Clonal Expansion of Macrolide-Resistant Sequence Type 3 Mycoplasma pneumoniae, South Korea

To investigate the genetic background for the emergence of macrolide resistance, we characterized the genetic features of Mycoplasma pneumoniae using multilocus sequence typing. Of the 146 M. pneumoniae strains collected during the 5 consecutive outbreaks of M. pneumoniae pneumonia during 2000–2016 in South Korea, macrolide resistance increased from 0% in the first outbreak to 84.4% in the fifth. Among the 8 sequence types (STs) identified, ST3 (74.7%) was the most prevalent, followed by ST14 (15.1%). Macrolide-susceptible strains comprised 8 different STs, and all macrolide-resistant strains were ST3 (98.3%) except 1 with ST14. The proportion of macrolide-resistant strains in ST3 remained 2.2% (1/46) until the 2006–2007 outbreak and then markedly increased to 82.6% (19/23) during the 2010–2012 outbreak and 95.0% (38/40) during the 2014–2016 outbreak. The findings demonstrated that clonal expansion of ST3 M. pneumoniae was associated with the increase in macrolide resistance in South Korea.

M ycoplasma pneumoniae is one of the major causes of community-acquired pneumonia in children and adolescents (1). M. pneumoniae pneumonia develops with a gradual onset of constitutional symptoms over several days to a week (2). Although most patients may have selflimited symptoms resembling those of an upper respiratory infection, M. pneumoniae is recognized for producing a broad array of extrapulmonary manifestations that include hemolysis, rash, and joint involvement (1,3). Previous studies have established that the P1 adhesin (P1), a 170-kd surface protein located at the tiplike structure of virulent M. pneumoniae, mediates its cytadherence to the surface of respiratory epithelial cells, which is a critical step in the infection process (4).
Epidemiologic studies have shown that outbreaks of M. pneumoniae pneumonia occur every 3-7 years, varying from region to region on the basis of underlying low-grade endemic activity (5,6). Because the P1 is a major determinant of virulence, most studies have targeted the genetic variations of the p1 gene to explain specific genotypes that link to the outbreaks (7)(8)(9). However, because M. pneumoniae has a small genome size, its genomic diversity is known to be limited among strains, and any associations of specific genotypes with disease outbreaks are rarely found (7,10,11). Since the first report of a macrolide-resistant M. pneumoniae isolated from a child in Japan in 2001 (12), several countries in Asia, including South Korea, Japan, and China, have reported increased prevalence of macrolide resistance (13)(14)(15)(16). Point mutations in domain V of 23S rRNA are responsible for macrolide resistance. High antimicrobial consumption may provide selective pressure for the development of macrolide resistance (13). However, rapid dissemination of multiple clones that exhibit macrolide resistance also can lead to the increase in macrolide resistance rates in the community (17).
The multilocus sequence typing (MLST) scheme was first applied in Neisseria meningitidis and is a tool widely used for strain differentiation in many types of bacteria (18). Recently, Brown et al. developed an MLST scheme for M. pneumoniae using 8 housekeeping genes with a relatively high discriminatory ability (19). MLST has the potential to be used as a tool to characterize strains isolated during epidemic outbreaks of M. pneumoniae pneumonia and to investigate relatedness of specific genetic background to the emergence of macrolide resistance.
To clarify the genetic diversity of M. pneumoniae strains between outbreaks, we conducted an MLST analysis of M. pneumoniae detected from children in whom community-acquired pneumonia was diagnosed during 5 consecutive epidemics of M. pneumoniae pneumonia during 2000-2016 in South Korea. We also sought to find the genetic background that may explain the emergence of macrolide resistance among M. pneumoniae strains.

M. pneumoniae Strains
This study comprised M. pneumoniae strains detected from children with pneumonia at 5 hospitals during 5 consecutive outbreaks of M. pneumoniae pneumonia in South Korea: 2000Korea: , 2003Korea: -2004Korea: , 2006Korea: -2007Korea: , 2010Korea: -2012Korea: , and 2014Korea: -2016. Epidemic periods were previously defined by an interval spanning an increase of >5 cases/2 months over a 4-month period to a decrease of <5 cases/2 months over a 4-month period in the primary site of this study (6,14). M. pneumoniae pneumonia was diagnosed using the following criteria: 1) the presence of rales on auscultation or infiltration of the lung demonstrated on chest radiograph and 2) a positive PCR result for M. pneumoniae or isolation of M. pneumoniae on culture. Five hospitals participated in this study: Seoul National University Children's Hospital (Seoul), Seoul National University Bundang Hospital (Seongnam), Eulji Hospital (Seoul), Chungnam University Hospital (Daejeon), and Chonnam University Hospital (Gwangju). Two hospitals in Seoul and 1 hospital in Seongnam cover the Seoul metropolitan area, where almost half of the South Korea population resides. Daejeon is representative of central South Korea, and Gwangju represents the south. M. pneumoniae strains detected from community-acquired cases were included for further analysis. We excluded healthcare-associated infections and intrafamilial infections.

DNA Extraction from M. pneumoniae and Macrolide Resistance
We extracted DNA directly from the cultivated M. pneumoniae or from nasopharyngeal aspirates using an extraction kit (DNeasy Kit; QIAGEN, Hilden, Germany), according to the manufacturer's instructions. We amplified the p1 gene by PCR for the detection of M. pneumoniae. Starting in 2010, M. pneumoniae was cultivated using pleuropneumonia-like organism broth and agar for nasopharyngeal aspirates or pleural fluid obtained from the patient as previously described (20). The mutations responsible for macrolide resistance were confirmed by sequencing analysis of the amplified PCR products for domain V of the 23S rRNA gene. Primers MP23SV-F 5′-TAACTATAAC-GGTCCTAAGG-3′ and MP23SV-R 5′-ACACTTAGAT-GCTTTCAGCG-3′ were used, and PCR products were sequenced to identify mutations (14).

MLST Analysis and P1 Typing
We performed MLST on the M. pneumoniae DNA samples as previously described. Each allele was assigned to the 8 housekeeping genes (ppa, pgm, gyrB, gmk, glyA, atpA, arc, and adk), and a corresponding sequence type (ST) was given for each sample (19). We submitted new alleles and allelic profiles to the PubMLST database for MLST assignment (http://pubmlst.org/mpneumoniae/). We used eBURST version 3 software (http://eburst.mlst.net/) to estimate the relationships among the strains and to assign strains to a clonal complex (CC) (21). We also conducted P1 typing for the samples from 2000-2016 by sequencing 2 of the repetitive elements located in the p1 gene of M. pneumoniae genome: RepMP2/3 and RepMP4. We assigned P1 subtypes and each subtype variant by comparison with previously published data (22).

Statistical Analysis
We conducted statistical analysis using IBM SPSS Statistics for Windows version 23.0 (IBM Corp., Armonk, NY, USA). A linear-by-linear association model was used for Pearson's χ 2 test for trend analyses. We considered a p value of <0.05 as significant.

Ethics Statement
The institutional review board of Seoul National University Hospital approved the study protocol (IRB no. H-1012-007-341). Informed consent was exempted because nasopharyngeal aspirates were obtained as a standard of patient care to identify the etiologic agents of acute lower respiratory tract infections.

M. pneumoniae Strains and Macrolide Resistance
Our study comprised 146 M. pneumoniae DNA samples. Samples included for each outbreak were selected as fol- Of the study population, 56.1% were male. The mean age of children was 6.5 years; 7.4% were <2 years of age, 32.1% were 2-5 years of age, 60.5% were >5 years of age. The remaining 187 samples that were not selected for this study did not differ significantly from selected samples with respect to mean patient age and geographic region.  Table 1). The epidemic distribution of STs is shown in Figure 2. During the study period, ST3 (109 [74.7%]) was the most commonly identified ST, followed by ST14 (22 [15.1%]). ST3 was also the predominant ST found during all 5 outbreaks. A total of 3-5 STs circulated during each outbreak, and several minor STs (ST7, ST15, and ST31) that circulated in the earlier outbreaks were not found in the recent outbreaks. The distribution of ST did not differ by geographic region.
We found a correlation between the increasing proportion of macrolide resistance and the proportion of macrolide-resistant strains within ST3. All of the strains in ST3 were macrolide susceptible until    (Table 3). These data strongly suggest that a clonal expansion of ST3 M. pneumoniae is responsible for the increasing proportion of macrolide resistance.

eBURST Analysis
Two CCs were identified on eBURST analysis (Figure 3 (23,24). The whole genome of M. pneumoniae is ≈820 kb and has up to 700 coding operons. On the basis of the data of the comparative analysis of 58 strains, the genome appears to be highly conservative among the strains (1). Because of relatively low sequence variations and many repetitive elements within the genome of M. pneumoniae, epidemiologic investigations on genetic diversity have focused on the sequence variations of p1 gene. Several studies explored an association between P1 subtypes and epidemic outbreaks. The postulation was based on the idea that development of temporary immunity to 1 type by an outbreak might enable reemergence of the other (7). Despite experimental grounds and the scientific reasoning, studies that followed did not support the hypothesis. A study from Germany that examined the P1 molecular typing of 467 M. pneumoniae did not support predominance of 1 of the 2 major P1 subtypes or switching of the subtypes during the endemic situation before and during the outbreak period (10). Furthermore, Diaz et al. examined 199 M. pneumoniae samples from 17 investigations of cases, small clusters, and outbreaks that were supported by the Centers for Disease Control and Prevention (Atlanta, GA, USA) to determine the association of P1 subtypes with macrolide resistance (11). In that study, the distribution of P1 subtypes did not differ between macrolide-resistant and macrolide-susceptible M. pneumoniae strains, suggesting that an individual P1 subtype is not associated with macrolide-resistant genotype.  (25). MLVA and MLST also were adopted for studying M. pneumoniae in recent years (19,26). MLVA uses naturally occurring variations in the number of tandem repeated DNA sequences found in many different loci of the genome. MLST characterizes the isolates of microbial species using DNA sequences from internal fragments of multiple housekeeping genes. Of the 2 molecular typing methods, the discriminatory power of MLST scheme with the 8 loci was 0.784 for the collection of 57 isolates, whereas MLVA scheme was 0.633 (19,27). This finding was due partly to removal of the Mpn1 locus in MLVA scheme because of inconsistency in nomenclature and identification of repeat regions (28).  (29). A study of MLVA typing of M. pneumoniae strains isolated during 2004-2014 in Yamagata, Japan, reported that the prevalence of macrolide resistance-associated mutations in type 4572 was 59.7% (108/181), which was significantly higher than in other MLVA strains (30). The prevalence of the A2063G mutation in type 4572 strains was 0.9% (1/107) during 2004-2010 but became 83.8% (62/74) during 2011-2014. A recent study from Hong Kong reported an increased prevalence in macrolide resistance as well and identified type 4572 strain as the contributor (31). In that study, the authors reported that the macrolide resistance rate for type 4572 significantly increased from 25.0% in 2011 to 100% in 2014. In contrast to those studies, a study of 152 M. pneumoniae strains conducted by Liu et al. suggested that macrolide-resistant strains were multiclonal origin (17). The results of that study clustered 137 macrolide-resistant strains into 15 MLVA types, indicating the high rate of macrolide resistance could result from dissemination of the multiple resistant clones. This conflicting result might have resulted from the 5 loci MLVA, which applied the earlier MLVA method, including the unstable Mpn1 locus (26).
We cannot, at this point, answer with confidence why an ST3 strain became the most prevalent strain among macrolide-resistant M. pneumoniae. We can, however, speculate. Mutation or some other mechanism could have caused the previously macrolide-susceptible ST3 strains to become macrolide resistant, and the new strain could have developed an ability to disseminate through high-density populations. Antimicrobial selective pressure could have aided this development. An alternative possibility is that the macrolide-resistant strains were introduced to and spread rapidly through the community. Our data demonstrate that ST3 and ST14 are not genetically related; they share 1 of 8 allelic loci and differ in P1 subtypes. Analysis with eBURST shows they exist in different clonal complexes. Further research with whole-genome sequencing can reveal the distinguishing characteristics of macrolideresistant and -susceptible strains within ST3 strains (32,33). In addition, whole-genome sequencing may reveal how the macrolide-resistant ST3 became predominant.
Antimicrobial pressure would have played a role to some extent because South Korea is a high antimicrobial drug use country. Trend analyses of the national data on antimicrobial drug consumption (expressed in defined daily   (19). Our study also shows a relationship between P1 subtype 1 and CC1 (ST1 and ST3), which is consistent with their report. This relationship could have occurred by acquisition of 23S rRNA mutation within certain STs from predisposing genetic factors or by an introduction of a macrolide-resistant strain. On the basis of the studies we have described and our own results, clonal expansion of certain molecular types, 4572 in MLVA and ST3 in MLST, is likely to be the reason for the marked increase in macrolide resistance.
Our study has several limitations. First, even though M. pneumoniae strains were collected from 5 hospitals, the data might not represent reality nationwide. However, this was the best possible multicenter-based study we could devise, given that no nationwide surveillance system exists. Second, despite our best efforts, the numbers of strains in the earlier outbreaks were smaller than those collected in the later outbreaks, when culture was performed prospectively. However, this study is of value because the distribution of STs in a certain region for a relatively long period of time is well described. The finding that an expansion of a single ST contributed to the increase in macrolide resistance is a potentially powerful insight for further research. Additional studies that investigate the epidemiology and mechanism of acquiring macrolide resistance will give further insight into better treatment strategies. In particular, further studies should be addressed to in silico methods for the analysis of genetic background that can explain macrolide resistance within ST3 strains. Recent advances in microbiology have made whole-genome sequencing a valuable investigation tool that can lead to the identification of a specific genotype associated with macrolide resistance or virulence of M. pneumoniae.
In summary, we found that, during outbreaks of M. pneumoniae pneumonia that showed substantial increase in macrolide resistance in South Korea, all but 1 macrolideresistant strain was ST3. These findings demonstrate that clonal expansion of an ST3 M. pneumoniae was associated with the increase in macrolide resistance in South Korea.