INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Streptococcus pneumoniae is an important human pathogen responsible for respiratory illnesses such as pneumonia and other serious diseases, including meningitis and otitis (4). Concern over the emergence of penicillin-resistant and multidrug-resistant strains has led to the development of antipneumococcal fluoroquinolones, such as sparfloxacin, levofloxacin, gatifloxacin, and moxifloxacin. These agents have greater activity than ciprofloxacin against S. pneumoniae, and several are now approved for first-line therapy of community-acquired pneumonia.

The likelihood that increased use of quinolones in the community will result in pneumococci with decreased drug susceptibility has focused attention on the mechanisms of quinolone action and resistance in this organism. Previous studies with Escherichia coli have shown that resistance often involves mutation of DNA gyrase and then of topoisomerase IV (15, 23, 26), two related ATP-dependent enzymes that act by a double-stranded DNA break (29, 46) and collaborate to ensure DNA unwinding during DNA replication and chromosome segregation at cell division (1, 49). Quinolones are thought to form a topoisomerase-drug-DNA ternary complex (10, 11, 14) that cellular processes convert into a lethal lesion, possibly a double-stranded DNA break (18, 44). Resistance mutations occur in a short discrete segment of the DNA gyrase gyrA and gyrB genes (and analogous parts of the topoisomerase IV parC and parE genes) termed the quinolone resistance-determining region (QRDR) (7, 33, 35, 47, 48). Hot spots for resistance involve changes of S. pneumoniae GyrA Ser81 to Phe or Tyr and of ParC Ser79 to Phe or Tyr (16, 17, 22, 32, 36-39, 41, 45). The precise effects of these changes remain to be examined at the enzyme level.

Attempts to understand the target preferences of quinolones in S. pneumoniae have centered on identifying the order of QRDR mutations acquired during stepwise drug challenge. Surprisingly, we found that ciprofloxacin selected parC (Ser79Phe or Tyr) QRDR changes (16, 22, 32, 36), whereas sparfloxacin selected gyrA (Ser81Phe or Tyr) QRDR mutants (38). We therefore proposed that the structure of the quinolone determines its target preference in S. pneumoniae (38). However, subsequent work showing that other quinolones, such as norfloxacin, levofloxacin, pefloxacin, and trovafloxacin, select parC changes (13, 41, 45) and the finding that S. pneumoniae topoisomerase IV is more sensitive than gyrase to inhibition by quinolones (including sparfloxacin) (31, 40) led Morrissey and George to suggest that topoisomerase IV is the quinolone target in S. pneumoniae (31) as in Staphylococcus aureus (8, 9, 34). Non-QRDR topoisomerase IV resistance mutations were invoked by them to explain the sparfloxacin resistance of our first-step gyrA mutants (31).

Though sparfloxacin is not alone in selecting gyrA mutants (2, 17, 39, 42), we have sought to clarify the roles of gyrase and topoisomerase IV in drug action. Here we exclude the participation of non-QRDR mutations in sparfloxacin resistance and, by characterizing recombinant Ser81Phe GyrA and Ser79Phe ParC proteins expressed in E. coli, establish that both mutations confer resistance to sparfloxacin (Spx^r) and to ciprofloxacin (Cip^r) in assays of enzyme inhibition and cleavable-complex formation. That the Cip^r phenotype of the corresponding gyrA allele and the Spx^r phenotype of the parC allele are silent in the wild-type S. pneumoniae background supports our proposed model that different quinolones act as cellular poisons through different topoisomerase targets. From these results and immunoblotting analysis of topoisomerase expression levels, we discuss the mechanism of selective quinolone action in S. pneumoniae.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Media and reagents. Luria broth and Luria agar were prepared as described previously (43), as was brain heart infusion medium containing 10% horse blood (37). Ciprofloxacin hydrochloride and sparfloxacin hydrochloride were kindly provided by Bayer UK, Newbury, United Kingdom, and by Dainippon Pharmaceutical Co., Suita, Japan.

Bacterial strains and plasmids. S. pneumoniae strain 7785 and its mutants 1C1, 1S1, 1S4, 2C6, 2C7, 2S1, 2GM1, and 3C4 have been described elsewhere (17, 36-38). S. pneumoniae R6 is a standard laboratory strain. E. coli strain DH5, used for cloning purposes, and strains BL21(DE3)pLysS and BL21(DE3)pLysE, used for protein expression, were from our laboratory collection. Plasmids pXP9, pXP10, pXP13, and pXP14, used to express S. pneumoniae GyrB, GyrA, ParC, and ParE proteins, have been described previously (40). Supercoiled and relaxed plasmid pBR322 DNAs were prepared as described earlier (40).

Drug susceptibilities. A twofold-dilution method was employed in which ~10⁵ CFU of bacteria was spotted on brain heart infusion-agar plates which were read after overnight aerobic incubation at 37°C. The MIC is the drug concentration at which no bacterial growth was seen under these conditions.

DNA sequence analysis. PCR was used to amplify the parE-parC region of strains 1S1 and 1S4. For each strain, three overlapping fragments were obtained that spanned 5.5 kb and included the entire parE and parC genes, the 400-bp intergenic region, and a sequence of about 300 bp upstream of the parE gene carrying the promoter region. The following primer pairs were used (37): H4023, 5'-GTCAATCACAAAGGTTG (nucleotide [nt] positions 316 to 300 upstream of parE); and reverse primer M0361, 5'-TCCGACTCTAATTTCC (nt positions 44 to 28 bp downstream of the parE termination codon); N6894, 5'-TGGGCTTTGTATCATATGTCTAAC (nt positions 15 to +9 at the start of the parC gene; underlining indicates an NdeI site overlapping the initiation codon); and reverse primer N6893, 5'-TGGCATCAAGAGATGGTC (nt positions 406 to 389 downstream of the parC termination codon); and finally, G8394, 5'-TGAAGCGATTGAGTTCC (nt positions 650 to 666 in the parE gene); and reverse primer M4721, 5'-TGCTGGCAAGACCGTTGG (nt positions 470 to 453 of the parC gene). Genomic DNA was prepared from strains 1S1 and 1S4 as previously described (38) and was used as the template for Taq DNA polymerase in the presence of 1.5 mM MgCl₂. PCR conditions were as follows: denaturation at 94°C for 1 min, annealing at 55°C for 1 min, and polymerization at 72°C for 3 min. Reactions were performed over 30 cycles. PCR products corresponding to the 2.3-kb parE fragment, 2.9-kb parC fragment, and 2.2-kb linking region were obtained and were each purified on Qiagen minispin columns. The DNA was sequenced directly on both top and bottom strands using an ABI Prism automated sequencer and a series of nested oligonucleotide primers.

GyrA(Phe81) and ParC(Phe79) expression plasmids. An overexpression construct containing the mutant gyrA gene was obtained by fragment exchange into gyrA plasmid pXP10 (40). PCR was used to amplify a 528-bp fragment corresponding to the 5'end of gyrA using chromosomal DNA from strain 1S1 as the template. The primers used were VGA35, 5'-ATGAGGCATTTACATATGCAGGATAAAAATTTAGTG (an NdeI site overlapping the ATG initiation codon is underlined) (40); and reverse primer VGA4, 5'-AGTTGCTCCATTAACCA (36). Strain 1S1 DNA and primers were incubated with Vent DNA polymerase (which has proofreading activity) in the presence of 1.5 mM MgCl₂ under the following conditions: denaturation at 94°C for 1 min, annealing at 48°C for 1 min, and polymerization at 72°C for 3 min. Reactions were performed over 30 cycles. The PCR product was digested with NdeI and with EcoRV (at an internal site in gyrA), purified by electrophoresis in low-gelling agarose, and recovered. The 405-bp fragment was ligated into NdeI-EcoRV-digested plasmid pXP10 and was used to transform E. coli DH5. Several fragment exchange plasmids were recovered, and following restriction analysis, the gyrA gene of one, pXP15, was sequenced in its entirety to confirm that the gyrA gene had the correct reading frame and encoded the Ser81Phe mutation.

Purification of gyrase and topoisomerase IV subunits and generation of antibodies. Conditions for the overexpression of recombinant S. pneumoniae GyrB, GyrA, ParC, and ParE subunits as His-tagged proteins in E. coli BL21 and detailed protocols for their purification to >95% homogeneity by nickel chelate column chromatography have been published elsewhere (40). These proteins were used individually as antigens to immunize rabbits and as ligands (absorbed to nitrocellulose) for affinity purification of rabbit antibodies. Mutant GyrA and ParC proteins were produced in E. coli BL21(DE3) by induction of plasmids pXP15 and pXP16 and were purified as described for the wild-type subunits (40).

Immunodetection and quantitation of gyrase and topoisomerase IV proteins in S. pneumoniae cell lysates. S. pneumoniae strains were grown overnight at 37°C either on brain heart infusion plates containing 10% horse blood or alternatively in Todd-Hewitt liquid medium in an atmosphere of 5% CO₂. Approximately 5 × 10⁹ CFU could be collected from confluent growth on three 90-mm-diameter petri dishes or from 10 ml of liquid medium. For cell lysis, 5 × 10⁹ cells were collected, washed once with 1 ml of phosphate-buffered saline, and then resuspended in 500 µl of phosphate-buffered saline containing 1% sodium dodecyl sulfate (SDS) and Complete protease inhibitor (Boehringer) (one tablet was dissolved in 1 ml of water, and 20 µl was added). The suspension was incubated at room temperature for 30 min, and then 500 µl of 2× SDS-polyacrylamide gel electrophoresis (PAGE) sample loading buffer (125 mM Tris-HCl, pH 6.8, 4% SDS, 20% [wt/vol] glycerol, 1.43 M 2-mercaptoethanol, and 0.2% bromophenol blue) was added, and the mixture was left at room temperature for 10 min. The mixture was centrifuged at 100,000 rpm for 20 min at 4°C in a Beckman TL-100 ultracentrifuge. The supernatant was used in SDS-PAGE analysis.

DNA supercoiling, kDNA decatenation, and DNA cleavage assays. Conditions for reconstitution of S. pneumoniae gyrase and topoisomerase IV from their subunits and assaying of supercoiling and kinetoplast DNA (kDNA) decatenation activity in the absence or presence of quinolone inhibitors have been described previously (40). The efficiency of wild-type and mutant gyrase complexes in mediating quinolone-promoted DNA breakage was determined as described earlier (40). The extent of DNA cleavage was quantitated from photographic negatives using a Molecular Dynamics Personal Densitometer SI and ImageQuant software.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of Phe81 GyrA and Phe79 ParC expression plasmids and purification of His-tagged proteins. To obtain the mutant proteins in quantity, we used a fragment exchange procedure to introduce the relevant mutation into gyrA expression plasmid pXP10 or parC plasmid pXP14, yielding constructs pXP15 and pXP16 (see Materials and Methods). The mutant gyrA and parC fragments were produced by PCR using proofreading Vent polymerase and chromosomal DNA from strains 1S1 and 2C7 as template (Table 1). The mutant gyrA and parC genes in pXP15 and pXP16 were sequenced in full to ensure that they differed from the 7785 parent sequence only at the expected codon. Transformation of the four plasmids into E. coli allowed isopropyl--D-thiogalactopyranoside-inducible expression of wild-type and mutant subunits as His-tagged proteins (40). After nickel chelate column chromatography, all four proteins were obtained in soluble form at >95% homogeneity (Fig. 1).

View larger version (60K): [in this window] [in a new window]   FIG. 1.   SDS-PAGE analysis of highly purified S. pneumoniae wild-type (wt) GyrA, GyrA(Phe81), wild-type ParC, and ParC(Phe79) proteins. The His-tagged proteins were overexpressed in E. coli, purified by nickel resin chromatography, and examined on an SDS-7.5% polyacrylamide gel. Lane M, marker proteins (sizes in kilodaltons).

Mutant gyrase and topoisomerase IV complexes are catalytically efficient and resistant to quinolone inhibition. When complemented with recombinant S. pneumoniae GyrB, the specific activity of Phe81 GyrA in a supercoiling assay was 2 × 10⁵ U/mg, which is comparable to that of the wild type (40). The specific activity of the Phe79 ParC protein when complemented with recombinant ParE in a kDNA decatenation assay was 2.5 × 10⁵ U/mg, which is also similar to that of wild-type ParC (40). It seems that neither mutation impairs catalytic activity.

View larger version (65K): [in this window] [in a new window]   FIG. 2.   DNA supercoiling activity of mutant S. pneumoniae DNA gyrase is highly refractory to inhibition by sparfloxacin (SPAR) (A) and by ciprofloxacin (CIP) (B). Relaxed pBR322 (0.4 µg) was incubated with gyrase activity (1 U) reconstituted from GyrB and either wild-type (wt) GyrA (left lanes) or GyrA(Phe81) (right lanes) in the presence of 1.4 mM ATP and the indicated amounts (in micromolar) of quinolones. Reactions were stopped, and the DNA products were separated by electrophoresis in 1% agarose. DNA was stained with ethidium bromide and photographed under UV illumination. Lanes a and b, supercoiled and relaxed pBR322 DNA, respectively. N, R, and S denote nicked, relaxed, and supercoiled DNA, respectively.

View larger version (53K): [in this window] [in a new window]   FIG. 3.   S. pneumoniae topoisomerase IV containing the ParC(Phe79) subunit is resistant to inhibition by sparfloxacin (SPAR). Topoisomerase IV activity (1 U), reconstituted from recombinant ParE subunit and either wild-type (wt) ParC or ParC(Phe79) protein, was incubated with kDNA (0.4 µg) in the presence of 1.4 mM ATP and quinolones at the concentrations indicated. DNA was analyzed by agarose gel electrophoresis as described in Fig. 2. Lane a, kDNA. N and M indicate kinetoplast network DNA and released relaxed minicircles, respectively.

Mutations markedly impede cleavable-complex formation with sparfloxacin and ciprofloxacin. To examine effects on cleavable-complex formation, supercoiled pBR322 was incubated with equal amounts of wild-type or mutant topoisomerases in the absence or presence of quinolones. After addition of SDS to induce DNA cleavage, samples were treated with proteinase K to remove GyrA (ParC) protein covalently linked to DNA. The products were then analyzed by electrophoresis in 1% agarose. Drug stabilization of the cleavable complex is expected to generate linear DNA on denaturation. For wild-type gyrase (Fig. 4), inclusion of ciprofloxacin or sparfloxacin produced a dose-dependent increase in linear DNA product, with values of 80 µM for the drug concentration producing 25% conversion of input DNA to the linear form. Unlike wild-type enzyme, the mutant gyrase exhibited a low level of DNA cleavage activity that was not further stimulated by drugs even at 640 µM. Moreover, although the wild-type enzyme showed weak DNA-relaxing activity, this activity was more marked with the mutant enzyme (Fig. 4, right panels). Inclusion of 1.4 mM ATP blocked DNA relaxation and produced a small (twofold) stimulation of sparfloxacin-mediated DNA breakage by both the wild-type and mutant enzymes, increasing DNA breakage by the mutant enzyme to 10% of total DNA at 640 µM sparfloxacin (Fig. 4C). Similar results were seen for ciprofloxacin (not shown). Thus, the mutant gyrase complex was at least 8- to 16-fold less efficient in cleavable-complex formation.

View larger version (84K): [in this window] [in a new window]   FIG. 4.   Ser81Phe mutation in GyrA impairs DNA cleavage by gyrase promoted by sparfloxacin (SPAR) (A and C) or ciprofloxacin (B). Supercoiled pBR322 DNA (0.4 µg) was incubated with S. pneumoniae GyrB (1.7 µg) and either wild-type (wt) GyrA or GyrA(Phe81) (0.45 µg) in the absence (A and B) or presence of 1.4 mM ATP (C) and quinolones at the concentrations indicated. After addition of SDS and proteinase K, DNA samples were analyzed by electrophoresis in 1% agarose. Lanes a and b, supercoiled pBR322 and EcoRI-cut pBR322. N, L, and S indicate nicked, linear, and supercoiled DNA.

View larger version (52K): [in this window] [in a new window]   FIG. 5.   DNA breakage by topoisomerase IV mediated by sparfloxacin (SPAR) (A) and by ciprofloxacin (B) is inhibited by the Phe79 mutation in ParC. DNA cleavage reactions were carried out as described in the legend to Fig. 4 using ParE (1.7 µg) and wild-type (wt) ParC or ParC(Phe79) (0.45 µg) but with a lower range of quinolone concentrations.

Cip^r phenotype of the gyrA(Ser81Phe) allele and Spx^r phenotype of parC(Ser79Phe) are silent in S. pneumoniae: evidence for target selection and a poison mechanism. The enzyme results can be compared with the resistance profiles of S. pneumoniae strains expressing the same mutant alleles (Table 1). Strain 1S1 is a first-step mutant selected with sparfloxacin which expresses Ser81Phe GyrA. From the in vitro enzyme studies, it appears that this mutation alone would be sufficient to account for the eightfold increase in resistance to sparfloxacin (Table 1). Interestingly, in contrast to expectations from enzyme studies, strain 1S1 was susceptible to ciprofloxacin. A second independent, first-step, sparfloxacin-selected mutant, 1S4, expressing Ser81Tyr GyrA, behaved similarly (Table 1). The silent Cip^r phenotype of these gyrA alleles can be explained if ciprofloxacin acts selectively through topoisomerase IV, not through gyrase. Similarly, strain 2C7 selected with ciprofloxacin and expressing the parC(Ser79Phe) allele exhibited the expected resistance to ciprofloxacin but, contrary to the enzyme studies, was susceptible to sparfloxacin (Table 1). Mutant 2C6 carrying a Ser79Tyr alteration showed similar behavior. The silent Spx^r phenotype of the parC allele would be expected if sparfloxacin acted preferentially through gyrase, consistent with results for strain 1S1. We note that the presence of mutations in both parC and gyrA in strains 2S1, 2GM1, and 3C4 generated a Cip^r Spx^r phenotype (Table 1). Taken together, the data in Table 1 show that the Cip^r phenotype of a mutant gyrA gene is masked by a wild-type parC gene, whereas the Spx^r phenotype of the mutant parC gene is masked by a wild-type gyrA gene. These epistatic effects of the gyrA and parC genes are not consistent with a killing mechanism involving simple enzyme inhibition but suggest that the quinolones act by a poison mechanism involving selective trapping of gyrase or topoisomerase IV as cleavable complexes (27).

Spx^r gyrA strains carry wild-type parE-parC genes: topoisomerase IV is not the preferred intracellular target of the drug. It has been suggested that topoisomerase IV is the intracellular target of all quinolones, including sparfloxacin (31). To explain the selection by sparfloxacin of first-step gyrA mutants such as 1S1 and 1S4 (Table 1), it was proposed by others that these strains must be double mutants carrying an additional undetected non-QRDR resistance mutation in topoisomerase IV (31). It was important to test this idea, as recent work on clinafloxacin resistance has suggested that some S. pneumoniae QRDRs may be more extensive than those defined in E. coli (39). Moreover, resistance to premafloxacin in S. aureus involves parC mutations that lie outside the conventional QRDR (20). Therefore, we amplified the entire 5.5-kb parE-parC locus of independent strains 1S1 and 1S4 in each case as three overlapping PCR products comprising a sequence 300 bp upstream of the parE gene to 300 bp downstream of parC. Direct sequence analysis of the PCR products revealed no differences in the parE and parC genes or their promoters in 1S1 or 1S4, compared to the known sequences of parental strain 7785. The absence of topoisomerase IV mutations in these strains indicates that sparfloxacin resistance accrues from a mutation at GyrA Ser81 and that gyrase (not topoisomerase IV) is the intracellular target.

Gyrase and topoisomerase IV subunits are differentially expressed in S. pneumoniae: relevance for ternary-complex formation. The relative amounts of ternary-complex formation will depend not only on drug-enzyme affinities but also on intracellular enzyme levels. To determine whether the apparently overwhelming in vitro preference of quinolones for topoisomerase IV might be counterbalanced by differential expression of gyrase in vivo, we developed and applied a quantitative immunoblotting procedure using rabbit polyclonal antisera raised against highly purified S. pneumoniae GyrA, GyrB, ParC, and ParE subunits produced in E. coli (40). Given the sequence homology between the 97-kDa GyrA and ParC proteins and between the 72-kDa GyrB and ParE subunits, cross-reacting antibodies were first removed from each antiserum by employing the purified recombinant proteins as ligands absorbed to nitrocellulose. By using quantitative immunoblotting with the recombinant subunits as standards, it was found that the anti-GyrA and anti-ParC antibodies were specific and did not cross-react with ParC and GyrA proteins, respectively (Fig. 6, panels 1 to 3). Similarly, the anti-GyrB and anti-ParE antisera were also specific (Fig. 6, panels 4 to 6). It was possible to detect 10 ng of each topoisomerase protein.

View larger version (14K): [in this window] [in a new window]   FIG. 6.   Polyclonal antisera specific for the S. pneumoniae GyrA, ParC, GyrB, and ParE proteins. Equimolar amounts of highly purified recombinant GyrA (A), ParC (C), GyrB (B), and ParE (E) proteins were run on SDS-6% polyacrylamide gels which were either stained with Coomassie blue (panels 1 and 4) or electrotransferred to nitrocellulose filters (panels 2, 3, 5, and 6). The filters were probed with rabbit antisera made to the recombinant His-tagged GyrA (panel 2) and ParC (panel 3) or recombinant His-tagged GyrB (panel 5) or ParE subunit (panel 6). In each case, GyrA- and ParC- and GyrB- and ParE- antisera were prestripped of cross-reacting antibodies using the other homologous recombinant protein as an affinity ligand. The binding of the first antibody was visualized by using an alkaline-phosphatase-conjugated anti-rabbit IgG second antibody with colorimetric development.

View larger version (26K): [in this window] [in a new window]   FIG. 7.   Quantitative immunoblotting of gyrase and topoisomerase IV subunits in protein lysates of S. pneumoniae and its quinolone-resistant mutants. (A) GyrA and GyrB are more abundant in cell extracts than ParC and ParE. Protein extracts (4 µg) from wild-type S. pneumoniae strains 7785 and R6 prepared by SDS lysis were run on four SDS-6% polyacrylamide gels alongside known amounts of recombinant GyrA, GyrB, ParC, or ParE protein used as the standard. Proteins were transferred to nitrocellulose filters and probed with specific antisera and a second antibody as described in the Fig. 6 legend. Due to the presence of histidine tags, the protein standards have a slightly lower mobility than the native proteins. (B) ParC and ParE proteins are detected using four- to eightfold higher amounts of extract. Protein extracts, prepared from strains 7785 and R6 and gyrA mutants 1S1 and 1S4, were loaded at 4 µg per lane for detection of GyrA and GyrB and at 16 µg and 32 µg per lane for quantitation of ParC and ParE, respectively.

Effects of gyrA resistance mutations on topoisomerase expression levels. The countervailing activities of gyrase and of relaxing enzymes such as topoisomerase IV are important in maintaining chromosomal supercoiling (28, 50). Moreover, the expression of some genes, including gyrase, is homeostatically regulated by DNA supercoiling. To determine whether gyrase mutations contribute to quinolone resistance by downregulating the expression of gyrase and/or topoisomerase IV subunits, we examined protein extracts from strains 1S1 and 1S4 by immunoblotting (Fig. 7B). In both cases, it is clear that the GyrA, GyrB, ParC, and ParE levels in the mutants were not visibly different from those present in wild-type strains 7785 and R6. Thus, the Ser81Phe or -Tyr mutations in GyrA do not cause resistance indirectly by affecting topoisomerase expression in S. pneumoniae but interfere directly with ternary-complex formation.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have applied a combination of biochemical and immunochemical approaches to analyze the properties of both wild-type S. pneumoniae GyrA and ParC proteins and mutants bearing respective Ser81Phe and Ser79Phe mutations, changes that are frequently observed in quinolone-resistant strains. By reconstitution of the mutant proteins with the complementary GyrB and ParE subunits, we have shown for the first time that the resulting gyrase and topoisomerase IV complexes were each at least some 8- to 16-fold less efficient in forming cleavable complexes (the relevant lesion) with quinolones. These biochemical studies establish the important result that the two mutations both confer resistance to both sparfloxacin and ciprofloxacin. That the Cip^r phenotype was silent in an Spx^r S. pneumoniae gyrA(Ser81Phe) strain expressing a wild-type topoisomerase IV, whereas Spx^r was silent in a Cip^r parC(Ser79Phe) strain, shows that sparfloxacin acts preferentially through gyrase, whereas ciprofloxacin acts through topoisomerase IV. The enzyme and genetic data therefore provide direct support for an earlier proposal that these quinolones act selectively through different targets in S. pneumoniae (38).

Recent transformation studies complement work on first-step resistant mutants in confirming the resistance phenotypes of mutant GyrA and ParC proteins in S. pneumoniae. Thus, defined parC PCR products encoding the Ser79Phe mutation have been shown to transform wild-type S. pneumoniae to ciprofloxacin resistance (32). Unfortunately, the sparfloxacin response of the transformants was not tested. Very recently, gyrA PCR fragments encoding the Ser81Phe alteration yielded transformants that were Spx^r but Cip^s (21). These results are in agreement with studies of first-step gyrA mutants (Table 1) and can be attributed directly to the Ser81Phe mutation. Thus, transformation, mutant characterization, and enzyme studies concur in providing unequivocal evidence for selective quinolone targeting.

The key to the enzyme studies has been the development of a reliable E. coli expression system for the S. pneumoniae gyrase and topoisomerase IV subunits using plasmid-borne genes whose sequence has been completely determined. By using a fragment exchange protocol, it was possible to introduce a single, validated codon change into the wild-type gyrA or parC gene used for protein expression. The reconstituted gyrase and topoisomerase IV complexes therefore differ from the wild type by a single amino acid change. The Ser81Phe mutation in GyrA and the Ser79Phe change in ParC lie at equivalent positions in the two proteins. Ser81 in S. pneumoniae GyrA is equivalent to Ser83 in the E. coli protein (7, 35), which X-ray structure analysis has shown lies in helix A'4, part of a helix-loop-helix motif thought to contact DNA and form the quinolone binding site (30). Presumably, mutation of Ser81 to Phe in the S. pneumoniae GyrA protein (or Ser79 in ParC) confers resistance by interfering with quinolone binding, thereby preventing assembly of the ternary complex. Consistent with this idea, we note that dose-dependent DNA cleavage was detected with the mutant enzymes (in the presence of ATP) at very high drug concentrations (e.g., Fig. 4C) suggesting that the two mutations operate by reducing drug affinity for the target. It is not known whether this effect is due to the steric bulk of the phenylalanine side chain, greater than that of serine (7).

Several interesting points emerge from the drug inhibition studies of the wild-type and quinolone-resistant topoisomerase complexes. First, topoisomerase IV is some 20- to 40-fold more efficient than gyrase in forming cleavable complexes with quinolones (Fig. 4 and 5). Second, despite their structural differences, the in vitro enzyme inhibitory properties of sparfloxacin and ciprofloxacin are very similar. These two features are difficult to reconcile with the clear in vivo evidence that sparfloxacin acts through gyrase and ciprofloxacin through topoisomerase IV. Obviously, if the strong predilection for quinolone action on topoisomerase IV holds under cellular conditions, then some other feature(s) must override drug-enzyme affinities so that sparfloxacin can kill through gyrase. In considering this point, we reasoned that ternary-complex formation favoring topoisomerase IV might be counterbalanced by differences in intracellular topoisomerase expression favoring gyrase. Small differences between quinolone target affinities affected by ATP, salt, DNA sequence, or chromosomal supercoiling might then allow selective killing through one or another target. However, by immunoblotting, it was found that gyrase levels are only three times higher than those of topoisomerase IV in S. pneumoniae cell extracts (Fig. 7), a ratio similar to that reported recently for Bacillus subtilis (19). It seems unlikely that these modest differences in expression alone could overturn the 20- to 40-fold preference for cleavable-complex formation with topoisomerase IV and allow selective drug action. Unfortunately, our attempts to examine cleavable-complex formation in S. pneumoniae have thus far been obfuscated by the relatively low levels of DNA breakage that can be induced in vivo (data not shown).

To account for selective intracellular drug targeting, we propose that there are drug-dependent differences in ternary-complex formation, processing, or repair in S. pneumoniae. This hypothesis requires that the ternary complexes formed by sparfloxacin and ciprofloxacin in vivo are somehow different from those formed in vitro, as no differences were seen in drug effects on the purified enzymes (Fig. 2 to 5). We note that quinolones are thought to act by forming ternary complexes upstream of the replication fork such that collisions between a component of the replication machinery, e.g., DNA polymerase or DNA helicase, and subsequent processing, e.g., abortive repair, convert the complex into an irreversible lethal lesion (24, 49). It appears that either gyrase or topoisomerase IV can act ahead of replication forks (6, 25). In principle, differential targeting could arise if quinolones differed in their partitioning into critical sites of enzyme-DNA complexes within the cell. This might be achieved by interactions with other proteins that differentially mask quinolone binding sites on the two topoisomerases. Alternatively, different drugs might have differential effects on the conversion of ternary complexes into lethal lesions through the agency of helicases or polymerases and other, as-yet-unknown factors that recognize and process the complex. This could occur directly through the presence of the quinolone in the ternary complex or indirectly by additional drug binding to accessory protein factors. Further studies will be needed on ternary-complex formation and processing in S. pneumoniae.

Finally, whatever its detailed explanation, selective quinolone action through one or another topoisomerase target may occur widely in bacteria with important consequences for mechanistic studies, drug design, and the circumvention of resistance. Recent data indicate that, as in S. pneumoniae, sparfloxacin and ofloxacin select respective gyrA and parC QRDR mutants of Mycoplasma hominis (5). Moreover, in S. aureus, though most quinolones act through topoisomerase IV, nalidixic acid is reported to act through gyrase (12). Quinolones with different modes of action may be useful probes of topoisomerase function in DNA replication. On the other hand, drugs that act equally through gyrase and topoisomerase IV in vivo would be desirable in requiring changes in two genes for resistance, thereby minimizing the development of resistant strains (17, 36, 38, 39). Studies with S. pneumoniae should provide a useful paradigm in exploring these areas.