Improved Identification of Rapidly Growing Mycobacteria by a 16S–23S Internal Transcribed Spacer Region PCR and Capillary Gel Electrophoresis

The identification of rapidly growing mycobacteria (RGM) remains problematic because of evolving taxonomy, limitations of current phenotypic methods and absence of a universal gene target for reliable speciation. This study evaluated a novel method of identification of RGM by amplification of the mycobacterial 16S–23S rRNA internal transcribed spacer (ITS) followed by resolution of amplified fragments by capillary gel electrophoresis (CGE). Nineteen American Type Culture Collection (ATCC) Mycobacterium strains and 178 clinical isolates of RGM (12 species) were studied. All RGM ATCC strains generated unique electropherograms with no overlap with slowly growing mycobacteria species, including M. tuberculosis. A total of 47 electropherograms for the 178 clinical isolates were observed allowing the speciation of 175/178 (98.3%) isolates, including the differentiation of the closely related species, M. massiliense (M. abscessus subspecies bolletii) and M. abscessus (M. abscessus sensu stricto). ITS fragment size ranged from 332 to 534 bp and 33.7% of clinical isolates generated electropherograms with two distinct peaks, while the remainder where characterized with a single peak. Unique peaks (fragment lengths) were identified for 11/12 (92%) RGM species with only M. moriokaense having an indistinguishable electropherogram from a rarely encountered CGE subtype of M. fortuitum. We conclude that amplification of the 16S–23S ITS gene region followed by resolution of fragments by CGE is a simple, rapid, accurate and reproducible method for species identification and characterization of the RGM.


Introduction
Rapidly growing mycobacteria (RGM) are ubiquitous environmental acid-fast bacilli that are important human and animal pathogens and are characterised by their growth in culture medium within 7 days post-inoculum. They cause infection in both immunocompetent and immunocompromised individuals, most commonly manifesting as pulmonary or skin and soft tissue infection, and less frequently as a cause of bacteremia, osteomyelitis, endocarditis, peritonitis, lymphadenitis, post-surgical infections, or catheter-related infection [1,2]. The speciation of the RGM facilitates the selection of the appropriate antimicrobial therapy, may predict disease progression and allows for the detection of novel species. In addition, speciation allows for the initial recognition of epidemiologically-linked cases that have been reported to occur in hospitals and in the community, with public health implications [3,4].
Historically, a combination of phenotypic biochemical and mycolic acid-based identification methods have been used to identify clinically relevant RGM, but these techniques are slow, lack discriminatory power to distinguish closely-related species and cannot identify novel species. Proteomic approaches, including matrix assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS), have been reported to speciate RGM; however, the ability of this technology to reliably distinguished closely related species remains uncertain [5][6][7]. Molecular methods include probe-based assays, PCR restriction enzyme analysis and gene sequence analysis, and have emerged as attractive approaches for species identification of mycobacteria [8]. Molecular targets that have been used include the 16S rRNA gene, the heat shock protein 65 (hsp65) gene, the b subunit of the RNA polymerase (rpoB) gene, as well as the 16S-23S internal transcribed spacer (ITS) [9]. The ITS region has been shown to be of particular utility in taxonomic assignment of mycobacteria because it is less highly conserved compared with the 16S rRNA gene and has potential to yield a greater degree of genetic polymorphism [9][10][11]. Yet no single gene target, including the ITS, has been shown to be sufficiently discriminatory to speciate all RGM [9].
The most commonly encountered RGM are species within the Mycobacterium fortuitum complex and the M. chelonae -abscessus group [1,2]. Before 2001 only two species, M. chelonae and M. abscessus, were recognised within the M. chelonae-abscessus group, however, more recently division of the M. abscessus group has been suggested. Some authors propose that M. abscessus be subdivided into three species -M. abscessus, M. massiliense and M. bolletii [12,13]. Others have indicated that these three organisms represent two subspecies of the M. abscessus group -ie. M. abscessus sensu stricto and M. abscessus subspecies bolletii -with the latter incorporating both M. massiliense and M. bolletii [14]. Importantly, M. abscessus, M. massiliense and M. bolletii cannot be distinguished using traditional phenotypic methods [14] and all three organisms share 100% sequence identity of the mycobacterial 16S rRNA gene [14,15]. Yet, clinical evidence has emerged that separating M. abscesses from M. massiliense is potentially useful as these organisms are associated with different disease phenotypes and progression [16][17][18]. In addition, inducible macrolide resistance is commonly encountered in M. abscessus sensu stricto, which harbours an intact erythromycin ribosomal methylase (erm) gene, while M. massiliense isolates are stably susceptible to macrolides as a result of two deletions within the erm gene [19,20]. The species names M. abscessus (M. abscessus sensu stricto) and M. massiliense (M. abscessus subspecies bolletii) continue to dominate the clinical literature and this is the terminology used within this study. Speciation of members within the M. fortuitum complex is seldom performed since the clinical utility of this practice is uncertain and because existing methods are time consuming and expensive [8].
In this study, we have developed and evaluated a method to identity RGM based on amplification of the Mycobacteria 16S-23S rRNA ITS region(s) followed by amplicon analysis by sequencerbased capillary gel electrophoresis (CGE). Unlike traditional gel electrophoresis, CGE uses 59-end fluorescein-labelled primers and a DNA analyzer, with rapid, accurate resolution of amplicons. It has shown good discriminatory power in the identification and characterization of other clinically significant bacteria including Clostridium difficile [21,22], Vibrio [23], and Nocardia species [24]. We applied this method to a representative set of clinical isolates identified by a combination of HPLC, real time PCR and 16S rRNA gene sequencing. The antimicrobial profiles of the M. abscessus group study isolates, where available, were also reviewed to investigate possible association, if any, between CGE profiles and antibiotic susceptibility.

Isolates
All RGM isolates referred to the New South Wales Mycobacterium Reference Laboratory, Westmead Hospital, Sydney, Australia between 1 March 2012 and 31 January 2013 were included in the study (n = 213). Duplicate isolates cultured from the same patient (n = 26), contaminated isolates (n = 6) and missing isolates (n = 6) were excluded. An additional 3 organisms (isolated after December 2011) representing uncommon RGM species were also retrieved. Study isolates were stored on Middlebrook 7H11 agar at 4uC. Nineteen American Type Culture Collection strains (ATCC, Manassas, VA) mycobacterial strains (4 RGM and 15 slowly-growing organisms) were also studied as shown in Table 1.

Identification of Mycobacteria
RGM were initially characterized by high-performance liquid chromatography (HPLC) analysis as previously described [25]. Isolates with HPLC profiles consistent with M. chelonae/M. abscessus group were confirmed by an in-house real-time multiplex PCR assay, using primers and species-specific TaqMan probes targeting the ITS region of the 16S-23S rRNA gene, adapted from the method of Xiong et al. [26]. Isolates identified as M. abscessus were then further speciated to M. abscessus and M. massiliense by sequencing the full ITS segment (primers described below) with reference to published sequences (GenBank Accession no. CU458896; GenBank Accession no. CP003699). Other study isolates, including M. fortuitum complex, M. mucogenicum, M. flavescens, M. elephantis, M. immunogenum, M. moriokaense, M. septicum, M. neoaurum and M. phlei were identified to species level using a combination of HPLC profile analysis followed by 16S rRNA sequencing when atypical HPLC traces were encountered (Table S1 and Figure S1) [27]. The 16S rRNA gene sequences of isolates were compared with sequences of type strains archived in the NCBI GenBank using NCBI BLASTn algorithm (www.ncbi.nlm. nih.gov/GenBank/).

DNA extraction
Bacterial DNA was extracted using the commercial InstaGene Matrix method (Bio-Rad Laboratories Inc, California, USA). Briefly, 2-4 individual mycobacterial colonies were placed into 1 mL of sterile water, mixed by vortex and then underwent sonication for 20 minutes. Suspensions were centrifuged at 11,000 rpm for 1 minute after which the supernatant was discarded. The residual pellet was resuspended in 200 mL of InstaGene Matrix (Bio-Rad), mixed by vortex and incubated at 56uC for 20 minutes. Samples were then boiled for 8 minutes and centrifuged (13,200 rpm for 2 minutes) prior to use. Samples were used within 24 hours or stored at 220uC. Routine protocols in our laboratory have demonstrated minimal degradation of extracted DNA under these conditions (data not shown).

Amplification of the 16S-23S rRNA internal transcribed spacer
The mycobacteria ITS region(s) were amplified prior to CGE analysis using a 6-FAM labelled ITS1F forward primer 59 GTGCGGCTGGATCACCTCCT 39 and a ITS1R reverse primer 59 AGCCTCCYACGTCCTTC(A/T)TCGGCT 39 (Sigma-Aldrich, Castle Hill, NSW, Australia). Sequencing of the ITS region was performed for M. chelonae, M. abscessus and M. massiliense using an ITS2F forward primer 59 GAAGTCGTAACAAGG-TAGCCG 39 and a ITS2R reverse primer 59 GACAGCTCCCC-GAGGC(A/T)TATCGCA 39 (Sigma-Aldrich). Each PCR mixture was made to a total volume of 25 mL consisting of 5 mL of DNA extract, 0.5 mM forward and reverse primers, 2.5 mM deoxynucleoside triphosphates (Roche, Castle Hill, Australia), 106 PCR buffer containing 15 mM MgCl 2 (Qiagen, Doncaster, Victoria, Australia), 0.5 U HotStarTaq DNA polymerase (Qiagen) and molecular biology-grade H 2 O (Eppendorf, North Ryde, Australia). PCR conditions were 95uC for 15 minutes and then 35 cycles of 95uC for 30 seconds, 62uC for 30 seconds, 72uC for 1 minute, followed by a further extension at 72uC for 10 minutes.
Sequencer-based capillary gel electrophoresis PCR fragment analysis was performed using the ABI 3730xl DNA analyzer employing a 48-capillary 50 cm POP-7 gel (Applied Biosystems, Forster City, USA). PCR products were diluted 1:30 with molecular biology-grade H 2 O (Eppendorf) to a final volume of 30 ml. Sample injection was at 1.6 kV over 15 seconds with a total running time of 6,200 seconds at 15 kV run voltage. A 20-to 1200-bp LIZ 1200 ladder (Chimerx, Madison, WI, USA) was used as an internal size control.

Result interpretation, reproducibility testing and data analysis
Amplified fragments, represented by one or more peaks according to fragment size, were analysed by GeneMapper software (Applied Biosystems). Smaller peaks were required to reach 10% of the height of the greatest peak and when double peaks were observed #1.0 bp apart, only the larger peak was analysed [21,22,24]. Fragment lengths found in .10% of a species were called 'common' and fragment lengths found only in one species were defined as 'unique'. Descriptive statistics were prepared using IBM SPSS Statistics 21.0 (IBM Corp., NY, USA). The pattern of peaks was noted in relation to the number and numerical value of each fragment rounded to the nearest whole number, with the exception of M. abscessus and M. massiliense isolates which were rounded to the nearest 0.5 bp.
A random set of the study DNA extracts (19/178, 10.7%), underwent repeat ITS-CGE testing to assess reproducibility. In addition, 3 patients with repeat isolates of RGM collected at different times during the study period (.1 month apart), had DNA extracted on each specimen with ITS-CGE results compared.

Ethics Statement
This study is not a research question but describes a laboratory validation of a novel method for the specific identification of mycobacterial pathogens grown on artificial media. In addition, the study does not involve the collection or reporting of patient data and no patient intervention occurred with the obtained results. Bacterial isolates were collected and stored as part of routine laboratory practice. No additional isolates were collected or stored for the purpose of this study. All bacterial samples were de-identified for the purpose of study experiments, data analysis and reporting.

Results
A total of 178 clinical isolates collected from different patients and representing 12 RGM were included in the study set.

Identification of Clinical Isolates
ITS region amplicons were obtained for all 178 clinical isolates. CGE analysis demonstrated single peaks for 118 (66.3%) isolates and double peaks for 60 (33.7%) isolates; ITS fragment sizes ranged from 332 to 534 bp. Overall, CGE analysis generated 47 unique electropherograms for the 178 isolates. Figure 1 (Panel A-E) highlights representative examples of the electropherograms for the study isolates.
Unique single peak electropherograms were evident for all 31 M. abscessus, 17 M. massiliense and 40 M. chelonae isolates, separating these closely related species. There were two patterns generated for the M. abscessus species, one with average measured amplicon length 387.98 bp (range 387.74-388.15; n = 29) and a second with length of 393 bp (n = 2). Both of these unique electropherograms were distinguished from the 17 M. massiliense isolates which generated an average measured CGE fragment size of 388.70 bp (range 388.44-388.93). Sequencing of the ITS fragments revealed 3 consistent point mutations between the M. abscessus and the M. massiliense isolates, including a single deletion of a cytosine base, accounting for the small but distinguishable fragment length differences. The 2 M. abscessus isolates with fragment length of 393 bp were epidemiologically distinct. The isolates were cultured from the sputum of patients collected from different jurisdictions of Australia (.1500 km apart). Sequencing of the ITS region revealed both isolates carried a 5 bp insertion polymorphism (-TTGTG-) not described in GenBank. Further sequencing of the full 16S gene for these two isolates demonstrated it was homologous with published sequences of M. abscessus/M. massiliense, (GenBank Accession no. CU458896; GenBank Accession no. CP003699), suggesting these 2 clinical isolates may represent a novel genotype of M. abscessus, and that CGE may be a potential approach for subtyping this species complex.
Amongst the 64 clinical M. fortuitum complex isolates, 27 different electropherograms were demonstrated. Five CGE types (CGE Type 1 to 5; Figure 2) accounted for 38 (59.4%) of the M. fortuitum complex isolates tested, with a high degree of heterogeneity seen with the remaining isolates ( Figure 2). Clustering by geography or specimen type was not observed. Almost all (175/ 178; 98.3%) clinical isolates could be assigned to the correct species by the ITS-CGE typing technique based on analysis of the number and size of the amplified fragments. The single M. moriokaense isolate could not be speciated as it produced an electropherogram with a single peak at 478 bp which was indistinguishable from those produced by 2 M. fortuitum isolates (CGE Type 8). Unique peaks were identified for the remaining 11/12 (92%) RGM species studied (summarized in Table 2).

Reproducibility
Reproducibility of the ITS-CGE technique was assessed by repeating the method on 19 (10.7%) isolates (with a total of 28 peaks) on different occasions with fresh PCR reagents. The average measured difference between the repeat fragment sizes was 0.12 bp (range of 0.0 to 0.39 bp). In addition, 3 patients from whom multiple clinical isolates were obtained during the study period, underwent comparison of CGE electropherograms derived from independent specimens. The average measured difference between the fragment lengths for the related isolates was 0.05 bp (range 0.01 to 0.08 bp).

Discussion
The reliable and accurate identification of RGM is challenging because of the lack of consensus in the current nomenclature as well as the absence of a robust and universal method for speciating all RGM. Traditional methods are labour intensive, time consuming and require the interpretation of subjective data [29]. Here we report a novel method of applying CGE to analyze the RGM 16S-23S ITS region, as a highly discriminatory and reproducible method for speciation of RGM. While other authors have utilized 16S-23S ITS restriction analysis [11,30,31] or probe hybrididization [26], this is the first study that has applied the high resolution CGE method to the intact ITS fragments and demonstrated it as a highly discriminating tool for the RGM group, including the delineation of the M. abscessus group. The method was initially validated using 19 ATCC strains, with 73.6% identified overall -100% of RGM and 66.7% of slowly growing mycobacteria. When applied to clinical isolates of RGM the diversity of electropherograms generated allowed for the allocation to species of 98.3% of the clinical RGM strains studied.
A key finding of the present study was that the ITS-CGE method clearly separates organisms belonging to the M. chelonae and M. abscessus complex on the basis of a 3-4 bp difference of the rRNA 16S-23S ITS fragment length. Notably, the high resolution of the CGE method was able to characterize the M. abscessus group into M. massiliense and M. abscessus on the basis of three consistent ITS polymorphisms resulting in a 1 bp difference in fragment length. Previously it has been reported that single gene sequences were inaccurate in separating the M. abscessus group, although the 16S-23S ITS was not considered by these authors [32]. Our finding that the mycobacteria ITS is sufficient to separate these closely related subspecies is consistent with previous reports of the utility of the ITS sequence in separating both the slowly and RGM groups [10,11].
The requests for clinical microbiology laboratories to delineate the M. abscessus group is likely to increase in the future with emerging evidence that sub-species within the group are associated with differing clinical outcomes [16][17][18], transmission within cystic fibrosis populations [33], and therapeutic implications, including the propensity for M. abscessus to develop macrolide resistance via activation of an intact erm gene [13]. This propensity to macrolide resistance is supported by the MIC data in our study population which demonstrated M. massiliense isolates have significantly lower MICs for clarithromycin (p,0.01). The single isolate of M. massiliense identified with an MIC of 16 may have harbored an alternate mechanism of resistance, such as point mutations of the 23S rRNA macrolide binding site [13,34], or may have expressed a functioning erm gene as has been recently described in a subset of M. massiliense isolates [35]. Nevertheless, the susceptibility data presented here supports the emerging view that laboratories need to delineate the M. abscessus group given the predictable clarithromycin susceptibility data along subspecies lines.
Electropherograms with distinct double peaks were identified for some RGM complex and species groups (Figure 1, Panel D; Table 2), presumably correlating with multiple genomic copies of the ITS. These double peaks increase the discriminatory power of the ITS-CGE method. For example, a fragment length of 429 bp was seen for a proportion of the M. fortuitum complex isolates as well as for one of the M. elephantis isolates. However, in each of these cases, the electropherogram revealed a distinct and unique second peak allowing for differentiation of the species. This study has highlighted the 16S-23S ITS sequence diversity of the RGM group, particularly within the M. fortuitum complex. The degree of resolution naturally lends this method to subtyping within this species complex where a high degree of hereogeneity was demonstrated ( Figure 2); however, the clinical utility of this approach in the tracking of patient isolates and in the investigation of nosocomial outbreaks, requires further evaluation.
Despite the excellent reproducibility of the ITS-GGE technique, electropherogram features need to be interpreted with clear criteria to allow consistent reporting [24]. Double peaks which were less than one bp apart were encountered in a small proportion of isolates which possibly arise due to variable secondary structure of the amplicons (see examples; Figure 1, Panel A and D). Sequencing of these fragments confirmed that the calculated 'true peak' consistently correlated with the larger peak. The accurate interpretation of these double peaks is especially important if using the method to resolve M. abscessus from M. massiliense as the distance between the 'true' unique peaks for these species (measured at 0.72 bp) may be less than the difference observed between the spurious double peaks. Variation was also seen in the relative height of fragments in electropherogram peaks, including some small peaks not analysed that were ,10% of the highest peak. The relative heights remained consistent on duplication of isolates and may be attributed to the variation in genomic repeats of 16S-23S ITS region. The relative variation in peak heights (when 2 peaks are present) may provide additional power to discriminate between closely related species.
There are several important benefits of ITS-CGE method when compared with tradition identification procedures [22][23][24]. The method can be completed within 1 to 2 working days and does not rely on time and labour-intensive agar gel techniques. Although complete cost analysis is yet to be performed, the ITS-CGE based method is less expensive (approximately $3 per isolate) compared to 16S rRNA sequence base method commonly employed for speciating the less frequently encountered species. The method has the capacity to distinguish novel species or subtypes which is in contrast to phenotypic and proteomic methods. The results generated by ITS-CGE are amenable to inter-laboratory comparison and allow portability across jurisdictions. Importantly, the method appears to be able to identify mixed cultures. For example, when we experimentally mixed M. abscessus and M. fortuitum colonies prior to extraction and then applied the ITS- CGE technique, the electropherogram contained the unique or 'signature' peaks for both species (Figure 1, Panel F). Limitations of this study include the relatively low number of the less common RGM species analyzed. In addition, it is possible that intra-species genetic variation in isolates circulating in disparate geographical settings may be associated with variation in electropherograms. This was demonstrated by the unique electropherogram generated by the M. fortuitum ATCC 6841 strain when compared to the clinical M. fortuitum study isolates. As a result of this variation, laboratories which choose to implement this method will need to validate the findings in their own geographic settings. This limitation could be overcome if laboratories were to generate a collaborative CGE database, incorporating well characterized clinical strains from disparate geographical locations. Various commercial sequencing companies provide CGE which may be useful for laboratories that do not have access to the required equipment and expertise.
In conclusion, the amplification of the Mycobacteria 16S to 23S rRNA ITS region(s) followed by fragment analysis by CGE enables the identification of clinically relevant RGM. This study has demonstrated that the ITS-CGE method can distinguish the taxonomically related but clinically important M. abscessus and M. massiliense species, as well as discriminate the heterogenous M. fortuitum complex. With the development of a comprehensive CGE-ITS database this method may complement existing approaches in the identification and characterization of RGM.