|Home | About | Journals | Submit | Contact Us | Français|
Intensive use of antimicrobial agents in health care settings not only leads to the selection of multiresistant nosocomial isolates of Staphylococcus aureus but may also promote endogenous, resistance-conferring mutations in bacterial genes that encode drug targets. We evaluated the spectrum of rifampin resistance-conferring mutations in cultures of methicillin-susceptible S. aureus (MSSA) or methicillin-resistant S. aureus (MRSA) strains exposed in vitro to sub-MICs of ciprofloxacin. Growth of ciprofloxacin-susceptible MRSA strain MRGR3 and ciprofloxacin-resistant MSSA strain RA1 (a NCTC 8325 derivative) in the presence of 1/2× or 1/4× MIC of ciprofloxacin led to higher frequencies of rifampin-resistant mutants on agar supplemented with rifampin (0.25 mg/liter) than under ciprofloxacin-free conditions. While rifampin-resistant mutants from ciprofloxacin-free cultures essentially showed single-amino-acid substitutions, a significant proportion of rifampin-resistant mutants from ciprofloxacin-exposed cultures displayed in-frame deletions or insertions in the rpoB gene at several positions of the rifampin resistance cluster I. In-frame deletions or insertions were also recorded in rpoB cluster I of rifampin-resistant mutants from ciprofloxacin-exposed cultures of mutS and mutL DNA repair mutants of ciprofloxacin-resistant S. aureus strain RA1. Frequencies of rifampin-resistant mutants grown under ciprofloxacin-free medium were higher for mutant strains RA1 mutS2 and RA1 mutL, but not RA1 recA, than for their parent RA1. In conclusion, ciprofloxacin-mediated DNA damage in S. aureus, as exemplified by the wide diversity of deletions or insertions in rpoB, suggests the occurrence of major, quinolone-mediated disturbances in DNA fork progression and replication repair. Besides promoting antibiotic resistance, accumulation of unrepaired DNA replication errors, including insertions and deletions, may also contribute to potentially lethal mutations.
The high prevalence of nosocomial infections due to multiresistant bacteria, such as found with methicillin-resistant Staphylococcus aureus (MRSA), is currently explained by intensive use of topical and systemic antimicrobial agents in health care settings, which represents a highly selective pressure for antibiotic-resistant bacterial clones. In addition, bacterial exposure to some antibiotic classes may potentially induce endogenous, resistance-conferring mutations in bacterial genes that encode drug targets. According to current models, increased antibiotic-mediated resistance mutation frequencies should ensue from impairment of error-free DNA repair mechanisms (8, 11, 12) that have to cope with extensive DNA damage or from triggering of error-prone DNA repair pathways such as the translesion synthesis evoked by the SOS response (14, 19). While DNA repair mechanisms have been characterized in detail in some model organisms such as Escherichia coli, characterization of their physiological role in S. aureus is still preliminary despite several studies (2, 9, 18, 23, 29, 35, 38).
The most commonly used antibiotic gene target for monitoring resistance mutation frequencies is rpoB, the gene encoding the β subunit of RNA polymerase, which determines rifampin susceptibility or resistance (4, 35, 38, 41). Previous studies listed a number of amino acid substitutions in resistance clusters of rpoB, which were linked with different levels of resistance to rifampin (3, 7, 34, 36, 37, 44, 45). Each type of amino acid substitution found in rpoB cluster I of clinical or laboratory rifampin-resistant S. aureus isolates was also linked with a specific fitness cost (37, 44).
Fluoroquinolones are DNA-targeting agents that interact with their target type II topoisomerases, thus increasing generation of single- and double-strand DNA breaks associated with stalled or collapsed replication forks (13, 21, 27, 30, 31). Topoisomerase IV is considered the primary target of ciprofloxacin in S. aureus (13, 16, 32). The stepwise emergence of ciprofloxacin resistance in S. aureus involves first parC mutations conferring low-level resistance, followed by gyrA mutations leading to high-level resistance (15, 32). Recent studies indicate that S. aureus exposure to ciprofloxacin can induce the SOS response (6, 9, 13, 19, 28, 29), which may lead to derepression of resident prophages (17), increased horizontal transmission of coresident pathogenicity islands (33, 43), or upregulation of fibronectin-binding proteins (6, 22). While increased frequencies of rifampin resistance were indeed observed in ciprofloxacin-exposed S. aureus, the spectrum of rifampin resistance-conferring mutations in those rpoB mutants was not reported (9, 28, 29).
The aims of our study were (i) to characterize at qualitative and quantitative levels genomic alterations induced in S. aureus by subinhibitory concentrations of ciprofloxacin, as detected by rifampin resistance mutations occurring in rpoB, and (ii) to evaluate the potential impact of some major DNA repair components on the spectrum of ciprofloxacin-triggered rifampin resistance-conferring mutations. The data provided may potentially explain how accumulation of unrepaired DNA replication errors, including insertions and deletions, may either promote resistance-conferring mutations or lead to quinolone-induced bacterial death (13).
(This study was presented in part at the 20th European Congress of Clinical Microbiology and Infectious Diseases, Vienna, Austria, April 2010.)
The properties of the double parC gyrA ciprofloxacin-resistant (Cipr) mutant RA1 (ciprofloxacin MIC, 32 μg/ml) of NCTC8325 strain ISP794 (ciprofloxacin MIC, 0.25 μg/ml), as well those of as its RA1 recA mutant (ciprofloxacin MIC, 2 μg/ml), were previously described (6). Strains RA1 mutS and RA1 mutL were constructed by transduction of mutS2::kan and mutL::kan (38) from RN4220 (provided by R. Leclercq, University of Caen) into RA1 and were verified by PCR.
MRSA strain MRGR3 is a previously described ciprofloxacin-susceptible (Cips) (ciprofloxacin MIC, 0.25 μg/ml) and rifampin-susceptible (Rifs) (rifampin MIC, 0.008 μg/ml) clinical isolate from a patient with catheter-related sepsis (26, 40) that is highly virulent in a rat model of chronic tissue cage infections and shows additional resistance to gentamicin, tetracycline, and chloramphenicol (24). Strain MRGR3 belongs to a different clonal family (sequence type 239 [ST239], clonal complex 5 [CC5]) than strains ISP794 and RA1 (ST8) (1).
MICs of ciprofloxacin or rifampin for all S. aureus strains were determined by broth macrodilution in cation-adjusted Mueller-Hinton broth (CAMHB) according to Clinical and Laboratory Standards Institute (CLSI) guidelines (10). S. aureus ATCC 29213 was used as a quality control strain for MIC assays.
To evaluate the impact of subinhibitory levels of ciprofloxacin on emergence of rifampin resistance, 2 × 107 CFU/ml of each S. aureus strain were incubated in CAMHB for 24 h at 37°C under shaking conditions (210 rpm), in the continuous presence or absence of subinhibitory concentrations (namely, 1/2×, 1/4×, and 1/8× MIC) of ciprofloxacin. Rifampin mutation frequencies were defined as the ratio of CFU counts determined by plating 100-μl portions of undiluted or serially diluted 24-h cultures (ca. 1010 CFU/ml) on MH agar plates supplemented with rifampin (0.25 μg/ml) to those on antibiotic-free MH agar plates.
For each strain and experimental condition, 10 to 20 individual colonies from rifampin-containing agar plates, for which rifampin resistance was confirmed by broth macrodilution MIC, were stored in skim milk at −70°C.
Two sets of oligonucleotide primers were used to amplify rpoB sequence fragments encompassing rifampin resistance clusters I and II: (i) 5′-TCTTTATGGGTGATTTCCCA-3′ and 5′-TAATAGCTGCACCAGAATCA-3′, yielding a 1,689-bp PCR fragment, and (ii) 5′-ACCGTCGTTTACGTTCTGTA-3′ and 5′-TCAGTGATAGCATGTGTATC-3′, yielding a 460-bp PCR fragment. Amplified DNA was sequenced in both directions at the nucleotide sequencing facility of the Faculty of Medicine of the University of Geneva.
The relative fitnesses of Rifr mutants and their Rifs parents were estimated by paired competition experiments (5, 37, 44). Dilutions (10−6 fold) of overnight cultures of Rifr and Rifs strains were mixed in CAMHB at a 1:1 ratio and grown at 37°C for 24 h. The numbers of viable Rifr and Rifs cells were determined at 0 and 24 h by plating the serially diluted mixed cultures onto selective (0.25 μg rifampin/ml) and nonselective MH agar. The relative fitness of each strain was calculated from the ratio of the number of generations (log B − log A/log 2) of Rifr strains to that of Rifs strains (5, 44) and expressed as the mean (± standard error of the mean [SEM]) determined from three independent competitive assays.
The significance of differences in mutation frequencies of strains grown under ciprofloxacin-free conditions was determined by comparing means (± standard deviations [SD]) of logarithmically transformed mutation frequencies, scored in at least three independent experiments, by one-way analysis of variance and post-analysis of variance pairwise comparisons between individual groups via the Tukey honestly significant difference (HSD) test (http://faculty.vassar.edu/lowry/VassarStats.html). A similar approach was used to evaluate the significance of ciprofloxacin-triggered increased mutation frequencies in repeated experiments. P values of <0.05 with two-tailed significance levels were used.
Average frequencies of Rifr mutants were not significantly different for the Cips parental strain ISP 794 (4.57 ×10−8), its Cipr parC gyrA mutant RA1 (8.41 ×10−8), or the unrelated Cips MRSA clinical isolate MRGR3 (1.44 × 10−7), grown in antibiotic-free CAMHB (Fig. 1).
In contrast, average frequencies of rifampin resistance were elevated in DNA repair mutants of Cipr strain RA1, namely, by ca. 7.5-fold in RA1 mutS2 (6.30 ×10−7) and 13.7-fold in RA1 mutL (1.15 ×10−6), compared to their parent RA1 (P < 0.01). In addition, rifampin MICs were elevated for RA1 mutS (0.06 μg/ml) and RA1 mutL (0.25 μg/ml) compared to RA1 (0.008 μg/ml). On the other hand, ciprofloxacin MICs (32 μg/ml) were equivalent for RA1 mutS, RA1 mutL, and RA1.
In contrast to the case for RA1 mutS and RA1 mutL, the average frequency of rifampin resistance (1.16 ×10−7) scored in the recA mutant of RA1 was not elevated compared to those displayed by ISP794 and RA1. While the rifampin MIC for RA1 recA was unchanged compared to that for RA1, its ciprofloxacin MIC (2 μg/ml) was markedly lower than that (32 μg/ml) for RA1, as previously reported (6).
Induction of rifampin resistance by S. aureus exposure to sub-MICs of ciprofloxacin was significantly influenced by ciprofloxacin levels and strain background (Fig. 1). Since pilot experiments indicated that S. aureus exposure to 1/8× MICs of ciprofloxacin led to rifampin resistance frequencies that were marginally increased compared to those under ciprofloxacin-free conditions (data not shown), this subinhibitory concentration was not further tested.
Frequencies of Rifr mutants were more strongly promoted for Cipr strain RA1 than for ISP794 by growth in the presence of 1/4× MICs (33-fold versus 3-fold) or 1/2× MICs (27-fold versus 2.3-fold) of ciprofloxacin, respectively, compared to antibiotic-free conditions. The higher ciprofloxacin-mediated induction of rifampin resistance in strain RA1 compared to its Cips parent remains unexplained but suggests the possibility that at the higher concentrations of ciprofloxacin used for RA1, there could be other mutagenic properties of ciprofloxacin independent of its interaction with its usual topoisomerase targets.
Compared to strain ISP794, exposure of the Cips MRSA clinical isolate MRGR3 to 1/4× or 1/2× MICs of ciprofloxacin led to higher, namely, 5-fold or 6-fold, respectively, increases (P < 0.01) in average frequencies of rifampin resistance.
Frequencies of Rifr mutants of RA1 mutS significantly (P < 0.05) increased by ca. 3.4-fold during growth in the presence of 1/2×, but not 1/4×, MIC of ciprofloxacin compared to growth in ciprofloxacin-free CAMHB. A 3.3-fold increase in mean frequencies of Rifr mutants, compared to that in ciprofloxacin-free cultures, was also recorded in RA1 mutL exposed to 1/2× MIC of ciprofloxacin, without reaching significance.
In contrast to the case for mutS and mutL mutants of strain RA1, we could not demonstrate increased frequencies of rifampin resistance in strain RA1 recA exposed to sub-MICs of ciprofloxacin. It should be mentioned that the irregular growth rate characteristics of strain RA1 recA exposed to sub-MICs of ciprofloxacin, which frequently led to low numbers of Rifr mutants on rifampin-supplemented agar (≤10 CFU/plate), prevented accurate estimates of ciprofloxacin-induced rifampin resistance frequencies.
Table 1 shows rifampin resistance levels and rpoB mutations assayed in Cipr RA1 and Cips MRGR3 strains, which showed the strongest response to ciprofloxacin exposure. To simplify the description of rifampin resistance levels among mutants that displayed a wide range of rifampin MICs (>0.25 to >128 μg/ml) in broth macrodilution assays, rifampin MIC data were pooled into three categories, arbitrarily designated low-level (rifampin MIC, >0.25 to 8 μg/ml), high-level (rifampin MIC range, 16 to 64 μg/ml), and very-high-level (rifampin MIC, ≥ 128 μg/ml) resistance. The distribution of rifampin resistance levels was influenced to some extent by ciprofloxacin exposure and strain background (Table 1). The proportion of Rifr mutants with very-high-level resistance decreased from 56 to 38% in Cipr strain RA1, but not Cips strain MRGR3, after growth in ciprofloxacin-containing compared to antibiotic-free CAMHB.
Sequence analysis of rifampin resistance-linked mutations in Cipr MSSA strain RA1 and Cips MRSA strain MRGR3 revealed two categories of rifampin resistance-conferring mutations at specific positions of rpoB cluster I (3, 7, 34, 37, 44): (i) amino acid substitutions and (ii) in-frame deletions or insertions resulting in single or multiple amino acid insertions or deletions (Table 1). In-frame deletions or insertions linked with rifampin resistance were observed exclusively in ciprofloxacin-exposed cultures. We found no occurrence of multiple mutations in, or mutations located outside, rpoB resistance cluster I of any Rifr mutant, in contrast to previous studies (3, 37, 44).
The overall distribution of each rpoB mutation among Rifr colonies was not uniform across the different strains and incubation conditions, but some mutations occurred more frequently than others. This nonuniform distribution is likely explained by the random-mutation model of Luria and Delbrück (25), in which early-occurring mutations generate more mutant progeny during growth than subsequently generated mutations.
Altogether, 14 different amino acid substitutions were identified at nine positions of rpoB cluster I, namely, Ser464, Gln468, Asp471, Asn474, Ala477, His481, Arg484, Ser486, and Leu488 (Table 1). Excepting for the rarely encountered Asp471/Asn and His481/Leu mutations that were not previously reported, all other mutations (n = 12) were already linked with rifampin resistance in S. aureus (3, 7, 34, 37, 44) and/or E. coli (3). The type of amino acid substitution determined the rifampin resistance level (Table 1), as previously reported (3, 7, 34, 37, 44). There were five amino acid substitutions leading to low-level, one amino acid substitution leading to high-level, and eight substitutions in six amino acids leading to very-high-level rifampin resistance.
In contrast to cultures grown in ciprofloxacin-free medium, which essentially yielded missense mutations leading to amino acid substitutions conferring rifampin resistance, a significant proportion (ca. 40%) of Rifr mutants selected from ciprofloxacin-containing cultures revealed in-frame mutations resulting in insertions or deletions of one or more amino acids in rpoB resistance cluster I (Table 1).
The most frequently encountered Rifr deletion mutation was ΔLeu485, which as far as we know was the only one to be previously detected in S. aureus unexposed to ciprofloxacin (37). Interestingly, this single-amino-acid deletion was generated by two different, independently generated nucleotide deletions (Table 1). Another Rifr mutant had a 4-amino-acid deletion (Leu479 to Lys482) that encompassed the critical residue His481. Finally, three additional Rifr mutants had 2-, 3-, and 4-amino-acid deletions starting from Leu488, respectively. Altogether, there were eight different nucleotide changes that led to single or multiple amino acid deletions and were linked with increased rifampin resistance.
In ciprofloxacin-exposed cultures, some Rifr mutants revealed insertions of single or multiple amino acids occurring at six different positions, namely, after Asp471, Ala473, Asn474, Thr480, Ser486, and Ala487 of rpoB cluster I (Table 1). Four of those insertions resulted from codon repeats that led to duplication of Ala473, Thr480, Ser486, or Ala487, respectively.
The high proportion and diversity of rifampin resistance-associated insertion and deletion mutations that were triggered by continuous exposure of bacterial cultures to 1/2× and 1/4× MICs of ciprofloxacin were unexpected. These genomic alterations could be the result of major quinolone-triggered disturbances in DNA fork progression, resulting in DNA slippage and accumulation of unrepaired DNA replication errors by unknown mechanisms during replication restart.
Rifampin resistance levels and rpoB cluster I genomic alterations were also recorded in Rifr mutants from ciprofloxacin-exposed or unexposed cultures of DNA repair mutant strains RA1 recA, RA1 mutS, and RA1 mutL (Table 2). While each type of amino acid substitution still predicted rifampin resistance levels, their distribution in Rifr mutants of the DNA repair mutants differed to some extent from that recorded in either their parental strain RA1 or the Cips MRSA strain MRGR3 (Table 1), with only two, one, and seven amino acid substitutions, leading to low, high, and very high levels of rifampin resistance, respectively. A striking feature of DNA repair mutants RA1 recA, RA1 mutS, and RA1 mutL was the high proportion of colonies showing the amino acid His481/Tyr substitution, especially after growth in ciprofloxacin-containing medium. A second important feature was the absence of any deletions in Rifr mutants of DNA repair mutants. In contrast, five different types of in-frame insertions were detected in rpoB cluster I of Rifr mutants selected from ciprofloxacin-exposed strains RA1 mutS and RA1 mutL (Table 2).
Fitness costs of the 13 most frequently encountered Rifr mutations were estimated by a previously described competitive growth assay, comparing each Rifr mutant to its respective Rifs parent. Eight of the Rifr-conferring amino acid substitutions had a significant fitness cost, ranging from 0.70 (±0.07) for Arg484/His to 0.89 (±0.06) for Ser484/Pro, which were similar to previously reported estimates (37, 44), except for the His481/Asn substitution, which showed no fitness cost in one study (44).
Fitness costs of some insertion or deletion mutations were also estimated but showed more extensive variations than those for the amino acid substitutions according to the nature and location of each insertion or deletion. While the fitness cost of the most frequently encountered deletion mutation, ΔLeu485, was 0.67 (±0.07), the fitness costs of mutations causing three different amino acid to be duplicated by codon insertions were 0.65 (±0.07) for Ala487[Ala], 0.69 (±0.03) for Thr480[Thr], and 0.80 (±0.06) for Ala473[Ala]. Further insertions or deletions at other amino acid sites showed more severe fitness costs, which could not be accurately estimated by the competitive growth assay (data not shown).
While the high prevalence of antibiotic resistance in S. aureus nosocomial isolates is currently explained by the selective action of intensively used topical and systemic antimicrobial agents in health care settings, a detailed analysis of molecular mechanisms linking resistance with bacterial exposure to each category of antimicrobial agents requires the use of simplified in vitro models. We evaluated the impact of subinhibitory concentrations of the DNA-targeting antimicrobial agent ciprofloxacin, as well as the contribution of DNA repair systems, on in vitro emergence of genomic alterations linked with rifampin resistance in ciprofloxacin-susceptible or -resistant strains of S. aureus.
Emergence of rifampin resistance is commonly used for evaluating resistance mutation frequencies in S. aureus and other bacterial species (3, 7, 35, 38). However, this sensitive detection method is labor-intensive due to the number of amino acid substitutions that are linked with rifampin resistance (3, 7). At least 23 different mutations have been previously recorded in 13 different positions of the rpoB rifampin resistance cluster I (3, 7, 34, 36, 37, 44, 45). Furthermore, each mutation has its own specific impact on rifampin resistance level (3) and is linked with a specific fitness cost (37, 44).
While our initial results confirmed the previously reported potential of ciprofloxacin to increase rifampin resistance frequencies in S. aureus (9, 29), sequencing studies revealed a high diversity of unanticipated ciprofloxacin-triggered insertions or deletions in rpoB resistance cluster I, which were absent in control cultures. The variety of amino acid positions at which insertion or deletion events took place within the very short 18-amino-acid segment of this resistance cluster was impressive and, by extension, may suggest a wider occurrence of similar alterations in other genomic regions of ciprofloxacin-exposed bacteria. In line with current models of ciprofloxacin antibacterial activity, those genomic alterations might reflect major disturbances in DNA fork progression, which may result in DNA slippage and accumulation of unrepaired DNA replication errors during replication restart (11, 12, 20, 21, 31). More extensive sequencing studies to evaluate the extent of genomic alterations resulting from ciprofloxacin exposure in S. aureus Rifr mutants are in progress.
Previous reports emphasized the role of the SOS response, involving an error-prone Y family DNA polymerase that lacks intrinsic proofreading activity (9), on ciprofloxacin-induced mutagenesis conferring rifampin resistance (9, 28, 29). However, the impact of the error-prone SOS response on the spectrum of rifampin resistance-conferring mutations triggered by ciprofloxacin exposure in S. aureus, especially the rarely reported in-frame insertions and deletions, is still uncertain. Whereas induction of key components of the SOS response by ciprofloxacin exposure was demonstrated in S. aureus by use of gene promoter-reporter fusions (28, 29) and microarray studies (9), neither a non-SOS-inducible lexA mutant nor a LexA-regulated, error-prone polymerase mutant displayed increased sensitivity to ciprofloxacin (9). While inactivation of the LexA-regulated Y family polymerase indeed prevented induction of rifampin resistance in response to UV damage (9), no similar effect was reported for ciprofloxacin-exposed S. aureus.
The high frequency of insertions and/or deletions occurring in rpoB cluster I of ciprofloxacin-exposed S. aureus is difficult to link with induction of the Y family DNA polymerase (9) when considering current models of DNA repair in prokaryotic organisms (11, 12, 20, 21, 31). According to these models, double-strand DNA breaks, including those resulting from ciprofloxacin exposure (13, 27), should be processed by recombinational DNA repair pathways that are essential for assembly of a repair primosome leading to completion of DNA replication (11, 21, 31, 42). In model organisms such as E. coli, postreplicative correction of DNA damage, including base mismatches and small insertions or deletions, is also influenced by the DNA mismatch repair (MMR) system, which removes nucleotides mispaired by DNA polymerases and insertion/deletion loops that result from slippage during replication or recombination (20). Accordingly, defects in the MMR system may lead to mutator phenotypes in different bacterial species (4, 8, 35, 38). In addition, the MMR system can downregulate the formation of heteroduplex tracts during genetic recombination (20).
Current models indicate that MMR proteins may be limiting in stress-exposed or stationary-phase bacterial cultures (8, 20). In S. aureus, mutSL expression was found to be weakly increased (28, 29) or even decreased (9) during ciprofloxacin exposure, which was explained by the absence of a defined SOS box in the mutSL promoter region (9). Defects in the MMR system were also linked with mutational emergence of aminoglycoside-induced small-colony variants of S. aureus (41) and hypermutability of such variants (4, 39).
Taken together, our data combined with those from previous reports support the view that induction of the SOS response, in particular its recA and lexA key components, is essential for initiating overall DNA repair in ciprofloxacin-exposed S. aureus. However, other key components of DNA repair, not directly controlled by the SOS response but belonging to the recombinational and mismatch repair systems, are likely required but potentially overloaded by the multiple ciprofloxacin-induced DNA damage events. This may indirectly contribute to accumulation of unrepaired DNA replication errors, in particular insertions and deletions, which may not only promote antibiotic resistance-conferring mutations but also contribute to quinolone-induced bacterial death. Further studies are needed to elucidate the functional role of major DNA repair systems in S. aureus and evaluate their contribution to mutational antibiotic resistance or bacterial death.
This study was supported by research grants 320000-116518 (to P.V.), 31003A-120428 (to W.K.), and 3200B0-108401 (to D.P.L.) from the Swiss National Science Foundation and by research grant R37-AI023988-20 (to D.C.H.) from the U.S. Public Health Service, National Institutes of Health.
We thank R. Leclercq for providing mutS2 and mutL mutants of S. aureus strain RN4220.
Published ahead of print on 28 February 2011.