Epidemiology and mechanism of drug resistance of Mycoplasma pneumoniae in Beijing, China: A multicenter study

Mycoplasma pneumoniae (M. pneumoniae) is one of the most common causes of community-acquired respiratory tract infections (RTIs). We aimed to investigate the prevalence of M. pneumoniae infection, antibiotic resistance and genetic diversity of M. pneumoniae isolates across multiple centers in Beijing, China. P1 protein was detected by Nested PCR to analyze the occurrence of M. pneumoniae in pediatric patients with RTI. M. pneumoniae isolates were cultured and analyzed by Nested-PCR to determine their genotypes. Broth microdilution method was used to determine the minimum inhibitory concentration (MIC) of antibiotics. Out of 822 children with RTI admitted to 11 hospitals in Beijing, 341 (41.48%) were positive for M. pneumoniae by Nested PCR and 236 (69.21%) samples had mutations in 23S rRNA domain V. The highest proportion of M. pneumoniae positive samples was observed in school-age children (118/190; 62.11%) and in pediatric patients with pneumonia (220/389; 56.56%). Out of 341 M. pneumoniae positive samples, 99 (12.04%) isolates were successfully cultured and the MIC values were determined for 65 M. pneumoniae strains. Out of these, 57 (87.69%) strains were resistant to macrolides, and all 65 strains were sensitive to tetracyclines or quinolones. M. pneumoniae P1 type I and P1 type II strains were found in 57/65 (87.69%) and 8/65 (12.31%) of cultured isolates, respectively. Overall, we demonstrated a high prevalence of M. pneumoniae infection and high macrolide resistance of M. pneumoniae strains in Beijing. School-age children were more susceptible to M. pneumoniae, particularly the children with pneumonia. Thus, establishment of a systematic surveillance program to fully understand the epidemiology of M. pneumoniae is critical for the standardized use of antibiotics in China.

lacking the cell wall, which was first isolated from a patient with primary atypical pneumonia [1,2]. M. pneumoniae is one of the most common pathogens causing community-acquired respiratory tract infections (RTIs) and has been recognized as a worldwide cause of primary atypical pneumonia [3][4][5]. It can affect people of all ages, especially the most vulnerable groups children and adolescents [6][7][8]. Nevertheless, as currently there are no reliable and rapid diagnostic tests for the detection of M. pneumoniae, the treatment of community-acquired pneumonia is empirical in most cases [9].
In addition to mild upper RTIs (URTIs; e.g., pharyngitis and sinusitis) and severe lower RTIs (LRTIs; e.g., bronchitis Epidemiology and mechanism of drug resistance of Mycoplasma pneumoniae in Beijing, China: A multicenter study

INTRODUCTION
Mycoplasma pneumoniae (M. pneumoniae) is a small, pliable, fastidious, and highly evolved pleomorphic bacteria and pneumonia), M. pneumoniae may also cause damage to extrapulmonary systems [10,11]. M. pneumoniae infections mainly occur in preschool children, however, a recent report suggested that the incidence in infants is also high [12]. Although the symptoms are mild to moderate in most cases of M. pneumoniae infections, hospitalization is occasionally required in severe cases. In addition, multiple organ system damage may occur [13]. M. pneumoniae infection appears as a cyclic epidemic disease with intervals of four to seven years worldwide and persists for one to two years [10,14,15]. This periodicity in M. pneumoniae infection may be related to changes in the sequence of P1 adhesin, which is the main method of M. pneumoniae typing. According to the differences in the sequence of P1 adhesin gene MPN141, M. pneumoniae is divided into P1 type I and P1 type II with several subtypes [16,17]. The prevalence of M. pneumoniae subtypes differs between countries and years, and the dominance of one subtype is followed by the dominance of another M. pneumoniae subtype [18].
Although M. pneumoniae is sensitive to macrolide, tetracycline, and quinolone antibiotics, which affect the synthesis of proteins and nucleic acids, macrolides remain the first choice for the treatment of M. pneumoniae infection in children [19]. However, the phenomenon of M. pneumoniae resistance continues to increase worldwide, especially in China and Japan, where the resistance rate is reported to be over 90% [20,21]. Since 2004, tetracycline and quinolone antibiotics have been approved for use in children over eight years of age with macrolide resistance or refractory M. pneumoniae infection [22]. Therefore, with the increasing use of these two drugs in clinical practice, the drug resistance becomes an even more serious problem. Moreover, only a few studies have focused on the M. pneumoniae tetracycline or quinolone resistance in China.
Several serology-based studies were conducted in Beijing [23][24][25], however, they were all single-center studies and thus limited by their design. To the best of our knowledge, there are no multicenter studies investigating the changes in the prevalence of M. pneumoniae infections in Beijing. Therefore, in the present study, we collected samples from several major children' s hospitals in Beijing area to study the prevalence of M. pneumoniae infection and the antibiotic resistance of M. pneumoniae. In addition, we used molecular assays to identify the genotypes of isolated M. pneumoniae and to infer their potential antibiotic (macrolide, tetracycline, and quinolone) resistance mechanism. Culture of M. pneumoniae-positive samples Culture of M. pneumoniae was performed using PPLO (pleuropneumonia-like organisms) basic medium (BD-Difco, USA) containing 15% newborn calf serum, 10% fresh yeast extract (OXOID, UK), 0.4% phenol red indicator (Sigma, USA), 1% glucose, and 50000 U/100 mL penicillin, as described previously [21]. Tracheal swab specimens were inoculated on the M. pneumoniae liquid medium and maintained in an incubator at 37°C with a 5% CO 2 atmosphere. The obtained isolates were stored at -80°C until further testing.

Measurement of minimum inhibitory concentration (MIC)
The measurement of MIC of antibiotics was performed using the standard broth microdilution method (standard MIC procedure) [27]. We used six antibiotics from three classes as follows: macrolides (erythromycin, azithromycin, and josamycin), tetracyclines (tetracycline and minocycline), and quinolones (levofloxacin). Antibiotic susceptibility test was conducted to distinguish between sensitive and clinically resistant strains according to the Clinical and Laboratory Standards Institute (CLSI) criteria [28]. The breakpoints of MIC were as follows: for macrolides, MIC ≥ 1.0 μg/ml was considered as resistant, 0. M. pneumoniae genotyping and detection of resistance-related genes P1 protein was detected by Nested PCR and classified according to the differences in the sequence of P1 adhesin gene MPN141 [29]. The reaction conditions were as follows: initial denaturation at 95°C for two minutes, followed by 30 cycles of 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, with a final extension step of 72°C for two minutes. PCR primer sequences are shown in Table 1.
The resistance-associated mutation in the 23S rRNA gene of M. pneumoniae was detected by Nested PCR. The reaction conditions were as follows: initial denaturation at 95°C for two minutes, followed by 35 cycles of 95°C for one minute, 55°C for one minute, and 72°C for 100 seconds, with a final extension step of 72°C for five minutes. The identification of nucleotide sequences was performed using BLAST against the NCBI GenBank database (http:// blast.ncbi.nlm.nih.gov/Blast.cgi). The sequences of the resistance-associated genes and PCR primers are shown in Table 1.

Statistical analysis
Statistical analysis was performed using IBM SPSS Statistics for Windows, Version 19.0 (IBM Corp., Armonk, NY, USA). Quantitative data were presented as mean ± standard deviation (SD). Qualitative data were described as numbers or percentages. The normality of data distribution was tested by the Kolmogorov-Smirnov test. Qualitative variables were analyzed using Chi-squared test. Differences were considered statistically significant at p < 0.05.

Detection of M. pneumoniae in pediatric patients with RTI
The mean age of children with RTI was 4.68 ± 3.23 years (range, one month to 17 years). Out of 822 patients with RTI admitted to 11 hospitals in Beijing, 341 (41.48%) had a positive PCR for M. pneumoniae. Ninety-nine (12.04%, 99/822) M. pneumoniae-positive samples were successfully cultured. The sex of 779 patients was available. There were 417 males and 362 females. There was no statistically significant difference with regard to the M. pneumoniae positivity rates between male and female samples ( Figure 1A).
The diagnosis of 775 children was available. The majority of M. pneumoniae-positive samples were found in patients with pneumonia (56.56%, 220/389), followed by those with URTI (28.71%, 60/209) and bronchitis (26.55%, 47/177). The M. pneumoniae detection rate in pneumonia samples was significantly higher than in URTI and bronchitis samples (p < 0.001), whereas there was no significant difference between URTI and bronchitis samples (p = 0.67).

Detection of mutations in domain V of the 23S rRNA gene
The 341 M. pneumoniae-positive PCR samples were analyzed by Nested PCR. Mutations in domain V of the 23S rRNA gene were found in 236 (69.21%) samples. The predominant mutation was A2063G (199/341, 58.36%), followed by A2064G (25/341, 7.33%). Twelve (3.52%) samples had the co-mutation of A2063G and A2064G. The mutation detection rate of

M. pneumoniae detection rate in different months
We analyzed the distribution of M. pneumoniae infection during all 12 months in 2014 ( Figure 2) and found that the sample size varied from month to month, ranging from There was no statistically significant difference in the M. pneumoniae positivity rates between boys and girls (p = 0.06). School-age children had the highest M. pneumoniae positivity rate, followed by preschool children and infants, with a significant difference (p < 0.001). Patients with pneumonia had the highest M. pneumoniae positivity rate, followed by those with URTI and bronchitis, with a significant difference (p < 0.001). B) Distribution of M. pneumoniae samples with mutations in domain V of 23S rRNA genes in children. No significant difference was found in the detection rate of mutant M. pneumoniae strains between boys and girls (p = 0.82). School-age children (7-17 years old) had the highest detection rate of M. pneumoniae mutant strains, followed by preschool children (3-6 years old) and infants (<2 years old), with a significant difference (p < 0.001). The majority of mutant strains were found in pneumonia samples, followed by URTI samples and bronchitis samples, with a significant difference (p < 0.001). *p < 0.05. RTI: Respiratory tract infection; URTI: Upper RTI. B A 2 to 181 positive samples. However, it should be noted that a smaller number of samples were collected during February and March, due to the Chinese Spring Festival. During the 12-month period, the M. pneumoniae positivity rate ranged from 22% (31/141) to 100% (2/2) and the resistance rate ranged from 38% (5/13) to 100% (3/3).
In terms of the quarterly distribution, the M. pneumoniae positivity rate was relatively higher in the first (63.64%) and second quarter (65.31%) of the year, followed by the third (56.33%) and fourth quarter (30.12%). Additionally, the resistance rate was the highest in the first quarter (85.71%), followed by the fourth quarter (72.79%).
Antibiotic resistance of M. pneumoniae isolates and related gene mutations Out of 99 M. pneumoniae cultured isolates, MIC measurement was successfully performed in 65 M. pneumoniae isolates. As shown in Table 2, among these 65 M. pneumoniae isolates, 57 (87.69%) strains with A2063G mutation in domain V of the 23S rRNA gene were resistant to macrolides, and the MIC90 values were 1024 μg/ml (erythromycin), 128 μg/ml (azithromycin), and 8 μg/ml (josamycin). The other eight wildtype M. pneumoniae isolates were macrolide-sensitive strains with MIC ≤ 0.5 μg/ml. With regard to tetracyclines, all M. pneumoniae isolates were sensitive with MIC ≤ 2.0 μg/ml. Besides, all M. pneumoniae isolates were sensitive to levofloxacin with MIC 0.25 to 1.0 μg/ml.
One macrolide-resistant strain harbored a missense mutation (K27N) in L4 ribosomal protein. Eight macrolide-sensitive strains harbored the mutation M144V in L4 ribosomal protein and S170P in L22, which was consistent with the reference strain FH (Table 3). No mutation was found in the tetracycline-resistance genes (tet and 16S rRNA). The mutations in the quinolone-resistance genes are summarized in Table 4. For all 65 M. pneumoniae-cultured isolates, no mutation was found in the gyrB gene. Two strains had D149G, A533S, and C705 mutations in the gyrA gene. One strain had R454C and G652R mutations in the parC gene. Another six strains had a co-mutation in the gyrA and parC genes.

M. pneumoniae genotyping and susceptibility
M. pneumoniae P1 type I and P1 type II strains were found in 87.69% (57/65) and 12.31% (8/65) of cultured isolates, respectively. One of the P1 type I strains was sensitive to macrolides and the remaining 56 (98.25%) strains were resistant to macrolides. One P1 type II strain was resistant to macrolides and The sample size varied from month to month, ranging from 2 to 181 positive samples. During the 12-month period, the M. pneumoniae positivity rate ranged from 22% (31/141) to 100% (2/2) and the resistance rate ranged from 38% (5/13) to 100% (3/3). In terms of the quarterly distribution, the M. pneumoniae positivity rate was relatively higher in the first (63.64%) and second quarter (65.31%) of the year, followed by the third (56.33%) and fourth quarter (30.12%). Resistance rate was the highest in the first quarter (85.71%), followed by the fourth quarter (72.79%).

DISCUSSION
M. pneumoniae causes upper and lower RTIs in children and adults. Although the symptoms are mild and self-limited in the majority of patients, or the infections are asymptomatic, approximately 25% of patients will be hospitalized due to extrapulmonary complications or severe pneumonia [30]. It is generally known that the incidence and prevalence of M. pneumoniae vary in different periods and regions. Thus, the surveillance of M. pneumoniae infection is particularly important in the monitoring and prevention of acquired pneumonia. Here, we conducted for the first time a multicenter study of M. pneumoniae infection in Beijing, China. Overall, 822 tracheal swab specimens were collected from pediatric patients with RTI symptoms in 11 hospitals in Beijing. Out of 822 samples, 341 (41.48%) were positive for M. pneumoniae by PCR. Previous study showed that the positivity rate of M. pneumoniae ranged from 19.13% to 29.07% in Beijing [5,25,31]. Thus, we may conclude that M. pneumoniae infection in Beijing has a gradual upward trend. This may be due to the rise of drug-resistant strains caused by the overuse of antibiotics in recent years.
Excessive or inappropriate use of antibiotics may lead to a selective pressure for the development of antibiotic resistance [32]. This greatly increases the number of resistant strains and, consequently, the severity of clinical symptoms, making the treatment more difficult. The prevalence of M. pneumoniae infections has obvious regional differences. Ishiguro et al. reported that the macrolide resistance in different cities in Hokkaido significantly varied, from 0.0% to 100% [33]. Therefore, enhanced surveillance of drug-resistant bacteria across a country is essential for the prevention of M. pneumoniae infection. In M. pneumoniae infections, macrolide resistance has become a potential threat worldwide, especially in China. According to a national survey on antibiotic usage, the rate of macrolide use in China has been around 70% [34] and the rate of macrolide sales reached 7.9 billion in 2014. In such circumstances, the high prevalence of M. pneumoniae macrolide-resistant strains observed in our study is not unexpected. Similarly, other studies showed that more than 90% of M. pneumoniae infections in China were caused by resistant strains [35][36][37]. Based on these data, we assume that the macrolide-resistant strains have developed in the southern or northern China. On the other hand, in Western countries, the proportion of resistant bacteria has been relatively low, ranging from 1% to 10% [10,16,32,38]. This phenomenon is mainly owing to the careful use of macrolides in these areas.   For that reason, a systematic and strict surveillance program is necessary to control the use of macrolides in our country and to manage the threat of emerging resistance of M. pneumoniae to the first-line antibiotic therapy.
Fortunately, tetracyclines and quinolones remained effective against clinical M. pneumoniae isolates. However, although we did not find tetracycline-or quinolone-resistant strains in the current study, an increasing number of Mycoplasma genitalium, Mycoplasma urealyticum, and Mycoplasma hominis strains resistant to these antibiotics has been isolated [39]. The drug resistance is primarily acquired and induced by the external environment. Therefore, with the increasing use of tetracycline and quinolone substitution therapy, the risk of development of clinically resistant strains should not be ignored. Additionally, it is worth noting that all cultured positive M. pneumoniae samples in our study had A2063G mutation in domain V of the 23S rRNA gene, which has been recognized as the most prevalent and highly associated with macrolide resistance. Although these M. pneumoniae positive strains harbored the same type of mutation, the correlation with resistance is not yet clear and further investigation is warranted to explain this phenomenon. In M. pneumoniae, resistance caused by 23S rRNA gene mutation is the most common resistance mechanism, because M. pneumoniae harbors only one copy of rRNA operon [40]. According to previous studies, mutations in the 23S rRNA gene, such as A-to-G transition or A-to-C transversion at position 2063 or 2064, respectively, predominantly cause the macrolide resistance in M. pneumoniae [41]. Furthermore, mutations at position 2617 in domain V of the 23S rRNA gene are associated with lower resistance to macrolides than the mutations at position 2063 or 2064 [41].
The positivity rate of M. pneumoniae infection and detection rate of 23S rRNA domain V gene mutations differed between the quarters of 2014 year and between different age groups of patients in our study. The highest M. pneumoniae infection rate was found in the second quarter and among school-age children. Additionally, we found no significant gender-related differences in the M. pneumoniae positivity rate. These results are consistent with previous epidemiological research on M. pneumoniae in China [5,42]. According to the age-based distribution, we speculate that the epidemiology of M. pneumonia in Beijing conforms to the traditional epidemiological model of infectious disease diffusion. In this model, the patients are linked by the characteristics of the bacterial strain, indicating a common exposure or person-to-person transmission [32]. Usually, school-age children live in a relatively closed environment and are in close contact with a large number of people, leading to a rapid transmission and outbreak of pathogens.
In our study, 57 (87.69%) clinical isolates were classified as P1 type I strains and 8 (12.31%) as P1 type II strains. The vast majority of P1 type I strains were macrolide-resistant, while most of the P1 type II isolates were macrolide-sensitive strains. The regular distribution of P1 subtypes among macrolide-resistant and macrolide-sensitive M. pneumoniae strains suggests an association between P1 subtype and the macrolide susceptibility. This potential correlation should be further investigated in studies with larger sample sizes. In addition, a more thorough analysis of the M. pneumoniae genome with whole-genome sequencing might provide more evidence for the resistance mechanism.
We found A2063G mutation in domain V of the 23S rRNA gene in 57 (87.69%) M. pneumoniae isolates, which was associated with high MIC90 of macrolides. Mutations in domain V of the 23S rRNA gene are the main cause of macrolides resistance, especially those at positions 2063 and 2064. Many studies have revealed that macrolides could inhibit protein synthesis by binding to large ribosomal subunits, so mutations in 23S rRNA may lead to decreased affinity of the ribosome for drug [21]. While many reports have investigated the role of the mutation in domain V of 23S rRNA gene, no study has analyzed mutations in the ribosomal proteins L4 and L22 [36]. Although we found several M. pneumoniae isolates with mutations in L4 and L22, the relationship between L4 and L22 mutations and macrolide resistance is yet unclear. Additionally, no mutation was detected in the tetracycline and quinolone resistance-associated genes, suggesting that tetracyclines and quinolones remain effective against clinical M. pneumoniae isolates.
There are several limitations in this study. First, this study only collected data for one year. Although we compared these data to the data from previous studies in Beijing, this may not be sufficient to observe changes in the prevalence of M. pneumoniae circulating types and epidemic trends over time. Second, due to the observational, non-randomized design of the study, we collected data in a retrospective manner, meaning that the information of some pediatric patients was unavailable and the quality of the data was dependent on the accuracy of the medical record.

CONCLUSION
We investigated M. pneumoniae infections in 11 hospitals in Beijing and found that school-age children were more susceptible to this disease, particularly pediatric patients with pneumonia. We also found that M. pneumoniae P1 type I may be the main cause of the epidemic in Beijing. The rates of macrolide resistance observed in this study were more than 85%. Fortunately, tetracyclines and quinolones remain effective against M. pneumoniae isolates in Beijing. In order to fully understand the biology and epidemiology of M. pneumoniae, the establishment of a systematic surveillance program is critical.