Examination of Methicillin-Resistant and Methicillin-Susceptible Staphylococcus aureus Mutants with Low-Level Fluoroquinolone Resistance

Abstract

For Staphylococcus aureus, stepwise mutations result in high-level quinolone resistance. Methicillin-resistant and -susceptible quinolone-resistant, first-step mutants generated in vitro were obtained and found to be no different than those recovered from murine abscesses. Approximately 10% of all first-step mutants were resistant to ethidium bromide, and selected strains had mutations that mapped to flqB. NorA-mediated resistance among first-step mutants may be more prevalent than previously reported.

Resistance to fluoroquinolones occurs in a stepwise fashion (24). The mutants initially obtained are resistant to low concentrations of quinolones (first-step mutants), and subsequent mutations lead to higher-level resistance. The most common mutation conferring low-level resistance occurs in the flqA locus (3, 4, 25). DNA topoisomerase IV, encoded at the flqA locus, is the primary target of fluoroquinolones (3, 4, 16, 26) for Staphylococcus aureus but not necessarily for other bacterial species. Some flqA mutants are also hypersusceptible to novobiocin (25) (Nov^hs), and mutations of selected strains map to grlB (6). Second-step mutations have been identified in gyrA. Mutations in gyrA, plus a mutation in topoisomerase IV (grlA/flqA) (3, 16, 26), confer high-level resistance to quinolone antibiotics. A mutation only in gyrA has no effect on susceptibility to quinolones (16, 26); thus, such mutations have not been found among first-step mutants (3, 4, 7, 8, 9, 19, 25). Low-level quinolone resistance due to increased expression of the NorA multidrug efflux pump are due to mutations in the flqB locus (17) and possibly other loci (10). However, it is unclear when, or at what stage in resistance, altered NorA expression occurs. Here we present data implicating a role for NorA in first-step quinolone resistance.

There are limited data regarding the effect of the passage of strains in vivo on the ability to detect first-step mutants from common staphylococcal infections such as those of the skin. In vivo passage without quinolone selection might result in positive or negative selection for different classes of fluoroquinolone-resistant mutants. In the present study, we found that passage in vivo did not alter the distribution of different classes of first-step mutants for both methicillin-resistant S. aureus (MRSA) and methicillin-susceptible S. aureus (MSSA) clinical strains.

Methods and reagents.

Tetracycline and erythromycin were used at 10 and 5 μg/ml, respectively. Resistance to fusidic acid (Fus^r), ciprofloxacin (Cfx^r), or ethidium bromide (Ebr^r) was determined when growth was evident on brain heart infusion agar (BHA) containing fusidic acid (10 μg/ml), ciprofloxacin (1 μg/ml), or ethidium bromide (20 μg/ml).

Broth microdilution MIC determinations were done with cation-adjusted Mueller-Hinton broth (1). BHA-novobiocin gradient plate assays were performed according to the method of Szybalski (21). Strains were considered Nov^hs when the length of confluent growth across a gradient plate containing novobiocin (0 to 0.25 μg/ml) was <66% of that of the parental strain. Identification to the species level was performed and clonality of mutant colonies was confirmed with an Acustaph kit (Carr-Scarborough Microbiologicals, Inc., Decatur, Ga.) and by pulsed-field gel electrophoresis of SmaI-digested genomic DNA (22), respectively.

Three-day murine abscess formation was performed by the method of Bunce et al. (2) using an initial inoculum of 10⁷ CFU of S. aureus.

High-molecular-weight DNA was obtained and transformations were done as described previously (20), with genetic mapping strains (18).

Results.

Using a modification of published methods (3, 8, 23, 25), we isolated quinolone-resistant mutants from broth cultures (in vitro mutants) and from murine abscesses (in vivo-passaged mutants). Consistent with previous reports (5, 12, 14), more quinolone-resistant mutants were recovered by using ciprofloxacin than by using either ofloxacin or levofloxacin. For in vitro mutants, 10⁸ CFU of 38 independent brain heart infusion broth cultures of the laboratory strain RN4220 (13) were inoculated onto BHA containing 1 μg of ciprofloxacin per ml (two times the MIC). A mean of 11 colonies/10⁸ CFU were obtained after 24 h of incubation at 37°C. Selection with ofloxacin at the same concentration (two times the MIC) yielded fivefold fewer mutants than those obtained by ciprofloxacin selection. Data similar to those above were obtained for MSSA (NB1720 and NB1722) and MRSA (NB1725 and NB1726) clinical strains. Except for a single mutant obtained from strain NB1726, no mutants were obtained with levofloxacin selection at two times the MIC (0.75 μg/ml).

Low-level-quinolone-resistant mutants were obtained from murine abscesses (in vivo-passaged mutants) infected with S. aureus RN4220 or MSSA or MRSA clinical strains at frequencies similar to those obtained for in vitro mutants (Table 1). It is possible that first-step mutants existed at the time of infection and persisted within the abscess because of the inoculum used to produce the abscess. It is also possible that some first-step mutants emerged during infection and were not differentially eliminated during growth.

TABLE 1.

Isolation of quinolone-resistant mutants derived from quinolone-susceptible strains present in murine abscesses

Strain	No. of abscesses	Mean CFU/25% of resuspended abscess	Mean no. of quinolone-resistant mutants/25% of resuspended abscessa			% of abscesses that yielded quinolone-resistant mutants upon selection witha:
			CFX	OFX	LFX	CFX	OFX	LFX
NB1721	8	2 × 107	80	1.1	0	100	13	0
NB1725	10	6.1 × 106	11	0.1	0b	100	10	0
RN4220	20	4.8 × 106	2.6	0.4	0	60	20	0

^aConcentrations of antibiotics (CFX, ciprofloxacin; OFX, ofloxacin; LFX, levofloxacin) used to select resistant mutants were two times the MICs for the parental strain. 

^bFirst-step mutants from NB1725-infected abscesses were selected on levofloxacin at three times the MIC. 

For all first-step mutants obtained, the mean MIC of ciprofloxacin increased approximately fourfold above the mean MIC for parental strains. For ofloxacin and levofloxacin, the increases were two- to threefold. Thus, compared to ciprofloxacin, the relative potency of levofloxacin and ofloxacin increased with single-step resistance—a result consistent with those of Thomson et al. (23).

Next, the phenotypes of first-step mutants were characterized. Ebr^r is an indicator of increased norA expression (10), and Nov^hs is an indicator of mutant alleles of flqA (topoisomerase IV). Both phenotypes were found among the in vitro- and the in vivo-passaged mutants. In 38 independent experiments, RN4220 was inoculated onto solid media containing 1 μg of ciprofloxacin per ml, and one resistant mutant per experiment was selected. Of the 38 chosen, 10 were Nov^hs and 1 was Ebr^r. A different sampling procedure yielded a higher percentage of Ebr^r mutants. Instead of sampling one mutant from many in an individual experiment, all mutants obtained on a plate in each of six experiments were examined. When all first-step mutants were sampled in each experiment, a mean of 13% of mutants were Ebr^r. For in vivo-passaged mutants, of 54 obtained from 18 mouse abscesses infected with RN4220, 30 were Nov^hs while 8 (derived from 5 abscesses) were Ebr^r. Of 27 MSSA isolates (NB1721) obtained from 8 abscesses, 8 were Nov^hs while none were Ebr^r, and for 27 MRSA isolates (NB1725) derived from 10 abscesses, 6 were Nov^hs and 3 were Ebr^r. Interestingly, although first-step mutants were heterogeneous based on susceptibilities to Ebr, MICs of the three quinolones tested were equivalent for all Ebr^s and Ebr^r mutants. The data suggest that for first-step mutants, resistance may be mediated by mutations affecting either topoisomerase IV or NorA.

Quinolone resistance caused by increased expression of norA (16) is detected by cross-resistance to other NorA substrates, such as ethidium bromide and cetrimide (10). Resistance to these substrates is reduced by the NorA inhibitor, reserpine (11, 15). Among the quinolone-resistant strains in Table 2, only the Ebr^r strains showed fourfold increases in resistance to cetrimide. In our hands, reserpine did not differentiate quinolone-resistant strains that were Ebr^r from those that were Ebr^s, because it reduced resistance levels in all strains tested. For EbrR first-step mutants, the data support the argument for mediation of quinolone resistance by NorA.

TABLE 2.

Susceptibilities of parental and first-step mutant strains to selected NorA substrates

Compound(s)	MICa (μg/ml) for:
	WTb RN4220	VA3 Ebrsc	VA352 Ebrsc	VA476 Ebrrc	VA331 Ebrrc	MRSA WT NB1725	VA430 Ebrrc
Ciprofloxacin	0.5	2	2	2	1	0.5	2
Ciprofloxacin-reserpine	0.1	0.5	0.5	0.3	0.3	0.06	0.1
Norfloxacin	1	16	8	8	4	1	8
Norfloxacin-reserpine	0.3	2	2	0.5	0.5	0.1	1
Ethidium bromide	64	64	64	256	256	64	512
Ethidium bromide-reserpine	4	4	4	32	16	8	32
Cetrimide	1	1	1	4	4	1	4
Cetrimide-reserpine	0.06	0.06	0.06	0.25	0.25	0.06	0.25

^aStandard broth microdilution MICs determined with and without the addition of reserpine (20 μg/ml). 

^bWT, wild type. 

^cCfx^r mutant. 

Mutations in the flqB locus, and possibly in other loci (10), result in the increased expression of norA (25), which confers ciprofloxacin, Ebr, and cetrimide resistance (11, 15). Selected Cfx^r Ebr^r VA476 and VA331 strains were found to have mutations that mapped by genetic transformation to the flqB locus (Table 3). The data suggest a gene order as follows: fus, 1108 locus, Cfx^r Ebr^r locus. The cotransformation frequencies and gene order are similar to those reported for 1108 and flqB (4%) by Ng et al. (17). Thus, the Cfx^r Ebr^r locus maps to a location similar to that for flqB. In addition, the absence of discordance between Cfx^r and Ebr^r phenotypes in genetic crosses suggests that the same locus is responsible for both phenotypes. Mutations in the grlA/flqA locus have been shown to confer resistance to ciprofloxacin (25). Here, mutations in our Cfx^r Ebr^s strains, VA3 and VA352, mapped to the flqA locus (Table 3). The ∼50 and 11% cotransformation frequencies of (Tn917lac)2 and (Tn917lac)1 with the Cfx^r locus are consistent with the 43 and 10% cotransduction frequencies with flqA, respectively, identified by Trucksis et al. (25).

TABLE 3.

Linkage of Cfx^r Ebr^r S. aureus with (chr::Tn916)1108 and Cfx^r S. aureus with (chr::Tn917lac)2 and (chr::Tn917lac)1

Donor strain (genotype and phenotype)	Recipient		Genetic transformant
	Strain	Phenotype	Phenotype	No. in class
ISP1461a (chr::Tn916)1108 fus Cfxs Ebrs Fusr	VA476	Cfxr Ebrr Novs	Ebrs Cfxs Fuss	12
			Ebrr Cfxr Fusr	2
			Ebrr Cfxr Fuss	162
	VA331	Cfxr Ebrr Novs	Ebrs Cfxs Fuss	19
			Ebrr Cfxr Fusr	2
			Ebrr Cfxr Fuss	259
ISP2133b (chr::Tn917lac)2 Cfxs	VA3	Cfxr Novs	Cfxs	95
			Cfxr	105
	VA352	Cfxr Novs	Cfxs	104
			Cfxr	96
ISP2134c (chr::Tn917lac)1 Cfxs	VA3	Cfxr Novs	Cips	32
			Cipr	267
	VA352	Cfxr Novs	Cips	33
			Cipr	267

^aTransformants of this strain were selected for (Tn916)1108. 

^bTransformants of this strain were selected for (Tn917lac)2. 

^cTransformants of this strain were selected for (Tn917lac)1. 

We conclude that first-step quinolone-resistant MRSA and MSSA strains—whether obtained in vitro or via in vivo passage—harbor either mutations in topoisomerase IV or mutations that alter NorA multidrug efflux pump activity.