Association of ISMav6 with the Pattern of Antibiotic Resistance in Korean Mycobacterium avium Clinical Isolates but No Relevance between Their Genotypes and Clinical Features

The aim of this study was to genetically characterize clinical isolates from patients diagnosed with Mycobacterium avium lung disease and to investigate the clinical significance. Multi-locus sequencing analysis (MLSA) and pattern of insertion sequence analysis of M. avium isolates from 92 Korean patients revealed that all isolates were M. avium subspecies hominissuis. In hsp65 sequevar analysis, codes 2, 15, and 16 were most frequently found (88/92) with similar proportions among cases additionally two isolates belonging to code N2 and an unreported code were identified, respectively. In insertion element analysis, all isolates were IS1311 positive and IS900 negative. Four of the M. avium subsp. hominissuis isolates did not harbor IS1245 and 1 of the M. avium isolates intriguingly harbored DT1, which is thought to be a M. intracellulare-specific element. M. avium subsp. hominissuis harboring ISMav6 is prevalent in Korea. No significant association between clinical manifestation and treatment response has been found in patients with the hsp65 code type and ISMav6, indicating that no specific strain/genotype among M. avium subsp. hominissuis organisms was a major source of M. avium lung disease. Interestingly, the presence of ISMav6 was correlated with greater resistance to moxifloxacin. Conclusively, the genotype of Korean M. avium subsp. hominissuis isolates is not a disease determinant responsible for lung disease and specific virulent factors of M. avium subsp. hominissuis need to be investigated further.


Introduction
A rise in the incidence of pulmonary disease caused by nontuberculous mycobacteria (NTM) has been reported worldwide [1,2]. Mycobacterium avium complex (MAC) is the most frequent etiology of NTM lung disease [3]. MAC initially included two species, M. avium and M. intracellulare. M. avium is the most clinically significant species for humans and animals within the MAC and is divided into four subspecies: M. avium subsp. avium, M. avium subsp. hominissuis, M. avium subsp. paratuberculosis, and M. avium subsp. silvaticum [4,5].
Although subspecies of M. avium in different geographic regions or populations may have different levels of virulence due to co-evolutionary processes, consequently leading to varying epidemiological dominance, most cases of M. avium human disease are due to M. avium subsp. hominissuis. Recently, lymphadenitis patients in France were found to be infected by only M. avium subsp. hominissuis among M. avium subspecies [6]. More recently, a subspecies identification analysis of M. avium clinical strains in the USA showed M. avium subsp. hominissuis to be the dominant M. avium subspecies (92.6%), followed by M. avium subsp. avium (7.4%) [7]. All German M. avium strains isolated from children and adults were identified as M. avium subsp. hominissuis [8].
Many studies have emphasized the importance of taxonomy in distinguishing species and subspecies of MAC because non-sequencing methods or 16S rRNA sequencing frequently fails to distinguish closely related species [9,10]. Multi-locus sequencing analysis (MLSA) has been suggested as the new standard method for identifying Mycobacterium species that are not well discriminated by 16S rRNA gene sequences alone [11][12][13][14].
The presence and distribution of various insertion sequences (IS) among M. avium subspecies have provided an unprecedented opportunity to define the genomic differences between M. avium subspecies as well as to develop molecular typing methods with sufficient discriminatory power to differentiate M. avium subspecies and isolates [15].
At our institution, the rpoB-PCR restriction fragment length polymorphism (RFLP) analysis [PRA] method was used for species identification and diagnosis of MAC lung disease until 2009 [16][17][18]. To gain better insight into M. avium lung disease in Korea, we used sequencingbased methods for subspecies identification and genotyping and compared clinical characteristics and treatment outcomes according to genotype. Furthermore, we investigated patterns of antibiotic resistance according to mycobacterial genotype as well as the presence or absence of ISMav6.

Study subjects
Clinical isolates from 92 patients with newly diagnosed M. avium lung disease from Jan. 2008 to Dec. 2009 at Samsung Medical Center (Seoul, Korea) were collected and stored. The data in the present study are part of an ongoing prospective observational cohort study investigating NTM lung disease (ClinicalTrials.gov Identifier: NCT00970801). The study protocol for isolates collection and genotyping analysis was approved by the institutional review board of the Samsung Medical Center (IRB approval 2008-09-016), and written informed consent was obtained from all participants. All patients met the diagnostic criteria for NTM lung disease [3]. All patients were immunocompetent and none of the patients tested positive for human immunodeficiency virus. All isolates were collected before initiating antibiotic treatment for NTM lung disease. Additionally, M. avium species initially identified by PRA based on the rpoB gene at the time of diagnosis as previously described were used for subsequent analysis. . MLSA including hsp65, rpoB, and 16S rRNA fragments was carried out using PCR primer sets as described previously [20][21][22]. The PCR products of target genes were subjected to sequence analysis. The nucleotide sequences of these genes were compared with data reported by BLAST analysis (http://www.ncbi.nlm.nih.gov) against sequences from M. avium subspecies type and related strains. M. avium subsp. avium ATCC 25291, M. avium subsp. hominissuis 104, M. avium subsp. paratuberculosis K-10, and M. avium subsp. silvaticum ATCC 49884 were used as reference strains. For phylogenetic analysis, sequences were trimmed using the CLUS-TAL-W multiple sequence alignment program [23]. Phylogenetic trees were obtained from DNA sequences utilizing the neighbor-joining method and Kimura's two parameter distance correction model with 1,000 bootstrap replications supported by MEGA 6.0 software [24].
hsp65 code analysis hsp65 code analysis was performed as previously described [25]. hsp65 gene PCR products were subjected to sequence analysis. The nucleotide sequences of the hsp65 gene were compared with data reported by BLAST analysis (http://www.ncbi.nlm.nih.gov) against the M. avium type and related strains. hsp65 codes were classified according to previously reported papers [25][26][27].

Insertion sequences element analysis
Multiplex PCR was performed to detect three target genes, IS900, IS1311, and DT1 using previously described methods [28]. A previously described primer set was used for the IS1245 insertion element [29]. The presence of ISMav6 was determined by PCR followed by sequencing analysis using a previously described primer set [26]. PCR products of insertion elements were sequenced and the existence of a specific insertion element in each strain was confirmed.  [30]. Statistical analyses were performed using SAS 9.1 (SAS Institute Inc., Cary, NC, USA) and a P-value less than 0.05 was considered statistically significant. ). Phylogenetic analysis based on the concatenated hsp65 and rpoB sequences from all isolates and from those of closely related species within the MAC showed that all isolates belong to M. avium subsp. hominissuis (Fig 2). Therefore, all isolates were identified as M. avium subsp. hominissuis using MLSA.

Subspecies identification of M. avium clinical isolates by MLSA
Distribution of hsp65 codes in M. avium subsp. hominissuis strains In total, the 92 isolates were classified into five different hsp65 sequevars. There were no isolates classified to hsp65 code 4, the M. avium subsp. avium sequevar. Four of these sequevars were well recognized as M. avium subsp. hominissuis type and clinical strains and 1 sequevar code was newly identified in this study. The new sequevar was coded N7 (following the code name given in the previous paper [N1-N3] [26] and accepted paper [N4-N6]) and two isolates were classified as code N7. The distribution of hsp65 sequevars in the 92 isolates is shown in Table 1.
The major codes were 2, 15, and 16.  Relatedness of clinical characteristics, treatment response, and drug susceptibility to hsp65 codes and presence/absence of ISMav6

Detection of insertion sequence elements
We analyzed clinical characteristics and treatment response among 3 major codes (code 2, 15, and 16). There were no significant differences in clinical features among the 3 groups (S1 and S2 Tables). We also analyzed clinical characteristics and treatment response according to the presence of ISMav6. There were no significant differences in clinical features between those with and without ISMav6 (S3 and S4 Tables). The association of genotype and the presence of ISMav6 with drug susceptibility patterns in the M. avium subsp. hominissuis isolates was evaluated for CLR and MXF. Drug susceptibility test were performed in 72 and 71 patients for CLR and MXF, respectively. None of the hsp65 codes showed trends in drug susceptibility levels (data not shown); however, the presence of ISMav6 was correlated with greater resistance to MXF (Table 2).

Discussion
In this study, clinical isolates from 92 patients previously diagnosed with M. avium lung disease over a two-year period were further analyzed. Species identification was initially performed by  a non-sequencing method and then species were re-identified using a sequencing method. Among the 92 isolates identified as M. avium by PRA at the time of diagnosis, all isolates were precisely identified as M. avium subsp. hominissuis.
ISMav6 is a novel IS recently reported in the genetic characterization of Japanese human clinical isolates [27]. In the present study, the prevalence of ISMav6 in Korean patients with M. avium lung disease was 61%. Interestingly, more clinical isolates with hsp65 code 15 harbored ISMav6 (84%, 21/25) than isolates with hsp65 code 2 and code 15 (47% and 58%, respectively). Also, both hsp65 code 15 and ISMav6 have rarely been reported in the literature except in Japan. In Germany, one M. avium subsp. hominissuis strain with hsp65 code 15 harboring ISMav6 was reported [8]. The high proportion of ISMav6 in M. avium subsp. hominissuis strains from Korea and Japan is thought to be a specific genetic feature. Thus, both hsp65 code 15 and ISMav6 may be related to the epidemiological diversity of M. avium clinical strains.
In general, DT1 is present in M. intracellulare and not present in M. avium subsp. hominissuis. One M. avium subsp. hominissuis isolates possessed DT1 in this study, which is a novel observation. Since a number of different IS elements have been described in various NTM species, species-specific IS elements have been revisited for MAC identification [28,29,35]. IS elements are mobile by nature, so there is a risk that similar elements will be found in unrelated bacteria because of mobility to or from MAC organisms. For example, natural occurrence of horizontal transfer of M. avium-specific IS1245 to M. kansasii has been reported [36]. Thus, the use of insertion sequences for species-specific markers should be more carefully conducted because it may influence molecular diagnosis and, consequently, treatment outcomes.
Kikuchi et al. reported that a variable number of tandem repeats (VNTR)-genotyping of 37 M. avium clinical isolates was associated with the progression of M. avium lung disease in Japan [37]. However, our study, which included more than 100 clinical isolates, did not identify any association between the M. avium VNTR genotype and disease progression of M. avium lung disease [38]. In the present study, disease progression was defined as when patients with M. avium lung disease require antibiotic treatment due to worsening symptoms, deteriorating chest radiograph features and microbiological findings within 2 years of diagnosis [39].
There was no difference in clinical characteristics and treatment response according to hsp65 sequevar codes and ISMav6, in agreement with previous VNTR-based observations that there was no association between the genotype and clinical characteristics of Korean patients [38]. Interestingly, the presence of ISMav6 was associated with drug resistance to MXF in this study. Tatano et al. reported an association between the VNTR genotype and susceptibility to quinolones and EMB [40]. Dvorska et al. found no relationship between IS1311 and IS1245based RFLP genotypes and drug susceptibility in MAC isolates [41]. These findings suggest that some genetic factors may influence the acquisition of drug resistance and ISMav6 may be a genetic factor associated with drug resistance. As far as we know, this is the first study to suggest an association between genotypes according to hsp65 codes and ISMav6 and clinical features with drug susceptibility. Our results indicate that specific genotypes among M. avium subsp. hominissuis organisms are not predominantly responsible for M. avium lung disease in Korea and further analysis of ISMav6 (i.e. the location of ISMav6 in the genome of M. avium isolates) will help identify relationships between genetic features and drug susceptibility.
The present study has some limitations. First, this study was conducted at a single center and was performed on a referral basis with final analysis of only a small number of Korean patients; therefore, caution should be used when attempting to generalize our findings. Second, this study was preliminary because we did not investigate the specific genes associated with drug resistance. Thus, further precise drug resistance typing of rpoB and gyrA/B with a large number of isolates will provide a better understanding of the association between M. avium subsp. hominissuis genotypes and drug resistance.
Nevertheless, to the best of our knowledge, this is the first report to investigate the link between ISMav6 and drug resistance to MXF in M. avium subsp. hominissuis strains from Korean patients. Future studies of informative and valuable genetic factors related to M. avium lung disease should be conducted in both the pathogen and host.
Supporting Information S1 Fig. rpoB sequence-based phylogenetic tree using the neighbor-joining method with Kimura's two-parameter distance correction model. Bootstrap analyses determined from 1,000 replicates are indicated at the nodes. Bar, 0.5% difference in nucleotide sequence. Gen-Bank accession numbers are given in parentheses. (TIF) S1