Broader Distribution of Plasmid-Mediated Quinolone Resistance in the United States

Abstract

The plasmid-encoded quinolone resistance gene qnrA confers low-level quinolone resistance, facilitating selection of higher-level resistance. Epidemiologic surveys for qnrA were extended to isolates of Enterobacter spp. and to quinolone-susceptible Enterobacteriaceae. Two (10%) of 20 ceftazidime-resistant quinolone-susceptible Klebsiella pneumoniae strains carried the gene, as did 12 (17%) of 71 ceftazidime-resistant Enterobacter strains from across the United States. One of these Enterobacter isolates was quinolone susceptible. Thus, qnrA is present in quinolone-resistant and quinolone-susceptible Enterobacter and Klebsiella strains in the United States.

Until 1998, it was believed that quinolone resistance could be acquired only through chromosomal mutation. Since then, it has become clear that the plasmid-borne gene qnrA can generate in Enterobacteriaceae an 8- to 32-fold increase in MICs of quinolones (8) by protecting DNA gyrase directly from quinolone inhibition (9). The originally described qnr gene is now referred to as qnrA, because of recent findings of other related qnr genes (1; G. A. Jacoby, K. Walsh, D. Mills, V. Walker, A. Robicsek, H. Oh, and D. C. Hooper, Abstr. 44th Intersci. Conf. Antimicrob. Agents Chemother., abstr. C2-1898a, 2004). This 8- to 32-fold elevation of MIC is insufficient, by itself, to convert most wild-type Enterobacteriaceae (baseline ciprofloxacin MICs are approximately 0.016 μg/ml) from susceptible to resistant by CLSI (formerly NCCLS) criteria (MIC ≥ 4 μg/ml) (2). However, the presence of qnrA facilitates the selection of chromosomal mutants in the presence of a quinolone (8). The gene does not appear to induce hypermutability. The higher baseline MIC likely increases the frequency of mutant selection by allowing chromosomal mutants with small, incremental MIC elevations to survive. The spread of a plasmid bearing this gene could thus potentiate the rapidity of development of higher-level fluoroquinolone resistance in organisms currently classified as susceptible. No previous survey has evaluated fluoroquinolone-susceptible clinical isolates of Enterobacteriaceae for the presence of qnrA.

Previous surveys have identified qnrA in 8% of quinolone-resistant clinical isolates of Escherichia coli in Shanghai (11) and in 1 of 297 such isolates in Paris (6). A recent study of ceftazidime-resistant, quinolone-resistant Klebsiella pneumoniae isolates from the United States detected qnrA in 11% of strains (10). qnrA has also recently been reported to be in Citrobacter freundii and Enterobacter spp. (4). A systematic examination of the gene's prevalence in genera other than Escherichia and Klebsiella, however, has not been reported.

In this study, we investigated the frequency of qnrA in recent isolates of fluoroquinolone-susceptible Klebsiella pneumoniae and in isolates of fluoroquinolone-resistant and fluoroquinolone-susceptible Enterobacter cloacae and Proteus spp. collected in the United States. Seventy-one clinical isolates of E. cloacae, 20 clinical isolates of K. pneumoniae, and 6 isolates of Proteus spp. from surveys conducted by Focus Technologies between 1999 and 2001 in the United States were tested. This survey gathers nonrepeat clinical isolates collected between January and March of each study year from participating clinical microbiology laboratories in all continental United States census regions. For these 3 years, the total numbers of isolates of Klebsiella pneumoniae, Enterobacter cloacae, and Proteus spp. collected were 1,616, 928, and 1,210, respectively. Because qnrA has consistently tracked with either extended-spectrum beta-lactamases or plasmid-encoded AmpC beta-lactamases, ceftazidime-resistant (MIC ≥ 16 μg/ml) strains were chosen for screening. Ceftazidime-resistant isolates represented 27% of the Enterobacter isolates, 6% of the Klebsiella isolates, and 2% of the Proteus sp. isolates collected during the same time period. The isolates of Enterobacter and Proteus spp. chosen had MICs of ciprofloxacin from 0.25 to >8 μg/ml, and the K. pneumoniae isolates had MICs from 0.25 to 1.0 μg/ml; 0.25 μg/ml is the lowest MIC found to be associated with the presence of qnrA (8, 11). The Enterobacter and Proteus sp. isolates represented 100% of those in the Focus Technologies collection that met the MIC selection criteria; among Klebsiella isolates, 2 out of 22 isolates meeting these criteria were unavailable for study. Additional strains used were E. coli strains V517 (5) and J53 containing plasmid Plac (3) as standards for plasmid size and E. coli J53AzR (resistant to azide) as a recipient for conjugation (8).

Screening was carried out by PCR amplification of qnr. Colonies were transferred to water in an Eppendorf tube and boiled to prepare DNA templates for PCR. qnrA was amplified with primers 5′-TCAGCAAGAGGATTTCTCA and 5′-GGCAGCACTATTACTCCCA to produce a 627-bp product. PCR conditions were 94°C for 45 s, 48°C for 45 s, and 72°C for 45 s for 30 cycles. Strains positive and negative for qnrA were included in each group of tested strains as controls. Products of amplification were detected by electrophoresis on a 1% agarose gel with ethidium bromide staining and photographed under UV light. All positives were confirmed by direct sequencing of the PCR products.

Conjugation experiments were carried out in LB broth with E. coli J53AzR as the recipient. Transconjugants were selected on Trypticase soy agar plates containing sodium azide (100 μg/ml) for counterselection and gentamicin (10 μg/ml) or ampicillin (100 μg/ml) to select for plasmid-encoded resistance. Colonies were then individually tested for quinolone resistance. From transconjugant plasmids, the sequence of approximately 1 kb of DNA adjacent to qnrA was determined with a set of outward-facing primers starting from both sides of the qnrA gene.

For the comparison of mutant selection frequencies in the presence and absence of qnrA, a qnrA-positive clinical Enterobacter strain (MIC, 0.5 mg ciprofloxacin per ml) was grown overnight in Luria-Bertani medium at 42°C to generate a variant that had spontaneously lost qnrA. One clone that had lost quinolone resistance but maintained other resistance phenotypes encoded on its plasmid was detected. Plasmid preparation revealed the persistent presence of the plasmid; Southern hybridization confirmed the loss of the qnrA gene from this plasmid (data not shown). The Enterobacter strains with and without qnrA were plated after 16 to 20 h of growth in LB broth at 37°C on LB agar containing 2 μg ciprofloxacin per ml (fourfold the MIC of the qnrA-containing clone) to select for resistant mutants. The MIC of ciprofloxacin was tested for several colonies selected in this way; all had eightfold increases in MIC, from 0.5 μg/ml to 4.0 μg/ml.

The qnrA gene was detected in 2 (10%) of the 20 fluoroquinolone-susceptible K. pneumoniae strains, both of which were from New York state (Table 1). qnrA was also present in 12 (17%) of 71 Enterobacter strains. In this group, 11 (24%) of 45 fluoroquinolone-resistant strains and 1 (4%) of 26 fluoroquinolone-susceptible strains carried qnrA. These 12 isolates came from five states (Table 1). Five isolates had sequences identical to the Alabama qnrA sequence, and seven had sequences identical to the Shanghai qnrA sequence (CTG instead of CTA at position 537, a silent polymorphism [11].) There was no demonstrable geographical clustering of either variant. None of the six Proteus sp. isolates was positive for qnrA.

TABLE 1.

Characteristics of qnrA-positive strains^a

Organism	Strain no.	Source	State	MIC (μg/ml)
				CIP	CTZ	CRO	IMI	GEN	SXT
E. cloacae	8	Outpatient	CA	4	>16	32	0.5	>8	>32
	17	Inpatient	AL	>4	>16	>32	0.12	>8	>32
	23	Inpatient	AL	>4	>16	>32	0.12	>8	>32
	28	Inpatient	WV	>4	>16	>32	0.5	8	>32
	33	Inpatient	WV	>4	>16	>32	0.25	8	>32
	55	Inpatient	AZ	4	>16	16	1	>8	>2
	59	Inpatient	CA	4	>16	16	0.5	>8	>2
	60	Inpatient	AZ	4	>16	32	0.5	>8	>2
	61	Inpatient	NY	0.5	>16	32	0.25	8	>2
	63	Outpatient	AZ	4	>16	>32	0.5	>8	>2
	64	Unknown	NY	>4	>16	16	1	>8	>32
	67	Inpatient	CA	8	>16	16	0.5	>8	>2
K. pneumoniae	39	Inpatient	NY	1	>16	8	0.12	8	>32
	57	Outpatient	NY	0.5	>16	8	0.12	8	>32

^aCIP, ciprofloxacin; CTZ, ceftazidime; CRO; ceftriaxone; IMI, imipenem; GEN, gentamicin; SXT, trimethoprim-sulfamethoxazole.

Quinolone resistance could be transferred by conjugation from each of four Enterobacter isolates selected for this purpose. All four transconjugants carried single plasmids, all greater than 150 kb in size, with identical resistance profiles (Table 2). The antibiogram appears the same as that for pMG252, the plasmid in which qnrA was first identified (10). DNA adjacent to qnrA was sequenced for two of these plasmids. In both, sequences identical to that surrounding qnrA in the original Alabama plasmid, pMG252, were found, suggesting that qnrA from the E. cloacae isolates was also contained within a class 1 integron in which orf513 lies upstream and repeats of the 3′ conserved sequence are found upstream and downstream of the gene (Fig. 1). These flanking sequences differ from those of the characterized In36 and In37 integrons found in Shanghai strains containing qnrA in that they contained an additional 76-bp noncoding element upstream of the gene and lacked ampR immediately downstream.

TABLE 2.

Resistance profile of transconjugants and recipient E. coli J53^a

Strain (state of origin)	MIC (μg/ml)
	CIP	CTX	FOX	CHL	SMZ	TET	GEN
E. cloacae
17 (Alabama)	0.5	8	≥32	32	≥512	1	2
55 (Arizona)	0.5	4	≥32	64	≥512	1	2
60 (Arizona)	0.5	8	≥32	64	≥512	1	2
61 (New York)	0.5	8	≥32	32	≥512	1	2
E. coli J53	0.008	0.125	4	≤16	32	1	0.5

^aCIP, ciprofloxacin; CTX, cefotaxime; FOX, cefoxitin; CHL, chloramphenicol; SMZ, sulfamethoxazole; TET, tetracycline; GEN, gentamicin.

FIG. 1.

Comparison of the regions flanking qnrA in pHSH2 from Shanghai E. coli isolates (a) with those from pMG252 and plasmids from E. cloacae isolates (b) in this study. The regions differ in that plasmids in panel b include a 76-bp noncoding element upstream of qnrA and lack ampR relative to pHSH2. qacEΔ1, a deletion mutant of the gacE gene, which encodes the QacE efflux pump.

The mean frequency of mutant selection in a clinical E. cloacae isolate with its native qnrA-containing plasmid was 3.4 × 10⁻⁷ (range, 2 × 10⁻⁷ to 5.7 × 10⁻⁷ in three independent trials). In contrast, in the same strain containing a variant of the same plasmid that had spontaneously lost qnrA, the mutant selection frequency was <2.9 × 10⁻⁹ in all three independent trials, a difference of over 100-fold. Thus, the presence of qnrA promotes selection of higher levels of quinolone resistance in clinical Enterobacter isolates, as it does in E. coli isolates (8) and likely in other enteric bacteria as well. This clinical isolate, with its relatively high baseline MIC (likely owing to underlying chromosomal mutations), yielded mutants that were resistant by CLSI breakpoints, supporting other in vitro data showing that qnrA and chromosomal mutations display an additive effect (7).

In this study, qnrA was found in 24% of ceftazidime-resistant, ciprofloxacin-resistant isolates of Enterobacter in the United States, a prevalence higher than that in any bacterial survey population reported to date. These strains came from widely disparate locations. The qnrA-bearing plasmids of geographically diverse Enterobacter strains showed strong similarities to pMG252, the large integron-carrying plasmid in which qnrA was originally identified. It is possible that a single plasmid or a family of related plasmids bearing multiple resistance markers, including qnrA, is being selected for extensively in diverse medical settings.

In addition, qnrA was detected in fluoroquinolone-susceptible Enterobacteriaceae for the first time. We showed that the presence of qnrA in one such clinical Enterobacter strain substantially facilitated the selection of fluoroquinolone-resistant mutants at a concentration of ciprofloxacin (2 μg/ml) normally achieved during a therapeutic course with this antimicrobial agent. Bacterial strains in which qnrA is the sole quinolone resistance determinant are concerning in that they are classified as susceptible but are capable of an easy transition to full resistance.

qnrA is prevalent in cephalosporin-resistant Enterobacter isolates from the United States and is also present in Enterobacter and Klebsiella isolates that are fluoroquinolone-susceptible. Such organisms may readily develop resistance when exposed to quinolones. More needs to be understood about the prevalence of qnrA in fluoroquinolone-susceptible bacteria and about the influence of this gene on the rise of quinolone resistance in Enterobacteriaceae.