Association of norB overexpression and fluoroquinolone resistance in clinical isolates of Staphylococcus aureus from Korea

Abstract

Objectives

Although the prevalence of fluoroquinolone resistance among methicillin-resistant Staphylococcus aureus (MRSA) is known to be higher than in methicillin-susceptible S. aureus (MSSA), the reasons have never been identified.

Methods

We randomly selected 115 isolates of S. aureus collected from 10 different hospitals in Korea between June 2009 and May 2011. To investigate the difference in fluoroquinolone resistance mechanisms between MRSA and MSSA, we evaluated gyrA and parC mutations and the relative expression of the multidrug efflux pump genes norA, norB and norC.

Results

All 62 ciprofloxacin-resistant S. aureus had either gyrA or parC mutations. The S84L mutation of gyrA (59/62, 95.2%) and the S80F mutation of parC (61/62, 98.4%) were the most common. Fifty-eight (93.6%) strains had both the S84L mutation of gyrA and the S80F mutation of parC. Among the 115 isolates, norB overexpression was the most common, occurring in 49 (42.6%) strains. There were only two (1.7%) strains with norA overexpression and none with norC overexpression. Strains overexpressing norB were more common among ciprofloxacin-resistant S. aureus (33/62, 53.2%) than ciprofloxacin-susceptible S. aureus (16/53, 30.2%) (P = 0.013). When we analysed 62 ciprofloxacin-resistant S. aureus strains, those overexpressing norB were more common in ciprofloxacin-resistant MRSA (28/46, 60.9%) than in ciprofloxacin-resistant MSSA (5/16, 31.3%) (P = 0.041).

Conclusions

Increased expression of norB can be a factor that contributes to ciprofloxacin resistance in MRSA strains.

Introduction

Fluoroquinolone resistance in Staphylococcus aureus has mainly been attributed to mutations occurring in the quinolone-resistance determining region (QRDR) of parC, encoding topoisomerase IV, and gyrA, encoding DNA gyrase A.¹ However, fluoroquinolone resistance can also be mediated by the chromosomally encoded multidrug resistance (MDR) efflux pumps NorA, NorB and NorC, which are widely present in different strains and are identified based on their ability to confer resistance to quinolones.^2–4 The contributions of active efflux mechanisms to resistance is less well defined, but can be additive with target alterations. Costa et al.⁵ demonstrated that mutations in the QRDR confer resistance up to a certain level, above which resistance is mainly driven by efflux.

Ciprofloxacin resistance in methicillin-resistant S. aureus (MRSA) is common worldwide, but resistance in methicillin-susceptible S. aureus (MSSA) is substantially less common.^6–8 According to US data from 2010, the rate of ciprofloxacin resistance was only 14.3% among MSSA isolates, but >70% among MRSA isolates, with the overall prevalence of MRSA being 50.8%.⁹ This phenomenon is similar in Korea. The rate of ciprofloxacin resistance among S. aureus was only 3.3% among MSSA, but 85.1% among MRSA.¹⁰ The mechanistic reasons for this difference have not yet been elucidated.^1,6,11

Although MRSA has been regarded as a nosocomial pathogen, new strains of MRSA causing serious infections have emerged in the community in recent years.¹² These community-associated (CA) MRSA isolates are more likely to be susceptible to antimicrobial agents than healthcare-associated (HA) MRSA, but differences in the mechanisms of resistance to fluoroquinolones have never been identified.¹³

This study therefore aimed to assess and characterize the patterns of fluoroquinolone resistance mechanisms in MRSA versus MSSA strains and CA versus HA strains among clinical isolates.

Materials and methods

Bacterial isolates and data collection

We randomly selected 115 non-duplicate clinical isolates of S. aureus collected from 10 different hospitals in Korea (Korean Invasive Staphylococcus aureus Infection Study Group; KOISA) between June 2009 and May 2011. Among them, 109 (94.8%) strains were isolated from blood, 4 (3.5%) from CSF and 2 (1.7%) from pleural fluid. Methicillin or oxacillin susceptibility data and clinical information were also collected. Seventy-one strains (61.7%) were MRSA and 44 (38.3%) were MSSA. The primary sites of infection were central venous catheters (22, 19.1%), skin and soft tissue (16, 13.9%), pneumonia (16, 13.9%), bone and joint (14, 12.2%), surgical site (6, 5.2%), urinary tract (4, 3.5%), intra-abdominal (2, 1.7%), CNS (2, 1.7%) and others (2, 1.7%). In 31 (27%) isolates the primary site of infection was not identified.

CA isolates were defined as those recovered from clinical cultures within 48 h of admission in patients lacking risk factors for MRSA acquisition.¹² Established risk factors included: (i) a history of hospitalization, surgery, dialysis or residence in a long-term care facility in the year before the date of culture; (ii) the presence of a permanent indwelling catheter or percutaneous medical device at the time of culture; or (iii) isolation of MRSA in the previous 6 months. The remaining isolates were deemed to be HA. According to this definition, 41 isolates (35.7%) were CA and 74 isolates (64.3%) were HA.

Antimicrobial susceptibility tests

MICs of ciprofloxacin (Sigma-Aldrich, St Louis, MO, USA) and moxifloxacin (LKT Laboratories, Minneapolis, MN, USA) were determined using the agar dilution method in accordance with CLSI guidelines with doubling drug dilutions ranging from 128 to 0.03 mg/L.¹⁴ The breakpoints for resistance were those recommended by the CLSI guidelines. A ciprofloxacin MIC of 2 mg/L and a moxifloxacin MIC of 1 mg/L were regarded as intermediate, while a ciprofloxacin MIC of ≥4 mg/L and a moxifloxacin MIC of ≥2 mg/L were regarded as resistant. Intermediate strains were regarded as resistant.

Screening for mutations in gyrA and parC

For ciprofloxacin-resistant strains, PCR amplification of the QRDRs of gyrA and parC was carried out using the primers shown in Table 1.^15,16 All primers used in this study were synthesized at the Massachusetts General Hospital DNA Core Facility, Cambridge, MA, USA. Colonies were suspended in 50 μL of water in a microcentrifuge tube and boiled to prepare DNA templates for PCR. The PCRs were conducted by a PTC-200 Peltier thermal cycler (MJ Research, Watertown, MA, USA). The amplification conditions were the same as previously described.¹⁵ PCR-amplified DNA was purified using a PCR purification kit (Qiagen, Valencia, CA, USA) and sequenced (Tufts University Core Facility, Boston, MA, USA). QRDR DNA sequences were compared with the sequences of S. aureus NCTC 8325 (GenBank accession numbers were: CP000253 for the complete genome; ABD29197.1 for GyrA; and ABD30448.1 for ParC).

Table 1.

Primers used in this study

Gene	Primer	Primer sequence (5′→3′)	Purpose	Product size (bp)	Comment
gyrA15	gyrA-F	AATGAACAAGGTATGACACC	PCR	223
	gyrA-R	TACGCGCTTCAGTATAACGC	PCR
parC15	parC-F	ACTTGAAGATGTTTTAGGTGAT	PCR	459
	parC-R	TTAGGAAATCTTGATGGCAA	PCR
gmk17	gmk-F	TCAGGACCATCTGGAGTAGGTAAAG	qRT–PCR	108	internal control
	gmk-R	TTCACGCATTTGACGTGTTG	qRT–PCR
norA17	norA-F	GACATTTCACCAAGCCATCAA	qRT–PCR	102	target gene
	norA-R	TGCCATAAATCCACCAATCC	qRT–PCR
norB17	norB-F	GCTACACCATCAACAGATACAGCAA	qRT–PCR	117	target gene
	norB-R	ACTCAATGCGACGCCAAA	qRT–PCR
norC	norC-RT-F	TGGGTTGGAGATGGATTTTC	qRT–PCR	130	target gene
	norC-RT-R	ACAATTAGCCCTGCAACGTC	qRT–PCR
spa21	spa-F	TAAAGACGATCCTTCAGTGAGC	PCR
	spa-R	CAGCAGTAGTGCCGTTTGCTT	PCR

RNA extraction and quantitative reverse transcription-PCR (qRT-PCR)

Bacteria were grown in Luria–Bertani broth to an optical density at 600 nm (OD₆₀₀) of 0.4–0.5. The pellets were resuspended in Tris-EDTA buffer containing 100 mg/L lysostaphin (Sigma-Aldrich) and incubated for 15 min at 37°C. RNA was extracted and purified using an RNeasy Mini Kit (Qiagen) and Turbo DNase I (Ambion Inc., Austin, TX, USA) following the directions provided by the manufacturers. cDNA synthesis was performed using the Verso cDNA Kit (Thermo Fisher Scientific, Rockford, IL, USA) with random hexamers, according to the manufacturer's instructions. qPCR was carried out with SsoFast EvaGreen^® Supermix (Bio-Rad, Hercules, CA, USA), 1 μL each of 10 μM forward primer and reverse primer (Table 1)¹⁷ and cDNA template in a 20 μL volume using a CFX96 qPCR detection system (Bio-Rad, Hercules, CA, USA). The amplification conditions were as follows: 30 s at 95°C, followed by 40 cycles of 5 s at 95°C and 5 s at 55°C. Relative expression levels of norA, norB and norC were calculated using the ΔΔCt method, with the expression levels of gmk used as internal control. An increase of 4-fold or more compared with that for S. aureus SH1000 was considered indicative of overexpression as previously decribed.^18,19

Effect of efflux pump inhibitor

To investigate the effect of the NorB efflux pump on fluoroquinolone resistance, we tested five control strains: SH1000, RN6390 (with and without vector pSK950), QT5′, a norB knockout mutant (RN6390 norB::cat), and RN6390 pQT8 (pSK950-norB), which overexpresses norB from plasmid pQT8.⁴ We tested the MICs of moxifloxacin, a more selective substrate for NorB, by the broth microdilution method in the presence and absence of 20 mg/L reserpine (Sigma-Aldrich), a known inhibitor of several efflux pumps. Strains harbouring plasmid pSK950 and pQT8 were grown in the presence of tetracycline (5 mg/L) at 30°C. We also tested moxifloxacin MICs for four clinical strains with the highest level of norB transcripts (18- to 22-fold).

spa typing

To exclude the possibility of specific clonal spread, we determined S. aureus spa types as previously described.^20,21 The polymorphic X region of the protein A gene (spa) was amplified using the primers spa-F and spa-R as shown in Table 1. The amplification conditions were as follows: DNA was denatured at 94°C for 4 min, followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 60 s and extension at 72°C for 60 s, followed by a step of final extension at 72°C for 10 min. Purification and DNA sequencing were carried out using the same methods as for the gyrA and parC genes. The assignment of spa type was made employing a web-based application at http://fortinbras.us/cgi-bin/spatyper/spatyper.pl, which utilizes the Ridom numbering scheme as described at http://spa.ridom.de/index.shtml.¹⁸

Statistical analysis

The relationship between ciprofloxacin resistance and expression of efflux pump genes was assessed by the χ² test or Fisher's exact test, and a P value <0.05 (two-sided) was considered as statistically significant. All statistical analyses were done using SAS Enterprise Guide version 4.1 (SAS Institute Inc., Cary, NC, USA).

Results

Susceptibility to ciprofloxacin and moxifloxacin

Among 115 strains, 62 (53.9%) were ciprofloxacin resistant and 53 (46.1%) were ciprofloxacin susceptible. The distribution of strains according to ciprofloxacin or methicillin susceptibility and site of acquisition is detailed in Table 2. All 53 ciprofloxacin-susceptible strains were also susceptible to moxifloxacin, and 2 ciprofloxacin-resistant strains and 1 ciprofloxacin-intermediate strain were susceptible to moxifloxacin, with MICs of 0.25 and 0.125 mg/L, respectively. Among all 115 strains, the MIC₅₀ and MIC₉₀ (MICs for 50% and 90% of the isolates, respectively) of ciprofloxacin were 8-fold higher (16 and >128 mg/L, respectively) than those of moxifloxacin (2 and 16 mg/L, respectively) (data not shown). In MSSA strains, the MIC₅₀ of both ciprofloxacin and moxifloxacin was 32–64 times lower than in MRSA (0.5 and 32 mg/L versus 0.125 and 4 mg/L, respectively) (Table 3). The MIC₅₀ values in CA strains showed a similar pattern to those in HA strains (1 and 32 mg/L for ciprofloxacin, 0.125 and 4 mg/L for moxifloxacin, respectively).

Table 2.

Distribution of 115 S. aureus isolates

	CIP-R (n = 62)		CIP-S (n = 53)		Total
	MRSA	MSSA	MRSA	MSSA
CA	14	2	12	13	41
HA	32	14	13	15	74
Total	46	16	25	28	115

CIP-R, ciprofloxacin resistant; CIP-S, ciprofloxacin susceptible.

Table 3.

MICs (mg/L) of ciprofloxacin and moxifloxacin for 115 S. aureus strains

	Ciprofloxacin			Moxifloxacin
	MIC50	MIC90	range	MIC50	MIC90	range
MSSA	0.5	>128	0.25 to >128	0.125	8	≤0.03 to 128
MRSA	32	>128	0.25 to >128	4	32	0.06 to >128
CA	1	>128	0.25 to >128	0.125	8	≤0.03 to >128
HA	32	>128	0.25 to >128	4	16	≤0.03 to 128

Mutations in gyrA and parC in ciprofloxacin-resistant isolates

All 62 ciprofloxacin-resistant strains had either gyrA and/or parC mutations that caused changes in amino acids. gyrA and parC mutations were found in 59 (95.2%) and 61 (98.4%) strains, respectively. The S84L mutation of gyrA (59/62, 95.2%) and the S80F mutation of parC (61/62, 98.4%) were the most common (Table 4). Fifty-eight (93.6%) strains had both the S84L mutation for gyrA and the S80F mutation for parC. A single mutation in either gyrA or parC was only found in one strain and three strains, respectively. gyrA mutations other than S84L included S85P (9/62, 14.5%), E88K and A126T (1/62, 1.6%, respectively). parC mutations other than S80F included E84K (13/62, 21.0%), S81P (2/62, 3.2%), E84G, S108N, P114S and A116Q (1/62, 1.6%). Within gyrA, four single or combined alterations were found: S84L single, combination of both S84L and S85P or E88K, and combination of S84L, S85P and A126T. Within parC, seven single or combined alterations were found: S80F or A116Q single, and combination of S80F with one of E84K, E84G, S81P, S108N or P144S.

Table 4.

QRDR mutations of 62 ciprofloxacin-resistant S. aureus strains

Genes and sites of mutation		No. of strains, n = 62	MRSA, n = 46	MSSA, n = 16	CA, n = 16	HA, n = 46
gyrA	parC
S84L	S80F	39 (62.9%)	32 (69.6%)	7 (43.8%)	10 (62.5%)	29 (63.0%)
S84L/S85P	S80F/E84K	7 (11.3%)	5 (10.9%)	2 (12.5%)	4 (25.0%)	3 (6.5%)
S84L	S80F/E84K	4 (6.5%)	4 (8.7%)	0	0	4 (8.7%)
S84L	S80F/E84G	2 (3.2%)	1 (2.2%)	1 (6.3%)	1 (6.3%)	1 (2.2%)
S84L	S80F/S81P	1 (1.6%)	1 (2.2%)	0	0	1 (2.2%)
S84L	S80F/S108N	1 (1.6%)	0	1 (6.3%)	0	1 (2.2%)
S84L	A116Q	1 (1.6%)	0	1 (6.3%)	0	1 (2.2%)
S84L/E88K	S80F/E84K	1 (1.6%)	1 (2.2%)	0	1 (6.3%)	0
S84L/S85P	S80F/S81P	1 (1.6%)	1 (2.2%)	0	0	1 (2.2%)
S84L/S85P/A126T	S80F/E84K	1 (1.6%)	1 (2.2%)	0	0	1 (2.2%)
S84L	—	1 (1.6%)	0	1 (6.3%)	0	1 (2.2%)
—	S80F	2 (3.2%)	0	2 (12.5%)	0	2 (4.3%)
—	S80F/P144S	1 (1.6%)	0	1 (6.3%)	0	1 (2.2%)

Expression of norA, norB and norC efflux pump genes

The number of strains that overexpressed each evaluated gene compared with S. aureus SH1000 is presented in Table 5. Among 115 S. aureus strains, the norB gene was found to be the most commonly overexpressed, occurring in 49 (42.6%) strains. Only two strains (1.7%) overexpressed norA and none overexpressed norC. One of the two strains overexpressing norA was a ciprofloxacin-resistant HA MRSA, and the other was a ciprofloxacin-resistant HA MSSA. The norA-overexpressing MRSA strain also showed increased expression of norB. norB overexpression was found most often in MRSA (33/49, 67.3%). When norB-overexpressing strains were analysed according to methicillin resistance, ciprofloxacin resistance and site of acquisition, norB overexpression was significantly more common in ciprofloxacin-resistant S. aureus (33/62, 53.2%) than ciprofloxacin-susceptible S. aureus (16/53, 30.2%) (P = 0.013). However, there was no difference in norB overexpression between MRSA and MSSA (P = 0.286) and only a trend in norB overexpression in HA versus CA strains (P = 0.078). When we analysed the 62 ciprofloxacin-resistant S. aureus strains, norB overexpression was more common among ciprofloxacin-resistant MRSA (28/46, 60.9%) than ciprofloxacin-resistant MSSA (5/16, 31.3%) (P = 0.041) (Table 6). By contrast, there was no difference in the overexpression of norB between ciprofloxacin-susceptible MRSA (11/28, 39.3%) and MSSA (5/25, 20.0%) strains (P = 0.127). When 71 MRSA strains were analysed, ciprofloxacin-resistant MRSA (28/46, 60.9%) were significantly associated with norB overexpression compared with ciprofloxacin-susceptible MRSA (5/25, 20.0%) (P = 0.001). In 44 MSSA strains, ciprofloxacin-resistant MSSA (5/16, 31.3%) did not differ from ciprofloxacin-susceptible MSSA (11/28, 39.3%) in the overexpression of norB (P = 0.594). Notably for moxifloxacin, which is a more selective substrate for NorB than ciprofloxacin, similar associations were found between moxifloxacin resistance and increased norB expression in MRSA strains [28/46 (60.9%) moxifloxacin-resistant versus 5/25 (20.0%) moxifloxacin-susceptible strains, P = 0.001], but not in MSSA strains [4/13 (30.8%) moxifloxacin-resistant versus 12/31 (38.7%) moxifloxacin-susceptible strains, P = 0.738].

Table 5.

Distribution of 115 clinical isolates of S. aureus overexpressing norA or norB efflux pump genes

	MRSA, n = 71	MSSA, n = 44	P value	CIP-R, n = 62	CIP-S, n = 53	P value	CA, n = 41	HA, n = 74	P value
norA	1 (1.4%)	1 (2.3%)	1.000	2 (3.2%)	0	0.499	0	2 (2.7%)	0.537
norB	33 (46.5%)	16 (36.4%)	0.286	33 (53.2%)	16 (30.2%)	0.013	13 (31.7%)	36 (48.6%)	0.078

CIP-R, ciprofloxacin resistant; CIP-S, ciprofloxacin susceptible.

Table 6.

Comparison of strains overexpressing norB in subgroups of ciprofloxacin-resistant S. aureus and MRSA

	CIP-R S. aureus (n = 62)		P value	MRSA (n = 71)		P value
	MRSA	MSSA		CIP-R	CIP-S
No. of norB overexpressers	28/46 (60.9%)	5/16 (31.3%)	0.041	28/46 (60.9%)	5/25 (20.0%)	0.001

CIP, ciprofloxacin; R, resistant; S, susceptible.

Distribution of spa types

Among the 115 S. aureus strains, 26 spa types were found. The distribution of spa types is shown in Table 7. Overall, t2460 was the most commonly identified spa type (28, 24.3%) followed by t324 (11, 9.6%), t002 (8, 7.0%), t189 and t264 (6, 5.2%). Forty-nine norB-overexpressing strains were assigned to 17 spa types. The most prevalent spa type was t2460 (16/59, 27.1%), which was also the most common among strains not overexpressing norB (12/66, 18.2%). The majority (25/28, 89.3%) of all t2460 strains were MRSA. In the norB-overexpressing group, t189 and t264 (10.2% each) were the next most common, followed by t324 (15.2%) in the non-overexpressing group. Among the 66 strains not overexpressing norB, 18 different spa types were identified. Nine spa types were identified in both norB-overexpressing and non-overexpressing groups, and the other nine spa types were found only in the norB non-overexpressing group. For 5 norB-overexpressing and 17 non-overexpressing strains, no matched spa types were found in SpaServer. These 22 strains had diverse numbers of spa repeats, with 5 different spa repeats among 5 norB-overexpressing strains, and 15 different spa repeats among 17 norB non-overexpressing strains. A diversity of strain types was thus found in both of the two groups of strains differing in norB expression.

Table 7.

spa types of 115 S. aureus strains

spa type	norB-overexpressing strains, n (%)	norB non-overexpressing strains, n (%)	Total, n (%)
t2460	16 (32.7)	12 (18.2)	28 (24.3)
t189	5 (10.2)	1 (1.5)	6 (5.2)
t264	5 (10.2)	1 (1.5)	6 (5.2)
t002	3 (6.1)	5 (7.6)	8 (7.0)
t601	2 (4.1)	1 (1.5)	3 (2.6)
t127	1 (2.0)	1 (1.5)	2 (1.7)
t324	1 (2.0)	10 (15.2)	11 (9.6)
t377	1 (2.0)	1 (1.5)	2 (1.7)
t688	1 (2.0)	1 (1.5)	2 (1.7)
t267	2 (4.1)	—	1 (0.9)
t2167	1 (2.0)	—	1 (0.9)
t3280	1 (2.0)	—	1 (0.9)
t1151	1 (2.0)	—	2 (1.7)
t019	1 (2.0)	—	1 (0.9)
t021	1 (2.0)	—	1 (0.9)
t251	1 (2.0)	—	1 (0.9)
t304	1 (2.0)	—	1 (0.9)
t037	—	3 (4.5)	3 (2.6)
t148	—	3 (4.5)	1 (0.9)
t664	—	2 (3.0)	3 (2.6)
t286	—	1 (1.5)	1 (0.9)
t342	—	1 (1.5)	1 (0.9)
t386	—	1 (1.5)	1 (0.9)
t128	—	1 (1.5)	2 (1.7)
t901	—	1 (1.5)	1 (0.9)
t2413	—	1 (1.5)	1 (0.9)
Novela	5 (10.2)	17 (25.8)	22 (19.1)
Untypeable	—	2 (3.0)	2 (1.7)
Total	49 (100)	66 (100)	115 (100)

^aNo matched spa type in SpaServer (http://SpaServer.ridom.de); there were five different spa repeats for norB-overexpressing strains and 15 different spa repeats for non-overexpressing strains.

Relationship of norB expression and the effect of reserpine on moxifloxacin MICs

Moxifloxacin is a substrate for NorB⁴ and NorC,² but not NorA. In defined laboratory strains RN6390 (with and without vector plasmid pSK950), QT5′, a norB knockout of RN6390, and SH1000, the reference to which pump gene expression was compared, the MIC of moxifloxacin was 0.06 mg/L and unaffected by reserpine, an efflux pump inhibitor. By contrast, the MIC of moxifloxacin for RN6390 containing norB cloned in plasmid pSK950 was 0.25 mg/L and was reduced 4-fold in the presence of reserpine. In four norB-overexpressing clinical strains (numbers 21, 22, 26 and 31), the moxifloxacin MICs were 8–16 mg/L and were reduced 2- to 4-fold in the presence of reserpine. Thus, increased norB expression is associated with a reduction by reserpine of the moxifloxacin MIC and, in the absence of increased norC expression noted above, probably contributes to the moxifloxacin resistance phenotype.

Discussion

In S. aureus, the chromosomally encoded MDR efflux pumps NorA, NorB and NorC are widely present in different strains and were identified based on their ability to confer resistance to quinolones.^2–4 NorA is the most commonly studied, and its increased expression resulted in low to moderate increases in the MICs of fluoroquinolones to predispose to the appearance of high-level target-based resistance in vitro.²² It has also been shown to play a role even in quinolone-susceptible strains, since norA disruption leads to a reduction in the MICs of ciprofloxacin.²³ In this study, norB was the most commonly overexpressed MDR pump gene in clinical isolates from Korea. norA overexpression was found in only two strains (3.2%) and norC overexpression was not seen. Although a previous study described the predominance of norB overexpression (25.4%) among a collection of S. aureus bloodstream isolates, there were also similar proportions of strains overexpressing norA (22.8%) and norC (16.7%) when using the same reference strain, SH1000.¹⁹ In that study, when a single efflux pump gene was overexpressed, it was predominantly norA, whereas norB and norC were the most common when two or more efflux pump genes were overexpressed. By contrast, in our study, most clinical isolates that overexpressed efflux pump genes showed overexpression of a single gene, and only one isolate (0.9%) showed overexpression of both norA and norB genes. norC was the only gene for which no overexpression was detected among clinical isolates. Kosmidis et al.¹⁸ reported that MDR pump gene overexpression in clinical isolates varied temporally and geographically. Notably in their report, the predominance of norB-only overexpression was also seen in strains from San Francisco, USA.

Our collection of isolates represented a diverse set of primary sites of infection. Since norB has been shown to contribute to bacterial fitness in murine subcutaneous abscess,²⁴ it is interesting to note that there was a trend towards increased norB expression in the subgroup of patients with primary skin and soft tissue (including surgical site) infections. In this subgroup, 13 of 22 (59.1%) overexpressed norB relative to 36 of 93 isolates (38.7%) from other primary sites of infection (P = 0.082).

Although increased expression of norB was not confined to ciprofloxacin-resistant strains, we found that norB overexpression was more common in ciprofloxacin-resistant strains (53.2%) than ciprofloxacin-susceptible strains (30.2%) (P = 0.013). Strains overexpressing norB were predominantly MRSA (33/49, 67.3%). A similar but non-significant trend was seen in another geographic area.¹⁸ norB overexpression was more common in HA (36/74, 48.6%) than CA strains (13/41, 31.7%); however, the difference was not statistically significant (P = 0.078). We also found a higher rate of norB overexpression in ciprofloxacin-resistant MRSA (60.9%) than ciprofloxacin-resistant MSSA (31.3%) (P = 0.041) and ciprofloxacin-resistant MRSA (60.9%) than ciprofloxacin-susceptible MRSA (20.0%) (P = 0.001). Although the level of ciprofloxacin resistance indicates that multiple mutations contribute to the resistance phenotype, these results implicate norB overexpression as a contributing factor to ciprofloxacin resistance in clinical isolates of S. aureus and possibly to a higher prevalence of ciprofloxacin resistance in MRSA strains. The role of norB expression in the resistance phenotype is further supported by the finding that in defined laboratory strains and clinical strains that overexpressed norB, reserpine reduced the level of resistance to moxifloxacin, a quinolone that is a more selective substrate for NorB relative to other pumps.

Among S. aureus, the primary target for fluoroquinolones varies depending on the particular drug; first-step mutations selected for resistance to ciprofloxacin are found in parC, encoding the ParC subunit of topoisomerase IV, and first-step mutations selected with moxifloxacin occur in gyrA, encoding the A subunit of DNA gyrase.^25,26 Most clinical isolates with resistance to fluoroquinolones have had mutations in both parC and gyrA.^5,15 Notably, the substrate profile for NorB includes both ciprofloxacin and moxifloxacin, whereas that for NorA includes ciprofloxacin but not moxifloxacin.^4,27 In this study, four strains that showed a single mutation in gyrA or parC were all HA MSSA. Among these strains, three with a single parC mutation were ciprofloxacin resistant but moxifloxacin susceptible, and one of the three strains overexpressed norB. One strain with a gyrA single mutation that did not overexpress norB was resistant to both ciprofloxacin and moxifloxacin. In this study, all 62 ciprofloxacin-resistant S. aureus had either gyrA or parC mutations, and 58 (93.6%) strains had both the S84L mutation for gyrA and S80F for parC, known as common mutations in gyrA and parC.^28,29 The sites of mutations in gyrA and parC were similar to previous reports.^5,15,30,31 As all strains had either S84L mutations for gyrA and/or S80F for parC, there was no difference in QRDR mutations between MRSA and MSSA or CA and HA strains among ciprofloxacin-resistant S. aureus. The higher prevalence of fluoroquinolone resistance in MRSA relative to MSSA cannot therefore be attributed to different patterns of QRDR mutations.

To exclude the influence of clonal spread on the results of this study, we tested spa types for all 115 strains. The t2460 type was the most common among strains overexpressing norB, but it accounted for only 32.7% of the group, and it was also the most common type among the strains not overexpressing norB (18.2%). The majority (25/28, 89.3%) of all t2460 strains were MRSA. The two norA-overexpressing strains were both t002, and this finding corresponded to a previous report that showed t002 to be the most common norA-overexpressing spa type, regardless of geographic origin.¹⁸ The low level of norA overexpression seen in our study may thus reflect a more diverse set of spa types studied. The distributions of major spa types are known to differ depending on the study period and geographic origin.^18,21 t2460 is thought to be one of the major spa types in Korea on the basis of the results of this and previous studies.^32,33 In this work, however, the spa types of clinical isolates were quite diverse, as >46 types (including novel types) and t2460 were found in strains both overexpressing and not overexpressing norB. Therefore, clonal spread of a single norB-overexpressing strain cannot account for the dominance of norB overexpression relative to that of norA and norC.

Increased expression of norB is thus significantly more common in ciprofloxacin-resistant MRSA than ciprofloxacin-resistant MSSA. To our knowledge, this study is the first to investigate the mechanism of higher ciprofloxacin resistance among MRSA than MSSA. This association could not be linked to differences in QRDR mutations or to clonal spread. The ability of NorB to provide a fitness advantage in S. aureus abscesses or its ability to confer reduced susceptibility to moxifloxacin and other fluoroquinolones are properties to consider as promoting its dominance among our patient population, but further studies are needed to strengthen these associations.

Funding

This work was supported by the Inje Research and Scholarship Foundation in 2011 (to Y. G. K.) from Inje University College of Medicine, by a grant (to H. B. K. on behalf of KOISA) from the Korea Centers for Disease Control and Prevention, and by a grant R37-AI23988 (to D. C. H.) from the National Institutes of Health, US Public Health Service.

Transparency declarations

None to declare.