Reversion From Methicillin Susceptibility to Methicillin Resistance in Staphylococcus aureus During Treatment of Bacteremia

Abstract

Approximately 3% of Staphylococcus aureus strains that, according to results of conventional phenotypic methods, are highly susceptible to methicillin-like antibiotics also have polymerase chain reaction (PCR) results positive for mecA. The genetic nature of these mecA-positive methicillin-susceptible S. aureus (MSSA) strains has not been investigated. We report the first clearly defined case of reversion from methicillin susceptibility to methicillin resistance among mecA-positive MSSA within a patient during antibiotic therapy. We describe the mechanism of reversion for this strain and for a second clinical isolate that reverts at a similar frequency. The rates of reversion are of the same order of magnitude as spontaneous resistance to drugs like rifampicin. When mecA is detected by PCR in the clinical laboratory, current guidelines recommend that these strains be reported as resistant. Because combination therapy using both a β-lactam and a second antibiotic suppressing the small revertant population may be superior to alternatives such as vancomycin, the benefits of distinguishing between mecA-positive MSSA and MRSA in clinical reports should be evaluated.

Resistance of methicillin-resistant Staphylococcus aureus (MRSA) to β-lactam antibiotics depends on the production of penicillin binding protein 2a (PBP2a), an alternative cell-wall transpeptidase. PBP2a, which binds β-lactams very poorly, replaces the transpeptidase function of the native β-lactam–susceptible enzyme PBP2 and is encoded by the gene mecA. mecA was acquired by S. aureus from nonpathogenic staphylococcal species [1, 2], and current MRSA strains are thought to be derived from a small number of original acquisition events [3, 4]. When MRSA is isolated from patients for whom methicillin-susceptible S. aureus (MSSA) was the original diagnosis, it is generally assumed that the original infection was either polymicrobial, with MSSA predominating, or that the MRSA was acquired during a secondary infection event. Because acquisition of mecA is very rare, the development of resistance in isolates highly susceptible to methicillin-like antibiotics is considered unlikely and has not been rigorously documented in a clinical setting.

Distinguishing MRSA from MSSA is complicated by phenotypic variability in methicillin susceptibility, known as heteroresistance, which can produce borderline results in clinical assays. To avoid misclassification of MRSA as MSSA due to this phenomenon, some clinical laboratories have begun using polymerase chain reaction (PCR) analysis to test S. aureus isolates for the presence of mecA. A small percentage of strains that, according to results of conventional phenotypic methods, are highly susceptible to methicillin-like antibiotics also have PCR results positive for mecA [5–19]. Because PCR is substantially faster and more accurate in predicting oxacillin resistance, Clinical Laboratory Standards Institute (CLSI) guidelines recommend such strains be reported as oxacillin resistant, although the nature of these strains is not understood. One potential explanation for this population of mecA-positive MSSA is that they are mutants in which mecA has been inactivated. Indeed, mecA expression in the absence of β-lactams has been shown to adversely affect in vitro fitness [20, 21], suggesting that selection would favor inactivation of mecA in the absence of antibiotics. Whether such strains should be treated clinically as MRSA depends on the nature of the inactivating mutations and, in particular, on their ability to revert and permit expression of resistance. Given the superior efficacy and low toxicity of β-lactam antibiotics as compared to alternatives such as vancomycin, treating mutants that do not revert or that revert at sufficiently low frequencies as MRSA may deny some patients significant therapeutic benefit.

Here we report the first clearly defined case of reversion of genetically inactivated mecA to an active form within a patient during antibiotic therapy. We describe the genetic lesion and the mechanism of reversion for this strain and for a second clinical isolate, from a separate patient, that reverted at a similar frequency but via a different mechanism.

METHODS

Bacterial Strains and Culture Conditions

Strains used and generated in this work are described in Supplementary Table 1. Mueller-Hinton II cation-adjusted broth and Mueller-Hinton cation-adjusted agar (MHA) supplemented with 4% NaCl or tryptic soy broth and tryptic soy agar (all media were manufactured by Difco) were used for cultivation of all strains, with incubation at 37°C unless otherwise specified.

PCR

Colony PCR [22] was used to amplify mecA for sequence analysis, using primers mecAUF and mecADR. mecA PCR products were sequenced by Genewiz, using primers mecAUR and mecADF, in addition to mecAUF and mecADR (Supplementary Table 2).

Western Blots

Log-phase cultures of S. aureus were diluted into broth containing 0 µg mL⁻¹, 0.01 µg mL⁻¹, 0.25 µg mL⁻¹, or 12 µg mL⁻¹ oxacillin. After incubation for 1 hour at 37°C, 900 µL of an OD₆₀₀ 0.5 equivalent were spun down, resuspended in 300 µL lysis buffer (15 µg mL⁻¹ DNase, 15 µg mL⁻¹ RNase, 100 µg mL⁻¹ lysostaphin, and 50 mM Tris pH 7.4), and incubated at 37°C for 30 minutes. Laemmli sample buffer was added to the lysates, and they were boiled for 5 minutes. Lysates were then separated on 10% sodium dodecyl sulfate polyacrylamide gels, transferred to nitrocellulose membranes, and probed with monoclonal antibody specific for PBP2a (PBP2a Monoclonal [catalog no. 10-P08A], Fitzgerald Industries International). Following reaction with a secondary antibody (goat anti-mouse IgG [H+L]) conjugated with the fluorescent dye cyanine 5 (Molecular Probes), bands were visualized by scanning with a 632-nm laser on a Typhoon 9410 (GE Healthcare). Imaging for gross detection of PBP2a was done with a photomultiplier tube voltage of 600, a 200-µm pixel size, and a high sensitivity. Imaging for quantification of relative amounts of PBP2a was done at a photomultiplier tube voltage of 300, a 50-µm pixel size, and a normal sensitivity.

Fluctuation Analysis

The contribution of random and induced mutation to reversion was assessed using fluctuation analysis as described by Luria and Delbrück [23]. Overnight cultures of each strain, grown in tryptic soy broth at 37°C, were diluted to 3.33 × 10³ cells mL⁻¹ and used to inoculate three 24-well plates with 300 µL per well. After incubation in a humidified shaking chamber at 37°C for 24 hours, 100 µL of each culture was plated to MHA with 4% NaCl and 12 µg mL⁻¹ oxacillin. The OD₆₀₀ was also measured for each culture. After incubation at 37°C for 48 hours, the number of colonies on each plate was counted. Counts among cultures were normalized to the OD₆₀₀ of the cultures measured at the time of plating.

We developed a custom software tool in C that calculates the number of pre-plating and post-plating mutation events using a maximum likelihood model. This program fit the OD₆₀₀-normalized counts to 3 distributions: (1) the Luria–Delbrück distribution, which allows only for random mutation [24, 25]; (2) a Gaussian binomial approximation, which allows only for induced mutation; and (3) joint distribution model, in which an additional fitted parameter representing the mean number of induced mutations per plate is added to the Luria–Delbrück distribution [26, 27]. Although induced mutation models are traditionally described by the Poisson distribution, we found that this model suffered from overdispersion as a result of plating error. To address this problem, we defined the induced mutation parameter as the mean of a Gaussian binomial approximation with a fixed standard deviation (calculated as the mean value divided by 4.72) determined by independent measurement of variation inherent in the plating process. The best fit was determined by the Akaike information criterion.

RESULTS

Case Description

A 76-year-old male who had undergone a left total knee arthroplasty 5 years previously presented with sudden onset of pain and swelling in that knee. Aspiration of synovial fluid showed 34 000 white blood cells with 84% polymorphonuclear leukocytes. The patient was admitted to the hospital and treated with intravenous vancomycin (1 g every 12 hours) and intravenous levofloxacin (750 mg every 24 hours). Cultures of blood and synovial fluid specimens grew S. aureus. Because susceptibility testing by Vitek AST-GP67 cards indicated oxacillin susceptibility and results of a cefoxitin screening test were negative, the isolate was classified as MSSA. Therapy was narrowed to vancomycin because of a reported allergy to penicillin. Five days after admission, the joint prosthesis was replaced with a tobramycin-impregnated spacer. Intraoperative cultures and repeat blood cultures on hospital days 3, 4, and 5 again identified MSSA. Neither transthoracic nor transesophageal echocardiography found evidence of endocarditis. On hospital day 7, therapy was changed to intravenous daptomycin (700 mg every 24 hours), but blood cultures on day 10 remained positive for MSSA. Computed tomography of the chest, abdomen, and pelvis revealed only left lower lobe pneumonia. Blood cultures on hospital day 12 identified MSSA, and antibiotic therapy was changed to intravenous linezolid (600 mg every 12 hours). On hospital day 13, desensitization to nafcillin was performed followed by treatment with intravenous nafcillin (2 g every 4 hours). Results of blood cultures on hospital day 14 were negative. The patient developed lower back pain, and magnetic resonance imaging of the lumbar spine revealed vertebral osteomyelitis at the level of T12–L1 vertebrae. The patient continued to receive intravenous nafcillin and completed 10 weeks of total antibiotic therapy, remaining afebrile and with resolution of back pain. Because of immobility, a urinary catheter was maintained, and a screening urine culture grew Klebsiella pneumoniae.

One week after discontinuing nafcillin treatment, results of repeat blood cultured were negative. A week later, the patient developed fever and sudden onset of severe back pain. Admission blood cultures were sterile, but urine culture grew Klebsiella pneumoniae. Intravenous vancomycin therapy was reinstituted. A biopsy specimen obtained from the T12–L1 disc interspace on hospital day 6 contained gram-positive cocci in clusters, and culture of the specimen grew S. aureus resistant to oxacillin. The patient was treated with a regimen of intravenous vancomycin, intravenous nafcillin, and oral rifampin for 10 weeks. His back pain resolved, and levels of inflammatory markers normalized. He has had no evidence of recurrence of osteomyelitis in the subsequent 4.5 years. A time line summarizing the significant clinical events is presented in Figure 1.

Figure 1.

Case time line. The time scale of the main bar is in weeks, whereas that for the 2 insets is days. Boxes below the time scale indicate antibiotic therapy type, duration, and route of administration. Boxes above the time scale identify times and sites of sample acquisition and identity of organisms recovered. The double box denotes detection of a methicillin-susceptible S. aureus (MRSA) isolate. Abbreviation: MSSA, methicillin-susceptible Staphylococcus aureus.

Reversion to Methicillin Resistance by Precise Excision of a Transposable Element

Microdilution trays for full-range minimum inhibitory concentration (MIC) testing, performed according to CLSI standards, yielded an oxacillin MIC of <0.25 µg mL⁻¹ for the blood isolate from first admission (B1) and >32 µg mL⁻¹ for the vertebral disc space isolate from readmission (DS1). A MRSA CHROMagar screening plate also confirmed susceptibility of B1. MSSA isolate B1 and MRSA isolate DS1 appeared indistinguishable by pulsed-field gel electrophoresis, and comparison with the fingerprint database maintained by the Massachusetts Department of Public Heath revealed 83% similarity with clonal group USA100 (Supplementary Figure 1). Multiplex PCR demonstrated that all isolates contained the expected markers for S. aureus, contained mecA, were distinguishable from the common MRSA USA300 clonal group, and were negative for Panton–Valentine leukocidin (Supplementary Figure 2).

As the above results indicated that the MRSA and MSSA isolates were clonal, we next asked whether the MSSA isolate could give rise to resistant clones. Three independent MSSA isolates from the patient (B1, B2, and SF1) were streaked heavily on MHA containing 4% NaCl and 6 µg mL⁻¹ oxacillin. In all cases, numerous colonies grew in the heavy zone of the streak pattern. These colonies grew well when restreaked on the same medium (Figure 2A). MICs of these resistant revertants were >64 µg mL⁻¹.

Figure 2.

Methicillin susceptibility in Staphylococcus aureus isolate B1 reverts to methicillin resistance by precise excision of a transposable element. A, Streak patterns of B1 on nonselective Mueller-Hinton agar (MHA; left) and MHA containing 6 µg mL⁻¹ oxacillin (center). Two colonies from the center plate were restreaked on oxacillin (right). B, Results of polymerase chain reaction (PCR) analysis of full-length mecA, illustrating the 1.5-kb size difference between susceptible and resistant isolates. Lanes 1–3, Replicate colonies from strain B1. Lanes 4–7, Resistant revertants of B1 obtained from selection on oxacillin as described in panel A. Lane 8, S. aureus isolate DS1. Similar results were obtained with S. aureus isolates B2 and SF1. C, Structure of mecA::IS1181, determined by sequencing. Pointed ends of genes indicate the direction of transcription. All DNA added by the insertion is shaded. The IS1181 transposase is flanked by 23–base pair imperfect inverted repeats (black hashed rectangles) that facilitate transposase binding and insertion at a target sequence. An 8–base pair duplication of the transposase target sequence in the mecA gene (ATATTAAC) is created when IS1181 is inserted. One copy of this direct repeat is removed when the transposable element undergoes precise excision, restoring the wild-type mecA sequence. See Supplementary Figure 3 for additional details. Abbreviation: PBP2a, penicillin-binding protein 2a.

The original MSSA isolates were positive for mecA in multiplex PCR with primers producing a short product (147 base pairs). To determine whether the full-length mecA gene was present, PCR primers homologous to the 5′ and 3′ ends of mecA were used to amplify the mecA coding region. This yielded a product of the expected size (2 kb) in all resistant revertants but a 3.5-kb product in all MSSA parents (Figure 2B). Sequencing revealed that mecA in the MSSA isolates was interrupted by insertion of the transposable element IS1181 after codon 545 of 669 and that the revertants had precisely excised this transposable element from mecA (Figure 2C). Western blots confirmed that this insertion prevents expression of full-length PBP2a (Figure 3).

Figure 3.

Revertants express penicillin-binding protein 2a (PBP2a) in an oxacillin-inducible manner. Western blots with PBP2a-specific monoclonal antibody were used to examine the products of mecA in the susceptible Staphylococcus aureus strain, B1; the resistant revertant, B1R; and the vertebral disk space methicillin-resistant S. aureus isolate, DS1. Because the cassette containing the mecA gene was determined to be a staphylococcal cassette chromosome mec (SCCmec) type II variant (Supplementary Figure 4) and expression of mecA in SCCmec type II cassettes is often induced by β-lactams, we treated cultures with the specified concentrations of oxacillin for 1 hour. Lane 1, Extract from B1 exposed to 0.25 µg mL⁻¹ oxacillin for 1 hour shows a faint band of the size predicted by the mecA::IS1181 sequence (solid circles; 63 kDa; Figure 1C) and several degradation products (open circles). No detectable PBP2a was produced by the B1 strain in the absence of induction by oxacillin (Supplementary Figure 5). Lanes 2 and 3, Full-length PBP2a produced by the B1R resistant revertant and DS1 isolate treated similarly.

We noticed that the frequency of revertants among independent cultures of B1 was remarkably constant, leading us to suspect that reversion was induced by exposure to oxacillin. To test this hypothesis, we performed fluctuation analysis, using 72 independent cultures of B1, to calculate the frequency of reversion to oxacillin resistance. The fluctuation test, first described by Luria and Delbrück [23], distinguishes between random and induced mutations. Cultures grown in nonselective liquid medium are plated on selective agar, and resistant colonies are counted. Resistant mutants that arise early during growth in a nonselective liquid culture will have many descendants and will therefore produce many colonies, while those that occur late will have fewer. Under these circumstances of random mutation in nonselective culture, the number of resistant colonies obtained follows the Luria–Delbrück distribution, which has a large standard deviation. However, if mutations occur primarily on the plate during exposure to the antibiotic, seeding replicate plates with a given number of bacteria should yield similar numbers of resistant colonies. Therefore, postplating mutation events result in a smaller standard deviation than that predicted by the Luria–Delbrück distribution [23, 28]. Induction of IS1181 excision by oxacillin in the plating medium would have this effect. Surprisingly, we found that 92% of reversion events occur after exposure to oxacillin, indicating that oxacillin significantly increases the rate of precise excision of IS1181 from mecA (Figure 4A). The frequency of reversion was 4 × 10⁻⁷. In comparison, when fluctuation analysis was performed to detect mutations conferring rifampicin resistance for the hypermutable E. coli ΔmutS, mutations were found to be entirely random, as expected (Figure 4B and Supplementary Note 1).

Figure 4.

Reversion to resistance is induced by oxacillin. A, Cumulative distribution of colony counts obtained by plating 72 independent cultures of Staphylococcus aureus strain B1 on Mueller-Hinton agar plates containing 12 µg mL⁻¹ oxacillin (solid circles). The dashed curve shows the Ma-Sandri-Sarkar maximum likelihood fit of the Luria–Delbrück distribution to these counts, which allows only for random mutation and fits the data poorly. To account for mutations occurring during selection on plates, we fitted a joint distribution model in which the additional parameter represents the mean number of induced mutations per plate [26, 27]. The solid curve shows the maximum likelihood fit of the joint model. B, The best fit (as determined by the Akaike information criterion) of the data for B1 estimated that 2 of 30 reversion events (8%) occurred before plating (random) and that 28 of 30 of reversion events (92%) occurred after plating in the presence of selection (induced; left). Mutations in a variety of genes related to precise excision, mecA transcription, and SOS had no effect on induced reversion to oxacillin resistance. Unlike reversion to oxacillin resistance, mutation to rifampicin resistance in the B1 isolate was a primarily random process (right). Postplating mutation to rifampicin resistance was mediated by the SOS response, as shown by the decrease in induced mutation events when SOS was constitutively repressed by the expression of noncleavable LexA^G94E [27]. In the hypermutable E. coliΔmutS strain, all mutations to rifampicin resistance occurred randomly.

Because antibiotic-mediated induction of precise excision has not been previously reported, we investigated the mechanism of induction. Although read-through transcription of the IS1181 transposase from the mecA promoter is induced by oxacillin (Supplementary Figure 6), deletion of the transposase from mecA::IS1181 had no effect on induced precise excision (Figure 4B). Consistent with this observation, virtually all studies of precise excision of classical transposable elements such as IS1181 have shown this process to be independent of transposase activity [29–31]. Previous reports have shown that precise excision depends on intrastrand hybridization of the inverted repeats, followed by slipped-strand mispairing during replication between the direct repeats formed by target site duplication or by resolution of the resulting cruciform structure via a pathway dependent on the exonuclease SbcD [32]. SbcD is upregulated during the SOS response, which is induced by some antibiotics, including β-lactams [33–36]. However, neither inactivation of SOS nor deletion of sbcD affected the rate of induced reversion to oxacillin resistance (Figure 4B), indicating that oxacillin induces precise excision of IS­1181 from mecA via an SOS-independent and SbcD-independent mechanism. In contrast, rifampicin-induced SOS responses in B1 increased mutation to rifampicin resistance, which requires a point mutation resulting in an amino acid substitution in rpoB.

Because transcription increases supercoiling and enhances cruciform formation [37], increased transcription over IS1181 during oxacillin exposure could promote precise excision and increase reversion rates. However, in a derivative of B1 constitutively transcribing mecA due to deletion of the regulatory genes mecI/mecRI and blaI/blaRI (Supplementary Figure 6), precise excision remained induced.

Taken together, these results have led us to reject the a priori hypotheses that appear most likely to explain the induction of precise excision. The mechanism underlying this phenomenon remains to be determined.

Reversion to Methicillin Resistance by Correction of a Frameshift Mutation

Our experience with strain B1 lead us to examine the properties of another clinical isolate for which, like strain B1, broth microdilution revealed high susceptibility to oxacillin (MIC, 0.5 µg mL⁻¹) and PCR detected mecA (Xpert MRSA/SA SSTI; Cepheid). The patient had a history of S. aureus infections associated with abscesses and line insertion sites and was treated at several hospitals, so tracing a definitive clinical history was difficult. At the time the MSSA isolate was collected, the patient was treated with vancomycin. Results of blood cultures performed the following month were negative.

The ability of this strain, J522BDU, to give rise to resistant clones was first assessed by streaking heavily on MHA containing 4% NaCl and 12 µg mL⁻¹ oxacillin. Like B1, numerous colonies grew in the heavy zone of the streak pattern. These colonies grew well when restreaked on the same medium. MICs of these resistant revertants were >64 µg mL⁻¹. A small amount of full-length PBP2a was present in J522BDU cultures, but the resistant revertant J522BDU-R produced 8-fold more PBP2a (Figure 5A).

Figure 5.

Reversion by Staphylococcus aureus strain J522BDU occurs randomly and increases expression of penicillin-binding protein 2a (PBP2a). A, Western blots with PBP2a-specific monoclonal antibody were used to examine the products of mecA in oxacillin-susceptible J522BDU and in J522BDU-R, a resistant revertant derived from J522BDU by streaking on Mueller-Hinton agar 4% NaCl containing 12 µg mL⁻¹ oxacillin. Because the cassette containing the mecA gene was determined to be staphylococcal cassette chromosome mec type II (Supplementary Figure 4), we treated cultures with the specified concentrations of oxacillin for 1 hour prior to preparation for electrophoresis as in Figure 3. Lane 1, Extract from J522BDU exposed to 0.25 µg mL⁻¹ oxacillin for 1 hour shows a faint band the size of full-length PBP2a. Lane 2, Full-length PBP2a produced by the J522BDU-R revertant treated similarly. The revertant J522BDU-R produced 8-fold more PBP2a than the susceptible parent, J522BDU. B, Fluctuation analysis performed as described for S. aureus strain B1 estimated that 4 of 14 reversion events (26%) occurred before plating (random) and that 10 of 14 reversion events (74%) occurred after plating in the presence of selection. This was similar to the pattern observed for B1 reversion to rifampicin, indicating that SOS responses after plating were likely responsible for these postplating reversion events.

Fluctuation analysis of J522BDU yielded a reversion frequency of 9 × 10⁻⁸, which was about 4-fold lower than that of B1. Unlike B1, reversion of J522BDU seems to be a random process that is not induced by oxacillin (Figure 5B).

To determine the mechanism of reversion, we sequenced the mecA coding region for the susceptible isolate J522BDU and 9 independent revertants selected by growth on MHA 4% NaCl with 12 µg mL⁻¹ oxacillin. As seen in Figure 6A, J522BDU has a single nucleotide deletion early in the coding region of mecA, resulting in an altered reading frame and premature termination at an early stop codon. The small amount of PBP2a protein still produced by this strain (Figure 5A) most likely results from misreading at the ribosome level, which is known to occur in stretches of repeated nucleotides.

Figure 6.

Staphylococcus aureus strain J522BDU reverts to oxacillin resistance by slipped-strand mispairing. Gray-shaded sequence indicates consensus with wild-type mecA and penicillin-binding protein 2a (PBP2a). Nucleotides inserted during reversion are bolded. A, Sequencing of oxacillin-susceptible J522BDU revealed a single adenine deletion from a stretch of 7 adenines at positions 255–261. This resulted in a frameshift and an early stop codon at positions 283–285 (boxed). Of the 9 independent revertants sequenced, 6 reverted to sequence I and 1 each reverted to sequences II, III, and IV. The frameshift and correcting insertions occurred in the N-terminal extension of PBP2a, a region that has no enzymatic activity and unknown function [38]. B, All revertants eliminated the early stop codon at residue 95 by correcting the reading frame.

We analyzed 9 independent revertants that all correct this frameshift mutation by inserting an additional nucleotide at or near the site of the deletion. Of these, 6 corrected the frameshift by extending the neighboring stretch of repeated nucleotides just downstream of the deletion (sequence I; Figure 6A). The other 3 revertants all corrected the frameshift by inserting a nucleotide in different positions nearby (sequences II, III, and IV; Figure 6A). All revertants restored the reading frame of mecA (Figure 6B). Because this stretch of repeated nucleotides is highly conserved in mecA genes, the insertion and deletion of single bases by slipped-strand mispairing could be a fairly common phenomenon to modulate PBP2a production in mecA-positive S. aureus.

DISCUSSION

Our clinical experience with the patient described above and our analysis of additional mecA-positive MSSA strains in which methicillin susceptibility reverts to methicillin resistance provide a framework for considering the implications of such strains for treatment strategies, antibiotic susceptibility testing, and reporting of resistance profiles. These strains have 3 properties that distinguish them as a class: frank sensitivity to oxacillin when assessed by the standard CLSI phenotypic methods used in clinical microbiology laboratories, presence of a mecA gene that is detectable by PCR but is genetically inactivated, and the generation of resistant revertants at modest frequencies (approximately 10⁻⁷). This reversion frequency falls within a troublesome window: it is low enough to escape detection by standard phenotypic methods because the bacterial inocula used in these assays are small (10⁵ colony-forming units) but high enough to allow significant risk for development of resistance, leading to treatment failure under monotherapy.

We observed 2 distinct mechanisms of reversion in strains with inactivated mecA: (1) precise excision of the transposable element IS1181 from within mecA and (2) slip-strand replication errors within mecA. Although the precise excision of IS1181 is a remarkable example of an induced mutation and gives rise to revertants at a higher rate, the slipped-strand mechanism is more concerning because it arises from sequence properties common to all mecA genes. In bacteria, cycles of positive and negative selection acting on expression of a particular gene often give rise to genetic systems that ensure that populations include some individuals in the currently disadvantaged state of expression, although these may be rare. In its simplest incarnation, this phenomenon, known as phase variation, is mediated by slipped-strand mutagenesis in low-entropy sequences characterized by stretches of repeated nucleotides [39]. This is essentially what we observe in strain J522BDU as it converts from mecA-positive MSSA to MRSA.

Although the case we describe here is the first rigorously documented instance in which a methicillin susceptible isolate reverted to high-level β-lactam resistance during antibiotic therapy, other reports suggest that similar events have occurred [8, 11]. Our analysis of the extant literature [5–16, 18, 19] estimates that the frequency of mecA-positive MSSA is approximately 3% of clinical S. aureus isolates (Supplementary Table 3), suggesting that such strains cause a large number of clinically significant infections annually in the United States [40, 41], which may currently be misidentified as MSSA/MRSA polymicrobial infections. The fitness cost associated with the presence of a functional mecA [20, 21] provides a selection pressure for mecA inactivation in the absence of antibiotic, predicting the presence of a population of mecA-positive MSSA. There is very little information regarding the fraction of mecA-positive MSSA that are capable of reversion to methicillin resistance. In vitro reversion has been observed previously [6, 10, 11, 13–15, 42], but the genetic mechanisms have not been investigated. The rates of reversion we observe are of the same order of magnitude as spontaneous resistance to rifampicin, which is often prescribed as part of combination therapy to prevent the emergence of resistant mutants.

Reporting mecA-positive MSSA as MRSA is likely to restrict treatment to β-lactam alternatives such as trimethoprim-sulfamethoxazole, doxycycline, or vancomycin, which are considered inferior [43, 44]. Laboratory methods that combine both genotypic and phenotypic testing to distinguish among MRSA, MSSA, and mecA-positive MSSA would provide clinicians with a more complete characterization of infecting strains. In the latter case, this would encourage the potentially more efficacious therapeutic strategy of including a β-lactam in combination with a second agent to control any small revertant population that may arise during therapy.

In addition to clinical implications, the existence of mecA-positive MSSA strains capable of reversion to methicillin resistance also has regulatory and legal ramifications. Multiple US states, the Veterans Affairs Healthcare system, and several nations have mandated screening and isolation programs for MRSA, although the efficacy of these programs is unclear [45–47]. If mecA-positive MSSA isolates ultimately fall under the legal definition of MRSA, truly comprehensive screening programs will need to incorporate genotypic testing, as phenotypic testing alone is insufficient to identify mecA-positive MSSA.

Our findings illustrate a generic principle inherent in the use of strictly genotypic information to predict resistance phenotypes: the presence of a gene does not guarantee that it is expressed, and withholding highly effective antibiotics solely on the basis of genotypic information may be harmful, especially when alternatives are limited. This potential for harm warrants explicit and careful consideration during development and implementation of molecular diagnostic assays.

Supplementary Data

Supplementary materials are available at http://jid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.