ABSTRACT
Eleven clinical isolates of Streptococcus pneumoniae, isolated in Finland during 1996 to 2000, had an unusual macrolide resistance phenotype. They were resistant to macrolides and streptogramin B but susceptible, intermediate, or low-level resistant to lincosamides. No acquired macrolide resistance genes were detected from the strains. The isolates were found to have mutations in domain V of the 23S rRNA or ribosomal protein L4. Seven isolates had an A2059C mutation in two to four out of the four alleles encoding the 23S rRNA, two isolates had an A2059G mutation in two alleles, one isolate had a C2611G mutation in all four alleles, and one isolate had a 69GTG71-to-69TPS71 substitution in ribosomal protein L4.
Streptococcus pneumoniae (pneumococcus) is an important pathogen in respiratory tract infections, meningitis, and septicemia both in children and adults. There has been an increase in the prevalence of macrolide-resistant pneumococci over the last decade. In Finland, the prevalence has increased from 0.6% in 1988 to 1990 to 11.2% in 2000 according to the Finnish Study Group for Antimicrobial Resistance (unpublished data [http://www.mmm.fi/elintarvikkeet_elaimet/julkaisut_tiedotteet/finres99en.htm]).
In streptococci, there are two well-characterized macrolide resistance mechanisms: target site modification and active drug efflux. Target site modification is mediated by the methylases encoded by the erm (erythromycin ribosome methylation) genes (22, 30). Methylation of A2058 of the peptidyl transferase loop of 23S rRNA causes resistance to 14-, 15-, and 16-membered ring macrolides; lincosamides; and streptogramin B: the macrolide-lincosamide-streptogramin B (MLSB) phenotype (30). The expression of the erm genes can be either constitutive or inducible (31). The active efflux mechanism, encoded by the mef (macrolide efflux) genes, is more specific and causes resistance only to 14- and 15-membered ring macrolides: the M phenotype (3, 25).
Among Finnish clinical isolates of pneumococci, we found a novel type of macrolide resistance, which did not fit into the two phenotypes described above. These strains were resistant to 14-, 15-, and 16-membered ring macrolides and to streptogramin B and were susceptible, intermediate, or low-level resistant to clindamycin (a lincosamide). Also, these strains did not carry mef(A) or erm(B) resistance genes (9). Tait-Kamradt et al. (26, 27) described mutations in the peptidyl transferase loop of the 23S rRNA and ribosomal protein L4 as a cause of a new resistance type in pneumococci with a similar phenotype. The mutations were obtained first in vitro after subsequent passages in azithromycin-containing broth, and later such mutations were found in clinical isolates. The number of alleles encoding 23S rRNA was confirmed to be four. Mutations in 23S rRNA causing macrolide resistance have been described also in other bacteria such as Propionibacterium and Mycobacterium species (13, 18).
In this study, we describe a new point mutation in a pneumococcal 23S rRNA-encoding gene as a cause of macrolide resistance. Also, three other point mutations in pneumococcal 23S rRNA- and ribosomal protein L4-encoding genes are described, one of which is described for a first time in a clinical isolate of pneumococcus.
MATERIALS AND METHODS
Bacterial strains.The pneumococci were identified by typical colony morphology and hemolysis on blood agar plates (Oxoid Ltd., Basingstoke, Hampshire, United Kingdom) supplemented with 5% sheep blood. The strains were further tested for Optochin sensitivity (Optochin disk; Oxoid Ltd.) and to confirm unclear results, they were tested with a Slidex Pneumo-Kit agglutination test (bioMérieux sa, Marcy l'Etoile, France).
MIC testing.The MIC testing was done using the agar plate dilution technique. The bacteria were cultured on Mueller-Hinton II (Becton Dickinson Microbiology Systems, Cockeysville, Md.) agar plates supplemented with 5% sheep blood and incubated for 20 h in 5% CO2 at 35°C. The antibiotics used were erythromycin, azithromycin, spiramycin, two ketolides (telithromycin [HMR3647] and HMR3004), dalfopristin (streptogramin A) and quinupristin (streptogramin B) (Hoechst Marion Roussel [Aventis Pharma], Romainville Cedex, France), clarithromycin (Abbott Laboratories Ltd., Queenborough, Kent, United Kingdom), clindamycin, lincomycin, and chloramphenicol (Sigma-Aldrich Chemie, GmbH, Steinheim, Germany). If available, NCCLS MIC breakpoints were used (16). The control strain S. pneumoniae ATCC 49619 was tested together with the studied strains.
Phenotyping.The double-disk method with erythromycin and clindamycin (Rosco Neo-sensitabs; A/S Rosco, Taastrup, Denmark) disks was used, in addition to the MIC data, for determination of macrolide resistance phenotypes (21). Bacteria were incubated for 20 h in 5% CO2 at 35°C after inoculation on Mueller-Hinton II agar plates supplemented with 5% sheep blood.
Resistance gene determinations.Isolation of DNA for PCR was done using the High Pure DNA isolation kit (Roche Diagnostics GmbH, Mannheim, Germany) according to the manufacturer's instructions. Macrolide resistance genes were determined by PCR (10). The macrolide resistance genes studied were erm(B), erm(A), erm(C), mef(A), msr(A), and erm(TR), a subclass of erm(A). Primers for detection of these genes, except for erm(TR), have been described previously (3, 19, 24). For erm(TR), primers 5"-CTTGTGGAAATGAGTCAACGG-3" [erm(TR) 1.] and 5"-TTGTTCATTGGATAATTTATC-3" [erm(TR) 2.] were used. Staphylococcus simulans 13044 with plasmid pPV142 [erm(C)] (23), Escherichia coli with plasmid pEM9592 [erm(A)] (14), Streptococcus pyogenes A200 [erm(TR)] (22), S. pyogenes A569 [mef(A)] (local strain), and Staphylococcus epidermidis A33 [msr(A)] (17) were used as positive controls.
Sequencing of 23S rRNA genes and ribosomal protein L4 genes.For the isolates with negative results from the PCR-based resistance gene detection, sequencing of the domain V of the 23S rRNA genes and ribosomal protein L4 genes was performed using ABI Prism BigDye Terminator Kit (Applied Biosystems, Foster City, Calif.). Sequencing primers were as presented earlier by Tait-Kamradt et al. (27). We also used one new primer starting at position 2303 of the 23S rRNA: 5"-GGTTGGAAATCATTCGCAGAG-3". Sequences were handled using SeqEd software (Applied Biosystems) and the GCG sequence software package (Wisconsin Package version 10.1; Genetics Computer Group, Madison, Wis.).
PFGE.Pulsed-field gel electrophoresis (PFGE) was done essentially as described earlier (5) with the following changes. (i) Bacteria were grown on a sheep blood agar plate and incubated overnight (18 to 20 h) at 35°C, and the growth was suspended in 0.6 ml of Tris-NaCl buffer. (ii) Plugs were washed three times with the Tris-EDTA buffer for 1 h each at 35°C on a platform rocker. (iii) The plugs were incubated with 50 U of SmaI for 24 h at 25°C. (iv) The chromosomal digests were separated by PFGE with a switch time of 0.2 to 25 s for 20 h.
Serotyping.Serotyping of pneumococcal isolates was performed by counterimmunoelectrophoresis, or, for the neutral serogroups or types 7 and 14, by latex agglutination. The capsular swelling test was used when needed to confirm an uncertain result. All antiserum pools, group- or type-specific antisera, and factor antisera for subtyping within groups containing the heptavalent vaccine serotypes (6B, 9V, 18C, 19F, and 23F) were purchased from Statens Seruminstitut, Copenhagen, Denmark.
RESULTS AND DISCUSSION
In the present work, we describe 11 clinical isolates of pneumococci with mutations in rRNA and ribosomal protein L4 as a cause of macrolide resistance. We found one mutation not previously published in pneumococcus and one mutation not previously described in a clinical isolate of pneumococcus.
All the mutated isolates were highly resistant to macrolides (except ketolides), but susceptible, intermediate, or low-level resistant to the lincosamides (clindamycin and lincomycin). Erythromycin did not induce clindamycin resistance in these isolates. PCR testing with primers detecting previously known acquired macrolide resistance genes gave negative results (Table 1).
23S and L4 mutations and characterization of mutated isolatese
Sequencing domain V from the four individual 23S rRNA genes and the whole ribosomal protein L4 of the isolates with the novel phenotype revealed point mutations at positions A2059 and C2611 (23S rRNA), and 69GTG71 (L4). Sequencing of the corresponding genes from two control strains—one erythromycin-susceptible and one erythromycin-resistant strain with the MLSB phenotype caused by the erm(B) gene—gave wild-type sequences. The mutated isolates belonged to the serotypes 6B, 14, 15, 19A, and 23 and had seven different kinds of PFGE profiles (Table 1).
Two kinds of mutations were found at position 2059 (E. coli numbering). Transversion A2059C was found in seven strains, and transition A2059G was found in two strains. In the A2059C mutated isolates, the number of mutated alleles varied from two to four (Table 1). Five of the seven isolates were molecularly similar (as studied by PFGE), including four isolates with four mutated alleles and one isolate with two mutated alleles. These five isolates were isolated from a small area in northern Finland.
The A2059G mutation has been found before both in clinical and laboratory-derived strains (26, 27, 29). The MICs of erythromycin for the two A2059G isolates were between 64 and 256 μg/ml, which was a little lower than those for the isolates with the A2059C mutation. The azithromycin MICs were high (128 to >512 μg/ml) for all A2059(C/G) mutated isolates. Also, MICs of spiramycin did not remarkably differ between these two kind of mutations, being between 512 and >512 μg/ml. The phenotypes of the isolates with the A2059G mutation correlate with the results described by Tait-Kamradt et al. (26, 27). Lincosamide (clindamycin and lincomycin) MICs for A2059(C/G) strains were elevated compared to those for the macrolide-sensitive strain but did not differ among each other.
We found one isolate with the mutation C2611G. This was an interesting finding, as this type of mutation has earlier been found only in a laboratory-derived pneumococcal strain, not in a clinical isolate. The MIC of erythromycin for this strain was high as it was for other strains in our study, but there were some differences concerning other antimicrobials. This isolate was not as highly resistant to azithromycin and spiramycin as the other mutated isolates (128 μg/ml and 16 μg/ml, respectively), but its resistance to streptogramin B was higher (>64 μg/ml), a phenomenon described earlier by Tait-Kamradt et al. (27). Streptogramin B MICs were also higher for this strain than for those with the erm(B) methylase gene.
There was one isolate with mutations in the ribosomal protein L4. The mutations in the following sequences are underlined: wild type, 202AAA GGA ACT GGA CGT216; mutated allele, 202AAA ACA CCT AGC CGT216 (giving rise to the L4 protein mutation); 3-amino-acid substitution 69GTG71 to 69TPS71 in a highly conserved region, 63KPWRQKGTGRAR74. This isolate was highly resistant to azithromycin and erythromycin but not as highly resistant to spiramycin as the isolates with mutations in A2059. Isolates containing any of the three 23S rRNA mutations had intermediate or low-level resistance to clindamycin (MIC, 0.5 to 2 μg/ml), while the isolate with the L4 mutation was susceptible to clindamycin (Table 1). Also, the MIC of lincomycin for the isolate with the L4 mutation was lower than those for the isolates with 23S rRNA mutations. Clarithromycin resistance levels did not differ between the strains with different mutations.
Ribosomal mutations have also been described in other bacteria (29). In Helicobacter pylori clinical isolates, the mutation A→G at positions 2059 and 2058 has been reported (28). The A2059G mutation caused the MICs of clarithromycin to vary from 2 μg/ml (intermediate) to 32 μg/ml (resistant), and the A2058G mutation caused the MICs of clarithromycin to vary from 32 to 128 μg/ml. Serial isolations of the same H. pylori clone from the patients suggested that the mutations were selected during antibiotic (clarithromycin) therapy.
An A2058G mutation was also described in a clinical isolate of Treponema pallidum causing high-level resistance to erythromycin, roxithromycin, and azithromycin (L. V. Stamm and H. L. Bergen, Letter, Antimicrob. Agents Chemother. 44:806-807, 2000).
In seven clinical isolates of macrolide (clarithromycin and azithromycin)-resistant Mycobacterium avium, adenine at position 2058 was replaced with either C, G, or T (15). Some of the strains occurred during monotherapy with clarithromycin or azithromycin, giving evidence for the hypothesis above that these mutations were enriched during antimicrobial therapy. Also, in propionibacteria this phenomenon has been seen and an A2059G mutation has been reported. Propionibacteria with this mutation differed from the pneumococci in that they were susceptible to streptogramin B (18).
In E. coli, a G2057A mutation caused resistance to both chloramphenicol and 14-membered macrolides (6). In our study, pneumococci with mutations at the nearby positions A2059 or C2611 were sensitive to chloramphenicol. The isolate with the mutation in protein L4 was chloramphenicol resistant. One can speculate whether the isolate with the L4 mutation is chloramphenicol resistant due to a coincidence or if this finding provides a clue as to how chloramphenicol interacts with the ribosome.
In the present work, we describe 10 pneumococcal strains with mutations in the peptidyl transferase loop of 23S rRNA. For the strains carrying A2059(C/G) mutations, the MICs of the ketolides were low (0.063 to 0.5 μg/ml) while those of erythromycin were high (64 to > 512 μg/ml), as were the MICs of the other macrolides and azithromycin (Table 1). The MICs of telithromycin for strains with A2059(C/G) mutations were about threefold higher than those for the strain with the wild-type alleles. This indicates that mutations have some effect, although weak, on the binding of the telithromycin to the pneumococcal ribosome (27).
We had one strain with a C2611G mutation in all four alleles. This strain was also highly resistant to 14- and 15-membered macrolides, but MICs of both ketolides remained low. However, in this strain, the telithromycin MICs were 10-fold higher than those for the strain with wild-type 23S rRNA genes. Tait-Kamradt et al. (27) reported one laboratory-produced pneumococcal strain with C-to-G mutation in two of the four alleles. This mutation increased the telithromycin MIC 100-fold compared to that for the parent strain.
In addition to the mutations mentioned above, other point mutations described in S. pneumoniae as conferring macrolide resistance are A2058G, C2611A, and A2062C (4, 27). These mutations as well as those described above result in resistance to macrolides, but MICs of telithromycin remain low. It has also previously been shown that methylation of adenine 2058 in domain V does not interfere with interactions between telithromycin and ribosomes of S. pneumoniae as much as it does in the case of S. pyogenes (9, 20) or some other gram-positive bacteria, such as Staphylococcus aureus or Enterococcus spp. (7). Although footprinting experiments done with E. coli ribosomes clearly indicate that ketolides protect positions 2058, 2059, and 2505 in domain V (8), based on the data presented above, it seems that in S. pneumoniae, interactions with domain V are not so important for the activity of ketolides as they are for other macrolides. Instead, interactions between ketolides and other parts of the ribosome seem to be more important for ketolide activity as indicated by L4 mutations (26) and 23S rRNA domain II mutations that do provide higher levels of ketolide resistance (32; A. Canu, B. Malbruny, M. Coquemont, T. A. Davies, P. C. Appelbaum, and R. Leclercq, Abstr. 40th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 1927, p. 118, 2000).
Based on 16S rRNA sequences, species in the genus Streptococcus have been divided into six groups (11). The in vitro susceptibility testing of telithromycin is available for species belonging to four of the six phylogenetically separated Streptococcus groups. Although the data are incomplete, it seems that in addition to that for S. pneumoniae (mitis group), the MICs of ketolides for at least Streptococcus mitis (mitis group) and Streptococcus agalactiae (pyogenic group) are low also when the strains are resistant to other macrolides (MLSB-type resistance) (1, 2, 9, 12). The MICs of ketolides for Streptococcus salivarius (salivarius group) and Streptococcus mutans (mutans group) are also low in general, but there are no data concerning the association between ketolide and erythromycin resistance in these species. Based on the present data available, S. pyogenes (pyogenic group) seems to be the only species in the genus Streptococcus for which the MICs of ketolides are high in the presence of constitutively expressed erm(B) macrolide resistance gene (MLSB phenotype).
In conclusion, we present 11 macrolide-resistant pneumococci with no known acquired macrolide resistance genes. The phenotypes were similar to those described previously for pneumococci carrying ribosomal mutations. We found four different kinds of mutations in the peptidyl transferase loop of the 23S rRNA or ribosomal protein L4 that probably are the cause of the macrolide resistance in these strains.
ACKNOWLEDGMENTS
This work was supported by a grant from the Finnish Academy and by a Special Government Grant (EVO-grant).
We thank Anna-Liisa Lumiaho, Saija Nylander, Hilkka Ohukainen, and Tuula Randell for their excellent technical assistance. We are grateful to the Finnish Study Group for Antimicrobial Resistance for the pneumococcal strains. We also thank George Somkuti for Staphylococcus simulans 130044 with the erm(C) resistance gene. We thank André Bryskier for providing antibiotics.
FOOTNOTES
- Received 13 June 2001.
- Returned for modification 3 October 2001.
- Accepted 23 November 2001.
- American Society for Microbiology
