Molecular Mechanisms of Intrinsic Streptomycin Resistance in Mycobacterium abscessus

ABSTRACT

Streptomycin, the first drug used for the treatment of tuberculosis, shows limited activity against the highly resistant pathogen Mycobacterium abscessus. We recently identified two aminoglycoside-acetylating genes [aac(2′) and eis2] which, however, do not affect susceptibility to streptomycin. This suggests the existence of a discrete mechanism of streptomycin resistance. M. abscessus BLASTP analysis identified MAB_2385 as a close homologue of the 3″-O-phosphotransferase [APH(3″)] from the opportunistic pathogen Mycobacterium fortuitum as a putative streptomycin resistance determinant. Heterologous expression of MAB_2385 in Mycobacterium smegmatis increased the streptomycin MIC, while the gene deletion mutant M. abscessus ΔMAB_2385 showed increased streptomycin susceptibility. The MICs of other aminoglycosides were not altered in M. abscessus ΔMAB_2385. This demonstrates that MAB_2385 encodes a specific and prime innate streptomycin resistance determinant in M. abscessus. We further explored the feasibility of applying rpsL-based streptomycin counterselection to generate gene deletion mutants in M. abscessus. Spontaneous streptomycin-resistant mutants of M. abscessus ΔMAB_2385 were selected, and we demonstrated that the wild-type rpsL is dominant over the mutated rpsL^K43R in merodiploid strains. In a proof of concept study, we exploited this phenotype for construction of a targeted deletion mutant, thereby establishing an rpsL-based counterselection method in M. abscessus.

INTRODUCTION

In the past several decades, Mycobacterium abscessus has emerged as one of the most clinically relevant nontuberculous mycobacteria (1–3). M. abscessus belongs to the rapidly growing mycobacteria (RGM) and was first described in 1953, isolated from a patient suffering from a knee infection (4, 5). M. abscessus skin and soft tissue infections, posttraumatic wound infections, or disseminated disease occurs after tattooing, mesotherapy, or surgical intervention. Sources of these infections include contaminated ink, multidose vials, surgical instruments, and even tap water (6–13, 70). Outbreaks of disseminated M. abscessus infections are described mainly in immunocompromised patients (12, 13). Additionally, M. abscessus accounts for more than 80% of all pulmonary infections with RGM (14). The pathogen poses a major threat for patients suffering from pulmonary disorders, like bronchiectasis or cystic fibrosis (1, 14). Occasionally, these pulmonary infections are fatal (14, 15). M. abscessus infections are difficult to treat due to the high-level antibiotic resistance conferred by an almost impermeable cell wall, drug efflux pumps, or drug-modifying enzymes (16–18).

Recently, various enzymes responsible for resistance to macrolides (Erm41), rifamycins (Arr_Mab), β-lactams (Bla_Mab), capreomycin (Eis2), and several aminoglycosides, such as amikacin, tobramycin, or kanamycin B [AAC(2′) and Eis2], were described (19–23). M. abscessus is naturally resistant to the aminoglycoside streptomycin (24), while streptomycin susceptibility was not affected by deletion of aac(2′) and eis2, indicating a discrete streptomycin resistance mechanism (22). In total, 12 putative aminoglycoside phosphotransferases, potentially having overlapping functions, are encoded in the genome of M. abscessus (18). In Mycobacterium fortuitum, another member of the RGM, an aminoglycoside, 3″-O-phosphotransferase [encoded by aph(3″)] was described as a streptomycin resistance determinant (25), while no streptomycin-modifying enzymes have been identified in other mycobacteria, like Mycobacterium tuberculosis or Mycobacterium smegmatis (26). These organisms are naturally susceptible to streptomycin (27–29). Streptomycin was the first drug used to treat tuberculosis (30). Streptomycin binds to the small ribosomal subunit and induces miscoding; it inhibits the initiation of translation and interferes with proofreading (31). Clinically acquired streptomycin resistance in M. tuberculosis is associated with single-point mutations in genes encoding the S12 protein of the small ribosomal subunit (rpsL) or the 16S rRNA (rrs) (32–35). RpsL and rRNA alterations interfere with streptomycin binding and ameliorate streptomycin-induced disturbance of ribosome dynamics and thereby confer streptomycin resistance (36–38).

Tools for the genetic manipulation of M. abscessus are still limited (22, 39, 40), as most of the conventional genetic tools for mycobacteria are not very effective and require high antibiotic concentrations due to the high level of intrinsic resistance (40). An rpsL-based counterselection strategy was previously exploited for allelic-exchange experiments in mycobacteria (41). rpsL⁺ can be used as counterselectable marker, since the wild-type rpsL is dominant over the mutated rpsL^mut allele in a merodiploid strain, resulting in a growth disadvantage in the presence of streptomycin, i.e., a streptomycin-susceptible phenotype (42). The rpsL⁺ counterselection strategy has been exploited to generate various allelic-exchange mutants for characterizing mechanisms of genome stability, protein homeostasis, posttranslational modification, virulence, and drug resistance in M. smegmatis, Mycobacterium bovis BCG, and M. tuberculosis (43–52), but also in other genera (53).

In this study, we identified the innate streptomycin resistance determinant in M. abscessus. We compared the susceptibilities of the wild type and the targeted gene deletion mutant devoid of the putative streptomycin resistance determinant to streptomycin and other aminoglycosides. In addition, we demonstrate that rpsL mutations confer streptomycin resistance and introduce rpsL⁺ as a novel counterselection marker for M. abscessus, based on streptomycin susceptibility.

RESULTS

In silico identification of a putative streptomycin resistance determinant.

M. abscessus shows lower streptomycin susceptibility than M. tuberculosis. Streptomycin susceptibility in M. tuberculosis is affected by single nucleotide polymorphisms in either rpsL or rrs. Sequence comparison of M. abscessus ATCC 19977^T RpsL (accession number YP_001704579) (18) and rrs (accession number NC_010397, region 1462398 to 1463901) (https://www.ncbi.nlm.nih.gov/nuccore/NC_010397.1?report=GenBank&from=1462398&to=1463901) (18) with homologous sequences in M. tuberculosis H37Rv (accession numbers: RpsL, NP_215196 [54]; rrs, NR_044826 [55]) and M. smegmatis mc² 155 (accession numbers: RpsL, YP_885784; rrsA, NC_008596, region 3823615 to 3825142 [https://www.ncbi.nlm.nih.gov/nuccore/NC_008596.1?from=3823615&to=3825142], and rrsB, NC_008596, region 5029475 to 5027948 [https://www.ncbi.nlm.nih.gov/nuccore/NC_008596.1?from=5027948&to=5029475]) did not reveal any alterations at codons 43 and 88 of RpsL or in relevant positions conferring streptomycin resistance in the 530 loop and the 915 region of 16S rRNA (Fig. 1). M. abscessus genome annotation, genome analysis, and aminoglycoside susceptibility testing suggested the presence of several putative aminoglycoside-modifying enzymes (18, 22). However, the already characterized aminoglycoside resistance genes aac(2′) and eis2 did not affect streptomycin resistance levels (22), indicating that another mechanism is responsible for the intrinsic streptomycin resistance in M. abscessus. In M. fortuitum, another RGM, a streptomycin resistance determinant, APH(3″), was identified (25). We used BLASTP to identify putative M. fortuitum FC1 APH(3″) (accession number ABC68330) (25) homologues in the genome of M. abscessus. Here, we identified MAB_2385 with an amino acid sequence identity of 55%.

FIG 1.

Sequence comparisons of RpsL (A) and rrs (B), the relevant streptomycin resistance determinants in M. tuberculosis and M. smegmatis. Published polymorphisms, each singly conferring streptomycin resistance, are listed below the relevant positions. The numbering is based on E. coli. Reference sequences of the E. coli K-12 substrain MG1655 RpsL (accession number NP_417801) (63) and rrsA (NC_000913, regions 4035531 to 4037072) (https://www.ncbi.nlm.nih.gov/nuccore/NC_000913.3?report=GenBank&from=4035531&to=4037072) (63) are shown. Amino acid substitutions in RpsL were observed at positions 43 and 88. The 530 loop and the 915 region of rrs are predominant for resistance mutations in M. tuberculosis H37Rv. M. smegmatis mc² 155 has two rrs copies (rrsA and rrsB), and thus, no resistance mutations were observed (29). In rare cases, nucleotide substitutions outside the two regions have been associated with streptomycin resistance (29, 33, 64–67). Comparison of the streptomycin resistance-conferring regions showed that amino acids and nucleotides, respectively, in M. abscessus ATCC 19977^T are identical to those in M. tuberculosis H37Rv and M. smegmatis mc² 155. Generated spontaneous streptomycin-resistant M. abscessus ATCC 19977^T mutants revealed substitutions in RpsL at amino acids 43 and 88 and mutations in the 530 loop of the single rrs gene. These polymorphisms have been described in other mycobacteria.

Heterologous expression of MAB_2385 and generation of a MAB_2385 gene deletion mutant.

MAB_2385 was cloned into the single-copy integrative vector pMV361-aac(3)IV and transformed into M. smegmatis. Transformation of pMV361-aac(3)IV-MAB_2385 into M. smegmatis increased the streptomycin MIC 8-fold (median, 4 mg/liter) compared to the wild type (median, 0.5 mg/liter) or strains carrying the empty vector pMV361-aac(3)IV (median, 0.5 mg/liter).

The suicide allelic replacement vector pSE-ΔMAB_2385 targeting MAB_2385 carrying an apramycin resistance cassette and M. tuberculosis katG for isoniazid negative selection was transformed into M. abscessus. M. tuberculosis KatG converts the prodrug isoniazid into its active form and thereby confers isoniazid susceptibility. Selection on apramycin was applied, and single-crossover transformants resulting from homologous recombination were identified by Southern blotting (Fig. 2). Counterselection of single-crossover recombinants with isoniazid resulted in the M. abscessus MAB_2385 gene deletion mutant (M. abscessus ΔMAB_2385). The complementation vector pMV361-aac(3)IV-MAB_2385 was transformed in M. abscessus ΔMAB_2385. The deletion of MAB_2385 and the gene's complementation in the mutant strain were verified by Southern blotting (Fig. 2).

FIG 2.

Southern blot analysis of the MAB_2385 gene deletion mutant and its complementation. Genomic DNA of the wild-type M. abscessus ATCC 19977^T (lane 1), the pSE-MAB_2385 recombinant after apramycin selection for site-specific homologous recombination (single crossover) (lane 2), the MAB_2385 gene deletion mutant after counterselection with isoniazid (lane 3), and the complemented ΔMAB_2385 transformed with the vector pMV361-aac(3)IV-MAB_2385 (lane 4) was isolated, digested with Tth111I, and probed with a flanking region upstream of MAB_2385. The pattern is consistent with the predicted wild-type fragment at 3.8 kbp, single-crossover fragments at 3.8 kbp and 6.3 kbp, the gene deletion mutant fragment at 5.4 kbp, and gene complementation with fragments at 1.2 kbp and 5.4 kbp.

Antibiotic susceptibility testing of M. abscessus strains.

Initially, Etests were performed to compare streptomycin susceptibilities of the wild-type M. abscessus, M. abscessus ΔMAB_2385, M. abscessus ΔMAB_2385 pMV361-aac(3)IV, and M. abscessus ΔMAB_2385 pMV361-aac(3)IV-MAB_2385 strains. The M. abscessus wild type had a MIC of 64 mg/liter. The M. abscessus ΔMAB_2385 deletion mutant exhibited significantly increased susceptibility (1.5 mg/liter), while the complementation of MAB_2385 restored the wild-type phenotype (96 mg/liter). The vector pMV361-aac(3)IV did not affect the streptomycin susceptibility of M. abscessus ΔMAB_2385 (Fig. 3).

FIG 3.

Streptomycin Etest of M. abscessus ATCC 19977^T, the MAB_2385 gene deletion mutant M. abscessus ΔMAB_2385, and transformants of the vector backbone pMV361-aac(3)IV and the complementation vector pMV361-aac(3)IV-MAB_2385. The bacterial suspension was evenly spread on the LB agar plate before the streptomycin Etest strip was placed on the surface of the plate. The plates were incubated for 5 days. The MIC of the wild type is 64 mg/liter, while the gene deletion mutant and the vector backbone transformant have a MIC of 1.5 mg/liter. The complemented mutant has a MIC similar that of the wild type, 96 mg/liter.

Susceptibility to streptomycin and other aminoglycosides was subsequently determined by broth microdilutions. Table 1 shows the median MICs evaluated on day 5. M. abscessus ΔMAB_2385 is 16-fold more susceptible to streptomycin than the wild type and the complementation mutant. Susceptibility to other aminoglycosides was not altered. In M. abscessus ΔMAB_2385 pMV361-aac(3)IV-MAB_2385, the wild-type phenotype was restored, while the M. abscessus ΔMAB_2385 pMV361-aac(3)IV control strain remained susceptible to streptomycin. On days 3, 7, and 12, similar differences in streptomycin MICs were observed (see Table S1 in the supplemental material).

TABLE 1.

Median MIC values after 5 days

Drug	MIC (mg/liter)a
	Wild type	ΔMAB_2385
		No vector	pMV361-aac(3)IV	pMV361-aac(3)IV-MAB_2385
Streptomycin	32	2	2	32
Spectinomycin	>256	>256	>256	>256
Apramycin	0.5	0.5	NAb	NA
Kanamycin A	1	1	NA	NA
Kanamycin B	8	8	NA	NA
Amikacin	2	1	2	2
Hygromycin B	256	256	256	256

^aBroth microdilutions in cation-adjusted Mueller Hinton broth.

^bNA, the aac(3)IV gene confers resistance to several aminoglycosides, including kanamycin A and kanamycin B, and thus, drug susceptibility testing was not applicable.

Generation and characterization of spontaneous streptomycin-resistant mutants.

The streptomycin-susceptible M. abscessus ΔMAB_2385 was plated on agar plates containing high streptomycin concentrations (32 mg/liter and 128 mg/liter) to select for spontaneous streptomycin-resistant colonies. Streptomycin-resistant colonies appeared at frequencies of approximately 10⁻⁶ and 10⁻⁷ on 32-mg/liter and 128-mg/liter streptomycin LB agar, respectively. Colonies were picked and screened by Sanger sequencing for alterations in rpsL and rrs. We identified mutations resulting in amino acid substitutions at codon 43 (K43N and K43R) and codon 88 (K88R) in RpsL. Other streptomycin-resistant colonies without any mutation in rpsL carried C491T mutations in rrs (Fig. 1). For further studies and to establish an rpsL-based counterselection marker in M. abscessus, a strain carrying an RpsL K43R alteration was used.

Reversion of the streptomycin-resistant phenotype.

In several bacterial species, merodiploid strains carrying an rpsL⁺ and an rpsL^mut allele are streptomycin susceptible, indicating that the wild-type allele is dominant (42). We transformed a single-copy integrating vector carrying the wild-type [pMV361-aac(3)IV-rpsL⁺] or the K43R [pMV361-aac(3)IV-rpsL^K43R] mutated rpsL into the streptomycin-resistant M. abscessus ΔMAB_2385 rpsL^K43R strain. As control, we transformed the empty backbone vector pMV361-aac(3)IV. The strains were plated on LB agar containing apramycin or a combination of apramycin and streptomycin and were incubated for 5 days before CFU were determined. Apramycin selection was applied to avoid vector loss through integrase-mediated excision (56). A one-tailed (H_A<H₀) two-sample Student's t test [t test with 4 degrees of freedom: t(4) = −7.3853; P < 0.001] revealed a significantly (10⁴-fold) reduced number of CFU of the strain carrying the pMV361-aac(3)IV-rpsL⁺ vector when it was plated on apramycin and streptomycin compared to apramycin alone. Bacteria carrying the vectors pMV361-aac(3)IV [t(4) = −0.1261; P = 0.4529] and pMV361-aac(3)IV-rpsL^K43R [t(4) = −1.3399; P = 0.1257] did not show reduced numbers of CFU when plated on apramycin and streptomycin compared to apramycin alone (Fig. 4). The data indicate that rpsL⁺ confers a streptomycin-susceptible phenotype on M. abscessus ΔMAB_2385 rpsL^K43R.

FIG 4.

CFU obtained after plating M. abscessus ΔMAB_2385 rpsL^K43R strains carrying pMV361-aac(3)IV (empty control), pMV361-aac(3)IV-rpsL⁺, or pMV361-aac(3)IV-rpsL^K43R on apramycin (gray bars) or apramycin and streptomycin (white bars) selective LB agar plates. Each strain was diluted to an OD₆₀₀ of 0.1, plated on apramycin- or apramycin- and streptomycin-containing LB agar plates, and incubated for 5 days. Bacterial counts on the different antibiotics were determined. The plot shows the means and standard deviations of the results of three independent experiments. One-tailed two-sample Student's t test: ***, P < 0.001 n.s., not significant.

Streptomycin as a counterselection marker.

Having shown that rpsL⁺ confers streptomycin susceptibility on M. abscessus ΔMAB_2385 rpsL^K43R, we wanted to test whether the rpsL⁺ counterselection method could be used for the generation of targeted deletion mutants in M. abscessus. For a proof of concept study, we chose the eis2 (MAB_4532c) locus, which was previously deleted by Rominski et al. (22). We constructed a novel allelic replacement vector by cloning the rpsL⁺ allele and the DsRed2 fluorescence marker into the eis2 suicide vector pSE-Δeis2 to produce pSE-aac(3)IV:DsRed2-Δeis2-rpsL⁺. The vector was transformed in M. abscessus ΔMAB_2385 rpsL^K43R. Transformants were selected on apramycin plates and identified by red fluorescence. Single-crossover recombinants identified by Southern blotting were plated on streptomycin LB agar to select for double-crossover recombinants. The CFU counts on streptomycin plates were compared to the CFU counts on apramycin-selective LB agar plates. In three independent experiments, the mean frequency was 8.25 × 10⁻⁴ based on CFU counts. Fluorescence imaging determined that on average 11.14% of the clones still carried the DsRed2 of the suicide vector, while almost 89% did not show red fluorescence. Twenty nonfluorescent colonies were picked and screened by Southern blotting for second intramolecular recombination. All the colonies had undergone this intramolecular recombination. Five were unmarked eis2 gene deletion mutants, while 15 were eis2 wild-type revertants. These results demonstrated that the streptomycin counterselection strategy can be applied to generate deletion mutants in the M. abscessus ΔMAB_2385 rpsL^K43R strain (Fig. 5).

FIG 5.

Southern blot analysis of the eis2 gene deletion mutant, with streptomycin used for counterselection. Genomic DNA of the streptomycin-resistant M. abscessus ΔMAB_2385 rpsL^K43R (lane 1); after transformation with pSE-(aac3)IV:DsRed2-Δeis2-rpsL⁺ vector (lane 2); and each of the two possible outcomes after counterselection with streptomycin, the gene deletion mutant (lane 3) or the parental strain (lane 4), was isolated and digested with Van91I. The pattern obtained is in accordance with the pattern reported by Rominski et al. (22); a 1.1-kpb wild-type fragment, a 1.1-kbp and a 3.4-kbp fragment for single-crossover transformants, and a 4.8-kbp fragment for the eis2 gene deletion mutant.

DISCUSSION

MAB_2385 is the intrinsic streptomycin resistance determinant in M. abscessus.

M. abscessus is a clinically relevant pathogen with several intrinsic resistance mechanisms. Increased interest in this highly resistant mycobacterium resulted in the identification of resistance mechanisms for macrolides (19), rifamycins (20), β-lactams (21), and capreomycin and aminoglycosides (22). The molecular mechanism of M. abscessus streptomycin resistance has not been identified yet. Alterations in the rpsL or the rrs gene associated with clinically acquired streptomycin resistance in M. tuberculosis have not been found in M. abscessus (29). In the genome of M. abscessus, several putative aminoglycoside-modifying enzymes, mostly aminoglycoside phosphotransferases, were annotated (18). By scanning the genome of M. abscessus for the presence of the streptomycin resistance determinant APH(3″) of M. fortuitum, we identified MAB_2385 as a candidate resistance gene. Heterologous expression of MAB_2385 in M. smegmatis increased MIC levels. Subsequent drug susceptibility testing in M. abscessus ΔMAB_2385 showed increased susceptibility to streptomycin compared to the wild type and the complemented strain. The susceptibility of M. abscessus ΔMAB_2385 to other aminoglycosides was not affected. Heinzel et al. had previously demonstrated that 3″-phosphotransferases specifically phosphorylate streptomycin, but not kanamycin (57). These findings support the assumption that MAB_2385 is a streptomycin 3″-O-phosphotransferase. However, additional biochemical studies should be performed to directly demonstrate streptomycin-modifying activity of MAB_2385.

Streptomycin as a counterselection marker in M. abscessus.

So far, the genetic toolbox for M. abscessus is ill equipped, and only a few efficient positive and negative selectable markers have been described (22, 39, 40). Sensitizing M. abscessus to streptomycin by deletion of MAB_2385 and subsequent selection of streptomycin-resistant rpsL mutants offers the possibility to exploit the rpsL⁺ counterselection strategy described previously for M. tuberculosis, M. bovis BCG, and M. smegmatis (41, 58). Hence, we selected rpsL mutants resulting in K43R substitutions, since high-level resistance was observed in other mycobacteria carrying the corresponding allele (32).

The introduction of a vector with the wild-type small ribosomal protein S12 allele (rpsL⁺) into M. abscessus ΔMAB_2385 rpsL^K43R significantly reduced the number of colonies grown on streptomycin-containing agar plates by a factor of approximately 10⁻⁴, indicating that rpsL⁺ is an efficient counterselection marker. Intramolecular recombination between the wild-type and the mutant rpsL most likely explains why a few bacteria still grew on streptomycin plates. The frequency is approximately 10⁻⁴ to 10⁻⁵ (59). Other events, such as the occurrence of spontaneous resistance mutations in the wild-type rpsL (10⁻⁶) or the loss of the vector catalyzed by the vector's integrase (8 × 10⁻⁵) (56) coupled with spontaneous apramycin resistance-conferring mutations (10⁻⁶) (in total, 8 × 10⁻¹¹) are less frequent. Finally, in a proof of concept study, we generated a gene deletion mutant by introducing rpsL⁺ in a modified eis2 allelic replacement vector and subsequent streptomycin counterselection, thereby establishing a new negative selection marker for M. abscessus.

In summary, we demonstrated that MAB_2385 specifically confers streptomycin resistance but does not have an effect on susceptibility to other aminoglycosides. Further, we were able to generate spontaneous streptomycin-resistant mutants in the M. abscessus ΔMAB_2385 strain carrying mutations in rpsL. This enabled us to introduce an rpsL-based streptomycin counterselection system in M. abscessus by exploiting the fact that in merodiploid strains the wild-type rpsL confers a streptomycin-susceptible phenotype.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

M. abscessus ATCC 19977^T was grown either in Middlebrook 7H9-OADC (oleic acid-albumin-dextrose-catalase) (BD Difco, Sparks, MD, USA) with 0.05% Tween 80 (7H9) (Hänsler AG, Switzerland) or on solid LB agar plates. Kanamycin A (50 mg/liter), apramycin (50 mg/liter), isoniazid (32 mg/liter), or streptomycin (32 mg/liter or 128 mg/liter) (Sigma-Aldrich, Switzerland) was added for primary selection of transformants and counterselection and isolation of spontaneous resistant mutants, respectively. Preparation of electrocompetent cells (22) and extraction of genomic DNA with phenol-chloroform-isoamyl alcohol (41) were described previously.

M. smegmatis mc² 155 was grown in LB with 0.05% Tween 80 or 7H9 broth and on Middlebrook 7H10-OADC or LB agar plates. For primary selection, kanamycin A (50 mg/liter) or apramycin (50 mg/liter) was supplemented. Electrocompetent cells were prepared as described previously, except that 7H9 was used as the growth medium (60). Escherichia coli XL1-Blue MRF′ was grown in LB broth or on LB agar plates containing either ampicillin (liquid, 100 mg/liter; solid, 120 mg/liter) or kanamycin A (50 mg/liter) (Sigma-Aldrich, Switzerland). All bacterial strains were grown at 37°C.

Vector construction and generation of strains.

The strains used in this study are listed in Table 2. Promega's (Switzerland) pGEM-T easy vector system was used for cloning of PCR fragments. The integrative single-copy vector pMV361-aac(3)IV and the suicide vector pSE-apr-katG (pSE) were described previously (22). For heterologous expression of MAB_2385 in M. smegmatis and for complementation, a 1,606-bp HindIII/NheI fragment of the MAB_2385 region was amplified from genomic DNA of M. abscessus ATCC 19977^T by PCR with the primers 5′-CGGAAGCTTCAAGGATTACCCGACACCAC-3′ and 5′-CCGCTAGCTGGATCAGCTCCAGGATG-3′ (restriction enzyme recognition sites are underlined). The PCR fragment was ligated into pGEM-T Easy and subsequently cut out with HindIII and NheI. pMV361-aac(3)IV was also cut with HindIII and NheI, and the MAB_2385 fragment was ligated into the linearized backbone, resulting in pMV361-aac(3)IV-MAB_2385. The vectors pMV361-aac(3)IV and pMV361-aac(3)IV-MAB_2385 were transformed into M. smegmatis mc² 155.

TABLE 2.

Strains used in this study

Strain	Vector	Description	Source
E. coli XL1-Blue MRF′		Cloning	Stratagene
M. smegmatis mc2 155		Wild type	68
	pMV361-aac(3)IV	Backbone control	This study
	pMV361-aac(3)IV-MAB_2385	Expression of MAB_2385	This study
M. abscessus ATCC 19977T		Wild type	69
M. abscessus ΔMAB_2385a		Gene deletion mutant	This study
	pMV361-aac(3)IV	Backbone control	This study
	pMV361-aac(3)IV-MAB_2385	Gene complementation	This study
M. abscessus ΔMAB_2385 rpsLK43Rb		K43R alteration in RpsL	This study
	pMV361-aac(3)IV	Backbone control	This study
	pMV361-aac(3)IV-rpsL+	Expression of rpsL+	This study
	pMV361-aac(3)IV-rpsLK43R	Expression of rpsLK43R	This study

^aM. abscessus ΔMAB_2385 was derived from M. abscessus ATCC 19977^T.

^bM. abscessus ΔMAB_2385 rpsL^K43R was derived from M. abscessus ΔMAB_2385.

The gene deletion suicide vector pSE-ΔMAB_2385 was generated by cloning two regions flanking MAB_2385. A 1,548-bp PscI/KpnI upstream fragment and a 1,484-bp KpnI/HindIII downstream fragment were amplified by PCR using genomic DNA of M. abscessus ATCC 19977^T. The primers for the amplification of the upstream fragment were 5′-GGGACATGTCGGGGATTTCTACACGCTTA-3′ and 5′-CGGGGTACCTTTCGCATAGCGAGAACC-3′, while the primers for the amplification of the downstream fragment were 5′-CGGGGTACCAACCGGTACGGAGCGTCT-3′ and 5′-CCCAAGCTTACACACACGTGATGCAGACC-3′. Both fragments were separately cloned into the pGEM-T Easy vector, subsequently cut with the appropriate enzymes, and stepwise ligated into the linearized pSE vector. The vector pSE-ΔMAB_2385 was transformed in M. abscessus ATCC 19977^T. Apramycin was used for positive selection and isoniazid for counterselection. Genomic DNA of M. abscessus transformants was isolated and cut with Tth111I to confirm gene deletion by Southern blotting. The vector pMV361-aac(3)IV-MAB_2385 was transformed into the M. abscessus ΔMAB_2385 deletion mutant to complement the strain. As a control, the vector backbone pMV361-aac(3)IV was transformed in parallel.

A suspension of mid-log-phase M. abscessus ΔMAB_2385 was plated on LB agar containing streptomycin (32 mg/liter and 128 mg/liter) and incubated for 5 days. Colonies were picked and screened for alterations in rpsL by Sanger sequencing with the primers 5′-GTTACCAGCTGCGAACCGTA-3′ and 5′-GAAAAACGCAGGACAACAGG-3′. A clone carrying a K43R amino acid substitution (M. abscessus ΔMAB_2385 rpsL^K43R) was used for further studies. Mutations in rrs were detected by sequencing with the primers 5′-ACCGGGAATTTGACTCAGGT-3′ and 5′-GAAAACGAGCGAGGCTATGT-3′.

The wild-type (rpsL⁺) and the mutated (rpsL^K43R) rpsL were PCR amplified with the primers 5′-CGGAAGCTTCTCCAGAGCGCCGTACAC-3′ and 5′-CCCGCTAGCCGCACTGTCGTTCCTGGAT-3′. This resulted in 1,375-bp HindIII/NheI fragments, which were cloned into pGEM-T Easy, cut with HindIII and NheI, and subsequently ligated into the HindIII/NheI-linearized pMV361-aac(3)IV vector. The resulting vectors, pMV361-aac(3)IV-rpsL⁺ and pMV361-aac(3)IV-rpsL^K43R, as well as the backbone vector pMV361-aac(3)IV, were transformed into M. abscessus ΔMAB_2385 rpsL^K43R.

A wild-type rpsL fragment (1,375 bp) with MunI (instead of HindIII/NheI) restriction sites was amplified by PCR with the primers 5′-CGGCAATTGCTCCAGAGCGCCGTACAC-3′ and 5′-CCCCAATTGCGCACTGTCGTTCCTGGAT-3′. The fragment was cloned into pGEM-T Easy, reisolated by digestion with MunI, and inserted into the MunI-linearized eis2 suicide targeting vector pSE-Δeis2 (22), resulting in pSE-Δeis2-rpsL⁺. This vector was further modified to facilitate identification of transformants (which are mostly single-crossover transformants) and double-crossover recombinants resulting from counterselection of single-crossover transformants: the apramycin cassette was replaced by an apramycin (aac3)IV:DsRed2 fusion cassette to produce pSE-(aac3)IV:DsRed2-Δeis2-rpsL⁺. This vector was transformed in M. abscessus ΔMAB_2385 rpsL^K43R, and positive selection was performed with apramycin to select transformants resulting from intermolecular homologous recombination between cloned eis2 sequences and the target gene. Subsequent counterselection of single-crossover transformants was performed with streptomycin (32 mg/liter). The second (intramolecular) recombination events resulted in the deletion of eis2 or reversion to wild-type eis2. This was confirmed by Southern blotting as described by Rominski et al. (22).

Etest.

Bacteria were grown on LB agar plates. After incubation, a bacterial suspension was adjusted to a McFarland standard of 0.50. Therefore, 1 to 10 colonies were picked with a cotton swab and transferred to a glass tube containing 0.9% NaCl. A small volume of the suspension was evenly spread on an LB agar plate with a cotton swab. A streptomycin Etest strip (concentration range, 0.064 to 1,024 mg/liter) was placed in the middle of the plate and incubated for 5 days until growth at the border was well defined before the MIC was determined.

MIC assays.

Streptomycin, spectinomycin, apramycin, amikacin, kanamycin A, and kanamycin B (all from Sigma-Aldrich, Switzerland) were dissolved in water and filter sterilized. Hygromycin B (Invivogen, San Diego, CA, USA) was delivered as a 100-g/liter solution. The MIC assays were in principle performed according to CLSI guidelines for antimicrobial susceptibility testing, with slight modifications, and were conducted essentially as described previously in 96-well microtiter plates (Corning Inc., NY, USA) (20, 61). For preparation of the inocula, M. abscessus colonies were picked from LB plates as described previously, while M. smegmatis was grown in liquid cultures. Growth of M. abscessus was visually inspected on days 3, 5, 7, and 12, while growth of M. smegmatis was visually inspected on day 3.

Selectivity in merodiploid rpsL strains.

M. abscessus ΔMAB_2385 rpsL^K43R carrying any of the three vectors pMV361-aac(3)IV, pMV361-aac(3)IV-rpsL⁺, and pMV361-aac(3)IV-rpsL^K43R was inoculated in 10 ml 7H9 with apramycin (50 mg/liter) and grown for 5 days. The optical density at 600 nm (OD₆₀₀) was adjusted to 0.1. Subsequently, 10-fold serial dilutions of the bacterial suspension were prepared, and 100 μl was spread on LB agar plates with apramycin (50 mg/liter) or a combination of apramycin (50 mg/liter) and streptomycin (32 mg/liter). After 5 days of incubation, CFU were counted. Statistical analysis was performed with the statistical software R (version 3.1.1) (62).