Effects of Mutations in GrlA of Topoisomerase IV from Staphylococcus aureus on Quinolone and Coumarin Activity

Abstract

The grlA genes of Staphylococcus aureus ISP794 (wild type), MT5224c4 (grlA [Phe-80]), MT5224c2 (grlA [Pro-116]), and MT111 (grlA [Glu-116]) were cloned in pSK950, a shuttle vector, and introduced into S. aureus strains derived from strain RN4220. The mutations at position 116 of GrlA (Ala→Pro or Glu) caused an increase in the level of fluoroquinolone resistance and a decrease in the level of coumarin susceptibility, whereas the mutation at position 80 (Ser→Phe) caused only an increase in the level of fluoroquinolone resistance. In multicopy alleles, both types of mutations were codominant for fluoroquinolone resistance, and mutations at position 116 were also codominant for coumarin resistance.

Bacterial DNA type II topoisomerases (DNA gyrase and topoisomerase IV) catalyze the modification of the topological state of DNA in cells. Topoisomerase IV appears to have a particular role in decatenation of daughter chromosomes (18). These enzymes are composed of two subunits which form A₂B₂ tetramers composed of GyrA and GyrB subunits in the case of DNA gyrase and GrlA (ParC in Escherichia coli) and GrlB (ParE in E. coli) subunits in the case of topoisomerase IV in Staphylococcus aureus. For both enzymes, the A subunit is responsible for DNA breakage and reunion (10), while the B subunit catalyzes the hydrolysis of ATP (5). DNA gyrase is the target of several antibiotics. The fluoroquinolones, such as ciprofloxacin and norfloxacin, inhibit the DNA breakage-reunion cycle by binding to the gyrase-DNA complex. The coumarins, such as novobiocin and coumermycin, act by inhibiting ATP hydrolysis mediated by the GyrB subunit of DNA gyrase. Topoisomerase IV is also the target of several fluoroquinolones (6, 8). The action of coumarins on topoisomerase IV has been demonstrated by showing the inhibition of the activity of the purified enzyme by novobiocin (2, 12) and by showing the increased coumarin susceptibility of a mutant with an altered GrlB subunit (Asn470Asp) (7). No mutations in GrlB homologous to those in the GyrB subunit that cause coumarin resistance have been reported.

The novel phenotype of quinolone resistance and coumarin hypersusceptibility of the GrlB (Asn470Asp) mutant was postulated to be due to a novel mechanism, possibly associated with altered catalytic function (7). We showed in the study described here that certain GrlA mutations also exhibit this phenotype. This class of GrlA mutations, in contrast to those localized in the common quinolone resistance-determining region, is closer to the active site of enzyme breakage of DNA, suggesting possible effects on enzyme function. The dominance of these mutations was also studied and was found to be similar to that of the GrlB (Asn470Asp) mutation.

In vitro mutants from strain MT5.

The strains and plasmids used in this study are described in Table 1. Strain MT5 carries the nov-142 locus (gyrB [Ile102Ser and Arg144Ile]), which is responsible for high-level resistance to coumarins (7). Some mutants of MT5 selected with ciprofloxacin or norfloxacin showed mutations in GrlA at position 80 or 116 (Table 2) (11, 14). For both kinds of mutants an increase in the level of fluoroquinolone resistance was observed, but only mutants carrying the mutation at position 116 exhibited a slight but reproducible decrease in the level of novobiocin resistance (twofold) (Table 2). In order to study the effects of these mutations on novobiocin susceptibility, cloning of the different grlA genes was done.

TABLE 1.

S. aureus strains and plasmids used in this study

Strain or plasmid	Genotype or characteristica	Source or reference
Strains
ISP2133	8325 pig-131 trp-489 Ω(chr::Tn917lac)2	14
MT5224c4	8325 pig-131 nov (gyrB142) hisG15 flqA (grlA542)	14
MT5224c2	8325 pig-131 nov (gyrB142) hisG15 flqA (grlA552)	11
MT111	8325 pig-131 hisG15 flqA (grlA548)	14
EN1	8325 pig-131 nov (gyrB142) hisG15 flqA (grlA542) Ω(chr::Tn917lac)2	11
BF3	8325 pig-131 hisG15 flqA (grlA548) Ω(chr::Tn917lac)2	This study: ISP2133 DNA × MT111
BF4	8325 pig-131 nov (gyrB142) hisG15 flqA (grlA552) Ω(chr::Tn917lac)2	This study: ISP2133 DNA × MT5224c2
RN4220	8325-4 r−	9
EN20	RN4220 flqA (grlA542) Ω(chr::Tn917lac)2	11
BF5	RN4220 flqA (grlA548) Ω(chr::Tn917lac)2	This study: BF3 DNA × RN4220
BF6	RN4220 flqA (grlA552) Ω(chr::Tn917lac)2	This study: BF4 DNA × RN4220
BF2	RN4220 nov (gyrB142)	7
BF7	RN4220 nov (gyrB142) flqA (grlA552) Ω(chr::Tn917lac)2	This study: BF4 DNA × BF2
BF8	RN4220 nov (gyrB142) flqA (grlA548) Ω(chr::Tn917lac)2	This study: BF3 DNA × BF2
BF9	RN4220 nov (gyrB142) flqA (grlA542) Ω(chr::Tn917lac)2	This study: EN1 DNA × BF2
Plasmids
pSK950	10.5-kb plasmid carrying the attP site of phage L54a, replicon of pSC101, Spr (E. coli) and temperature-sensitive replicon of pE194, Tcr (S. aureus)	G. L. Archer
pSKISA	2.7-kb BamHI fragment containing the grlA gene from ISP794 cloned into pSK950	This study
pSKMTA	2.7-kb BamHI fragment containing the grlA gene from MT111 cloned into pSK950	This study
pSKC2A	2.7-kb BamHI fragment containing the grlA gene from MT5224c2 cloned into pSK950	This study
pSKC4A	2.7-kb BamHI fragment containing the grlA gene from MT5224c4 cloned into pSK950	This study
pBFISA	100-bp EcoRI fragment containing the promoter of grlB cloned into pSKISA	This study
pBFMTA	100-bp EcoRI fragment containing the promoter of grlB cloned into pSKMTA	This study
pBFC2A	100-bp EcoRI fragment containing the promoter of grlB cloned into pSKC2A	This study
pBFC4A	100-bp EcoRI fragment containing the promoter of grlB cloned into pSKC4A	This study

^aSp, spectinomycin, Tc, tetracycline. 

TABLE 2.

Activities of ciprofloxacin and novobiocin against mutants of MT5

Mutant strain	Amino acid mutation in GrlA	MIC (μg/ml)
		Ciprofloxacin	Novobiocin
MT5		0.25	40.0
MT5224c4	Ser80Phe	2.0	40.0
MT52222	Ser80Phe	2.0	40.0
MT5224c2	Ala116Pro	2.0	20.0
MT52184	Ala116Pro	4.0	20.0
MT5224c3	Ala116Glu	2.0	20.0

Cloning of grlA genes.

The grlA genes of strain ISP794 (wild-type), MT5224c4 (grlA [Phe-80]), MT5224c2 (grlA [Pro-116]), and MT111 (grlA [Glu-116]) were amplified by PCR with Vent DNA polymerase and primers containing a BamHI site as described previously (7). The PCR product was cloned into the BamHI site of pGEM3-zf(+). In order to verify that no mutation was introduced by the polymerase, the sequences of the entire grlA gene for each allele were determined, and no changes were found. The grlA genes were then subcloned into the BamHI site of pSK950, a shuttle vector carrying the thermosensitive replicon of plasmid pE194 from S. aureus. Plasmids pSKISA (grlA⁺), pSKC2A (grlA [Pro-116]), pSKC4A (grlA [Phe-80]), and pSKMTA (grlA [Glu-116]) were obtained (Table 1). pSKC4A (grlA [Phe-80]) was introduced into wild-type strain S. aureus RN4220 (r⁻), and pSKISA (grlA⁺) was introduced into S. aureus EN20 (a derivative of RN4220 with the grlA [Phe-80] mutation on the chromosome) (11). MICs were measured with Mueller-Hinton agar supplemented with serial twofold increasing concentrations of drugs, and the cells were grown at 30°C. Unexpectedly, for cells carrying pSKISA and pSKC4A the MICs were similar in comparison to the MICs for cells containing the vector plasmid alone (data not shown), suggesting that the grlA genes were not expressed. The region that was amplified by PCR included the putative promoter previously described by Yamagishi et al. (17). Thus, this promoter appears to be weak or inefficient, suggesting that expression of grlA is dependent on the promoter of grlB, which is upstream of grlA. The initiation codon of grlA overlaps the stop codon of grlB (6, 17).

In order to express the grlA genes, a 100-bp fragment containing the grlB promoter was amplified by PCR with primers containing an EcoRI site (primer 5′-ATA TAT GGA ATT CAG CTA TGA AAG T-3′, with the 5′ nucleotide at position 264 in the oligonucleotide coordinates used by Yamagishi et al. [17], and primer 5′-ATG AAT TCG GCA CCT GCA AAC GTA-3′ [position 379]) and was cloned into the EcoRI site of pGEM3-zf(+). This fragment was also subcloned into the EcoRI site of the previously constructed pSK950 derivatives containing different grlA alleles, pSKISA, pSKC2A, pSKC4A, and pSKMTA. The EcoRI site of pSK950 is localized 10 bp upstream of the BamHI site. The obtained plasmids were pBFISA (grlA⁺), pBFC2A (grlA [Pro-116]), pBFC4A (grlA [Phe-80]), and pBFMTA (grlA [Glu-116]) (Table 1).

Expression of grlA genes in multicopy alleles.

The plasmids with the mutated grlA genes (pBFMTA, pBFC2A, and pBFC4A) were introduced into strain RN4220. Strains derived from RN4220 carried the three different alleles and were obtained by transformation of high-molecular-weight chromosomal DNA as described previously (13): EN20 (grlA [Phe-80]), BF5 (grlA [Pro-116]), and BF6 (grlA [Glu-116]). The plasmid with the wild-type gene (pBFISA) was introduced into each of these strains. The MICs were determined and are presented in Table 3. First, for BF6 and BF5, which carry the mutation at position 116 on the chromosome, the MICs of fluoroquinolone (ciprofloxacin and norfloxacin) were increased (two- to eightfold) and the MICs of coumarins (novobiocin and coumermycin) were decreased (fourfold) in comparison to the MICs for the parent strain RN4220. Second, when the plasmids carrying the mutated grlA genes were introduced into wild-type strain RN4220, an increase in the fluoroquinolone MICs was observed. In addition, RN4220 carrying pBFMTA (grlA [Glu-116]) and pBFC2A (grlA [Pro-116]) exhibited decreased levels of resistance to coumarins (two- to fourfold), in contrast to RN4220 carrying pBFC4A (grlA [Phe-80]), for which no change in the coumarin MICs was found. For the merodiploids of mutant strains EN20, BF5, and BF6 containing plasmid pBFISA (grlA⁺), the opposite effects were observed: decreases in the fluoroquinolone MICs for all strains and increases in the coumarin MICs only for BF5 and BF6 (Table 3). These results confirmed that the mutation at position 116 is responsible for the quinolone resistance and the coumarin hypersusceptibility phenotype of these mutants.

TABLE 3.

Susceptibility patterns of S. aureus gyrB⁺ strains diploid for grlA

Strain/plasmid	grlAa		MIC (μg/ml)b
	Chromosome	Plasmid	Ciprofloxacin	Norfloxacin	Novobiocin	Coumermycin
RN4220/pSK950	Wild	None	0.5	1.0	0.16	0.02
RN4220/pBFMTA	Wild	Ala116Glu	1.0	4.0	0.08	0.01
RN4220/pBFC2A	Wild	Ala116Pro	1.0	4.0	0.04	0.01
RN4220/pBFC4A	Wild	Ser80Phe	1.0	8.0	0.16	0.02
BF6/pSK950	Ala116Glu	None	1.0	8.0	0.04	0.01
BF6/pBFISA	Ala116Glu	Wild	0.25	1.0	0.08	0.01
BF5/pSK950	Ala116Pro	None	1.0	8.0	0.04	0.01
BF5/pBFISA	Ala116Pro	Wild	0.25	1.0	0.08	0.01
EN20/pSK950	Ser80Phe	None	2.0	16.0	0.16	0.02
EN20/pBFISA	Ser80Phe	Wild	0.25	1.0	0.16	0.01

^aWild, wild-type gene; none, plasmid vector only. 

^bThe MICs are the mean values for at least three different transformants. 

In our previous study, the coumarin hypersusceptibility phenotype of a grlB mutant was also seen in the gyrB142 (coumarin resistant) genetic background. To determine if this effect was also seen with the grlA mutations at position 116, the plasmids were then introduced into strains carrying the gyrB142 allele (nov-142). These strains were obtained by introducing the different grlA alleles into strain BF2 (gyrB [Ile102Ser and Arg144Ile]) (Table 1). The results, presented in Table 4, obtained for fluoroquinolone resistance were similar to the results observed for strain RN4220 (gyrB⁺). For coumarin resistance, the presence of plasmids pBFC2A (grlA [Pro-116]) and pBFC4A (grlA [Glu-116]) in BF2 also increased the level of susceptibility to novobiocin and coumermycin (fourfold). Conversely, pBFISA (grlA⁺) increased by fourfold the coumarin MICs for strains carrying the mutation at position 116 (BF7 and BF8). These results confirmed that the mutations in grlA at position 116 are responsible for decreasing the level of susceptibility to coumarins either in a gyrB⁺ or in a gyrB142 background, whereas the mutation grlA (Phe-80) only modifies fluoroquinolone susceptibility.

TABLE 4.

Susceptibility patterns of S. aureus RN4220 gyrB142 strains diploid for grlA^a

Strain/plasmid	grlAb		MIC (μg/ml)c
	Chromosome	Plasmid	Ciprofloxacin	Norfloxacin	Novobiocin	Coumermycin
BF2/pSK950	Wild	None	0.5	1.0	20.0	5.0
BF2/pBFMTA	Wild	Ala116Glu	1.0	4.0	5.0	1.25
BF2/pBFC2A	Wild	Ala116Pro	1.0	4.0	5.0	1.25
BF2/pBFC4A	Wild	Ser80Phe	1.0	4.0	10.0	2.5
BF7/pSK950	Ala116Glu	None	1.0	8.0	5.0	1.25
BF7/pBFISA	Ala116Glu	Wild	0.25	0.5	20.0	5.0
BF8/pSK950	Ala116Pro	None	1.0	4.0	5.0	1.25
BF8/pBFISA	Ala116Pro	Wild	0.25	0.5	20.0	5.0
BF9/pSK950	Ser80Phe	None	2.0	8.0	20.0	5.0
BF9/pBFISA	Ser80Phe	Wild	0.25	0.5	20.0	5.0

^aBF2, BF7, BF8, and BF9 carry the double-mutation (Ser102 and Ile144) gyrB gene. 

^bWild, wild-type gene; none, plasmid vector only. 

^cThe MICs are the mean values for at least three different transformants. 

Determination of dominance.

Our results also indicate that both mutations in grlA are codominant for quinolone resistance in multicopy alleles (Table 3 and 4). The codominance of fluoroquinolone resistance was also observed for mutations in grlB (7), parC (in E. coli, equivalent to grlA) (8), and parE (in E. coli, equivalent to grlB) (3). For coumarin hypersusceptibility, mutation at position 116 is codominant in multicopy alleles in both gyrB⁺ and gyrB142 backgrounds, as described previously for the mutation grlB (Asp470) (7). Thus, the grlB mutation and grlA mutations at position 116 are also epistatic for the coumarin resistance of GyrB.

Implications for mechanisms of quinolone resistance.

Mutations at positions 80 and 116 of GrlA probably have different mechanisms of quinolone resistance. In the crystal structure of a fragment of the GyrA subunit from E. coli (10), Ser83, which is homologous to Ser80 of GrlA, and other previously described quinolone resistance mutations in GyrA, which have not been associated with alterations in coumarin susceptibility, are clustered on a negatively charged surface that bridges the upper (or head) interface of the two subunit monomers. Modeling suggests that this surface is a principal site of binding of the “gate” strand of DNA, which is broken and resealed by enzyme action. The model further predicts that DNA distortion is required for interaction of the active-site tyrosine residues (Tyr122 in E. coli GyrA and Tyr119 in S. aureus GrlA) with the target phosphoryl residues of DNA during initial DNA cleavage. The residues of the quinolone resistance-determining region (QRDR) are localized in the area where this distortion is predicted to occur, suggesting that the binding of quinolones to the gyrase-DNA complex may occur in this region. The finding that complexes of DNA and gyrase reconstituted with a resistant GyrA (Ser83Trp) subunit exhibit substantially reduced levels of quinolone binding (16) is consistent with this model. The catalytic activity of topoisomerase IV reconstituted with GrlA (Ser80Tyr) has also been shown to be similar to that of the wild-type enzyme (2). Thus, the Ser80Phe mutation in GrlA, which is homologous to Ser83 of GyrA, also likely confers resistance by causing a reduction in the level of binding of quinolones to the enzyme-DNA complex and is predicted to have normal or near normal catalytic efficiency.

Quinolone resistance mutations at position 116 of GrlA, in contrast, have an additional phenotype of coumarin hypersusceptibility. This phenotype was first reported in a GrlB (Asn470Asp) mutant (7). The localization of position 470 of GrlB, based on the crystal structure of the homologous yeast topoisomerase II (1), is distant from the homologous putative sites of quinolone binding in GrlA and GyrA, suggesting a distinct mechanism of quinolone resistance and one possibly associated with altered enzyme catalytic efficiency (7, 15). GrlA116 mutations are also outside the previously defined QRDR and are close to the active site, Tyr119. In the E. coli GyrA structure, the Tyr122 residues at the head-dimer interface are at the periphery of the putative DNA-binding surface, the center of which contains Ser80 and other residues of the QRDR (10). Thus, these tyrosine residues may not themselves be involved in quinolone binding to the complex. The proximity of the mutations at position 116 to the amino acid that mediates DNA cleavage (Tyr119) further suggests that catalytic efficiency might also be affected. The replacement of Ala by Pro might be predicted to distort the helical structure in this region (α6 in the yeast structure) (1), and replacement by Glu would provide an additional negative charge in a region of positively charged residues and would thereby alter DNA binding, which these residues are thought to mediate. Conformational changes in the putative region of quinolone binding might also be envisioned to occur secondarily to the changes at position 116.

Future studies of drug binding and the catalytic activity of the mutant enzymes will be necessary to test the hypothesis that resistance in GrlB and GrlA116 mutants occurs by mechanisms distinct from alteration of a quinolone-binding site, but mutations in a domain homologous to that harboring the mutation in GrlB cause amsacrine resistance in yeast topoisomerase II (15). If impaired enzyme function mediates quinolone resistance and coumarin hypersusceptibility in our S. aureus GrlB and GrlA116 mutants, then this impairment is insufficient to alter the growth of either mutant under laboratory conditions (7) (data not shown). Normal growth patterns, however, have been described for coumarin-resistant gyrB mutants of E. coli that exhibit impaired gyrase function (4).

In conclusion, the mutations at position 116 of GrlA (Ala→Pro or Glu) cause an increase in the level of fluoroquinolone resistance and a decrease in the level of coumarin susceptibility, whereas the mutation at position 80 (Ser→Phe) causes only an increase in the level of fluoroquinolone resistance. Both types of mutations are codominant for fluoroquinolone resistance in multicopy alleles.