New Mycobacteroides abscessus subsp. massiliense strains with recombinant hsp65 gene laterally transferred from Mycobacteroides abscessus subsp. abscessus: Potential for misidentification of M. abscessus strains with the hsp65-based method

It has been reported that lateral gene transfer (LGT) events among Mycobacteroides abscessus strains are prevalent. The hsp65 gene, a chronometer gene for bacterial phylogenetic analysis, is resistant to LGT events, particularly among mycobacterial strains, rendering the hsp65-targeting method the most widely used method for mycobacterial detection. To determine the prevalence of M. abscessus strains that are subject to hsp65 LGT, we applied rpoB typing to 100 clinically isolated Korean strains of M. abscessus that had been identified by hsp65 sequence analysis. The analysis indicated the presence of 2 rough strains, showing a discrepancy between the 2 typing methods. MLST analysis based on the partial sequencing of seven housekeeping genes, erm(41) PCR and further hsp65 PCR-restriction enzyme and polymorphism analysis (PRA) were conducted to identify the two strains. The MLST results showed that the two strains belong to M. abscessus subsp. massiliense and not to M. abscessus subsp. abscessus, as indicated by the rpoB-based analysis, suggesting that their hsp65 genes are subject to LGT from M. abscessus subsp. abscessus. Further analysis of these strains using the hsp65 PRA method indicated that these strains possess a PRA pattern identical to that of M. abscessus subsp. abscessus and distinct from that of M. abscessus subsp. massiliense. In conclusion, we identified two M. abscessus subsp. massiliense rough strains from Korean patients with hsp65 genes that might be laterally transferred from M. abscessus subsp. abscessus. To the best of our knowledge, this is the first demonstration of possible LGT events associated with the hsp65 gene in mycobacteria. Our results also suggest that there is the potential for misidentification when the hsp65-based protocol is used for mycobacterial identification.


Introduction
Rapidly growing mycobacteria (RGM) are ubiquitous organisms that have gained increasing attention as important human pathogens [1,2]. Among RGMs, infections due to the Mycobacteroides abscessus strains have shown increased worldwide clinical importance, and their incidence in cystic fibrosis patients has increased [3]. In South Korea, the incidence of lung diseases caused by M. abscessus has also been increasing, and this organism accounts for 70-80% of the lung disease caused by RGM [4][5][6][7]. M. abscessus can cause lung disease in immunocompetent individuals and shares a number of characteristics with M. tuberculosis, including the ability to induce granulomatous lesions or caseous necrosis [8]. Infections involving M. abscessus are notorious for being difficult to treat due both to the natural broad-spectrum antibiotic resistance of this species and to its acquired resistance, with disparate antibiotic susceptibility patterns being observed among clinical strains [9].
The taxonomy of M. abscessus strains remains problematic. Currently, these strains are divided into two subspecies, M. abscessus subsp. abscessus (the former species Mycobacteroides abscessus) and M. abscessus subsp. bolletii. M. abscessus subsp. bolletii was proposed to combine the two former species, M. massiliense and M. bolletii [10,11]. M. massiliense can be further subdivided into two genotypes (Type I and Type II) based on hsp65 sequence analysis [12][13][14]. Also, recent phylogenomics and comparative genome analyses on 150 genomes of Mycobacterium species revealed that the genus of Mycobacterium was divided four novel genera. In the case of M. abscessus-chelonae complex, their genus was emended into Mycobacteroides [15].
Lateral gene transfer (LGT) has been proposed as the major driving force for the acquisition of prokaryotic genetic diversity, an attribution that leads to better survival of prokaryotic organisms under harsh environmental conditions [16,17]. The recent increase in available information on mycobacterial genomes supports the idea that LGT plays an important role in the evolutionary transition of mycobacteria from saprophytic organisms into opportunistic or specialized, highly persisting pathogens [18,19]. In particular, interspecies or intraspecies LGT events among members of the M. abscessus strains have been reported to be very prevalent [20]. And it has reported that M. abscessus evolution is sporadically punctuated by dramatic genome wide remodelling events [21]. We recently identified six M. abscessus subsp. massiliense strains isolated from Korean patients in which the rpoB gene was laterally transferred from M. abscessus subsp. abscessus [22], suggesting the potential for misidentification when the rpoB tying method is used to the differentiate among M. abscessus strains. In addition, other reports have presented evidence for LGT events involving the rpoB gene, including the observation that the rpoBC operon of the Type I genotype of M. yongoenense has been laterally transferred from a distantly related strain, Mycobacterium parascrofulaceum [23].
The hsp65 (groEL2) gene, another chronometer molecule, has been widely used as a targeting molecule for mycobacterial identification and detection [24,25] together with the rpoB gene [26,27]. In this study, we sought to address issues associated with the possibility that LGT events involving the hsp65 gene occur among members of the M. abscessus strains. To this end, we applied rpoB typing (711 bp) to 100 clinically isolated Korean strains of M. abscessus that had already been identified by hsp65 sequence analysis (603 bp). In the strains in which the two methods yielded discordant results, additional phylogenetic analysis was conducted to confirm the authenticity of potential LGT events in their hsp65 genes.

Mycobacterial strains and culture conditions
Of the total 206 clinical isolated strains, 106 strains were used in the previous paper. These strains were identified as M. massiliense using the hsp65-based method. Among these 106 strains, 6 strains showed different rpoB sequence from M. massiliense, however, these strains showed similar sequence homologies with M. massiliense. So, the possibility of lateral gene transfer in the rpoB gene of these strains were described in the previous study [22].
We used other 100 strains of the total 206 strains in this study which were identified as M. abscessus using the hsp65-based method [12] (Table 1). These M. abscessus strains included both rough and smooth morphotypes (28 and 72 strains, respectively). Among them, two strains, 55262 and 55184 showed different rpoB gene sequence from M. abscessus. Although both studies used samples collected at the same time period and applied similar experimental methods, the two studies were considered to be different, because different groups of samples (which were divided by hsp65-based method) were used and putative lateral gene transfer events were considered to be applied to different genes, respectively. All

DNA extraction, PCR and sequencing
Bacterial DNA was extracted from individual clinical isolates using the bead beater-phenol extraction method [24] and used as templates for PCR amplification. Partial hsp65 (603 bp) and rpoB (711 bp) gene-targeted PCR was conducted in a total of 100 M. abscessus strains as described previously [12,24]. MLST analyses targeting seven housekeeping genes were applied to investigate the genetic diversity of the two putative recombinant strains. The seven target genes were argH (argininosuccinate lyase), cya (adenylate cyclase), glpK (glycerol kinase), gnd (6-phosphogluconate dehydrogenase), murC (UDP N-acetylmuramate-L-alanine ligase), pta (phosphate acetyltransferase) and purH (phosphoribosylaminoimidazole carboxylase ATPase subunit) [28,29]. Also, erm(41)-targeted PCR was applied to the reference strain and to two Rec-mas-H strains [30,31]. Additionally, to confirm the putative recombination site of the Rec-mas-H strain within the hsp65 gene sequence, the entire hsp65 gene sequence of Asan 55262 was sequenced and compared with the hsp65 gene sequences of the M. abscessus subsp. abscessus CIP 104536 T , M. abscessus subsp. massiliense CIP 108297 T and M. abscessus subsp. massiliense Asan 50594 strains. The complete hsp65 gene sequence of each selected isolate was amplified using 5 primer sets. Detailed information on the primers is provided in S1 Table. All the PCR reactions were conducted as described previously [22].

Nucleotide sequence accession numbers
The hsp65, rpoB and 7 MLST gene sequences determined in this study were deposited in Gen-Bank under the accession numbers MH430895-MH430913 and are listed in S2 Table. Results

Phylogenetic analysis of two Rec-mas-H strains based on seven different MLST genes
For precise species delineation of the two strains, MLST analyses based on the partial sequencing of seven housekeeping genes, argH, cya, glpK, gnd, murC, pta and purH, that have been previously used for the elucidation of recombination events in M. abscessus strains [29] were also performed in this study. Each single-gene-based tree was built from the sequences of each of the seven genes in the MLST scheme for the separation of the two strains at the subspecies level of the M. abscessus strains (Fig 2). With the exception of the glpK gene, which has a 1-bp difference in the sequence, the sequences of all the MLST genes were identical in the two Recmas-H strains. Most of the constructed phylogenetic trees (those for the argH, cya, glpK, gnd, murC, and pta genes) showed a topology similar to that of the rpoB gene sequence-based tree. With the exception of the purH tree, all of the MLST gene-based trees (argH, cya, glpK, gnd, murC and pta genes) indicated that the two strains were closely related to the M. abscessus subsp. massiliense group, and these topologies were strongly supported by analysis based on  the maximum parsimony algorithm (Fig 2). The calculated sequence similarities of all 9 genes (7 MLST genes, the hsp65 gene and the rpoB gene) and their concatenated sequences between the reference strains of the M. abscessus strains and two Rec-mas-H strains are shown in Table 2 and S3 Table.

Phylogenetic analysis of two Rec-mas-H strains using trees based on concatenated sequences
The phylogenetic tree based on the concatenated sequences of the seven MLST genes showed that the two Rec-mas-H strains belong to the M. abscessus subsp. massiliense group and not to M. abscessus subsp. abscessus or M. abscessus subsp. bolletii (Fig 3A) as shown in the rpoBbased analysis (Fig 1B). This suggests that the Rec-mas-H strains may be members of M. abscessus subsp. massiliense rather than M. abscessus subsp. abscessus and that their hsp65 gene may have undergone lateral gene transfer from M. abscessus subsp. abscessus. In the tree based on the concatenated sequences of the seven MLST genes, the two Rec-mas-H strains were closer to the M. abscessus subsp. massiliense Asan 50594 strain than to the M. abscessus subsp. massiliense type strain; this result may be due to the differences in the purH gene sequence ( Fig  2G, Table 2 and S3 Table).
Addition of the hsp65 and/or rpoB gene sequences to the 7 MLST concatenated gene sequences did not affect the overall topology of the tree obtained using the 7 MLST concatenated sequences (Fig 3B and 3C). However, the addition of the hsp65 gene sequence slightly affected the sequence similarity between the two Rec-mas-H strains and the M. abscessus subsp. abscessus and M. abscessus subsp. massiliense group strains. The sequence similarity with M. abscessus subsp. abscessus was slightly increased (from 97.6-97.7% to 97.9%), and the similarity with M. abscessus subsp. massiliense group strains was slightly decreased (from 99.5-99.6% to 99.4%) after addition of the hsp65 gene sequence ( Table 2).

Separation of 2 Rec-mas-H strains by erm(41) PCR at the subspecies level
The M. abscessus subsp. massiliense erm (41) gene is reported to have a large C-terminal deletion. Therefore, erm(41) PCR can be used as a simple method to differentiate M. abscessus  [30,31]. To further confirm the authenticity of the two Rec-mas-H strains, we applied erm (41) PCR; unlike the M. abscessus subsp. abscessus and M. abscessus subsp. bolletii type strains, which produced a full-sized product (approximately 700 bp), the two Rec-mas-H strains produced a shorter product (approximately 350 bp) that was identical to the product of M. abscessus subsp. massiliense type strain (Fig 4).

Differentiation of 2 Rec-mas-H strains from M. abscessus subsp. massiliense based on restriction patterns of the partial hsp65 gene sequence
PCR-based restriction enzyme and polymorphism analysis (PRA) was performed for further differentiation of the 2 Rec-mas-H strains from M. abscessus subsp. massiliense. Using the two restriction enzymes, BstEII and HaeIII, the partial hsp65 gene sequence was digested [25], and the obtained fragments were compared to those obtained fragments from the partial hsp65 gene sequences of M. abscessus subsp. abscessus and M. abscessus subsp. massiliense by gel electrophoresis. All the BstEII restriction digests showed a similar pattern of fragments 220/245 bp in size. However, HaeIII restriction digests of the 2 Rec-mas-H strains showed patterns identical to those of M. abscessus subsp. abscessus (160/60 bp) but distinct from those of M. abscessus subsp. massiliense (200/60 bp) (Fig 5).

Discussion
The occurrence of LGT events among species or subspecies in genes encoding chronometer molecules that are used for the diagnosis or identification of pathogenic bacteria may compromise the results obtained when attempting to identify the disease-causing organisms present in infected patients, potentially leading to treatment failure. This problem is especially true in the case of NTM infections, which require a long culture period and often show species-or subspecies-dependent disparities in treatment regimens. Therefore, investigation of LGT events that may affect the diagnosis or identification of target molecules used in the differentiation of mycobacteria is necessary. The general molecular target for differentiation among species or subspecies in bacterial taxonomy is the 16s rRNA gene [36][37][38]. However, 16s rRNA sequence-based diagnostic and taxonomic methods have some limitations for species differentiation within the genus Mycobacterium, mainly due to the lack of 16S rRNA sequence diversity [39][40][41]. In particular, differentiation of RGM species, including the M. abscessus strains, is almost impossible. Therefore, instead of the 16s rRNA gene, alternative chronometer molecules such as the rpoB and hsp65 genes have been widely used as targets for mycobacterial identification. However, LGT events associated with the rpoB gene have been reported to occur among mycobacteria species or subspecies. For example, the M. yongonense Type I strain carries an rpoBC operon that was laterally transferred from a distantly related species, M. parascrofulaceum [23,42]. In addition, we recently isolated six M. abscessus subsp. massiliense strains from Korean patients with specific hybrid rpoB genes that were laterally transferred from M. abscessus subsp. abscessus [22]. This suggests that there is a risk of mis-identification when rpoBbased methods are used in mycobacterial diagnosis.
To date, no LGT events associated with another chronometer gene that is used as a mycobacterial target, hsp65, have been reported either in mycobacterial strains or in species within other genera, suggesting that the hsp65 gene is more resistant to LGT than the rpoB gene. However, in this study, we identified for the first time two M. abscessus subsp. massiliense rough strains with hsp65 genes that might have been laterally transferred from M. abscessus subsp. abscessus. These results suggest that the use of hsp65-based diagnosis in mycobacteria also creates a risk of misidentification, at least when attempting to differentiate subspecies within the M. abscessus strains. Indeed, we verified that two hsp65 recombinant M. abscessus subsp. massiliense strains were mis-identified as M. abscessus subsp. abscessus by the hsp65-PRA method targeting the 441-bp Telenti fragment, the most widely used method for NTM differentiation [25]. To the best of our knowledge, this is the first reported case of NTM misidentification by the hsp65-PRA method, and it strongly supports the above notion.
In contrast to the phylogenetic similarity of six recombinant M. abscessus subsp. massiliense strains carrying a hybrid rpoB gene to the smooth M. abscessus subsp. massiliense Type I genotype reported in our previous study [22], our phylogenetic analysis based on MLST sequences indicated that the two M. abscessus subsp. massiliense strains with recombinant hsp65 genes from M. abscessus subsp. abscessus are more closely related to M. abscessus subsp. massiliense Type II than to Type I. The results suggest that these two strains may have descended from the M. abscessus subsp. massiliense Type II genotype. Given that all Type II strains have a rough morphotype due to deletion of the glycopeptidolipid (GPL) gene, the rough colony morphotypes of the 2 hsp65 recombinant strains also support the above notion. However, the two putative recombinant strains showed unique sequences in hsp65, rpoB, glpK, murC, pta and purH genes which were differentiated from M. abscessus subsp. massiliense Type II strain (Figs 1 and 2, S3 Table).
Combined analysis of the results obtained in the present study and in our previous study [22] indicated that LGT events occurred in a total of 8 (3.9%) strains (2 strains with hsp65 hsp65 recombinant M. abscessus subsp. massiliense recombination and 6 strains with rpoB recombination) in a sample of 206 M. abscessus strains from Korean patients (100 strains in the present study and 106 strains in the recent study). Of note, further MLST analysis showed that all these strains belonged to the M. abscessus subsp. massiliense subspecies and were not M. abscessus subsp. abscessus, suggesting that M. abscessus subsp. massiliense may be more vulnerable to LGT events than M. abscessus subsp. abscessus. This provides a possible explanation of the fact that the genetic and taxonomic diversity among M. abscessus subsp. massiliense strains is higher than that among M. abscessus subsp. abscessus strains.
In conclusion, we identified two M. abscessus subsp. massiliense rough strains from Korean patients with hsp65 genes that might have been laterally transferred from M. abscessus subsp. abscessus. To the best of our knowledge, this is the first report to demonstrate LGT events associated with the hsp65 gene in mycobacteria. This report also suggests that there is potential for misidentification when hsp65-based protocols are used for mycobacterial identification.
Supporting information S1