Prevalence and Clonal Characterization of Streptococcus pyogenes Clinical Isolates with Reduced Fluoroquinolone Susceptibility in Spain

Abstract

The aim of this study was to determine the prevalence and characteristics of non-fluoroquinolone (FQ)-susceptible Streptococcus pyogenes isolates and to study their mechanisms of resistance. We performed a prospective prevalence study with 468 isolates collected from 2005 to 2007 and a retrospective study that was based on the examination of existing data collected from 1999 to 2008. The retrospective study included data for isolates with high-level resistance (HR) to ciprofloxacin (MIC ≥ 32 μg/ml) (HR isolates) and isolates with the same emm types as those reported in the literature with low-level resistance (LR) to ciprofloxacin (MICs, 2 to 8 μg/ml) (LR isolates, n = 205). Genetic characterization of the isolates was performed by means of emm typing and multilocus sequence typing. The prevalence of LR ranged from 1.9% in 2005 to 30.8% in 2007. This increase was mainly due to the circulation of an emm6 subtype (emm6.4) that represented 77.1% of the LR isolates in 2007. Notably, another emm6 subtype, also detected in 2007 (emm6.37), showed coresistance to 14- and 15-membered macrolides mediated by the mefA gene. Only three HR isolates were detected (isolates emm68.1/ST247/T3,13,B3264, emm77/ST399/T28, and emm28/ST52/T28), and all were identified in the retrospective study. Overall, the 673 isolates represented 25 emm types. All LR isolates were clustered into two emm types: emm6 (six emm6 subtypes) and emm75. All the 156 emm6 isolates had LR, harbored the Ser79/Ala mutation in the parC gene product, and had the same sequence type (ST), ST382. Most (21/33) of the emm75 isolates had LR, showed the Ser79/Phe plus Asp91/Asn double mutation in the parC gene product, and were ST150. The Asp91/Asn mutation by itself did not confer resistance to FQs.

Streptococcus pyogenes is among the most prevalent pathogens in humans. These pathogens cause a broad spectrum of infections, ranging from streptococcal sore throat, skin, and soft tissue infections to severe invasive disease. S. pyogenes also causes severe postinfectious syndromes, such as glomerulonephritis and acute rheumatic fever (8). S. pyogenes is an important cause of morbidity and mortality worldwide.

S. pyogenes remains susceptible to penicillin, which is still the drug of choice for the treatment of many infections caused by this pathogen. Macrolides and lincosamides are the primary treatment for S. pyogenes infections in patients with hypersensitivity to beta-lactam antibiotics, and clindamycin is the first choice for the treatment of patients with life-threatening soft tissue infections, such as necrotizing fasciitis, because it halts exotoxin production (7, 24). Resistance to macrolides and lincosamides is not infrequently found in S. pyogenes (18, 20, 25), and when it is detected, fluoroquinolones (FQs) are a useful therapeutic alternative. FQs have been used with great success for the treatment of pneumococcal and staphylococcal infections in adults (5, 14).

The mechanisms of FQ resistance among Streptococcus spp. are mainly mediated by point mutations in the quinolone resistance-determining region of the bacterial topoisomerase II enzymes, namely, DNA gyrase and topoisomerase IV (16). DNA gyrase and topoisomerase IV enzymes are homologues, with each enzyme consisting of a tetramer with two subunits (two gyrA and two gyrB molecules in DNA gyrase, two parC and two parE molecules in topoisomerase IV). In the last few years, S. pyogenes isolates with low-level resistance to FQs have been reported by several authors (1, 3, 8, 11, 12, 15, 23, 27), but isolates with a high level of resistance have been detected very infrequently (2, 13, 21, 22, 26). To our knowledge, S. pyogenes isolates with low- or high-level resistance to ciprofloxacin (LR and HR isolates, respectively) without a mutation in the parC gene have not been described. Although isolates with high-level resistance to FQs (levofloxacin or ciprofloxacin MICs ≥ 16 μg/ml) are easily detected in routine laboratory work, those with reduced susceptibility frequently remain undetectable.

The serologically based M-protein typing scheme was replaced approximately 12 years ago by a highly correlated emm typing scheme based on sequence analysis of the portion of the emm gene that encodes the M-protein serospecificity of S. pyogenes. More than 200 emm types are currently listed in the online Centers for Disease Control and Prevention database (http://www.cdc.gov/ncidod/biotech/strep/strepindex.htm). Multilocus sequencing typing (MLST) is routinely used to define S. pyogenes clones and has a precision greater than that of other typing methods (6).

The aim of this study was to determine the susceptibilities of S. pyogenes isolates to FQs and their mechanisms of FQ resistance and to characterize clinical S. pyogenes clones in Spain, a country with a high percentage of S. pyogenes isolates resistant to several antimicrobial agents (4, 17-19).

MATERIALS AND METHODS

One isolate from the first 13 patients infected by S. pyogenes in Gipuzkoa, Spain, each month was selected for inclusion in a prospective 3-year (2005 to 2007) prevalence study. Only one isolate per patient was selected (13 isolates × 12 months × 3 years = 468 isolates). We also performed a retrospective study (1999 to 2008) that included HR isolates from Madrid and Gipuzkoa (MICs ≥ 32 μg/ml) detected as part of routine work in two hospital laboratories and isolates from a sample from our collection of strains with the same emm types as those reported in the literature and for which the ciprofloxacin MICs were ≥2 μg/ml (8-10, 12-14, 17, 19, 20). Most of the isolates were obtained from throat swab specimens.

The emm types were determined by sequencing the emm genes according to the guidelines of the Division of Bacterial and Mycotic Diseases, Centers for Disease Control and Prevention (available at http://www.cdc.gov/ncidod/biotech/strep/doc.htm). MLST was performed according to the recommendations provided at http://spyogenes.mlst.net.

MICs were determined and interpreted by use of the agar dilution procedure recommended by the Clinical and Laboratory Standards Institute (CLSI) (9) with Mueller-Hinton 2 agar (bioMerieux, Mercy l'Etoile, France) supplemented with horse blood (5%, vol/vol). As the CLSI has not described an interpretive standard for the MIC of ciprofloxacin for Streptococcus spp., we defined LR to be an MIC of 2 to 8 μg/ml, HR to be an MIC of ≥16 μg/ml, and ciprofloxacin sensitivity (Cip^s) to be an MIC of ≤1 μg/ml. The susceptibilities to six additional FQs (ofloxacin, norfloxacin levofloxacin, grepafloxacin, moxifloxacin, and gemifloxacin) of a sample of 89 isolates (all HR, ≥20% of each emm type showing LR and ≥10% of each emm type showing sensitivity) was studied. The antimicrobial agents were obtained as dry powders from Fluka/Sigma (Steinheim, Germany) (ofloxacin, norfloxacin, ciprofloxacin, and levofloxacin) or their respective pharmaceutical manufacturers (gemifloxacin and grepafloxacin were from GlaxoSmithKline [Hertfordshire, United Kingdom], and moxifloxacin was obtained from Bayer [West Haven, CT]). A Kirby-Bauer disk diffusion assay was performed with 10-μg norfloxacin disks. The isolates were investigated for the presence of an efflux pump that confers FQ resistance by determination of the MICs of ciprofloxacin and norfloxacin in the presence of 20 mg/liter of reserpine (Sigma, Steinheim, Germany). A fourfold decrease in the MIC in the presence of reserpine (2 dilution steps) was considered evidence of the presence of an efflux mechanism.

For the clinical isolates, the characterization, type identification, and detection of the tet and mef genes were performed as reported previously (20).

To determine the mechanisms of resistance to FQs in S. pyogenes, the quinolone resistance-determining region of parC of all isolates with ciprofloxacin MICs of ≥2 μg/ml was sequenced. The gyrA genes of all HR isolates and a sample of 17 LR isolates were sequenced. For the remaining LR isolates, the gyrA gene was studied by PCR-restriction fragment length polymorphism (PCR-RFLP) assay (16, 26). To verify that the Cip^s isolates included in this study had no mutations in the parC or the gyrA gene, a PCR-RFLP assay (16, 26) was performed with a sample of 128 isolates, and the parC genes of 15 of these isolates were sequenced for technical confirmation. The nucleotide sequences were aligned by use of an amino acid sequence of the parC gene product (GenBank accession number AF220946) and the gyrA gene product (GenBank accession number AF220945) from a wild-type strain by using the National Center for Biotechnology Information tBLASTx program (http://www.ncbi.nlm.nih.gov/BLAST/Blast.cgi).

RESULTS

Of the 468 isolates collected in the 3-year study of the prevalence of FQ susceptibility (369 isolates from children <15 years old and 99 isolates from adults), 61 (13%) showed ciprofloxacin MICs of 2 to 8 μg/ml and none of the isolates showed a ciprofloxacin MIC of >8 μg/ml. Similar percentages of LR isolates were found in children and adults (13.6% and 11.1%, respectively; P = 0.5). The percentage of LR isolates increased throughout the 3 years of the study: 1.9% (3/156) in 2005, 6.4% (10/156) in 2006, and 30.8% (48/156) in 2007. The increase was nonsignificant during the first 2 years of the study but was highly significant (P < 0.0001) between each of the first 2 years and the third year. This increase was mainly due to the circulation of an emm6 subtype (emm6.4) of sequence type 382 (ST382), which represented 77.1% (37/48) of the LR isolates recovered in 2007, most of which were isolated from children (n = 31). All emm6.4 subtype isolates were susceptible to macrolides and tetracycline.

Among the 61 LR isolates detected in the prevalence study, 7 (11.5%) showed coresistance to macrolides (1 emm75 isolate was also resistant to lincosamides). The emm6.37 subtype showing coresistance to 14- and 15-membered macrolides was first detected in 2007. None of the LR isolates showed coresistance to tetracycline. Twenty-four different emm types were detected in the prevalence study, but five types (emm4, emm6, emm11, emm12, and emm87) accounted for more than 50% (258/468) of the isolates (Table 1). All LR isolates were clustered into only two emm types: emm6 and emm75.

TABLE 1.

emm types, ciprofloxacin and norfloxacin susceptibilities, and parC and gyrA mutations in Streptococcus pyogenes isolates included in this study

emm type	Total no. of isolates (total no. in prevalence study)	No. of isolates with:			Amino acid substitution in:		Reference(s) to emm types with FQ resistance
		Ciprofloxacin MIC of ≥2 μg/ml	Norfloxacin MIC of ≥16 μg/ml	Mutation in parC	parC gene product	gyrA gene product
emm1a	35 (35)	0	0	0			9, 13
emm2	22 (22)	0	0	0			18
emm3	22 (22)	0	0	0			7, 8, 13
emm4	73 (73)	0	0	0			9, 13
emm6	156 (55)	156	156	156	Ser79/Ala	No mutation	7, 8, 9, 11, 12, 13, 18; this study
emm9	20 (2)						9
emm11	45 (45)	0	0	0			13
emm12	45 (45)	0	0	0			7, 8, 9, 13, 16, 18, 19
emm22	27 (12)	0	0	0			7, 9
emm28	47 (27)	1b	1	1	Ser79/Phe + Asp91/Asn	Ser81/Phe	7, 9, 17; this study
emm44/61	1 (1)	0	0	0			11
emm48	6 (6)	0	0	0
emm58	8 (8)	0	0	0
emm68	2 (1)	1b	1	1	Ser79/Phe	Ser81/Phe	This study
emm73	14 (0)						7
emm75	33 (14)	21	21	21	Ser79/Phe + Asp91/Asn	No mutation	7, 9, 11, 13; this study
emm77	24 (14)	1b	1	1	Ser79/Phe	Ser81/Phe	This study
emm78	9 (2)						7
emm87	40 (40)	0	0	0
emm89	22 (22)	0	0	0			16, 18
Otherc	22 (22)	0	0	0
Total	673 (468)	180	180	180

^aemm types marked in boldface are those that, in this or other studies, have been described to have a ciprofloxacin MIC of ≥2 μg/ml.

^bUnderlined emm types have a ciprofloxacin MIC of ≥32 μg/ml.

^cemm types with 5 Cip^s isolates or less (emm14.3, emm71, emm94, emm102, emm103) and 13 isolates whose emm types could not be amplified.

Throughout the last decade (1999 to 2008), three HR isolates were detected in routine laboratory work, and all showed a ciprofloxacin MIC of ≥32 μg/ml (one of them had an MIC of 32 μg/ml and the other two had MICs of 64 μg/ml). These isolates were characterized as emm68.1/ST247/T3,13,B3264, emm77/ST399/T28, and emm28/ST52/T28 and were obtained in 2001, 2005, and 2007, respectively.

The susceptibilities of Cip^s, LR, and HR isolates to seven FQs are shown in Table 2.

TABLE 2.

Susceptibilities to seven FQs of three HR isolates and a representative sample of LR isolates and Cip^s isolates

FQ	MIC (μg/ml)
	HR isolates (n = 3)	LR isolates (n = 39)			Cips isolates (n = 47)
		50%	90%	Range	50%	90%	Range
Norfloxacina	64	32	>32	16->32	4	8	4-8
Ofloxacin	64	4	8	4-8	1	2	≤0.25-4
Ciprofloxacina	32	4	8	2-8	<0.5	1	<0.5-1
Levofloxacin	32	2	2	<0.5-4	≤0.5	1	≤0.5-1
Grepafloxacin	16	2	2	<0.5-4	≤0.5	1	≤ 0.5-2
Moxifloxacin	4	0.5	0.5	<0.5-1	≤0.25	≤0.25
Gemifloxacin	1	≤0.25	≤0.25	≤0.25-0.25	≤0.25	≤0.25

^aSusceptibility in the presence of reserpine (20 mg/liter) did not decrease the MIC by more than 1 dilution for any isolate.

No changes in the MICs of ciprofloxacin and norfloxacin were observed when the medium was supplemented with reserpine. Overall, 673 isolates were studied, and 180 had ciprofloxacin MICs of ≥2 μg/ml; of these, all had mutations in the Ser79 codon of the parC gene product (Table 1). A total of 177 isolates were LR (ciprofloxacin MICs, 2 to 8 μg/ml) and belonged to emm types emm6 (156/177) and emm75 (21/177). All 156 emm6 isolates harbored the Ser79/Ala mutation. Sixty-four percent (21/33) of the emm75 isolates were LR, and all showed the Ser79/Phe plus Asp91/Asn double mutation, which differs from the amino acid substitution found in emm6 isolates. No mutations in the gyrA gene were found in the 177 LR isolates. All emm6 and emm75 isolates were susceptible to tetracycline, but 25% (39/156) of the emm6 isolates showed coresistance to macrolides. Only one LR emm75 isolate showed coresistance to macrolides (it carried the ermB gene). Among the 156 isolates of emm type emm6 (55 of which were isolated in the prevalence study), six emm subtypes were found: emm6.0, emm6.34, emm6.36, emm6.37, emm6.4, and emm6.69. All of these were related to only one T surface antigen-agglutination type, T type 6, and one MLST (ST382). All emm6.36, emm6.37, and emm6.69 isolates were erythromycin resistant but clindamycin susceptible and carried the mefA gene. The mefA gene was also detected in some isolates of the emm6.0 subtype. All 33 emm75 isolates were T8,25. Twelve of them were Cip^s isolates and 21 were LR isolates. Only one ST was detected (ST150) among the latter 21 LR isolates, while the Cip^s isolates belonged to two distinct STs (ST150 and ST49). Five of the six ST150 Cip^s isolates showed the Asp91/Asn mutation in the parC gene product.

Norfloxacin was a useful tool for screening for LR isolates, as all of these isolates showed a null inhibition zone (0 mm) around the 10-μg disk and MICs of ≥16 μg/ml. All three isolates with ciprofloxacin MICs of ≥32 μg/ml had the Ser79/Phe amino acid substitution in the parC gene product and harbored the Ser81/Phe mutation within the gyrA gene product (Table 1). These isolates were macrolide susceptible and, with the exception of the emm77/ST399/T28 strain, which contained the tetM gene (tetracycline MIC, 8 μg/ml), were also tetracycline susceptible. In isolate emm28/ST52/T28, the Asp91/Asn mutation in the parC gene product was also found.

DISCUSSION

In S. pyogenes, as in S. pneumoniae, FQ resistance develops in a stepwise fashion (10, 16). Mutations appear first in the topoisomerase IV gene, parC, leading to LR (ciprofloxacin MICs, 2 to 8 μg/ml), while HR results from additional mutations in the DNA gyrase gene, gyrA, which is associated with ciprofloxacin MICs of ≥16 μg/ml. The activity of an efflux pump contributes only minimally to FQ resistance in Streptococcus spp. (10, 28). In this study, no significant change in the FQ MICs was observed when testing was done in the presence of reserpine. All strains in this study showing a ciprofloxacin MIC of ≥2 μg/ml had a mutation in the Ser79 codon of the product of the parC subunit. In addition, none of the strains without this mutation studied showed a ciprofloxacin MIC of >1 μg/ml. Mutations can occur in all four subunits, but to our knowledge, no LR or HR S. pyogenes isolates with no mutation in parC has been described. In practice, ciprofloxacin resistance does not occur without mutations in the Ser79 codon of the product of the parC subunit. FQ susceptibility is not usually studied in S. pyogenes, and 5-μg ciprofloxacin or levofloxacin diffusion disks are not useful for the detection of LR. Consequently, LR is often undetected among S. pyogenes isolates. The retrospective study aimed to correct the poor identification of LR in routine laboratory work. Therefore, strains that were possibly resistant (those with the same emm type as the emm types of LR isolates reported in the literature) and that might have gone undetected were included in this study to be analyzed in greater depth. No LR isolate with an emm type other than emm6 or emm75 was found in this study.

Orscheln et al. (15) reported that 100% of isolates of emm type emm6 harbored a polymorphism with the Ser79/Ala mutation in the parC gene product, which inherently confers LR on these isolates. All emm6 isolates were LR. To our knowledge, to date, among LR isolates of the emm6 type, there was no specific reference to their subtypes or STs, although Yan et al. reported that two different resistant strains among the M/emm6 isolates were found in pediatric patients in the United States (27). Six different emm6 subtypes were found in this study, with one of them (emm6.69) being described for the first time in the present study. The plural composition of the emm6 type found in this study supports the original observation of Orscheln et al. (15) and suggests a longer-standing origin of resistance in these isolates.

The Asp91/Asn mutation by itself did not confer resistance to FQ, as it was detected in Cip^s isolates. Between 1999 and 2008, only three HR strains were detected in this study, and all showed mutations in parC and gyrA. The three strains belonged to distinct emm types, and apart from emm28/ST52/T28, the other two are described to be HR for the first time in the present study.

Unlike S. pneumoniae, the isolation of S. pyogenes with HR, to date, occurs infrequently throughout the world (2, 13, 21-23, 26). However, given the high prevalence of isolates with a mutation in the parC gene, in the near future this number may increase since only one new mutation in the gyrA gene is required, as recently shown by Malhotra-Kumar et al. (13).

Thirteen percent of the strains analyzed in the prevalence study showed LR. This percentage is similar to that found in other surveys, which reported frequencies of between 5.4% and 10.9% (11-13, 15, 27).

The increase in LR observed in the last year of the prevalence study was due to the increased circulation of an emm6.4/T6/ST382 clone. It is surprising that most of the LR isolates were detected in children who had not taken an FQ. In our experience, the detection of LR isolates in routine laboratory work can be aided by the use of norfloxacin, since 100% of LR isolates had norfloxacin MICs of ≥16 μg/ml and no zone of inhibition around a 10-μg disk was observed with these isolates. The marked increase in the number of isolates with LR observed in 2007 in our region shows that the mutation found in these isolates does not seem to cause a fitness cost that hinders its diffusion and serves as a reminder that continuous surveillance for S. pyogenes resistance to FQs is mandatory.