The Thymidine-Dependent Small-Colony-Variant Phenotype Is Associated with Hypermutability and Antibiotic Resistance in Clinical Staphylococcus aureus Isolates

Abstract

Thymidine-dependent small-colony variants (TD-SCVs) of Staphylococcus aureus can be isolated from the airway secretions of patients suffering from cystic fibrosis (CF) and are implicated in persistent and treatment-resistant infections. These characteristics, as well as the variety of mutations in the thymidylate synthase-encoding thyA gene which are responsible for thymidine dependency, suggest that these morphological variants are hypermutable. To prove this hypothesis, we analyzed the mutator phenotype of different S. aureus phenotypes, in particular CF-derived TD-SCVs, CF-derived isolates with a normal phenotype (NCVs), and non-CF NCVs. The comparative analysis revealed that the CF isolates had significantly higher mutation rates than the non-CF isolates. The TD-SCVs, in turn, harbored significantly more strong hypermutators (mutation rate ≥ 10⁻⁷) than the CF and non-CF NCVs. In addition, antimicrobial resistance to non-beta-lactam antibiotics, including gentamicin, ciprofloxacin, erythromycin, fosfomycin, and rifampin, was significantly more prevalent in TD-SCVs than in CF and non-CF NCVs. Interestingly, macrolide resistance, which is usually mediated by mobile genetic elements, was conferred in half of the macrolide-resistant TD-SCVs by the point mutation A2058G or A2058T in the genes encoding the 23S rRNA. Sequence analysis of mutS and mutL, which are involved in DNA mismatch repair in gram-positive bacteria, revealed that in hypermutable CF isolates and especially in TD-SCVs, mutL was often truncated due to frameshift mutations. In conclusion, these data provide direct evidence that TD-SCVs are hypermutators. This hypermutability apparently favors the acquisition of antibiotic resistance and facilitates bacterial adaptation during long-term persistence.

Cystic fibrosis (CF), the most common life-shortening autosomal recessive disorder in populations of European origin, is caused by alterations in the cystic fibrosis transmembrane conductance regulator (45). Mutations in the gene encoding this protein disrupt electrolyte secretion, leading to hyperosmolar viscous mucus (23). While the gene defect results in a myriad of medical problems, ultimately 80 to 95% of CF patients succumb to respiratory failure brought on by chronic bacterial infection and concomitant bronchopulmonary inflammation (39). Pseudomonas aeruginosa, Staphylococcus aureus, and Haemophilus influenzae are considered the major pathogens that chronically infect the airways of CF patients (10).

Appropriate antibiotic therapy directed against the pathogens isolated from the respiratory tract reduces the morbidity of CF lung disease (12). However, with increasing duration of infection, antibiotic treatment becomes more and more challenging. Bacterial populations from CF airways develop resistance to several antibiotics (43), acquire the capacity for biofilm formation (33), and display a wide spectrum of different morphotypes, including small-colony variants (SCVs) (5, 13, 16, 17, 20).

It has been suggested that hypermutable strains, defined as cells that display higher mutation frequencies than their wild-type counterparts, adapt specifically to the highly compartmentalized and stressful environment of the CF lung, as well as to the challenges imposed by host defenses and prolonged antibiotic therapies (31). Indeed, high percentages of S. aureus, P. aeruginosa, and H. influenzae strains isolated from the chronically infected airways of CF patients show higher mutation frequencies than bacterial strains isolated from non-CF patients (7, 18, 36, 38).

Hypermutability requires the occurrence of mutations in genes that are responsible for DNA repair, i.e., genes of the oxidized guanine repair system (mutM, mutT, and mutY) and/or genes of the methyl-directed mismatch repair (MMR) system (mutS, mutL, mutH, and uvrD) (29). In S. aureus, a relationship between inactivation of MutS or MutL, the essential components of the MMR system in gram-positive bacteria, and emergence of a hypermutator phenotype was demonstrated recently (35). Moreover, it has been speculated that hypermutability favors the emergence of macrolide resistance in CF-derived S. aureus isolates and that this resistance is unusually conferred by point mutations in the ribosomal genes encoding the 23S rRNA (36, 37).

However, no data are available regarding the mutator phenotype of thymidine-dependent S. aureus SCVs (TD-SCVs), which can be isolated frequently in the context of CF, especially after prior use of trimethoprim-sulfamethoxazole (SXT). These versatile variants are implicated in persistent and recurrent infections, and some of them have the ability to revert to normal-colony variants (NCVs) (5, 13, 22, 34). In contrast to S. aureus NCVs, TD-SCVs grow only in the presence of extracellular thymidine, form smaller colonies, are less pigmented, show decreased hemolytic activity, and are generally resistant to SXT (5, 13, 21). All these characteristics as well as the observation that random mutations in the highly conserved thymidylate synthase-encoding gene thyA are involved in the formation of the TD-SCV phenotype (4, 6) suggest that TD-SCVs are hypermutators, possibly evolving from hypermutable S. aureus NCVs. To verify this hypothesis we analyzed clinical S. aureus isolates comprising CF-derived TD-SCVs, CF-derived NCVs, and non-CF NCVs with regard to (i) mutator phenotype, (ii) antibiotic susceptibility and presence of ribosomal mutations conferring macrolide resistance, and (iii) integrity of the major components of the MMR system, i.e., mutS and mutL.

MATERIALS AND METHODS

Bacterial strains.

CF-derived S. aureus isolates, TD-SCVs (n = 20) as well as NCVs (n = 20), were recovered from routine respiratory specimens from CF patients attending the University Hospital of Frankfurt am Main, Frankfurt am Main, Germany, between January 2004 and January 2007. All TD-SCVs showed characteristics of an SCV phenotype (21) and were subjected to species confirmation as described previously (5). Seventy percent (n = 14) of the patients testing positive for TD-SCVs had definitely received prior therapy or long-term prophylaxis with SXT. ATCC 29213, nasal isolates (n = 10), and blood culture isolates (n = 10) were used as non-CF S. aureus control strains.

DNA techniques.

S. aureus DNA was isolated with a QIAamp DNA minikit (Qiagen, Hilden, Germany) following digestion of the bacterial cell wall with 10 U/ml lysostaphin (Sigma-Aldrich, Steinheim, Germany). PCR was carried out with Phusion (Finnzymes, Espoo, Finland), a DNA polymerase with strong proofreading activity. Amplification of the genes conferring macrolide resistance (ermA, ermB, ermC, and msrA) was done with the primers described by Lina et al. (26) and amplification of the MMR system components mutS and mutL with the primers listed by Prunier and Leclercq (35). In addition, domain V of the 23S rRNA was amplified with the primers P3 (5′-CTGTCTCAACGAGAGACTCGG-3′) and P4 (5′-CGCTCACGTTTCAAAGGCTCC-3′). Nucleotide sequences of DNA fragments were determined by cycle sequencing using an ABI Prism DNA sequencer (Applied Biosystems, Foster City, CA). Clonal relatedness of S. aureus isolates was analyzed by pulsed-field gel electrophoresis after SmaI restriction of whole chromosomal DNA. The procedure was performed as described previously (47).

Determination of antibiotic susceptibility and mutation frequency.

Antimicrobial susceptibility was determined by the disk diffusion method according to CLSI (8) and Comité de l'Antibiogramme de la Société Française de Microbiologie (9) (fosfomycin only) guidelines. In addition, MICs were measured for rifampin using Etest strips (AB Biodisk, Solna, Sweden). Isolates with an NCV phenotype were tested on Mueller-Hinton (MH) agar (Heipha Diagnostika, Eppelheim, Germany), whereas TD-SCVs were tested on MH agar supplemented with 5% defibrinated sheep blood (Heipha Diagnostika), as they cannot be cultured in the absence of thymidine (5, 13). Plates were incubated at 37°C for 24 h. For the measurement of mutation frequencies (31) of SCV and NCV isolates, one bacterial colony was resuspended in 20 ml of brain heart infusion (BHI) broth (Becton Dickinson, Heidelberg, Germany) and incubated overnight with shaking (200 rpm) at 37°C. Bacterial cells were then collected by centrifugation at 3,000 rpm for 5 min and resuspended in 1 ml of BHI broth. A 100-μl sample of this suspension as well as samples from successive dilutions was plated onto BHI agar plates with and without rifampin (100 μg/ml). After 48 h of incubation at 37°C, the CFU were counted, and the mutation frequencies were determined by dividing the number of CFU on antibiotic-supplemented agar by the number of CFU on antibiotic-free agar (46). Three colonies growing in high antibiotic concentrations were streaked onto another plate with the antibiotic agent in order to prove the stability of the mutants. All experiments were performed in triplicate, and mean values ± standard deviations were reported. The isolates were originally susceptible to the concentration of rifampin used. For the two isolates with elevated rifampin MICs (16 and 32 μg/ml), the mutation frequency measurement was confirmed with streptomycin (50 μg/ml) as described above.

Homology searches and sequence alignments.

Sequences of mutS, mutL, and the ribosomal genes encoding the 23S rRNA of S. aureus were retrieved from the NIH GenBank database (http://www.ncbi.nlm.nih.gov) and aligned using DNASTAR Lasergene software version 7.2 and Chromas software version 1.45.

Statistical analysis.

The two-tailed Fisher's exact test was used to analyze binary variables, whereas continuous scaled variables were evaluated with the nonparametric Kruskal-Wallis test. P values of <0.05 were considered statistically significant. Statistical analysis was performed using BiAS software version 8.1.

Nucleotide sequence accession numbers.

The sequences determined in this study have been deposited in the EMBL nucleotide sequence database under the following accession numbers (mutS and mutL, respectively): S33, AM980462 and AM980674; S68, AM980463 and AM980673; S231, AM980464 and AM980672; S93, AM980465 and AM980671; S96, AM980466 and AM980670; S34, AM980467 and AM980669; S187, AM980468 and AM980668; S196, AM980469 and AM980667; S165, AM980470 and AM980666; S178, AM980471 and AM980665; S163, AM980472 and AM980664; S181, AM980473 and AM980663; BK119, AM980474 and AM980662; NA03, AM980475 and AM980661; BK156, AM980476 and AM980660; NA29, AM980477 and AM980659; NA11, AM980478 and AM980658; BK106, AM980479 and AM980657; ATCC 29213, AM980480 and AM980656.

RESULTS

It was recently shown that mutations in the thymidylate synthase gene thyA are involved in the formation of the TD-SCV phenotype in S. aureus (4, 6). Sequence analysis of thyA of clinical TD-SCVs revealed that these mutations are highly variable. Various mutations located at different positions occurred in this highly conserved gene, including point mutations resulting in nonsense or missense mutations, in-frame deletions or insertions, and frameshift mutations (4, 6). To verify that TD-SCVs are hypermutable, we analyzed the mutation frequencies, the antibiotic resistance patterns, and the integrity of the MMR system components mutS and mutL of clinical S. aureus isolates comprising 20 CF-derived TD-SCVs, 20 CF-derived NCVs, and 20 non-CF NCVs.

Mutator phenotype.

The mutation frequencies of S. aureus isolates are shown in Fig. 1. Significantly higher mutation rates were determined for the CF-derived S. aureus isolates, i.e., NCVs and TD-SCVs, than for the non-CF NCVs (P < 10⁻⁴, respectively, Kruskal-Wallis test). While the non-CF NCVs showed mutation frequencies (mean ± standard deviation, [8.5 ± 15.8] × 10⁻¹⁰) similar to that of the reference strain ATCC 29213 ([2.1 ± 1.4] × 10⁻¹⁰), the mutation frequencies of the CF NCVs ranged from 8.8 × 10⁻¹⁰ to 4.8 × 10⁻⁸ and those of the TD-SCVs from 1.3 × 10⁻⁹ to 3.8 × 10⁻⁶. Although the mean mutation frequency determined for the TD-SCVs ([3.4 ± 9.0] × 10⁻⁷) was higher than that for the CF NCVs ([8.0 ± 12.7] × 10⁻⁹), the observed difference was not statistically significant (P = 0.12, Kruskal-Wallis test). However, with regard to strong hypermutability, the TD-SCVs harbored significantly more strong hypermutators than the CF and non-CF NCVs (P = 0.047, Fisher's exact test). Strong hypermutators were defined as isolates with mutation rates of ≥10⁻⁷ (29, 36). According to this definition, we detected five strong hypermutators among the TD-SCVs and none among the CF NCVs or the non-CF NCVs.

FIG. 1.

Mutation frequencies of S. aureus isolates. CF-derived TD-SCVs, NCVs, and non-CF NCVs cultured from nasal swabs or blood cultures were analyzed. Each dot represents the mean mutation frequency plus standard deviation calculated from three independent experiments for one isolate. CF isolates (TD-SCVs and NCVs) showed significantly higher mutation frequencies than non-CF isolates (P < 10⁻⁴, Kruskal-Wallis test). The TD-SCVs harbored significantly more strong hypermutators (mutation rate ≥ 10⁻⁷) than the CF and non-CF NCVs (P = 0.047, Fisher's exact test). Culture conditions are defined in Materials and Methods.

Antibiotic susceptibility.

The resistance profiles of S. aureus isolates are presented in Fig. 2. Antibiotic resistance to the non-beta-lactam antibiotics gentamicin, ciprofloxacin, erythromycin, fosfomycin, and rifampin was significantly more prevalent in TD-SCVs than in CF NCVs (P = 0.037, Fisher's exact test) and non-CF NCVs (P = 0.004, Fisher's exact test). In the group of the TD-SCVs, the resistance to non-beta-lactam antibiotics was mainly due to isolates with a distinctly elevated mutation frequency (Table 1). Thus, all gentamicin-resistant (n = 4), all fosfomycin-resistant (n = 2), and all rifampin-resistant (n = 2) TD-SCVs as well as five of the six erythromycin-resistant and four of the five ciprofloxacin-resistant TD-SCVs showed mutation frequencies of ≥10⁻⁸. Analysis of the mechanism of macrolide resistance revealed that in all three non-CF NCVs and in three of four CF NCVs, erythromycin resistance was conferred by the methylase gene ermA or ermC, whereas only three of six erythromycin-resistant TD-SCVs harbored these mobile genetic elements. In the remaining three TD-SCVs, macrolide resistance was mediated by the ribosomal point mutations A2058G (S33 and S93) and A2058T (S231). In one CF NCV isolate, neither methylase/efflux genes nor point mutations in domain V of the 23S rRNA could be detected.

FIG. 2.

Antimicrobial resistance of S. aureus CF-derived TD-SCVs (n = 20), CF NCVs (n = 20), and non-CF NCVs (n = 20).

TABLE 1.

Alterations of MutS and MutL in S. aureus isolates

Isolate	Alteration(s) of MutSa	Alteration(s) of MutLb	Mutation frequency (mean ± SD)	Resistance phenotypec
CF TD-SCVsd
S33	I700V	N343I, 345YYIHSNNKKSNLNKDKTQRITKRRRFHLKKVTVSHLWQKIKTMR388, I389Stop	(3.8 ± 4.0) × 10−6	AMP, OXA, CIP, ERY, SXT, FOF, RIF
S68	None	N343I, 345YYIHSNNKKSNLNKDKTQRITKRRRFHLKKVTVSHLWQKIKTMR388, I389Stop	(5.5 ± 3.0) × 10−7	ERY, SXT
S231	None	N343I, 345YYIHSNNKKSNLNKDKTQRITKRRRFHLKKVTVSHLWQKIKTMR388, I389Stop	(1.2 ± 0.9) × 10−7	CIP, ERY, SXT, RIF
S93	None	141RNRYTCRIIIL151, N152Stop	(3.2 ± 3.6) × 10−8	AMP, CIP, ERY, SXT
S96	None	G63E	(5.1 ± 7.1) × 10−7	AMP, SXT
S34	None	None	(1.6 ± 1.5) × 10−6	AMP, GEN, CIP, SXT
CF NCVsd
S187	S201P	N343I, 345CYIHSNNKKSNLNKDKTQRITKRRRFHLKKVTVSHLWQKIKTMR388, I389Stop	(4.8 ± 3.4) × 10−8	AMP
S196	N588D, S814C	N364D, S377R, N418D	(4.1 ± 0.9) × 10−8	AMP
S165	None	N364D	(1.2 ± 1.2) × 10−8	AMP
S178	None	None	(8.7 ± 6.1) × 10−9	AMP
S163	None	None	(7.9 ± 9.1) × 10−9	AMP
S181	None	None	(6.4 ± 5.1) × 10−9
Non-CF NCVse
BK119	S201P	None	(1.8 ± 1.5) × 10−10	AMP
NA03	None	None	(4.1 ± 2.7) × 10−10
BK156	None	None	(2.7 ± 1.0) × 10−10	AMP
NA29	None	None	(2.6 ± 0.8) × 10−10	AMP
NA11	None	None	(1.7 ± 1.0) × 10−10
BK106	None	None	(1.1 ± 0.5) × 10−10	AMP, CIP
ATCC 29213	None	None	(2.1 ± 1.4) × 10−10	AMP

^aThe full length of the predicted mutS product is 872 amino acids.

^bThe full length of the predicted mutL product is 669 amino acids.

^cAntibiotic susceptibility was determined for ampicillin (AMP), oxacillin (OXA), gentamicin (GEN), ciprofloxacin (CIP), erythromycin (ERY), trimethoprim-sulfamethoxazole (SXT), fosfomycin (FOF), and rifampin (RIF) by the disk diffusion method. Zones of inhibition were evaluated according to CLSI (8) and Comité de l'Antibiogramme de la Société Française de Microbiologie (FOF only) (9) guidelines.

^dSix isolates with the highest mutation rates are shown as examples.

^eSix isolates with normal mutator phenotypes (mutation frequency < 1.0 × 10⁻⁹) are shown as examples.

Analysis of the mutS and mutL genes.

To investigate whether hypermutability is associated with defects in the MMR system, we sequenced the entire mutS (2,619 bp) and mutL (2,010 bp) genes of selected S. aureus isolates. CF isolates with the highest mutation frequencies (TD-SCVs, n = 6; CF-NCVs, n = 6), non-CF NCVs with a normal mutator phenotype (n = 6; mutation frequency, <1 × 10⁻⁹), and reference strain ATCC 29213 were chosen for comparative sequence analysis (Table 1). Alignment of the sequences of S. aureus strains N315, Mu50, MW2, MRSA252, MSSA476, COL, and NCTC8325 revealed that MutS and MutL were highly conserved (protein sequence identity, 99.2 to 100% and 98.9 to 100%). Accordingly, no mutations could be detected in the mutSL operon of reference strain ATCC 29213. Five of the six SCV isolates analyzed, however, showed alterations in the deduced MutS and/or MutL amino acid sequences. In particular, four of these isolates contained either a 1-bp deletion (S33, S68, and S231) or a 1-bp insertion (S93) which resulted in premature termination of the mutL coding sequence due to the corresponding frameshift. Interestingly, the truncation occurred in three isolates (S33, S68, and S231) at the same amino acid position within MutL, and one of these isolates (S33) harbored an additional amino acid substitution (I700V) of unknown significance in MutS. One isolate (S96) showed a single amino acid substitution (G63E) in a region of MutL that is implicated in the ATP-binding site in Escherichia coli (1-3). With regard to the NCV isolates, however, only one CF NCV (S187) with an elevated mutation frequency showed a truncation of MutL. Again, a 1-bp deletion implemented a stop codon at amino acid position 389. Furthermore, two CF NCVs with elevated mutation frequencies (S196 and S165) and one non-CF NCV with a normal mutator phenotype (BK119) showed amino acid substitutions of unknown significance in MutS, MutL, or both. No alterations were found in the deduced MutS/MutL amino acid sequences of one TD-SCV (S34), three CF NCVs (S178, S181, and S163), and five non-CF NCVs (NA03, NA11, NA29, BK106, and BK156). Clonal unrelatedness of the CF isolates with implementation of a stop codon at the same amino acid position within MutL was established by pulsed-field gel electrophoresis (Fig. 3).

FIG. 3.

Pulsed-field gel electrophoresis patterns of SmaI digests of total DNA of the clinical CF isolates S33, S68, and S231 (TD-SCVs) and S187 (NCV), which all showed a premature termination of the mutL coding sequence at the same amino acid position. Lane M, DNA size marker. Molecular sizes of the marker fragments (in kilobases) are given on the left.

DISCUSSION

Hypermutability is considered a driving force in bacterial evolution and the key factor in the development of antimicrobial resistance in bacterial pathogens causing chronic lung infections (27, 31). Despite, or precisely because of, attenuated virulence and reduced fitness in vitro, mutator strains are implicated in long-term persistence and treatment-resistant infections (7, 18, 28, 30), especially in the context of CF. The proportion of mutators in the clinical setting of CF ranges from 14.5% to 54.4% depending on the method used for detection, the mutator definition applied, and the bacterial species analyzed (7, 15, 36, 38). Although various species, including P. aeruginosa, S. aureus, and H. influenzae (7, 36, 38), have been analyzed, there are no data regarding the mutator strength of TD-SCVs, the S. aureus phenotype that in contrast to the normal phenotype is isolated primarily from respiratory specimens from older CF patients with advanced disease (5). In line with previous studies on CF-derived S. aureus NCVs (36), our comparative mutator phenotype analysis determined significantly higher mutation frequencies for both CF-derived S. aureus phenotypes, TD-SCVs and NCVs, than for the non-CF isolates. The TD-SCVs, in turn, showed a significantly higher prevalence of strong mutators than all NCV isolates analyzed in this study, both CF and non-CF derived. In accordance with previous studies on P. aeruginosa and S. aureus CF NCVs (7, 31, 44), antibiotic resistance was more prevalent among isolates with elevated mutation frequencies. Thus, TD-SCVs with a mutation frequency of ≥10⁻⁸ had acquired antibiotic resistance. Interestingly, resistance to non-beta-lactam antibiotics, including gentamicin, ciprofloxacin, erythromycin, fosfomycin, and rifampin, was significantly more prevalent in TD-SCVs than in CF and non-CF NCVs. We have no general explanation for this phenomenon, but it may indicate acquisition of antimicrobial resistance via point mutations due to the hypermutable phenotype. Our observation that macrolide resistance in TD-SCVs was conferred not only by methylase or efflux genes but also by the emergence of ribosomal point mutations as described for hypermutable CF NCVs (37) corroborates this hypothesis. In line with previous studies, all 20 TD-SCVs were resistant to SXT, an antimicrobial agent that interferes with the tetrahydrofolic acid pathway (13, 20). Tetrahydrofolic acid acts as a cofactor for thymidylate synthase, which is essential for intracellular thymidine supply. As TD-SCVs have detrimental mutations in the thymidylate synthase gene thyA and are able to use extracellular thymidine from the environment, this SXT resistance is probably a result of the ability to bypass the folic acid pathway, rather than a result of acquired resistance genes (34, 49).

The mutator phenotype has been linked in various bacterial genera to defects in the MMR system (14, 25, 31, 32, 48), which maintains genome stability by counteracting recombination between homologous but diverged DNA sequences and rectifying base-pair mismatches and small insertions or deletions introduced by DNA polymerase (19). In addition, the emergence of various morphotypic variants has been described for a P. aeruginosa strain with a disrupted mutS gene (42). While the component MutS is generally implicated in the detection of DNA mismatches and initiation of the MMR machinery, the function of MutL is to make a connection between the recognition of a mismatch and the excision of the mismatch from the strand within which it is contained (19). Accordingly, inactivation of MutS or MutL, the major MMR system components of gram-positive bacteria, in Bacillus subtilis and S. aureus has been shown to result in a hypermutator phenotype in vitro (14, 35). In addition, various alterations have been detected in MutS and/or MutL of hypermutable clinical S. aureus NCVs and one menaquinone-dependent clinical S. aureus SCV, but the significance of these amino acid substitutions or small deletions has still to be proven, as alterations of this type can also be found in isolates with a normal mutator phenotype (35, 36, 41). However, it is of note that we detected the amino acid substitutions N588D and S814C in MutS as well as the substitutions N364D, S377R, and N418D in MutL of a CF NCV isolate with an elevated mutation frequency, as described by Schaaff et al. (41) and Prunier et al. (35, 36). In contrast, hypermutability in clinical TD-SCVs seems to be primarily associated with a truncation of the mutL gene, as shown in this study. MutL can be generally dissected into a highly conserved N-terminal ATPase domain of approximately 300 residues (3), a flexible linker, and a slightly divergent C-terminal domain including a conserved biological dimerization interface (11, 24). The results of our sequence analysis revealed that in four of six TD-SCVs analyzed, the function of MutL was most likely impaired by a premature termination of the coding sequence by frameshift mutations. Intriguingly, three of four TD-SCVs showed a truncation at the same amino acid position (I389Stop). Such premature terminations in the C-terminal region of MutL have been reported to result in a dominant mutator phenotype in E. coli (1). In addition, one CF NCV isolate with an elevated mutation frequency and none of the isolates with a normal mutator phenotype harbored a truncation in this probable hot spot mutation region. Since all S. aureus isolates analyzed in this study were genetically unrelated, a patient-to-patient transmission of the CF isolates with a MutL truncation at the same amino acid position can be excluded. The significance of the additional amino acid substitution I700V in MutS of TD-SCV isolate S33 and the single substitution G63E in MutL of TD-SCV isolate S96 remains speculative, although it is of note that the substitution G63E is located in a region that is implicated in the ATP-binding site in E. coli (3). Thus, it has been reported that several mutations in the N-terminal domain of MutL affect the binding of ATP and result in a strong mutator phenotype in E. coli (1-3). There was one hypermutable TD-SCV isolate (S34) without alterations in the mutSL operon. Alterations of other mutator genes are conceivable in this context. Other putative mutator genes have already been identified in the B. subtilis chromosome (40), and some have homologues in the S. aureus chromosome. Alterations of other putative mutator genes might also be responsible for the occurrence of TD-SCVs with only slightly elevated mutation frequencies, since none of these isolates harbored the central MutL truncation (our unpublished data). For example, it has been demonstrated in E. coli that inactivation of components of the oxidized guanine system (MutM, MutY, and MutT) gives rise to weak, moderate, and strong mutator phenotypes (29).

In conclusion, these data provide direct evidence that TD-SCVs are hypermutable. The prevalence of strong hypermutators is significantly higher for TD-SCVs than for CF NCVs. Thus, it is tempting to speculate that in the context of CF, the TD-SCVs evolve primarily from hypermutable NCV isolates. Replication errors due to a defective DNA mismatch repair system and SXT selective pressure will favor the emergence of detrimental mutations in the thyA gene and therewith the formation of a TD-SCV phenotype. The hypermutability enables the TD-SCVs to survive in the hostile environment of the bronchial CF habitat and to resist antibiotic therapy in spite of their metabolic impairments.