|Home | About | Journals | Submit | Contact Us | Français|
We selected in vitro erythromycin-resistant strains of Bartonella henselae. The mutants obtained had point mutations in domain V of 23S rRNA and/or in ribosomal protein L4. One lymph node of a patient with cat-scratch disease had such a mutation in 23S rRNA, suggesting that natural resistant strains may infect humans.
Bacteria of the genus Bartonella are fastidious, facultative intracellular bacilli, belonging to the alpha-2 subgroup of Proteobacteria. Bartonella henselae is the causative agent of cat-scratch disease (CSD) and is also involved in other clinical situations such as endocarditis, bacillary angiomatosis (BA) and peliosis hepatitis (PH) in immunocompromised patients (1). Interestingly, erythromycin has become the drug of first choice and has been successfully used to treat many patients with BA (8) and PH (14). However, when the treatment duration is less than 15 days, relapses after antibiotic withdrawal are common, and therefore treatment should be given for 3 to 4 months (8). Macrolide compounds inhibit protein synthesis by binding to domains II and V of 23S rRNA (7). The first mechanism of macrolide resistance described was due to posttranscriptional modifications of the 23S rRNA by the adenine-N6-methyltransferase. Modification of the ribosomal target confers cross-resistance to macrolides and remains the most frequent mechanism of resistance. Two other mechanisms of resistance to macrolides have been described: mutations in 23S rRNA gene, L4 and L22 ribosomal proteins (4, 6, 15, 17) and active efflux (16).
For B. henselae, potential mechanisms of resistance to macrolides are not known, and the objective of the present study was to select in vitro erythromycin-resistant (ER) strains to examine the molecular mechanism of resistance. Moreover, the sequences of the macrolide resistance target genes of 15 B. henselae PCR-positive lymph nodes from patients with CSD (18) have been examined to look for possible natural resistance in clinical isolates.
The seven strains of B. henselae used here are described in Table Table1.1. Selection of ER mutants was performed by serial passages of isolates on blood agar plates containing a disk of erythromycin (15 μg) initially placed in the corner of the plate. The plates were incubated at 37°C, and the diameters of growth inhibition (in mm) were measured every 2 weeks. The confluent growth outside the zone of inhibition was harvested with an inoculation loop and subcultured every 2 weeks until the strain became completely resistant. Resistance to erythromycin was defined as an isolate with growth in contact to the disk of erythromycin. The ER and erythromycin-susceptible strains were screened by PCR amplification and DNA sequencing using specific oligonucleotide primers of the entire23S rRNA gene, the L4 and L22 ribosomal proteins described previously (9). The nucleotide sequences obtained were compared by using the CLUSTAL W program supported by the Infobiogen website to look at possible mutations known to be associated with macrolide resistance. The known original sequences of Escherichia coli (K-12 MG1655 [for 23S rRNA gene comparison]) and Streptococcus pneumoniae (TIGR4 [for L4 and L22 genes comparison]) were retrieved from the KEGG Web site and used for sequence alignments. To study a possible heterogeneity of the sequences of the two copies of the domain V of the 23S rRNA gene, PCR products of the ER mutants obtained were cloned in pGEM-T Easy Vector (Promega, Charbonniéres, France), as described by the manufacturer, and 10 clones were sequenced twice.
Seven ER mutants were obtained after 2 to 16 subcultures (Table (Table1).1). Interestingly, for one strain (BhM), in the first experiment of in vitro selection, we found a resistant subpopulation after the first passage with several colonies growing in close contact to the antibiotic disk and other colonies around the zone of growth inhibition. The strain was homogeneously resistant to erythromycin after the second passage. The seven ER mutants obtained were also resistant to telithromycin, clarithromycin, and azithromycin (data not shown). The ER mutants were stable for 10 passages after antibiotic withdrawal. Compared to the parental strains, the ER mutants harbored mutations either in the domain V of the 23S rRNA gene and/or in the highly conserved stretch of amino acids in ribosomal protein L4 (Table (Table1).1). Most of the mutations in 23S rRNA gene (A2058G, A2058C, A2059G, and C2611T) have been previously reported to confer erythromycin resistance in other bacteria (3, 10, 15, 16). We found a heterogeneity of sequences of domain V of 23S rRNA gene for the resistant subpopulation of the strain BhM (Table (Table1)1) obtained after the first passage, as demonstrated after cloning with half sequences with A and half sequences with G at position 2058. This was not observed for the other ER mutants. Interestingly, the electropherogram obtained for this subpopulation showed a double peak (A and G) at position 2058 (data not shown). This means that a change in one of the two copies of the 23S rRNA gene could be sufficient to confer erythromycin resistance, as previously reported for other bacteria such as Mycoplasma pneumoniae, Streptomyces ambofaciens, and S. pneumoniae (5, 11, 12). Finally, the A2059G transition was also detected in one of the 15 lymph nodes from patients with CSD (Table (Table1).1). This node was excised from a 10-year-old female that was not treated with antibiotics prior to excision, suggesting that naturally occurring ER strains may infect humans. Thus, we believe that an ER strain was acquired from an animal that was either exposed to macrolides or with a natural isolate harboring this mutation. This information is likely to be clinically relevant, since failures and relapses have been reported with erythromycin in the treatment of CSD (2, 13). For the ribosomal protein L4, we did not find nucleotide insertion, as recently reported for B. quintana (9), but amino acid mutations at two different positions (Table (Table1).1). Although we found that the G71R mutation alone was sufficient for resistance (BhFr4), the significance of H75Y (BhFr2) mutation remains uncertain because it has never been reported in other bacteria. In summary, the present study showed that mutations in 23S rRNA and ribosomal protein L4 were responsible for in vitro macrolide resistance in B. henselae selected with erythromycin. Since resistant mutants of B. henselae could be obtained easily in vitro and also occurred in natural isolates, we believe that the probability to select resistant isolates in nature is high. Thus, it seems reasonable to avoid the use of erythromycin alone in the treatment of CSD.
We thank Paul Newton for help with the manuscript.