Identiﬁcation and Characterization of Fluoroquinolone Non-susceptible Streptococcus pyogenes Clones Harboring Tetracycline and Macrolide Resistance in Shanghai, China

analysis of the parC QRDR sequences suggested the possibility that FQ resistance may be acquired through inter-species lateral gene transfer. This study reports the emergence of macrolide, tetracycline, and ﬂuoroquinolone multidrug-resistant clones across several GAS emm types including emm 1 and emm 12, warranting continual surveillance given the extensive use of ﬂuoroquinolones in clinical use.


INTRODUCTION
Streptococcus pyogenes, ranked as one of the top 10 infectious causes of death worldwide, is responsible for more than 517,000 deaths annually (Carapetis et al., 2005). Also known as group A Streptococcus (GAS), it can cause various clinical infections, such as pharyngitis, impetigo, scarlet fever, necrotizing fasciitis, streptococcal toxic shock syndrome, and the immune mediated post-infectious manifestations of acute post-streptococcal glomerulonephritis, acute rheumatic fever and rheumatic heart disease (Walker et al., 2014). Since 2011, an unexpected pediatric scarlet fever epidemic has occurred and sustained in Hong Kong and throughout mainland China, with an incidence of 22-31 cases per 100,000 people (Chen et al., 2012;Lau et al., 2012;Tse et al., 2012;Yang et al., 2013;Davies et al., 2015). Furthermore, an ongoing scarlet fever outbreak has been reported in the United Kingdom since 2013/2014 season (Chalker et al., 2017), with an incidence of 25 cases per 100,000 people.
Penicillin and macrolides are primary antibiotic therapeutics administered for suspected S. pyogenes infections (Montes et al., 2010). Although S. pyogenes remains susceptible to penicillin, resistance to macrolides is increasing and a major cause for concern in China, with a frequency >93% (Chen et al., 2012;Yang et al., 2013). Fluoroquinolones (FQs) are an attractive alternative when patients are hypersensitive to beta-lactam antibiotics (Montes et al., 2010). In China, consumption of FQs increased by up to 2 × 10 8 standard units during 2001-2010, becoming the third most consumed antibiotic in the adult population (Van Boeckel et al., 2014). Increasing frequencies of S. pyogenes isolates with reduced susceptibility to FQs have been observed in many countries (Reinert et al., 2004;Malhotra-Kumar et al., 2005;Smeesters et al., 2009;Montes et al., 2010;Pires et al., 2010;Van Heirstraeten et al., 2012;Petrelli et al., 2014), yet data on FQ-nonsusceptible S. pyogenes in China remain scarce despite extremely high resistance to primary interventions such as macrolides.
In Gram-positive bacteria, FQ resistance is mainly mediated through point mutations within the quinolone resistancedetermining region (QRDR) of topoisomerase IV ParC and ParE and /or the topoisomerase II DNA gyrase GyrA and GyrB and efflux pump (Hooper, 2002). In S. pyogenes, FQ resistance is reported to be conferred by either parC or gyrA (Yan et al., 2000;Albertí et al., 2005;Montes et al., 2010), and develop in a stepwise manner (Malhotra-Kumar et al., 2009). At first, mutations occurring in parC can lead to low-level resistance. Then, additional mutations in gyrA can lead to high-level resistance. Mutations in the parC and gyrA QRDRs can occur not only spontaneously but also by means of horizontal gene transfer from other streptococcal species (Pletz et al., 2006;Duesberg et al., 2008;Pinho et al., 2010).
The aim of this study was to investigate the frequency, mechanism, and epidemiological association of FQ nonsusceptibility in S. pyogenes during 2011 and 2016 from Shanghai, China.

Isolate Database
In order to generate a S. pyogenes isolate database representative of the Shanghai population, carriage, and clinical isolates were prospectively collected from different geographical locations in Shanghai between June 2011 to June 2016. A total of 2,258 isolates were collected, of which 2,230 were from children ≤15 years old and 28 from adults. The database including 2,094 clinical isolates and 164 carriage isolates mostly from healthy students in schools. Clinical isolates were primarily from pediatric patients presenting with scarlet fever (n = 1,717), tonsillitis (n = 355), invasive infections (n = 1), or Henoch-Schonlein purpura (n = 1). S. pyogenes isolates were confirmed by latex-agglutination with the Diagnostic Streptococcal Grouping Kit (Oxoid, Hampshire, UK) and Vitek 2 system (bioMérieux, Marcy l'Etoile, France). All 2,258 isolates were tested for fluoroquinolone resistance.

Molecular Typing and Genomic Fingerprinting of S. pyogenes
All isolates were characterized by emm type according to the standard protocol of Centers for Disease Control and Prevention (CDC; http://www.cdc.gov/streplab/protocol-emmtype.html). Pulsed-field gel electrophoresis (PFGE), a method involving chromosomal DNA macrorestriction, was performed to describe the genomic relationships among the GAS population using restriction endonuclease SmaI (TaKaRa, Dalian, China). PFGE patterns were analyzed using BioNumerics software package (version 6.5; Applied Maths, Austin, TX, USA) under the unweighted pair group method and an arithmetic averages (UPGMA) clustering algorithm, with settings as 1.0% optimization and 1.5% band tolerance (Carrico et al., 2006) As recommended in Hong Kong scarlet fever outbreak investigation (Tse et al., 2012), isolates with 80% similarity of PFGE bands were assigned to the same cluster. In the same cluster, isolates with indistinguishable PFGE pattern (no different bands) were assigned to the same clone.

FQs Susceptibility Test
Minimum inhibitory concentration (MIC) of levofloxacin was determined and interpreted with the broth microdilution method procedure recommended by the Clinical and Laboratory Standards Institute (CLSI) guidelines published in 2015 (Clinical and Laboratory Standards Institute, 2015). According to the breakpoints in CLSI 2015, FQ susceptibility were defined as susceptible, intermediate-resistant, and resistant when the isolates with MIC to levofloxacin as ≤2µg/ml, 4µg/ml, and ≥8µg/ml. S. pyogenes FQ non-susceptibility was defined as intermediate-resistant and resistant isolates. Efflux pump activity in the FQ non-susceptible isolates was investigated by the agar dilution method according to CLSI 2015 guidelines, comparing the MICs to levofloxacin and ciprofloxacin in the presence or absence of 30 µg/ml reserpine (Sigma-Aldrich), an efflux pump blocker (Jones et al., 2003). When a FQ-non-susceptible isolate showed a four-fold or greater decrease in the MIC to levofloxacin or ciprofloxacin, the efflux phenotype was deemed to be positive, which supported the involvement of the efflux mechanism (Jones et al., 2003).

Sequencing of Quinolone Resistance-Associated Genes
The QRDR sequences of genes encoding the topoisomerase IV(parC and parE) and the gyrase (gyrA and gyrB) were amplified and sequenced as previously described (Jones et al., 2003). Sequences of parC (accession number AF220946: position 64-561), parE (accession number AE004092: 752342-754294), gyrA (accession number AF220945: 79-540) and gyrB (accession number AE004092: 581675-583627) from the quinolonesusceptible S. pyogenes strain SF370 (ATCC 700294) were defined as reference alleles (Malhotra-Kumar et al., 2005). The genome sequences of S. pyogenes isolates from Hong Kong and mainland China and the QRDR sequence of parC from 2 S. pneumonia, 7 S. agalactiae, 21 S. dysgalactiae, 1 S. canis, 1 S. iniae, 1 S. porcinus, 2 S. equi, and 1 S. difficilis isolates were retrieved from GenBank with the accession number listed in Tables S1, S2, respectively. The Lasergene software package (version 7.1; DNASTAR, Wisconsin, USA) was used to analyze nucleotide sequences and the deduced amino acid sequences. MEGA 5 (http://www.megasoftware.net/) was used to generate multi-sequence alignments and Neighbor-joining phylogenetic analysis. Clades were determined with the bootstrap values >75% in the bootstrap test with 1,000 replicates. The evolutionary distances were computed by the Kimura two-parameter method and are in the units of the number of base substitutions per site.

Detection of Other Resistance-Associated Genes
Genes associated with macrolide resistance (ermB, ermA, and mef ) and tetracycline resistance (tetM and tetO) were screened by PCR using primers and reaction parameters as previously described (Pérez-Trallero et al., 2007).

Statistical Analysis
Statistical analysis was performed using SPSS (version 20.0; IBM). Fisher's exact test was used for the comparison of FQ resistance frequency between isolates from children and adults. Statistical significance was assessed at P < 0.05.

Accession Numbers
The sequences of the parC alleles defined in this study were submitted to GenBank under accession numbers MF278797 to MF278806, while the parE QRDR sequences of the 30 FQ-nonsusceptible isolates as MG894399 to MG894428.

Ethical Aspects
All GAS isolates characterized in this study were collected from GAS infection cases and their close contacts, as part of routine clinical management of patients and Shanghai surveillance system for scarlet fever, according to guidelines of infectious diseases used in China nationwide. Informed consent from patients and close contacts was approved by Shanghai Municipal Center for Disease Control and Prevention ethical review committee (No: 2016-4) (Chen et al., 2017).

Characterization of FQ Resistance Mechanisms
To examine the genetic basis of FQ-non-susceptibility, the parC, parE, gyrA, and gyrB sequences were sequenced and analyzed.
Of the 30 FQ-non-susceptible isolates, all harbored at least one amino acid alteration in ParC, 11 (36.7%) harbored ParE mutations, and none had mutations in GyrA and GyrB QRDRs ( Table 2). In ParC, alteration at the amino acid site 79 (Ser) was found in 26 (86.7%) FQ-non-susceptible isolates, including S79A (emm6), S79F (emm1, emm11, emm12, and emm87), and S79Y (emm12). The mutation A121V was found in emm12 isolates irrespective of the FQ susceptibility. Less common mutations included D78A (1), D91N (2), S140P (2), and a novel mutation, D83V (2). A total of 9 parC nucleotide alleles were found in the 30 FQ-non-susceptible isolates, designated as SHparC1-SHparC9 ( Table 2). With the exception of SHparC6 that was represented by isolates of emm1 and emm87, each parC allele was represented by isolates of the same emm type. In ParE QRDRs, mutations FIGURE 2 | PFGE patterns of 30 fluoroquinolone-non-susceptible S. pyogenes strains isolated between 2011 to 2016 in Shanghai, China. Clusters A-F were assigned according to 80% similarity, and, isolates with indistinguishable pulsed-field gel electrophoresis (PFGE) pattern (no different bands) were assigned to the same clone. The numbers of similarity are labeled next to the branches.
of 7 amino acid sites were identified in 11 FQ-non-susceptible isolates ( Table 2) (2). To further expand on these findings, we tested the efflux phenotype on all 30 FQ-non-susceptible isolates. None of the FQ-non-susceptible isolates showed a four-fold or greater decrease in the MIC to levofloxacin or ciprofloxacin after adding reserpine (30 µg/ml) indicating that the FQ tolerance could not be explained by the efflux pump mechanism.

Conservation of ParC Mutations Within Whole Genome Sequences From Wider Chinese Clinical Isolates
One hundred and seventy S. pyogenes genome sequences of isolates from Hong Kong (152) and mainland China (18)

Evolution of parC Mutations Within Streptococcaceae
Phylogenetic analysis involving 51 QRDR parC sequences from S. pyogenes, S. pneumonia, S. agalactiae, S. dysgalactiae, S. canis, S. iniae, S. porcinus, S. equi, and S. difficilis was performed to examine evidence of lateral gene transfer and FQ resistance. All S. pyogenes and S. dysgalactiae isolates clustered together into a single clade (Clade S. pyogenes & S. dysgalactiae, Figure 3). The clade can be further divided into two sub-branches: the S. pyogenes sub-branch and the S. dysgalactiae sub-branch. Most of the S. pyogenes parC alleles (8/10) defined in this study were grouped into S. pyogenes sub-branch, while alleles SHparC8 and SHparC9 were found in the S. dysgalactiae sub-branch. Five parC sequences of S. dysgalactiae, including FQ-susceptible and FQ-non-susceptible isolates, were included in the S. pyogenes sub-branch.

DISCUSSION
This study is the first report that investigates and characterizes FQ-non-susceptibility within S. pyogenes from mainland China during 2011-2016, the period overlapping an ongoing multidrug resistant GAS scarlet fever epidemic in China (Tse et al., 2012;Yang et al., 2013;Davies et al., 2015). We identified FQ nonsusceptibility to be 1.3% in 2,258 GAS isolates from Shanghai, China, including within emm12 and emm1 scarlet fever clones resistant to macrolides and tetracycline. The appearance of these clones has further narrowed the choices of treatment for GAS infections. We observed a significant difference in FQ resistance frequencies in GAS isolated from adults (14.3%) compared to children (1.2%). This may in part be attributable to clinical practice where administration of fluoroquinolones is largely restricted to adults due to cartilage toxicity observed in children (Stahlmann et al., 1990).
Since 2006, very high frequencies (>90%) of resistance to macrolide and tetracycline in GAS isolates have been reported in mainland China (Jing et al., 2006;Chen et al., 2012;Yang et al., 2013). Coincidently, fluoroquinolone consumption has increased over the last two decades and is currently the third most consumed antibiotic in China (Van Boeckel et al., 2014). Our study identified that 80% of FQ-non-susceptible isolates were also resistant to both macrolides (ermB) and tetracycline (tetM). In previous GAS studies, only three isolates harboring this multidrug resistance profile have been reported in a Belgian study (Malhotra-Kumar et al., 2005). With the emergence of antimicrobial resistance in respiratory pathogens intrinsically linked with antibiotic consumption , the appearance of multidrug resistant clones through selective pressure is highly probable.
Compared with the prevalence of GAS FQ-non-susceptibility in Taiwan (11.1% in 2005Taiwan (11.1% in -2012, Japan (14.1% in 2010-2012), Belgium (11.4% in 2007), Spain (13% in 2005, and USA (9% in 2003) (Yan et al., 2008;Montes et al., 2010;Van Heirstraeten et al., 2012;Wajima et al., 2013;Lin et al., 2015), the overall prevalence of FQ-non-susceptible S. pyogenes in Shanghai (1.3%) and Beijing (3.4%) (Yang et al., 2013) are relatively low. This may be attributed to the limited administration of FQ to pediatric patients in China, which constitute most of isolates in this study (98.8%) and all the isolates in the Beijing study. Correspondingly, low frequencies of FQ resistance was found in the most frequently isolated GAS emm type in this study, emm12 (1.1%). The frequencies of FQ resistance in emm12 isolates varies in different countries, from 0 (0/45) in Spain to 5% (1/20) in Japan, and to 50% (33/66) in Taiwan (Montes et al., 2010;Wajima FIGURE 3 | Phylogenetic analysis of parC quinolone-resistance-determining region from different streptococcal species. Strains are shown as; species, strain number or parC allele; (emm type); country or district; and fluoroquinolone susceptibility. All fluoroquinolone-non-susceptible isolates were labeled with "R." Phylogenetic analysis was conducted in MEGA 5 using Neighbor-joining. Bootstrap values are indicated when support is >75% (from 1,000 replicates). The tree is drawn to scale, with the branch lengths being in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed by the Kimura two-parameter method and are in the units of the number of base substitutions per site. Lin et al., 2015). Based on whole genome analysis of scarlet fever isolates in Hong Kong (Davies et al., 2015), we estimate the frequency of emm12 FQ resistance in Hong Kong was 9.0% (12/134), of which 66.7% (8/12) were also resistant to macrolide and tetracycline possessing the corresponding molecular markers ermB and tetM (Table S1); however, validation of these findings through FQ susceptibility testing is required to confirm this genetic observation. The emergence of FQ resistance within ongoing scarlet fever epidemics in Hong Kong and mainland China (Luk et al., 2012;Yang et al., 2013) calls for more precaution when prescribing fluoroquinolones for suspected GAS infections.
In this study, FQ-non-susceptible emm6 isolates were discovered in China for the first time. Orscheln et al. proposed that FQ resistance is an intrinsic property of the emm6 lineage, with mutations in ParC driving FQs resistance long before the introduction of FQ antibiotics (Orscheln et al., 2005). The global dissemination of emm6 strains was suggested to be the main reason for the increased proportion of FQ-non-susceptible S. pyogenes in Belgium and Spain (Montes et al., 2010;Van Heirstraeten et al., 2012). All six emm6 isolates identified in this study were a single FQ-non-susceptible clone on the basis of PFGE, and harbored the same ParC QRDR S79A mutation (parC allele of SHparC7) with those emm6 isolates from USA and Spain (Orscheln et al., 2005;Rivera et al., 2005). In this study, the emm6 clone was also resistant to erythromycin, clindamycin, and tetracycline. This observation is mirrored for the global emm11 clone (Wajima et al., 2008(Wajima et al., , 2013Van Heirstraeten et al., 2012) with all four emm11 GAS isolated in this study representing a single FQ-non-susceptible clone harboring an S79F ParC mutation and the macrolide and tetracycline resistance markers. Collectively, monitoring for the global dissemination of these multidrug resistant GAS clones is warranted.
All FQ-non-susceptible S. pyogenes in this study showed a lowlevel resistance (MIC range, 4-16 µg/ml) and molecular analyses of the classic FQ-non-susceptibility genetic markers suggests that FQ-non-susceptibility in China is primarily linked to ParC QRDR mutations, for they all harbored ParC mutation(s), occasionally with ParE mutation(s), and none with GyrA and GyrB mutations or involvement of efflux pump, in addition, isolates with ParE mutation(s) showed similar MIC values to levofloxacin and ciprofloxacin compared to those only possessing ParC mutation(s). This supports the stepwise theory of FQ resistance: without GyrA mutation, where mutations in ParC only confer a low-level resistance (Billal et al., 2007). The most common alterations occurred in the amino acid site Ser79, a potential site to bind FQs (Laponogov et al., 2007), which is supported by findings from American and European countries (Orscheln et al., 2005;Smeesters et al., 2009). Polymorphisms of parC QRDR were found in FQ-non-susceptible isolates, and they were mostly linked with emm types in this study. This suggested most of the mutations might occur spontaneously within the strains' evolutionary process, which was validated by the phylogenetic analysis of the parC QRDR however two parC alleles may have arisen through horizontal gene transfer from other streptococcal species such as S. dysgalactiae. Some parC sequence of FQ-susceptible and FQ-non-susceptible S. dysgalactiae were also grouped into the S. pyogenes sub-branch in this study, which provided more evidence for a global gene pool shared by S. dysgalactiae and S. pyogenes (Pinho et al., 2010). A number of ParE mutations were identified in this study, including three mutations of E360D, V377L, and H380N concurrently present in a emm87 isolate and an A378T mutation identified only in emm6 isolates as documented previously (Malhotra-Kumar et al., 2005;Lin et al., 2015), while S402L mutation was only found in emm11 isolates, and mutations of D438N and D493N were only identified in emm89 isolates, which were first reported in this study. These phenomena suggest most of the ParE mutations might be evolutionary linked with emm type rather than drivers of FQ resistance.
One limitation of this study is the small sample size of isolates from adults, which might lead to a possible bias when comparing the FQ-non-susceptibility frequency between isolates from children and adults, but the appearance of adult FQ-nonsusceptible isolates in different years (2012 and 2013) supported, to some extent, the validation of the high frequency in adults. The lack of adult isolates can largely be attributed to the fact that only scarlet fever, which is primarily a disease of children, is the only notifiable disease among all the GAS infections in China (Yang et al., 2013). Another possible attribution might be that the morbidity of GAS infections is low in adults; however, more data on adult GAS infections in China is required to support this hypothesis (Chen et al., 2017). Our finding of a higher rate of FQ-non-susceptibility in isolates from adults (14.3%) is alarming and warrants further investigation at a population level.
In conclusion, the emergence of macrolide, tetracycline and fluoroquinolone-non-susceptible S. pyogenes in Shanghai, China during the period of scarlet fever epidemics was revealed in this study. Although FQ resistance was infrequent in mainland China, the presence of macrolide, tetracycline and fluoroquinolone multidrug-resistant clones across multiple S. pyogenes emm sequence types is alarming and the spread of these isolates should be monitored globally.