Inducible and Acquired Clarithromycin Resistance in the Mycobacterium abscessus Complex

Purpose Clarithromycin was considered the cornerstone for the treatment of Mycobacterium abscessus complex infections. Genetic resistance mechanisms have been described and many experts propose amikacin as an alternative. Nevertheless, clarithromycin has several advantages; therefore, it is necessary to identify the non-functional erm(41) allele to determine the most suitable treatment. The aims of this study were to characterize the molecular mechanisms of clarithromycin resistance in a collection of Mycobacterium abscessus complex isolates and to verify the relationship between these mechanisms and the antibiogram. Materials and Methods Clinical isolates of M. abscessus complex (n = 22) from 16 patients were identified using four housekeeping genes (rpoB, secA1, sodA and hsp65), and their genetic resistance was characterized by studying erm(41) and rrl genes. Nine strains were recovered from the clinical isolates and subjected to E-test and microdilution clarithromycin susceptibility tests, with readings at 3, 7 and 14 days. Results We classified 11/16 (68.8%) M. abscessus subsp. abscessus, 4/16 (25.0%) M. abscessus subsp. bolletii, and 1/16 (6.3%) M. abscessus subsp. massiliense. T28 erm(41) allele was observed in 8 Mycobacterium abscessus subps. abscessus and 3 Mycobacterium abscessus subsp. bolletii. One strain of M. abscessus subsp. bolletii had an erm(41) gene truncated and was susceptible to clarithromycin. No mutations were observed in rrl gene first isolates. In three patients, follow-up of initial rrl wild-type strains showed acquired resistance. Conclusions Most clinical isolates of M. abscessus complex had inducible resistance to clarithromycin and total absence of constitutive resistance. Our findings showed that the acquisition of resistance mutations in rrl gene was associated with functional and non-functional erm(41) gene. Caution is needed when using erm(41) sequencing alone to identify M. abscessus subspecies. This study reports an acquired mutation at position 2057 of rrl gene, conferring medium-low clarithromycin constitutive resistance.


Introduction
Mycobacterium abscessus complex subspecies are rapid-growing mycobacteria (RGM) responsible for chronic pulmonary infections, cutaneous infections and, in some cases, bacteraemia [1,2]. These subspecies are considered as the most drug-resistant of all the RGM [3][4][5], which complicates treatment of the infections they cause. Macrolides such as clarithromycin and azithromycin were considered the cornerstone for the treatment of Mycobacterium abscessus complex infections [5]. Molecular resistance mechanisms have been described and many experts propose amikacin as an alternative [6]. Nevertheless, clarithromycin has several advantages; therefore, it is necessary to identify the non-functional erm(41) allele to determine the most suitable treatment.
Clarithromycin resistance in the M. abscessus complex can be constitutive or inducible, depending on two mechanisms. Constitutive resistance involves spontaneous point mutations, selected during macrolide-based chemotherapy, at positions 2058 and 2059 of the rrl gene, which encodes for the 23S rRNA [7]. The second mechanism is associated with inducible resistance [8], involving the presence of a functional erm(41) gene and following methylation of the ribosome. Constitutive resistance is characterized by in vitro high level resistance values to clarithromycin at 3 days of culture [4,7], while inducible resistance associated with the erm(41) gene is characterized by in vitro MIC resistant values only at 7 and even 14 days [9].
The taxonomy of the M. abscessus group is complex and still unresolved. In 2011, M. bolletii and M. massiliense were classified as a single subspecies (M. abscessus subsp. bolletii) due to their low genetic diversity [10]. In 2013, whole genome sequencing data analysis supported the differentiation of the M. abscessus complex into 3 subspecies: M. abscessus subsp. abscessus, M. abscessus subsp. bolletii and M. abscessus subsp. massiliense [11]. In the present study, we use this nomenclature for the three subspecies.
It is widely accepted that several housekeeping genes are needed to identify the three subspecies in the complex [12,13]. These three subspecies are closely related, and recombination and gene transfer has been described [11]. The genes most commonly used to classify M. abscessus subspecies are the rpoB, secA1, sodA, recA and hsp65 genes [12,13]. Even when powerful new techniques such as matrix-assisted laser desorption/ionization time-of-flight (MAL-DI-TOF) [14] are used, identification of M. abscessus complex isolates at the subspecies level is not 100% accurate.
We collected M. abscessus complex clinical isolates from 1995 to 2014 at a university hospital in Spain. The aims of this study were to characterize the molecular mechanisms of clarithromycin resistance in a collection of Mycobacterium abscessus complex initial isolates, identified at the subspecies level, and to verify the relationship between the genetic resistance mechanisms and the antibiogram results. We also studied 10 isolates in the follow-up of 4 patients by molecular typing and phenotypic and genotypic antibiograms.

Ethics statement
This study was retrospective without interaction with patients and all the patients' information was de-identified prior to analysis. Therefore the Fundació de Gestió Sanitaria del Hospital de la Santa Creu i Sant Pau ethics committees waived the need for informed consent and approved this research study (IISBP-CLA-2014-23).

Samples
Twenty-two clinical isolates were obtained between 1995 and 2014 from 16 patients: 3 from skin and 19 from lung infections. Ten isolates from 4 patients were used for follow-up: 3 from patient 1, 3 from patient 2, 2 from patient 3, and 2 from patient 4. DNA from all isolates was recovered for genotypic identification and molecular clarithromycin resistance analysis. Nine first recovered isolates from 9 individual patients were subjected to clarithromycin susceptibility tests.

Subspecies classification
Total DNA was extracted from mycobacterial clinical isolates by thermal shock [18] or Insta-Gene matrix (Bio-Rad Laboratories, Hercules, CA, USA). Analysis of rpoB, secA1, sodA and hsp65 genes was performed for identification at the subspecies level, as described previously [12,13]. PCR products were sent to Macrogen for sequencing (Meibergdeef, 1105AZ Amsterdam, The Netherlands). The sequences obtained were analyzed using the BLAST algorithm (NCBI).
For erm(41) and rrl amplification, the PCR conditions were 5 min at 95°C, then 35 cycles at 95°C for 60 s, 62°C (for erm(41) but 55°C for rrl) for 60 s, and 72°C for 90 s, followed by 72°C for 10 min in the final extension.
The reference sequence used for the analysis of erm(41) and rrl genes was the M. abscessus subsp. abscessus ATCC 19977T (GenBank accession number NC_010397).

Clarithromycin susceptibility testing
Clarithromycin MIC determination was performed in 9/16 initial isolates by the E-test method (bioMérieux, bioMérieux España S.A., Madrid, Spain) in Mueller-Hinton agar plates at 30°C [19]. The clarithromycin susceptibility test was also done in cation-adjusted Mueller-Hinton medium using the broth microdilution method [5] in Sensititre RAPMYCOI plates (Sensititre, Trek Diagnostic Systems, East Grinstead, United Kindom) according to CLSI guidelines [9]. Strains were incubated for 14 days, and clarithromycin MIC was read at 3, 7 and 14 days. Clarithromycin MIC cut-off for microdilution ranges from 2 susceptible to 8 resistant, and 1 susceptible to 4 resistant for the E-test [19].

Patient follow-up
Strain typing in the patient follow-up was performed by variable-number of tandem-repeat analysis (VNTR). The primers used were TR45, TR109, TR116, TR150, TR155 and TR172. The six pairs of primers and loci were the same as used by Wong et al. [20]. The PCR conditions for VNTR were as follows: a initial denaturalization step of 15 min at 95°C; 30 cycles of denaturalization at 95°C for 60 s, annealing at 59°C for 60 s and an extension at 72°C for 90 s, with a final elongation step at 72°C for 10 min. PCR amplicons were analyzed using the EPOCH capillary electrophoresis system (BioTek, Germany). Results were analyzed using GeneMapper (Applied Biosystems).

Sequences accession numbers
The obtained sequences for each gene were submitted to GenBank and given the following accession numbers: KT185536 to KT185557 for rpoB, KT185514 to KT185535 for hsp65, KT185603 to KT185624 for sodA, KT185581 to KT185602 for secA1, KT185492 to KT185513 for erm41 and KT185558 to KT185580 for rrl.
Genetic basis of resistance to clarithromycin in M. abscessus complex Table 1 also shows rrl and erm(41) results and the type of resistance in initial isolates. No mutations were observed in the rrl gene in any of these isolates. For the erm(41) T28 polymorphism,  [16]. For the first time, to our knowledge, in this study was found an erm(41) gene truncated in 1 M. abscessus subsp bolletii. Data of the similarity of the rpoB, secA1, sodA and hsp65 gene sequences of this strain and the reference strains of M. abscessus complex is shown in the Table 2.

Clarithromycin susceptibility testing
We recovered 9 strains of the M. abscessus group for phenotypic analysis. E-test and microdilution methods were performed with recovered strains to compare the two methods and relate the in vitro MIC results with our genotypic analysis. Table 1 shows clarithromycin MIC for the E-test and microdilution. All strains were clarithromycin-susceptible with both techniques at 3 days. Five out of nine strains were resistant at 7 days for both methods and two strains were susceptible. One strain was resistant for E-test and susceptible for microdilution, the remaining strain was susceptible for E-test and resistant for microdilution (Table 1). At day 14, 7/9 strains were resistant with both methods. In both methods, susceptible and resistant strains agreed at day 14. Table 1 also describes the follow-up results for 4 patients: M. abscessus subsp. abscessus strains obtained from 3 patients (Patients 1-3) showed the acquisition of a point mutation in the rrl gene. Patient 1, erm(41) C28 sequevar, showed an acquired resistance at position A2058G. Patient 3, erm(41) T28 sequevar, showed an acquired resistance at position A2059G. The erm (41) T28 sequevar from patient 2 showed a double peak at position A2057G/A in the DNA sequence, indicating a new selected mutation for clarithromycin constitutive resistance. In contrast, the M. abscessus subsp. bolletii strain from patient 4 showed no acquired mutation in the rrl gene.

Patient follow-up
All the strains studied in the follow-up were identical to the initial isolates for each patient. Each patient had a different strain. Table 1 also shows the profile of the variable-number of tandem-repeat assay.

Discussion
Clarithromycin is a key agent in the treatment of M. abscessus complex infections [9]. The main cause of treatment failure is inducible resistance [8,21]. Significant differences in the erm(41) gene associated with inducible resistance are found among the M. abscessus complex [15,22], suggesting that an accurate erm(41) allele identification is important to predict treatment outcomes. As the three M. abscessus complex subspecies are closely related and show horizontal gene transfer [11,23], more than one housekeeping gene is needed for subspecies identification [13]. Nevertheless, the genes to be used are not defined [24]. This work aimed to characterize the molecular mechanisms of clarithromycin resistance in the M. abscessus complex and to verify the relationship between these mechanisms and the clarithromycin susceptibility test.
In contrast with some authors [12,13] and in agreement with Nie et al. [25], we found each gene (rpoB, secA1, sodA and hsp65) was able to classify M. abscessus subspecies, with the exception of one strain, which was identified as M. abscessus subsp. abscessus by rpoB, sodA and hsp65, and M. abscessus subsp. bolletii by secA1 gene. In contrast with our results, Tan et al. [26] were able to correctly classify all their strains by sequencing the secA1 gene. M. abscessus subsp. abscessus was the most predominant subspecies of the complex (68.8%), in keeping with findings in the literature [12,21,27], where it ranges from 51.2% to 78.5%. Next in predominance was M. abscessus subsp. bolletii (25.0%), in contrast with many papers where the second most frequent subspecies is M. abscessus subsp. massiliense [16,28].
Regarding genetic resistance mechanisms, rrl gene sequencing in the initial isolates showed an absence of mutations. This is in contrast with previous reports [21,29], where the presence of constitutive clarithromycin resistance ranges from 2.7% to 28.6%. In agreement with the literature, however, we observed inducible resistance due to nucleotide T28 of erm(41) in 68.8% (11/16) of M. abscessus complex isolates [8,21].
As previously reported, erm(41) is a subspecies-specific gene in the M. abscessus group [7,16]. We found only one discrepancy with respect to the published criteria [7,17], attributed to an M. abscessus subsp. bolletii strain with a truncated erm(41) gene. To our knowledge, this is the first description of M. abscessus subsp. bolletii with a non-functional truncated erm(41) gene. The strain (07I004) matched M. abscessus subsp. bolletii reference strains CCUG 50184 and CIP108541 for the four housekeeping genes studied but not for the erm(41) gene. This finding supports other reports suggesting that gene transfer and recombination occurs between subspecies [11,28]. Several cases of M. abscessus subsp. massiliense with the functional erm(41) gene have been reported [30]. Our data thus support the notion that erm(41) is not a subspecies-specific gene, and that erm(41) gene sequencing should not be used as the only technique to classify M. abscessus complex subspecies.
The data produced by our study revealed the acquisition of constitutive clarithromycin resistance. Constitutive resistance to clarithromycin due to mutations in rrl was selected in both erm(41) C28 (as shown in Patient 1) [7] and erm(41) T28 strains (as shown in Patients 2 and 3) [21]. With the advantage of having the initial rrl wild-type strains, our results support the idea, in opposition to Bastian et al. [7], that selection of rrl mutants is similar in T28 and C28 strains.
We described the acquisition of a 2057 point mutation in a T28 M. abscessus subsp. abscessus strain with an initial rrl wild-type isolate. A double peak in the 2057 DNA sequence was observed, which suggests the coexistence of two mixed DNA populations (an rrl wild-type and a 2057 mutated population), since these mycobacteria possess only a single copy of the rrl gene in their genome [5]. The amplification and DNA sequencing was repeated three independent times to discard amplification or sequencing errors. Further work should include cloning to determine whether two variants exist in this strain.
It is important to emphasize that mutations at position 2057 of rrl in other microorganisms are related to medium-low levels of clarithromycin resistance in vitro [31], as seen in this study. Vester et al. [31] also reported that mutations at positions 2057 and 2611 of the rrl gene caused resistance to macrolides and ketolides due to their proximity to the action centre. Mutations at these positions obstruct the link between the antibiotic and its target, and cause lowlevel resistance to 14-membered-ring macrolides and no resistance to 16-membered-ring macrolides. In contrast, recently, Luo et al. [32] reported a 2057 point mutation in a M. abscessus subsp. abscessus which showed a high level resistance (MIC >256 μg/mL).
In our study we found complete agreement between erm(41), rrl sequencing and clarithromycin susceptibility testing. After a 3-day culture, all strains were susceptible as was observed with both methods (E-test and microdilution). Following a 7-day culture, 85% (6/7) of inducible resistance [erm(41) T28 sequevar] was detected, and after a 14-day culture, we detected all inducible resistance strains. Although we did not find significant differences between the two methods, one strain with the E-test and one strain with the microdilution method took 14 days to reveal resistance. This is in contrast with other authors who report differences between Etest and microdilution performance [33]. Although the microdilution is the standard method according to the CLSI [9], in our study, the E-test showed the same results as the microdilution, and MIC reading at early stages was easier.
We would like to emphasize that one of the limitations of our study is the relatively small amount of isolates included. This is important to consider when drowning conclusions on epidemiology, but it does not invalidate the molecular observations.

Conclusion
In conclusion, our findings identified one strain of M. abscessus subsp. bolletii with a truncated and non-functional erm(41) gene. Caution is needed if erm(41) gene sequencing is the only technique used for subspecies identification. Although clarithromycin constitutive resistance is not detected in the initial isolates, it can be acquired in T28 and C28 erm(41) strains of the M. abscessus complex. Finally, the acquisition of a mutation at position 2057 in the rrl gene confers medium-low level resistance to clarithromycin.