Molecular Analysis of Linezolid-Resistant Clinical Isolates of Mycobacterium abscessus

A total of 194 Mycobacterium abscessus isolates were collected from patients, and the whole genomes were sequenced. Eighty-five (43.8%) isolates showed linezolid (LZD) resistance.

ABSTRACT

A total of 194 Mycobacterium abscessus isolates were collected from patients, and the whole genomes were sequenced. Eighty-five (43.8%) isolates showed linezolid (LZD) resistance. Only 8.2% of resistant isolates harbored 23S rRNA mutations. Quantitative real-time PCR (qRT-PCR) revealed higher transcriptional levels of efflux pumps lmrS and mmpL9 in LZD-resistant isolates. Genome comparative analysis identified several new LZD resistance-associated genes. This study highlights the role of efflux pumps in LZD-resistant M. abscessus and proposes potential target genes for further studies.

INTRODUCTION

Mycobacterium abscessus, one of the important nontuberculous mycobacterial (NTM) pathogens (1), causes human infections with high morbidity and mortality (2, 3). However, chemotherapeutic options against infections caused by M. abscessus are very limited due to its innate resistance to multiple antibiotic classes (4).

Linezolid (LZD), the first member of the oxazolidinone class, has been reported to be one of the most potent antibiotics against infections caused by M. abscessus (3, 5). Unfortunately, LZD-resistant M. abscessus strains are emerging worldwide (6, 7). Almost all resistance mechanisms against LZD reported to date involve alterations of LZD binding sites, including mutations in 23S rRNA and ribosomal proteins (L3, L4, and L22), or modifications of 23S rRNA, which were mainly investigated in M. tuberculosis, Staphylococcus spp., and Enterococcus spp. (8–10).

To date, knowledge on LZD resistance mechanisms in M. abscessus is limited. In this study, we collected 194 M. abscessus clinical isolates and sequenced all the genomes. Further investigation of resistance mechanism was performed in 85 LZD-resistant clinical isolates.

Screening of LZD-resistant isolates.

One hundred ninety-four M. abscessus isolates were collected in Shanghai Pulmonary Hospital from sputum and bronchoalveolar lavage fluid samples between January 2012 and December 2017. LZD MICs were determined by a broth microdilution method according to CLSI guidelines, and the breakpoints were interpreted according to CLSI document M24-A2 (≤8 mg/liter, susceptible; 16 mg/liter, intermediate resistant; ≥32 mg/liter, fully resistant) (11). Mycobacterium peregrinum ATCC 700686 and Staphylococcus aureus ATCC 29213 served as the control reference strains.

The MICs of LZD against 194 M. abscessus isolates ranged from 0.5 to 64 mg/liter, with an MIC₅₀ of 8 mg/liter and an MIC₉₀ of 32 mg/liter (Fig. 1A). Eighty-five (43.8%) isolates were resistant to LZD, 44 (22.6%) of which were intermediate resistant and 41 (21.2%) which were fully resistant. The remaining 109 (56.2%) isolates were susceptible to LZD. The LZD resistance rate of M. abscessus was high (43.8%), which is consistent with findings from previous studies (6, 7, 12–16).

FIG 1.

(A) Distribution of LZD MICs of 194 M. abscessus clinical isolates. The number and proportion of isolates are labeled on the top of each bar. (B) Schematic diagram of mutations in the 23S rRNA among 194 M. abscessus clinical isolates. Green indicates that the mutation is present only in LZD-susceptible isolates, red indicates that the mutation is present only in LZD-resistant isolates, and yellow indicates that the mutation is present in both LZD-susceptible and -resistant isolates.

Alternations in the LZD target sites.

Whole genomes of the 194 strains were sequenced (BioProject PRJNA488058 from this study and PRJNA448987 and PRJNA448987 from our previous studies), including 96 isolated in 2017 and 98 isolated during 2012 to 2016 (13, 17). The sequences of the entire 23S rRNA, L3, L4, and L22 proteins were extracted from the whole-genome sequence data of each strain and compared with those from reference strain ATCC 19977. A total of 26 mutation types were observed in 23S rRNA. Detailed information about the mutations is listed in Table S1 in the supplemental material. Nine mutations were found in 7 (8.2%) LZD-resistant strains, indicating that these mutations contributed to LZD resistance (Fig. 1B, red). Other 17 mutations in 23S rRNA were present in either susceptible strains or in both susceptible and resistant strains, suggesting that they do not contribute to LZD resistance. No meaningful mutations were found in L3, L4, and L22 in LZD-resistant strains. These results suggest that a mutation in ribosomal proteins is not responsible for LZD resistance in most of the strains isolated in this study.

The methyltransferase genes cfr, rlmN, and spr033 and the pseudouridine synthase gene rulC that modify the 23S rRNA at the LZD binding sites are known to affect LZD susceptibility (18–21). However, none of them were found in our 194 isolates.

Efflux pumps play an important role in LZD resistance of M. abscessus.

Several efflux pumps, including drrABC, rv0987, lmrS, acrAB, mmpL9, acrF, and optrA, have been reported to extrude LZD (22). Therefore, efflux pump inhibition tests were conducted with a combination of phenylalanine-arginine β-naphthylamide (PaβN, 20 mg/liter), carbonyl cyanide 3-chlorophenylhydrazone (CCCP; 5 mg/liter), and reserpine (12 mg/liter) (23, 24). As shown in Table 1, these inhibitors could decrease MICs of LZD in over 50% of resistant strains, supporting the role of efflux pumps in LZD resistance of M. abscessus.

TABLE 1.

MIC fold changes of 41 linezolid-resistant M. abscessus strains upon addition of efflux pump inhibitors^a

Treatment	No. (%) with MIC fold change decrease of:
	1	2	4
Linezolid + PAβN	12 (29.3)	26 (63.4)	3 (7.3)
Linezolid + CCCP	10 (24.4)	25 (61.0)	6 (14.6)
Linezolid + reserpine	19 (46.3)	20 (48.8)	2 (4.9)

^a1 represents no MIC fold change.

Sequence alignment showed that the homologs of drrABC, rv0987, lmrS, acrAB, mmpL9, and acrF were present in all of the 194 M. abscessus isolates, except for optrA.

Therefore, LZD-resistant isolates with a significant MIC fold change (4-fold) upon efflux pump inhibition (n = 6), along with 6 randomly selected LZD-susceptible isolates (MICs, 0.5 to 4 mg/liter), were selected and subjected to quantitative real-time PCR (qRT-PCR) analysis, as previously described (17). Primer pairs for amplification of each gene were as follows: mmpL9, ACGTCATTTCAGCTCTGCCA/AAGGGGCGGGTGATACTTTG; drrC, GTCGAGTACAGCACGCGATA/TAATCCGACCAGCAACCCAC; drrA, GTCCCGGATTGGCGAAATTG/GCTGCTTTTCCATCTCGCTG; lmrS, TGGTCAATGCTCGCATTCCT/ATCGGGTATCCCCTTGGTCA; acrF, ACTTCGTTGCGTTCCTCGAT/AGCGTTGTCACTCAACACCA; and acrB, GATTCGGTATCGGTGGCTGT/CCGGATTCTCCTCGACGAAC. As shown in Fig. 2, the LZD-resistant strains had >50-fold (P = 0.004) and >5-fold (P = 0.04) increased transcriptional levels of lmrS and mmpL9, respectively, compared to the LZD-susceptible strains. These results indicated that efflux pumps lmrS and mmpL9 play an important role in LZD resistance in M. abscessus. No difference in the transcription levels of drrABC, rv0987, acrAB, or acrF was observed between the LZD-susceptible and resistant groups (data not shown).

FIG 2.

qRT-PCR analysis of transcript levels of lmrS (A) and mmpL9 (B). Error bars represent the standard errors of each data point. A t test was used to test the differences among groups. Triangles (▲) indicate the strains whose LZD MIC decreased 4-fold after treatment with the inhibitor PAβN. Circles (●) indicate the strains whose LZD MIC decreased 4-fold after treatment with the inhibitor reserpine.

Whole-genome comparative analysis.

For 25% of the LZD-resistant M. abscessus isolates in this study, resistance could not be explained by known mechanisms, suggesting the presence of novel mechanisms for LZD resistance. Accordingly, genome comparative analysis was conducted and identified 24 genes that were highly associated with LZD resistance (P < 0.01), such as genes encoding MmpL10, which is known to mediate drug resistance in M. tuberculosis (25), and FabG, which is required for antibiotic resistance in P. aeruginosa (26). Detailed information for these genes is listed in Table S2.

In conclusion, this study suggests that rather than mutations or modifications of LZD target sites, efflux pumps played a predominant role in LZD resistance of M. abscessus. Whole-genome sequencing and comparative analyses also identified new LZD resistance-associated genes, which set the foundation for elucidation of the mechanism of LZD resistance in M. abscessus.

Accession number(s).

Whole-genome sequences have been deposited under BioProject no. PRJNA488058.