Genetic Analyses of Mutations Contributing to Fluoroquinolone Resistance in Clinical Isolates of Streptococcus pneumoniae

Abstract

Twenty-one clinical isolates of Streptococcus pneumoniae showing reduced susceptibility or resistance to fluoroquinolones were characterized by serotype, antimicrobial susceptibility, and genetic analyses of the quinolone resistance-determining regions (QRDRs) of gyrA, gyrB, parC, and parE. Five strains were resistant to three or more classes of antimicrobial agents. In susceptibility profiles for gatifloxacin, gemifloxacin, levofloxacin, moxifloxacin, ofloxacin, sparfloxacin, and trovafloxacin, 14 isolates had intermediate- or high-level resistance to all fluoroquinolones tested except gemifloxacin (no breakpoints assigned). Fluoroquinolone resistance was not associated with serotype or with resistance to other antimicrobial agents. Mutations in the QRDRs of these isolates were more heterogeneous than those previously reported for mutants selected in vitro. Eight isolates had amino acid changes at sites other than ParC/S79 and GyrA/S81; several strains contained mutations in gyrB, parE, or both loci. Contributions to fluoroquinolone resistance by individual amino acid changes, including GyrB/E474K, ParE/E474K, and ParC/A63T, were confirmed by genetic transformation of S. pneumoniae R6. Mutations in gyrB were important for resistance to gatifloxacin but not moxifloxacin, and mutation of gyrA was associated with resistance to moxifloxacin but not gatifloxacin, suggesting differences in the drug-target interactions of the two 8-methoxyquinolones. The positions of amino acid changes within the four genes affected resistance more than did the total number of QRDR mutations. However, the effect of a specific mutation varied significantly depending on the agent tested. These data suggest that the heterogeneity of mutations will likely increase as pneumococci are exposed to novel fluoroquinolone structures, complicating the prediction of cross-resistance within this class of antimicrobial agents.

Streptococcus pneumoniae is the leading cause of community-acquired pneumonia in the United States and a major cause of meningitis and otitis media (44). For decades, the treatment of pneumococcal infections relied primarily on the use of β-lactam and macrolide agents. However, with the emergence and spread of penicillin- and multidrug-resistant (MDR) strains (44, 46) the development of new antimicrobial agents has become essential in providing effective treatment options.

Fluoroquinolones are broad-spectrum antimicrobial agents introduced in the 1980s. The early fluoroquinolones, such as ciprofloxacin and ofloxacin, had limited activity against S. pneumoniae (48). Newer fluoroquinolones, with enhanced activity against pneumococci and other gram-positive organisms, have been developed and will provide useful alternative therapies for pneumococcal infections as long as the development of resistance can be minimized and controlled. However, with increased use of fluoroquinolone therapy, resistant pneumococci have emerged and treatment failures have been documented (5, 31, 40, 43, 49). Although current data indicate that the incidence of resistant strains is low, recent reports have noted an increase in the prevalence of fluoroquinolone-resistant pneumococci (8, 21; C. García-Rey, L. Aguilar, F. Baquero, and the Spanish Surveillance Group for Respiratory Pathogens, Letter, Antimicrob. Agents Chemother. 44:3481–3482, 2001).

Resistance to fluoroquinolones is acquired in stepwise fashion with the introduction of chromosomal mutations that alter the target proteins, DNA gyrase and topoisomerase IV, or decrease intracellular drug accumulation by active drug efflux. DNA gyrase, an A₂B₂ tetramer encoded by gyrA and gyrB, introduces negative supercoils in DNA to relieve the topological stress generated during DNA replication and transcription (45, 51). Topoisomerase IV, a C₂Ε₂ tetramer encoded by parC and parE, is important in decatenation and partitioning of daughter chromosomes following DNA replication (1, 30). These activities require the introduction and repair of double-strand breaks in the chromosomal DNA by both enzymes. The primary target of a fluoroquinolone, either gyrase or topoisomerase IV, depends on the chemical and structural properties of the agent.

High-level resistance is associated with point mutations in discrete regions of the gyrase and topoisomerase IV genes, designated as the quinolone resistance-determining regions (QRDRs) (50). Early studies of these mutations were based on the analysis of mutants selected in vitro on subinhibitory concentrations of ciprofloxacin (25, 33, 36). The resulting mutations occurred most frequently in the codons for ParC/S79 and GyrA/S81. Only recently, however, have studies included analysis of gyrB and parE (10, 20, 28, 34, 39), although mutations in these loci have been described as rare and insignificant (26, 27, 47).

In this study, recent clinical isolates of fluoroquinolone-resistant pneumococci from the United States were analyzed to determine the molecular basis of resistance, the activity of newer fluoroquinolones against the mutant strains, and the patterns of cross-resistance among fluoroquinolones. Twenty-one previously undescribed isolates, with reduced susceptibility or resistance to fluoroquinolones, were characterized by serotype, susceptibility profiles, and genetic analyses of the QRDRs of gyrA, gyrB, parC, and parE. Contributions to fluoroquinolone resistance by individual mutations in gyrB and parE of selected isolates and a novel mutation in parC were investigated by sequential genetic transformation of the susceptible S. pneumoniae strain R6.

(This study was presented in part at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, Calif., September 1999.)

MATERIALS AND METHODS

Bacterial strains and antimicrobial susceptibility testing.

Pneumococcal strains included in this study were isolated from 1997 to 1999 by hospital laboratories located in 11 states in the United States. Isolates with resistance (MIC of ofloxacin or levofloxacin, ≥8 μg/ml) or reduced susceptibility to fluoroquinolones (MIC of ofloxacin, 4 μg/ml; MIC of levofloxacin, 2 μg/ml) were sent to the Centers for Disease Control and Prevention (CDC) for confirmation of resistance phenotype and genetic analysis. S. pneumoniae R6, a fluoroquinolone-susceptible, nonencapsulated strain derived from Avery's strain R36A (2), was used as the recipient for genetic transformations. Isolates were confirmed as pneumococci on the basis of their susceptibility to ethylhydrocupreine (optochin) and bile solubility. Serotyping was performed at the CDC by Neufeld's quellung reaction using type-specific antisera prepared at the CDC as described by Facklam and Washington (11).

MICs of selected antimicrobial agents were determined by the broth microdilution method using cation-adjusted Mueller-Hinton broth containing 5% lysed horse blood according to the guidelines recommended by the NCCLS (35). S. pneumoniae ATCC 49619 was used for quality control. Reagent powders of antimicrobial agents were obtained from manufacturers as follows: gatifloxacin, Bristol-Meyers Squibb (Wallingford, Conn.); gemifloxacin, SmithKline Beecham (Philadelphia, Pa.); levofloxacin and ofloxacin, Ortho-McNeil Pharmaceutical (Raritan, N.J.); sparfloxacin, Rhône-Poulenc Rorer (Collegeville, Pa.); trovafloxacin, Pfizer (New York, N.Y.); moxifloxacin, Bayer Corp. (West Haven, Conn.); and penicillin, erythromycin, tetracycline, chloramphenicol, ceftriaxone, sulfamethoxazole, and trimethroprim, from Sigma Chemical Co. (St. Louis, Mo.).

Preparation of chromosomal DNA.

S. pneumoniae cells were grown to exponential phase in Todd-Hewitt broth (Difco Laboratories, Detroit, Mich.) supplemented with 0.5% yeast extract (Difco) and harvested by centrifugation. Each cell pellet was washed and resuspended in 50 mM Tris-HCl (pH 8.0)–10 mM EDTA, and lysis was induced by incubation with 1 mg of lysozyme per ml (37°C for 15 min) and 0.5% deoxycholate (25°C for 30 min). Cell lysates were incubated with 0.1 mg of RNase per ml (37°C for 15 min) and with proteinase K and buffer AL (56°C for 30 min) and applied to the QIAamp column (QIAamp Tissue Kit, Qiagen, Chatsworth, Calif.). Chromosomal DNA was eluted according to the manufacturer's protocol.

PCR and DNA sequence analysis.

Oligonucleotide primers PNC6 and PNC7, or PNC10 and PNC11 (25), were used to amplify a 232- or 329-bp gene fragment (excluding primers) of gyrA and parC, respectively, from chromosomal DNA of each of the 21 clinical isolates and strains R6 and ATCC 49619. A 321-bp gene fragment of parE (excluding primers) was amplified with oligonucleotide primers SPPARE7 and SPPARE8 as described by Perichon et al. (41). Oligonucleotides H4025 and H4026, described by Pan et al. (36), were used to amplify a 422-bp gene fragment of gyrB. An additional set of oligonucleotide primers, described by Morrissey and George (32), was utilized to amplify a larger (608-bp) parC gene fragment for genetic transformation studies.

Amplification products were purified with the QIAquick PCR purification kit (Qiagen). DNA sequences were determined with ABI Prism dRhodomine terminator cycle sequencing (PE Biosystems, Foster City, Calif.) and the ABI 377 automated sequencer (PE Biosystems). DNA sequences were confirmed by using products of independent PCRs to determine the sequence of each strand. DNA and inferred amino acid sequences were analyzed with DNASIS 2.6 sequence analysis programs (Hitachi Software Engineering Co., Ltd., San Francisco, Calif.).

Genetic transformations.

The contribution of selected mutations to fluoroquinolone resistance was determined by sequential transformation of S. pneumoniae R6 with the relevant amplified QRDR gene fragments. To induce competence, recipient cells (R6 or R6 transformants) were grown in competence medium (17) to an optical density at 650 nm of 0.18 followed by incubation with 1 μg of competence-stimulating peptide per ml (18) for 15 min at 37°C. Competent cells were transformed with either a purified PCR-generated gene fragment (10 μg/ml) or chromosomal DNA (1 μg/ml) and incubated at 30°C for 30 min followed by 37°C for 2 to 4 h for phenotypic expression of resistance. Transformants were selected on Mueller-Hinton agar supplemented with 5% defibrinated sheep blood and the appropriate concentration of levofloxacin or sparfloxacin. Competence-stimulating peptide was a generous gift from Don Morrison (University of Illinois, Chicago).

Active efflux.

The role of active drug efflux was determined for selected strains by the method of Beyer et al. (6) with the following modification: cells were incubated for 5 min at 37°C in medium containing 10 μg of reserpine per ml before the addition of fluoroquinolone.

RESULTS

Characterization of pneumococcal isolates.

Twenty-one clinical isolates with resistance or reduced susceptibility to fluoroquinolones were included in this study. The serotype, state of origin, and antimicrobial resistance profile of each isolate are shown in Table 1. Serotypes included 6A (four strains), 6B (one strain), 11A (one strain), 15B (one strain), 19A (two strains), 19F (one strain), 20 (one strain), 22F (two strains), 23B (one strain), 23F (four strains), 31 (one strain), 35B (one strain), and one nontypeable strain. Based on breakpoints established by the NCCLS (35), five strains were multidrug resistant (MDR), i.e., resistant to three or more classes of antimicrobial agents, and two of the MDR strains (4566 and 4567) were either intermediate or resistant to all classes of agents represented in the initial antibiogram except vancomycin (data not shown). MDR was associated with serotype 23F and was also noted for one nontypeable strain. Resistance to fluoroquinolones was not associated with serotype or geographical region. Three of the 16 fluoroquinolone-resistant strains (ofloxacin MIC, ≥8 μg/ml) were also penicillin resistant (penicillin MIC, ≥2 μg/ml), and 10 fluoroquinolone-resistant strains were not resistant to any class of agents tested except fluoroquinolones.

TABLE 1.

Characteristics of S. pneumoniae clinical isolates analyzed in this study

Strain	Serotype	Stateb	MIC (μg/ml)a
			OFX	PEN	ERY	TET	SXT	CHL	CRO
R6	-c	-	2	≤0.03	0.12	0.5	≤0.06	4	0.06
4563	6A	NC	4	1	8	≤0.25	4	4	0.25
4566	23F	CA	4	2	>8	8	4	8	1
4562	6A	KY	4	1	4	≤0.25	4	2	0.25
4567	23F	TN	4	4	>8	16	4	16	8
4570	11A	WV	4	2	≤0.06	≤0.25	4	2	1
4571	NTd	GA	8	4	0.12	≤0.25	4	4	4
4554	6A	WA	16	≤0.03	≤0.06	≤0.25	8	2	≤0.03
4555	6A	AL	16	0.06	≤0.06	≤0.25	0.12	4	0.06
4556	20	KY	16	0.06	≤0.06	≤0.25	0.25	4	0.06
4569	23F	GA	32	2	≤0.06	2	4	16	2
4557	35B	CA	32	≤0.03	0.12	≤0.25	0.25	4	≤0.03
4553	6B	WV	32	0.06	0.12	≤0.25	0.25	4	0.25
4561	19A	NY	32	0.12	≤0.06	≤0.25	0.12	2	0.12
4560	19A	NY	32	0.5	≤0.06	≤0.25	0.25	2	0.12
4558	31	TX	32	≤0.03	0.12	≤0.25	0.25	4	≤0.03
4552	23B	WA	32	≤0.03	≤0.06	≤0.25	0.12	2	0.12
4568	23F	PA	32	2	1	2	4	8	2
6274	15B	AZ	32	≤0.03	0.25	≤0.25	0.25	2	≤0.03
4559	19F	AL	64	0.5	≤0.06	≤0.25	4	4	0.5
6275	22F	AZ	64	≤0.03	0.12	≤0.25	0.12	4	≤0.03
6276	22F	AZ	64	≤0.03	≤0.06	≤0.25	4	≤1	≤0.03

^aOFX, ofloxacin; PEN, penicillin; ERY, erythromycin; TET, tetracycline; SXT, trimethroprim-sulfamethoxazole; CHL, chloramphenicol; CRO, ceftriaxone. NCCLS breakpoints, S/I/R: OFX, ≤2/4/≥8; PEN, ≤0.06/0.12-1/≥2; ERY, ≤0.25/0.5/≥1; TET, ≤2/4/≥8; SXT, ≤0.5/1-2/≥4; CHL, ≤4/-/≥8; CRO, ≤0.5/1/≥2. 

^bState abbreviations: AL, Alabama; AZ, Arizona; CA, California; GA, Georgia; KY, Kentucky; NC, North Carolina; NY, New York; PA, Pennsylvania; TN, Tennessee; TX, Texas; WA, Washington; WV, West Virginia. 

^c–, not applicable (R6 is a nonencapsulated laboratory strain). 

^dNT, nontypeable. 

Fluoroquinolone susceptibility.

Fluoroquinolone susceptibility profiles were determined using gatifloxacin, gemifloxacin, levofloxacin, moxifloxacin, ofloxacin, sparfloxacin, and trovafloxacin (Table 2). Gemifloxacin had the highest level of antipneumococcal activity on a per-gram basis. The MICs of this investigational agent were 8- to 128-fold lower than the other fluoroquinolones, when tested against the resistant strains, and did not exceed 1 μg/ml. NCCLS breakpoints have not been established for gemifloxacin.

TABLE 2.

Amino acid changes in the QRDR of DNA gyrase and topoisomerase IV subunits and associated fluoroquinolone susceptibility profiles for clinical isolates of S. pneumoniae

Strain	Amino acid change in:									MIC (μg/ml)b
	GyrAa		GyrB		ParC			ParE		OFX	LVX	TVA	SPX	MOX	GAT	GMX
R6	S81	E85	D435	E474	A63	S79	D83	D435	E474	2	1	0.12	0.25	0.12	0.25	0.03
4563	–c	–	–	–	–	F	–	–	–	4	2	0.5	0.5	0.25	0.5	0.06
4566	–	–	–	–	–	Y	–	–	–	4	2	0.5	0.5	0.25	0.5	0.06
4562	–	–	–	–	–	F	–	–	–	4	2	0.5	0.5	0.25	0.5	0.06
4567	–	–	–	–	–	F	–	–	–	4	2	0.5	1	0.5	0.5	0.06
4570	–	–	–	–	–	F	–	–	–	4	2	0.5	1	0.25	0.5	0.12
4571	–	–	E	–	T	–	–	–	K	8	4	0.5	2	0.5	2	0.12
4554	–	–	–	K	–	F	–	–	–	16	8	2	1	0.25	2	0.06
4555	F	–	–	–	–	–	N	–	–	16	8	2	8	2	2	0.25
4556	F	–	–	–	–	–	Y	–	–	16	8	2	8	2	2	0.25
4569	–	G	–	–	–	–	–	N	–	32	16	2	8	2	4	0.25
4557	F	–	–	–	–	F	–	–	–	32	16	4	8	4	4	0.25
4553	F	–	–	–	–	Y	–	–	–	32	16	4	8	4	4	0.25
6274	F	–	–	–	–	F	–	–	–	32	16	4	8	4	4	0.5
4561	F	–	–	–	–	F	–	–	–	32	16	8	8	4	8	0.25
4560	F	–	–	–	–	F	–	–	–	32	16	8	8	4	8	0.25
4558	F	–	–	–	–	F	–	–	–	32	16	8	16	4	8	0.25
4552	F	–	–	–	–	F	–	–	–	32	16	8	16	4	4	0.25
4568	F	–	–	–	–	Y	–	–	–	32	16	8	16	4	8	0.25
6276	–	K	–	–	–	F	–	–	–	64	32	8	32	8	8	1
6275	F	–	–	–	–	F	Y	–	–	64	32	16	16	4	8	0.5
4559	F	–	–	–	–	Y	V	–	–	64	32	16	32	8	16	0.5

^aAll amino acid positions correspond to S. pneumoniae sequences (4, 37). 

^bOFX, ofloxacin; LVX, levofloxacin; TVA, trovafloxacin; SPX, sparfloxacin; MOX, moxifloxacin; GAT, gatifloxacin; GMX, gemifloxacin. NCCLS breakpoints, S/I/R; OFX, LVX, ≤2/4/≥8; TVA, MOX, GAT, ≤1/2/≥4; SPX, ≤0.5/1/≥2; (GMX, no breakpoints). 

^c–, no change from the fluoroquinolone-susceptible laboratory strain R6. 

For strains with intermediate levels of resistance to ofloxacin, the MICs of gatifloxacin, gemifloxacin, levofloxacin, and trovafloxacin increased two- to fourfold compared with those for the susceptible control strain but did not exceed NCCLS susceptibility breakpoints. There was no increase in the MIC of moxifloxacin for four of the five ofloxacin-intermediate strains. For strains with high-level resistance to ofloxacin (MIC ≥ 8 μg/ml), the extent of cross-resistance among the newer fluoroquinolones was not consistent and depended on the mutations involved (see below).

DNA sequence analysis of the QRDRs of gyrA, gyrB, parC, and parE

DNA sequences and the inferred amino acid sequences were determined for amplified gene fragments that included the QRDRs of gyrA, gyrB, parC, and parE from each isolate. The sequences were aligned and compared with corresponding sequences from the susceptible laboratory strain R6 (Table 2) and the NCCLS quality control strain, ATCC 49619. Amino acid sequences of the QRDRs and susceptibility profiles for ATCC 49619 were identical to those of R6 (data not shown). In general, mutations resulting in amino acid changes in the QRDRs could be categorized as follows: (i) single mutants involving S79 of ParC (five strains); (ii) double mutants involving S79 of ParC and S81 of GyrA (eight strains); (iii) triple mutants that included S79 and D83 of ParC, plus S81 of GyrA (two strains); and (iv) mutants with multiple changes that involved other positions in ParC or GyrA, or amino acid changes in GyrB or ParE (six strains). Mutations in gyrB, parE, or both loci were identified in three strains (4571, 4554, and 4569). None of the isolates examined had a single mutation in gyrA without a mutation in parC or parE.

Correlation of QRDR mutations and fluoroquinolone resistance.

In isolates with a single amino acid change, the mutation was consistently found in the codon for S79 of ParC. Although an amino acid substitution for ParC/S79 resulted in a two- or fourfold increase in the MICs of most fluoroquinolones, MICs remained within the susceptible range for all agents except ofloxacin and sparfloxacin. Strains with double and triple mutations that involved S79 and D83 of ParC and S81 of GyrA were consistently resistant to all fluoroquinolones tested. However, MICs of fluoroquinolones were more variable for isolates with amino acid changes at positions other than ParC/S79 and GyrA/S81. The MICs for the double mutant 4569 (GyrA/E85G, ParE/D435N) were two- to eightfold greater than for the double mutant 4554 (ParC/S79F and GyrB/E474K) for all fluoroquinolones tested except trovafloxacin, but two- to fourfold less than the double mutant 6276 (GyrA/E85K and ParC/S70F). Strain 4571, which contained three mutations, was fully resistant to ofloxacin and sparfloxacin but remained susceptible to trovafloxacin and moxifloxacin.

The MICs of the two 8-methoxyquinolones moxifloxacin and gatifloxacin varied among isolates with different amino acid changes. Resistance to moxifloxacin, but not gatifloxacin, appeared to be associated with alteration of GyrA. Conversely, an amino acid substitution in GyrB (strain 4554) was associated with a fourfold increase in the MIC of gatifloxacin, but not moxifloxacin, compared with those for strains having the same S79F change in ParC.

Alteration of E85 in GyrA (strains 4569 and 6276) was associated with significant decreases in susceptibility to most of the fluoroquinolones tested. Strain 4569 (GyrA/E85G, ParE/D435N) had no ParC mutation; however, the MICs of fluoroquinolones tested against strain 4569 were equal to those of ParC/S79 and GyrA/S81 double mutants. Also, the MICs of most agents tested were two-or fourfold higher in strain 6276 (GyrA/E85K combined with ParC/S79F) compared with MICs of ParC/S79 and GyrA/S81 double mutants. Alteration of the analogous D83 in ParC (triple mutants 6275 and 4559) was associated with twofold increases in the MICs of gemifloxacin, levofloxacin, ofloxacin, and trovafloxacin, compared with double mutants lacking this change. However, in the triple mutants, the MICs of gatifloxacin, moxifloxacin, and sparfloxacin appeared to depend upon the amino acid substituted for D83. When the aspartic acid residue was changed to valine (strain 4559), the MIC of each of these agents increased two-to fourfold, but when the change was to tyrosine (strain 6275), MICs were the same as those of double mutants lacking this alteration.

Transfer of fluoroquinolone resistance by genetic transformations.

The contribution of each amino acid change found in strains 4554 and 4571 was determined by sequential transformations (Table 3). Individual mutations were introduced into the susceptible laboratory strain R6 by transformation with purified PCR-amplified QRDR gene fragments, and several attempts were made to introduce each fragment as a first mutation. Transformation was confirmed by DNA sequence analysis. The QRDRs of gyrA, gyrB, parC, and parE from selected transformants were analyzed to verify the presence of the expected mutation and the absence of any additional mutations that may have been introduced as a result of polymerase errors during PCR amplification.

TABLE 3.

Susceptibility profiles of S. pneumoniae R6 transformants

Strain	Donor gene(s)/mutationa	MIC (μg/ml)b
		OFX	LVX	TVA	SPX	MOX	GAT	GMX
R6	none	2	1	0.12	0.25	0.12	0.25	0.03
R64554parC	parC/S79F	4	2	0.5	0.5	0.25	0.5	0.06
R64554parC+gyrB	parC/S79F, gyrB/E474K	16	8	2	1	0.25	2	0.06
R64554chr	4554chrc (parC/S79F, gyrB/E474K)	16	8	2	1	0.25	2	0.06
4554		16	8	2	1	0.25	2	0.06
R64571parE	parE/E474K	4	2	0.12	0.25	0.12	0.25	0.03
R64571parE+gyrB	parE/E474K, gyrB/D435E	8	4	0.25	0.5	0.25	1	0.06
R64571parE, gyrB+parC	parE/E474K, gyrB/D435E, parC/A63T	8	4	0.5	2	0.5	2	0.12
R64571chr	4571chr (parE/E474K, gyrB/D435E, parC/A63T)	8	4	0.5	2	0.5	2	0.12
4571		8	4	0.5	2	0.5	2	0.12

^aS. pneumoniae R6 transformed with QRDR gene fragment of specified gene(s) from the designated strain. 

^bOFX, ofloxacin; LVX, levofloxacin; TVA, trovafloxacin; SPX, sparfloxacin; MOX, moxifloxacin; GAT, gatifloxacin; GMX, gemifloxacin. NCCLS breakpoints, S/I/R; OFX, LVX, ≤2/4/≥8; TVA, MOX, GAT, ≤1/2/≥4; SPX, ≤0.5/1/≥2; (GMX, no breakpoints available). 

^cchr, transformed with chromosomal DNA from the designated strain. 

Amplified QRDR gene fragments of gyrB (E474K) and parC (S79F) from strain 4554 were introduced independently into R6; however, no gyrB transformants could be selected on levofloxacin without prior introduction of the parC mutation. MICs of all fluoroquinolones increased two- to fourfold for R6/4554^parC and were consistent with those of clinical isolates with the same single mutation. Subsequent introduction of the gyrB QRDR (E474K) into R6/4554^parC resulted in an additional twofold increase in the MIC of sparfloxacin and a fourfold increase in the MICs for ofloxacin, levofloxacin, trovafloxacin, and gatifloxacin. The MICs of moxifloxacin and gemifloxacin were not affected by the presence of the gyrB mutation. The fluoroquinolone resistance profile of R6/4554^parC,gyrB was equivalent to that of the parent strain, 4554, and R6/4554^chr. Reserpine had no effect on the growth of strain 4554 in the presence of levofloxacin, indicating that active efflux does not contribute to resistance in this strain (data not shown).

In transformations of R6 with QRDR gene fragments of gyrB, parC, and parE from strain 4571, only parE transformants (E474K) could be selected on levofloxacin as single mutants. Introduction of the gyrB gene fragment (D435E) into R6/4571^parE resulted in R6/4571^parE,gyrB transformants for which the MICs increased fourfold for gatifloxacin and twofold for each of the other fluoroquinolones. Transformants with the 4571 parC mutation (A63T) could be selected only after the parE and gyrB mutations were introduced. For this final transformation, sparfloxacin was used for selection of R6/4571^{parE,gyrB,parC} because the MICs of levofloxacin for the donor strain, 4571, and the recipient, R6/4571^parE,gyrB, were the same. In addition, alternative PCR primers were used to shift the position of the parC mutation farther downstream from the 5′ end of the gene fragment. Numerous transformations using the 329-bp fragment of parC (amplified with oligonucleotide primers PNC10 and PNC11) were successful for introducing mutations in codons for S79 and D83, located 130 to 145 bp from the 5′ end of the gene fragment. However, efforts to introduce the mutation in the codon for ParC/A63, located at nucleotide position 83, were unsuccessful. This problem was resolved by increasing the size of the amplified gene fragment, shifting the mutation downstream to a position 181 bp from the 5′ end. For the resulting R6/4571^{parE,gyrB,parC} transformants, the MIC of sparfloxacin increased fourfold and MICs of gatifloxacin, gemifloxacin, moxifloxacin, and trovafloxacin increased twofold. The fluoroquinolone susceptibility profiles of all R6/4571^{parE,gyrB,parC} transformants were identical to that of the donor strain, 4571, and of R6/4571^chr. No active drug efflux was detected in growth inhibition assays of strain 4571 with reserpine (data not shown).

DISCUSSION

The increasing prevalence of penicillin- and multidrug-resistant pneumococci has resulted in extensive use of fluoroquinolones for antimicrobial therapy of pneumococcal infections. However, the emergence of fluoroquinolone-resistant isolates (8, 21) threatens to further limit the number of agents available for effective treatment of pneumococcal disease. A more thorough understanding of the molecular mechanisms of resistance will aid in designing strategies to minimize the emergence of resistant strains and to predict patterns of cross-resistance among this class of antimicrobial agents. Thus, the purpose of this study was to characterize recent clinical isolates of fluoroquinolone-resistant pneumococci, determine the activity and patterns of cross-resistance of newer fluoroquinolones against these isolates, and investigate the contribution of selected gyrA, gyrB, parC, and parE mutations to fluoroquinolone resistance.

Analysis of the susceptibility profiles revealed an association of multidrug resistance, but not fluoroquinolone resistance, with specific serotypes of S. pneumoniae. Four of the five MDR strains in this study were serotype 23F, which is consistent with a previous report that MDR is associated with pneumococcal serotypes 6, 14, 19, and 23 in U.S. isolates (46). However, 11 serotypes and one nontypeable strain were represented among the 16 isolates with high-level resistance to fluoroquinolones, precluding any correlation between serotype and fluoroquinolone resistance. There was also no correlation between resistance to fluoroquinolones and resistance to other classes of antimicrobial agents. An association of penicillin resistance and reduced susceptibility to fluoroquinolones in pneumococci has been proposed and is a subject of debate (13, 15, 16). However, among the 16 isolates with high-level fluoroquinolone resistance, only 3 were resistant to penicillin, and those were MDR strains. Ten isolates were not resistant to any of the agents tested except fluoroquinolones, suggesting that fluoroquinolone resistance is not associated with resistance to other antimicrobial agents. This diversity of susceptibility profiles and serotypes also suggests that mutations associated with resistance to fluoroquinolones occur de novo in multiple strains, as is proposed in the recent report on a survey of fluoroquinolone-resistant pneumococci in Canada (8).

Numerous studies of mutants selected from cultures containing subinhibitory levels of ciprofloxacin have led to the hypothesis that a single amino acid change in ParC results in decreased susceptibility to fluoroquinolones and double mutations involving ParC and GyrA result in high-level resistance. However, based on data from clinical isolates characterized in this study and previous reports (10, 28, 29), such a generalization appears to be valid only for double mutants of ParC/S79 and GyrA/S81. Among more heterogeneous mutants, the level of resistance varied, depending on the fluoroquinolone tested and the mutations present.

For all types of mutants, the most active agent was gemifloxacin. This new broad-spectrum fluoronaphthyridone, with a novel pyrrolidine substituent at the C-7 position, has been reported to have potent antipneumococcal activity (9, 19). Although breakpoints have not been established for this investigational agent, the exceptional activity (all MICs were ≤1 μg/ml) was consistent for all highly resistant isolates, including double mutants of ParC/S79 and GyrA/S81 and strains with more heterogeneous mutations.

For fluoroquinolones other than gemifloxacin, resistance and cross-resistance were consistent only among the subset of double mutants with amino acid changes in both ParC/S79 and GyrA/S81. Strain 4569 did not have a mutation at either ParC/S79 or GyrA/S81, but this isolate had high-level resistance to four fluoroquinolones. A triple mutant, strain 4571, had lower levels of fluoroquinolone resistance than did any of the double mutants. Therefore, when amino acid changes occurred at alternate sites in GyrA or ParC, or when changes were found in GyrB and ParE, the associated resistance appeared to be more dependent on the amino acid position than on the number of mutations.

As in previous reports, most isolates had mutations at conventional sites in parC (codons for S79 or D83) and in gyrA (codons for S81 and E85). However, 4 of 16 isolates with high-level resistance to one or more fluoroquinolones did not contain a mutation in the codon for S79 of ParC, the amino acid position most frequently reported to be associated with resistance of pneumococci to this class of agents. In addition, 3 of these 16 isolates had multiple mutations that included sites in gyrB, parE, or both loci. Our results suggest that the occurrence of mutations in these alternate sites may be more common in clinical isolates than in mutants selected in vitro. This may be a result of in vivo selection during clinical therapy with newer fluoroquinolones, or by exposure of pneumococcal strains to multiple fluoroquinolone structures over time.

The molecular basis of fluoroquinolone resistance due to mutations in gyrB and parE is unclear. GyrB has an N-terminal ATPase domain (residues 2 to 393 in Escherichia coli) and a C-terminal domain (residues 394 to 804) designated as B′ (7). Recent X-ray crystallographic and mutational analyses of type II topoisomerase structures suggest that the B′ domain of GyrB and the homologous region of ParE play an active part in the catalysis of DNA cleavage (3, 12). In the model of topoisomerase activity proposed by Fass et al. (12), ATP binding by the N-terminal domain of GyrB results in a conformational change that rotates the B′ domain to expose the DNA-binding groove. As a result, the QRDRs of GyrA and GyrB are brought into juxtaposition. Based on this model, mutations in the B′ domain could contribute to fluoroquinolone resistance by spatial and chemical alteration of the target site in the DNA-enzyme complex. A GyrB (or ParE) mutant would be expected to interact differently with each fluoroquinolone based on the spatial and chemical characteristics of both the agent and the amino acid substitution.

Initial studies of mutants selected in vitro with ciprofloxacin identified an amino acid change of D435N in gyrB and parE of S. pneumoniae. This region has been designated as the QRDR of these genes based on amino acid changes in analogous positions in E. coli and Staphylococcus aureus (22, 36, 41). More recently, mutants with changes in the E474 position have been selected in vitro on agents other than ciprofloxacin, including clinafloxacin (39), gemifloxacin (19), levofloxacin (10), and trovafloxacin (34). In GyrB, alterations of E474K have been reported in clinical isolates of pneumococci (10, 20). The impact of this amino acid change on fluoroquinolone resistance was not determined but was considered to be minimal because it lies outside the conventional QRDR (19).

The contributions of E474K alterations in GyrB and ParE were determined in this study and support the model proposed by Fass et al. (12). In both loci, the effect of the mutation on the MIC of a fluoroquinolone varied significantly depending on the agent. Only the MICs of levofloxacin and ofloxacin were increased with the introduction of this mutation into parE. However, when the same mutation occurred in gyrB, MICs were increased for all fluoroquinolones except moxifloxacin and gemifloxacin. Interestingly, GyrB mutants (changes of either E474K or D435E) were important for resistance to gatifloxacin but not to moxifloxacin. The chemical structures of these 8-methoxyquinolones differ only in their C-7 substituents (23, 42). Thus, for these agents, the chemical and structural characteristics of their C-7 substituents mediate unique and significantly different interactions with their target proteins.

In addition to gyrB and parE mutations, a novel amino acid change of A63T in ParC was detected in one isolate. Unlike most parC mutations in pneumococci, this alteration could not be selected on levofloxacin as a first-round transformant. However, when the parC gene fragment encoding the A63T change was introduced after the 4571 parE and gyrB mutations, MICs of five fluoroquinolones increased two- to fourfold, indicating that mutations affecting this amino acid position make significant contributions to resistance. No alterations have been reported for this amino acid position in S. pneumoniae. However, a similar change has been described for the corresponding position in GyrA (A67S) of E. coli (50).

No single mutants of gyrA were found in the isolates included in this study. Single mutants of gyrA have been detected among strains selected in vitro on newer fluoroquinolones, such as sparfloxacin or gatifloxacin (14, 38), but are rarely reported for clinical isolates. This may reflect the relatively recent use of newer fluoroquinolones that select for gyrA mutations first. Also, isolates with a single mutation in gyrA may be overlooked if the MIC of the fluoroquinolone agent(s) used to screen for resistance is unchanged or only modestly increased.

Mutations selected with newer fluoroquinolones appear to be more diverse than those selected with ciprofloxacin. For example, exposure of pneumococci to levofloxacin results in increased selection of GyrB mutants (10), and the primary target of sparfloxacin is GyrA instead of ParC (38). Newer fluoroquinolones also select mutations outside the conventional QRDR. In a study described by Ince and Hooper (24), mutants of S. aureus were selected with premafloxacin, an 8-methoxyquinolone. Several amino acid changes that contributed to resistance were found outside the conventional QRDR of grlA (the staphylococcal gene analogous to parC). In addition, data from the present study have confirmed that the QRDR of gyrB and parE should include E474. Thus, the heterogeneity of mutations should be expected to increase as newer fluoroquinolones are introduced for therapeutic use and the QRDRs are extended to accommodate these changes.

In summary, the characterization of 21 fluoroquinolone-resistant clinical isolates confirms that fluoroquinolone resistance is not associated with serotype in pneumococci. Molecular analyses of pneumococcal clinical isolates continue to reveal novel amino acid changes and a greater heterogeneity of mutations than is reported for mutants selected in vitro on ciprofloxacin. Significant contributions to fluoroquinolone resistance by gyrB and parE mutations were determined by genetic transformation, and gyrB mutations were demonstrated to be particularly important in resistance to gatifloxacin, but not moxifloxacin. The emergence of clinical isolates resistant to six fluoroquinolones, especially among serotype 23F MDR strains, is cause for considerable concern and emphasizes the need for judicious use of these agents. Although the investigational agent, gemifloxacin, retained a high level of activity against pneumococcal strains with the most common gyrA and parC mutations, it is likely that clinical isolates with novel amino acid changes will render even this new fluoroquinolone ineffective in the near future. Continued surveillance and molecular analysis of resistant isolates of pneumococci will provide data needed for a rational approach to fluoroquinolone susceptibility testing and therapeutic use of this class of agents.