Prevalence , genotyping and macrolide resistance of Mycoplasma pneumoniae among isolates of patients with respiratory tract infections , Central Slovenia , 2006 to 2014

In this retrospective study we employed real-time polymerase chain reaction (PCR) to analyse the occurrence of Mycoplasma pneumoniae among upper and lower respiratory tract infections (RTI) in the Central Region of Slovenia between January 2006 and December 2014. We also used a culture and pyrosequencing approach to genotype strains and infer their potential macrolide resistance. Of a total 9,431 tested samples from inand out-patient with RTI, 1,255 (13%) were found to be positive by M. pneumoniae PCR. The proportion of positive samples was 19% (947/5,092) among children (≤16 years-old) and 7% (308/4,339) among adults (>16 years-old). Overall, among those PCR tested, the highest proportions of M. pneumoniae infections during the study period were observed in 2010 and 2014. In these two years, 18% (218/1,237) and 25% (721/2,844) of samples were positive respectively, indicating epidemic periods. From the 1,255 M. pneumoniae PCR-positive samples, 783 (614 from paediatric and 169 from adult patients) were successfully cultured. Of these, 40% (312/783) were constituted of strains belonging to the P1 type II genomic group, while 60% (469/783) contained strains of the P1 type I group. Two isolates comprised both P1 type I and II strains. Results of a genotype analysis by year, showed that the dominant M. pneumoniae P1 type during the 2010 epidemic was P1 type II (82% of isolates; 81/99), which was replaced by P1 type I in the 2014 epidemic (75%; 384/510). This observation could indicate that the two epidemics may have been driven by a type shift phenomenon, although both types remained present in the studied population during the assessed period of time. Only 1% of strains (7/783) were found to harbour an A2063G mutation in the 23S rRNA gene, which confers macrolide resistance, suggesting that the occurrence of M. pneumoniae macrolide resistance still seems to be sporadic in our geographic area.

M. pneumoniae incidence usually increases during epidemics, which occur at intervals of four to seven years [8][9][10].This pattern is presumably linked to the alternation of P1 adhesin types, which tend to change dominance over the course of time and, consequently, allow M. pneumoniae to elude the human host immune response [11].According to nucleotide (nt) differences in two repetitive regions (RepMP2/3 and RepMP4) in the MPN141 gene, which codes for P1 adhesin, M. pneumoniae strains can be divided into two genomic groups, called P1 type I and P1 type II [12], with several subtypes [13].
In Slovenia, M. pneumoniae infections are not notifiable and M. pneumoniae genotyping as well as antibiotic resistance testing are not yet provided as a routine service.Moreover, limited data are so far available for longitudinal studies looking into changes in the prevalence of M. pneumoniae infections, genetic diversity and antibiotic resistance.
In the only two published serology-based studies available from Slovenia, one from April 1996 to March 1997 [29] and the other from November 1999 to April 2001 [30], M. pneumoniae was frequently found as the causative agent of community-acquired pneumonia in hospitalised patients (5.7% [29] and 24.8% [30]).Based on these reports and the observation of an increased number of M. pneumoniae cases from routine laboratory testing all over Slovenia in 2010 and 2014 (data not shown), we decided to analyse the occurrence of M. pneumoniae infections in in-and out-patients with RTIs.At the end of 2005, real-time polymerase chain reaction (PCR) was introduced as the main diagnostic tool for M. pneumoniae infections in the Institute of Microbiology and Immunology at the University of Ljubljana, which covers the entire Central Slovenian Region.This allowed us to analyse tested samples from January 2006 to December 2014.We also used culture and molecular assays to further characterise the genotype of isolated strains during this period, and infer their potential macrolide resistance.

Patient samples and Mycoplasma pneumoniae testing
Our study was conducted between January 2006 and December 2014.Patients considered, were inhabitants of the Central Region of Slovenia, which on average represents 41% (843,528/2,038,281) [31]

Culture of Mycoplasma pneumoniae positive samples
Culture was performed by using Mycoplasma selective broth and agar plates (OXOID, United Kingdom (UK)) enriched with Mycoplasma supplement G or P (OXOID, UK) according to standard methods described elsewhere [32].The obtained isolates were stored at -80 °C until further testing.

Mycoplasma pneumoniae genotyping and macrolide resistance detection
Neither mutations in the 23S rRNA gene domain II, ribosomal proteins L4 and L22 [26], nor genomic or plasmid erm genes have been shown to be implicated in macrolide resistance in M. pneumoniae [33].Moreover as M. pneumoniae harbours only one copy of the 23S rRNA gene [34], molecular tests targeting the 23S rRNA gene domain V offer valuable tools for quick identification of resistant strains.Such tests have been developed, as well as some allowing rapid and reliable genotyping, and these can even be applied directly to clinical samples [11,15,[35][36][37][38][39].Recent publications by Spuesens et al. [40,41] describe a pyrosequencing approach for genotyping M. pneumoniae and determination of macrolide resistance.Based on single nt polymorphisms (SNPs) in the MPN141 gene's constant region, and on additional information from the constant region of the MPN528a gene, the method enables quick classification of strains into P1 type I and P1 type II.Macrolide resistance implicated mutations in the two regions of the 23S rRNA gene V domain can also be detected.This pyrosequencing approach was used for our study with some modifications which are further described.
Mycoplasma pneumoniae DNA was purified from 200 µL of liquid culture (OXOID, UK) using the MagNA Pure Compact automated DNA extraction system (Roche, Germany).A total of 5 µL of the 200 µL eluate was used in all PCR reactions.
All primers, reaction conditions and master mix compositions used for the amplification of M. pneumoniae MPN141, MPN528a and two parts of the 23S rRNA domain V (designated as Assay 1 -region encompassing macrolide resistance important positions 2,063, 2,064, 2,067 and Assay 2 -region encompassing macrolide resistance important position 2,617) were as described by Spuesens et al. [40], with several modifications.Briefly, for all PCR assays the number of cycles was increased to 50, with the denaturation step prolonged to 15 s and the extension step shortened to 15 s.The annealing step was 30 s with the temperature set to 55 °C for all PCR assays.The master mix was as previously published [40].The resulting biotinylated amplification products were checked for expected sizes and suitable amounts by 1.5% agarose gel electrophoresis stained with 1X SYBR Safe (Invitrogen, Germany).
All remaining amplified fragments of MPN141, MPN528a, 23S rRNA Assay 1 and Assay 2 were pyrosequenced on a PyroMarkID instrument (Biotage, Sweden) with PyroMark Gold Q96 SQA Reagents (Qiagen, Germany) using previously designed sequencing primers (TIB MOLBIOL, Germany) [40].For the detection of P1 type specific SNP, we adopted the dispensation order developed earlier [40], but we constructed our own specific nt dispensation orders that cover all so far detected mutations linked with macrolide resistance in respective portions of the M. pneumoniae 23S rRNA gene domain V.Moreover, the dispensation order for the M. pneumoniae 23S rRNA pyrosequencing Assay 1 was extended to produce 79 bases long pyrosequencing products, which were later used to additionally check the specificity of the PCR assay, by National Center for Biotechnology Information (NCBI) basic local alignment tool (BLAST) search.
Genotype and possible macrolide resistance designations were attributed using IdentyFire software (Biotage, Sweden) loaded with custom built libraries.

Mycoplasma pneumoniae 23S rRNA polymerase chain reaction Assay 1 and Assay 2 specificity
NCBI BLAST search of the primers [40]  A more detailed analysis on a year-to-year basis showed that although both P1 types remained present during the whole study period, the most observed M. pneumoniae P1 type during the 2010 epidemic was P1 type II (82%; 81/99), while in the 2014 epidemic P1 type I dominated (75%; 384/510) (Table 2).Moreover it seems that between these two epidemic years, the proportion of isolates with P1 type II decreased in favour of P1 type I (Table 2).
Seven samples (1%; 7/783), all from inpatients, were found to contain an A2063G mutation (Table 2).Six mutated strains were cultured from respiratory samples of paediatric patients and one from an adult patient.Three of the presumed macrolide resistant strains belonged to the P1 type I and four to the P1 type II.All other isolated M. pneumoniae strains had no mutation in the two sequenced areas of the 23S rRNA gene and were therefore presumed to be macrolide sensitive.

Mycoplasma pneumoniae 23S rRNA polymerase chain reaction Assay 1 and Assay 2 specificity
As predicted by the NCBI BLAST search, no tested organism except M. genitalium produced a positive PCR result in either M. pneumoniae 23S rRNA Assay 1 or Assay 2. However, after pyrosequencing the PCR product from Assay 1, the sequence of M. pneumoniae had six differences (73/79; 92% identity) distinguishing it from M. genitalium.

Discussion
From our data, we observed that in 2006,2007,2008,2009,2012  winter and spring, suggesting that in these years the occurrence of M. pneumoniae infections was most likely endemic.On the other hand, a different pattern with unusual increases in M. pneumoniae infections was seen from November 2009 to December 2011 and from May 2014 to December 2014.Based on these results, we suggest that a M. pneumoniae epidemic took place in Slovenia from November 2009 to December 2011 similarly to other European countries [42].Interestingly, our results also seem to show that another, even more extensive, outbreak of M. pneumoniae infections began in May 2014, reached its peak in November 2014, when 37% (214/580) of patients were positive for M. pneumoniae DNA and persisted at least until December 2014, when 29% (171/598) of patients were infected.Comparable results from other countries are not yet available, so we cannot say whether such a scenario is limited to Slovenia or correlates with a new European-wide epidemic.
We do not believe that the rises in the number of M. pneumoniae positive patients were due to increased awareness or testing availability, since no changes were made in the healthcare system in Slovenia during the time of the study.Additionally, the same test was used throughout and increased numbers of positives cannot therefore be ascribed to methodological changes in sensitivity.Moreover, a similar increase in the number of M. pneumoniae cases was also observed by other laboratories in Slovenia (data not shown).Finally, the number of tested patients did increase from 2006 to 2014 but most likely as a result of more patients seeking help at healthcare facilities during epidemics.

Analysis of M. pneumoniae P1 type distribution in
Central Slovenia from 2006 to 2014 shows that both P1 types were present in the studied population during the whole period.No complete type shift phenomenon, as described in a Japanese study [43], could be observed in our population but it would seem that the 2010/11 epidemic was caused mostly by M. pneumoniae P1 type II and the 2014 epidemic by M. pneumoniae P1 type I.Although we observed a large increase in M. pneumoniae cases in 2014, we cannot be sure that the reason for the new outbreak was an increase in the number of P1 type I strains.Further studies using multilocus variable-number tandem repeat analysis (MLVA) [44] are needed to assess whether the 2010/11 and 2014 epidemics were caused by a specific MLVA type or multiple types, as seen in a recent study from France and Israel [45].On the basis of P1 typing, however, it might be concluded that the prevalence of the two major M. pneumoniae genetic types in Slovenia does oscillate during time.
The results of our study suggest that macrolide-resistant M. pneumoniae strains, although represented by small numbers of isolates, are already present in Slovenian patients, which seems to be in concordance with most European studies, which also show low level of macrolide resistance in M. pneumoniae [11,14,15,17].An exception is a study from 2011, in which macrolide resistance was detected in 26% of children presenting with M. pneumoniae pneumonia or bronchitis in an Italian city [16].
We noted that all mutated strains harbour an A2063G substitution, which is recognised as the most prevalent and high level macrolide resistance mediating mutation [27].Although all strains harboured the same type of mutation, we did not find any connection among the patients from whom these were derived.The patients were from different geographical parts of Central Slovenia and the dates of sample collection were several months or even years apart.Moreover, basic molecular typing studies showed that three presumed macrolide-resistant strains belonged to P1 type I and four to P1 type II subgroup, which additionally supports a lack of connection between the strains.Studies using more accurate genetic methods (MLVA) could shed further light on the phylogeny of these strains.
To summarise, we observed two epidemics of M. pneumoniae infections.The first was, similarly to some other European countries, in 2010/11.The second epidemic seems to have started in Slovenia in 2014.We also found that M. pneumoniae P1 type II may have been the main cause of the 2010 epidemic while P1 type I may have been responsible for the new 2014

Table 1
Samples of patients with upper and lower respiratory tract infections tested by Mycoplasma pneumoniae real-time polymerase chain reaction and annual positivity rate, CentralSlovenian region, 2006Slovenian region,  -2014 (n=9,431 samples)   (n=9,431 samples) and 2013 the number of M. pneumoniae positive cases usually rose during the summer but quickly returned to low values during the autumn, Figure Rates of samples testing positive for Mycoplasma pneumoniae among samples from patients with upper and lower respiratory tract infections, Central Slovenian region, January 2006-December 2014 (n=9,431 tested samples)

Table 2
Annual number of Mycoplasma pneumoniae isolates, P1 type distribution, percentage of inferred macrolideresistant strains and detected 23S rRNA mutations, Central Slovenian region, 2006-2014 outbreak.Additionally, in concordance with the situation in Europe, we documented the presence of macrolide-resistant M. pneumoniae strains in children and in adults at very low level.