-18 was assessed. Both concentrations of PAβN had no influence on either the growth or the Ery^r of isolate 001A-18. Although the presence of 10 μg/ml PAβN had no effect on the growth of isolate 001B-17, it did result in a slight increase in the inhibition zone diameter around the Ery disk (from 0 to 15 mm). Neither the growth nor the Ery^r of isolate 001B-17 was affected by the presence of 5 μg/ml PAβN.

Stability of macrolide resistance.

To examine whether the level of macrolide resistance, the type of mutation in the 23S rRNA gene, and the number of the mutated copies of the target gene (in the case of C. jejuni) play a role in the stability of Ery^r in Campylobacter in vitro, seven representative isolates (001B-15, UA749, 001B-22, 001A-168, UA697, UA709, and UA40) (Table 4) were examined after 15, 35, and 55 repeated subcultures on drug-free media. The representative isolates demonstrated a wide range of Ery MICs (128 to >1,024 μg/ml) as well as different types of macrolide-associated mutations (four had an A2059G mutation, two carried an A2058C mutation, and one had an A2058G mutation). The isolates remained remarkably resistant to Ery except for isolate 001A-168, which reverted back to Ery and Cla susceptibility after 55 subcultures (MICs of 2 μg/ml and 4 μg/ml, respectively). In addition, sequencing of the three copies of the 23S rRNA gene of the Ery-susceptible isolate 001A-168 revealed no point mutation at position 2058. This finding confirmed that the A2058G mutation was responsible for macrolide resistance in isolate 001A-168.

Transformation studies and effect of the 23S rRNA mutation on the growth rate of Campylobacter isolates.

Transformation studies were used to confirm that the A2058C and A2059G mutations observed are associated with the resistance of Campylobacter isolates to macrolides. Transformation of a part of the 23S rRNA gene (300-bp PCR product amplified by the use of the CJ1-DP1 primer pair) from isolates UA695 and UA37 (with A→C and A→G mutations at positions 2058 and 2059, respectively) yielded Ery^r transformants at a frequency of 10⁻⁶ and 5 × 10⁻⁷ transformants per recipient cell, respectively. The resulting transformants exhibited levels of Ery^r comparable to those of isolates UA695 and UA37 (MICs of >1,024 μg/ml). Subsequent sequencing of the region carrying domain V of the 23S rRNA gene (amplified by the CJ1-DP1 PCR primers) in the transformants indicated that the resistance-associated mutation was homozygous and identical to that of the corresponding donor isolate. This suggests that the emergence of the transformants involved acquisition of the mutant sequence by all 23S rRNA gene copies, presumably one after the other.

To assess the impact of these mutations on the growth of their host, we compared the growth pattern of each type of the Ery^r transformants to that of its parent isolate (C. jejuni 81116). Also, the growth pattern of the Ery^r isolate 001A-168 (carrying the A2058G mutation in the 23S rRNA gene) was compared to that of its susceptible counterpart. The growth of each type of the Ery^r transformants exhibited no difference compared to that of the parent C. jejuni 81116, as indicated by viable counts over a period of 54 h (data not shown). Samples of Ery^r isolate 001A-168, collected after 48 and 54 h, demonstrated about a 2-log viability decrease compared to the corresponding samples of its susceptible counterpart (data not shown). In addition, there was a marked reduction in the sizes of the colonies of Ery^r isolate 001A-168 relative to those of the corresponding colonies of its susceptible isogenic isolate.