Efflux Pumps Contribute to Intrinsic Clarithromycin Resistance in Clinical, Mycobacterium abscessus Isolates

Introduction

Infections caused by nontuberculous mycobacteria (NTM) have increased dramatically worldwide in recent years.^1–4 Mycobacterium abscessus is one of the most common NTM detected. M. abscessus infections are difficult to manage because M. abscessus is intrinsically resistant to a variety of antimicrobials currently available in clinical practice; infections that are refractory to antibiotic therapy frequently result in morbidity and mortality.^5–7 Recently, human-to-human M. abscessus transmission was reported, making the problem even more disconcerting.⁸

Clarithromycin (CLA), which is effective in treating M. abscessus lung disease, is recommended as the core agent for treatment of infections although effective therapeutic options are evolving.^9,10 The emergence of CLA resistance, however, is challenging. Currently, a 2270/2271 point mutation (Escherichia coli 2058/2059) in the 23S rRNA (rrl) gene and a full-length erm(41) gene with thymine located at position 28 [erm(41)full-T28 sequevar] are the best described mechanisms conferring CLA resistance. The erm(41)full-T28 sequevar confers induced CLA resistance and a rrl 2270/2271 point mutation confers intrinsic CLA resistance in the M. abscessus isolates assessed.¹¹ The rrl mutation, however, fails to account fully for the intrinsic resistance to CLA exhibited by M. abscessus.^12–15 In addition to a rrl 2270/2271 mutation, the expression of a variety of efflux pump genes correlates with CLA resistance in mycobacterium.^16–24 Almost all the homologous, efflux pump genes are found in M. abscessus; whether these efflux pump genes are involved in the resistance of clinical M. abscessus isolates to CLA remains to be determined.

In this study, the MIC of CLA was determined for 194 clinical M. abscessus isolates collected from patients with lung diseases. A comprehensive exploration of the molecular mechanisms of CLA resistance was performed by combining comparative genome sequence analysis and qRT-PCR; the efflux pump genes were a specific focus. In addition to mutations in the CLA target-site genes, efflux pump genes MAB_2355c, MAB_1409c and MAB_1846 were found to play an important role in CLA resistance. To our knowledge, this is the first comprehensive mechanistic investigation of intrinsic CLA resistance in a large number of clinical, M. abscessus isolates. This work extends our understanding of the factors that affect the resistance of M. abscessus to CLA and suggests novel approaches to treating CLA-resistant infections.

Materials and Methods

Bacterial Isolation and Identification

Between January 2014 and December 2017, 194 M. abscessus isolates were collected at Shanghai Pulmonary Hospital from sputum and bronchoalveolar lavage fluid samples of patients with M. abscessus lung infections. All isolates were preserved in the Clinical Microbiology Laboratory. Species were preliminarily screened for NTM by MGIT960 medium culture and the p-nitrobenzoic acid test, followed by molecular identification of M. abscessus by sequencing the rpoB and erm(41) genes.²⁵ Each isolate was then stored at −80°C until use.

Clarithromycin Susceptibility Test

CLA susceptibility was determined by the broth microdilution method according to Clinical and Laboratory Standards Institute (CLSI) guidelines; the breakpoints were interpreted according to (CLSI)-M24-A2, as follows: susceptible, MIC ≤ 2 mg/L; intermediate, MIC = 4 mg/L; and resistant, MIC ≥ 8 mg/L. The intrinsic resistance of M. abscessus to CLA was assessed after 3 days exposure. Mycobacterium peregrinum (ATCC 700686; American Type Culture Collection, Manassas, VA, USA) and Staphylococcus aureus (ATCC 29213; American Type Culture Collection, Manassas, VA, USA) served as control reference strains.

Whole-Genome Sequencing and Comparison of Resistance Genes

One hundred ninety-four isolates were sequenced. DNA extraction, library construction and sequencing were conducted by methods previously reported by us.²⁶ The full genome sequence of each isolate has been published and is available at DDBJ/ENA/GenBank (96 sequences found under bioproject PRJNA488058, 35 under bioproject PRJNA448987 and 63 under bioproject PRJNA398137).^26,27 Sequence homologs were identified by BLAST and aligned with the homologous sequences of the M. abscessus reference strain ATCC19977 (NC_010397.1). Ribosome structural genes, rrl, rplC, rplD and rplV, and rrl modification gene erm(41) were also included in this analysis.

RNA Extraction and Quantitative Reverse Transcription PCR (qRT-PCR)

Seven homolog efflux pump genes and the regulatory gene whiB7 (MAB_3508c) correlate with macrolides resistance in mycobacteria, were selected and analyzed. Relative expression of the genes was assessed by comparing the quantity of mRNA expressed by the organism cultured in the presence and absence of CLA using the same technical approach reported previously.²⁸ A culture incubated in the presence of half its MIC of CLA was shaken at 37°C for 3 h; the RNA was then extracted according to the protocols described by Medjahed et al²⁹ cDNA was synthesized using the HiScript III RT SuperMix with gDNA wiper (Vazyme Biotech Co., Ltd). qRT-PCR was performed using ChamQ Universal SYBR Master Mix (Vazyme Biotech Co., Ltd) on a QS6 Real-Time PCR System (Applied Biosystems, Carlsbad, CA). SigA was chosen as the endogenous reference gene. All PCR primer pairs used for amplification are shown in Supplementary Table 1. Calculation of fold change was described previously in detail.³⁰ Reactions were repeated in triplicate; genes with expression levels ≥4 were considered overexpressed

Efflux Pump Inhibition Assay

The MIC of CLA used in combination with efflux pump inhibitors, phenylalanine-arginine β-naphthylamide (PaβN), carbonyl cyanide 3-chlorophenylhydrazone (CCCP) or verapamil (VP), was determined as previously described.^31,32 A 2-fold decrease in MIC in the presence of an inhibitor was considered significant. The efflux pump inhibition test was performed in triplicate.

Results

Clarithromycin Susceptibility and Resistance Profiles

One hundred ninety-four clinical, M. abscessus isolates were collected; 148 isolates belonged to subsp. abscessus and 46 isolates belonged to subsp. massiliense. The MICs of CLA ranged from 0.06 to 256 mg/L for these isolates; MIC₅₀ and MIC₉₀ at 3 days were 0.5 and 2 mg/L, respectively.

A rrl 2270/2271 mutation confers intrinsic resistance to CLA (MIC ≥8 mg/L accessed after 3 days). Accordingly, a comprehensive analysis of the distribution of this resistant genotype and resistance profile among the 194 isolates was performed. A total of 13 (6.7%, 8 subsp. abscessus and 5 subsp. massiliense) resistant isolates were found (Table 1). Among these, only 7 (53.8%) of the isolates harbored a rrl 2270/2271 mutation while 6 isolates with no rrl 2270/2271 mutation also exhibited intrinsic resistance.

Table 1 MIC of Clarithromycin for 194 Clinical M. abscessus Isolates

Gene Comparison of M. abscessus Isolates

CLA target-site genes rrl, rplC, rplD and rplV, encoding the ribosome structure, were further analyzed among all 194 isolates. Besides the 2270/2271 mutation, 23 additional mutations were observed in rrl although none specifically presented in CLA-susceptible or -nonsusceptible isolates (Supplementary Table 2). Moreover, no meaningful mutation in ribosome protein genes rplC, rplD and rplV was found in the resistant isolates. The rrl modification gene erm(41) was also analyzed; no specific sequence changes related to CLA resistance were found.

Transcriptional Analysis of Efflux Pumps

The putative efflux pump genes in M. abscessus were identified by literature review and analysis of homology to genes found in other Mycobacterium strains (Table 2). Six CLA-resistant isolates (MIC ≥8 mg/L) harboring a rrl mutation and 6 resistant isolates with no rrl mutation, along with 6 CLA sensitive (MIC ≤2 mg/L) and 6 CLA intermediate (MIC ≤4 mg/L) isolates were selected at random. The expression of efflux pump genes was quantified by qRT-PCR. Among the 7 putative efflux pump genes analyzed, the transcriptional levels of MAB_2355c (Figure 1A), MAB_1409c (Figure 1B) and MAB_1846 (Figure 1C), as well as their positive regulatory gene whiB7 (Figure 1D), showed consistent increases that correlated inversely with decreases in clarithromycin susceptibility (Figure 1). No significant changes in expression of the four remaining putative efflux pump genes were observed after CLA exposure (data not shown). The efflux pump genes in the unexplained (rrl mutation-negative) intrinsic resistant isolates showed the highest level of expression. Among the efflux pump genes, for example, MAB_2355c exhibited >40-fold and >20-fold increases in transcription in the rrl mutation-negative and rrl mutation-positive isolates, respectively.

Table 2 Blast Search of the M. abscessus Genome with Reference Gene Sequences Corresponding to Efflux Pumps Conferring or Putatively Involved in CLA Resistance

Figure 1 Efflux pump gene expression by clinical, M. abscessus isolates following exposure to CLA stress. Isolates that were CLA-susceptible, -intermediate and resistant [without (unexplained) or with a rrl 2270/2271 mutation] are shown. MAB_2355c (A), MAB_1409c (B), MAB_1846 (C) and whiB7 (D), message levels were quantified by qRT-PCR. Open circles, triangles, diamonds, and solid circles indicate the distribution of mRNA expression levels in isolates comprising the sensitive, intermediate, unexplained drug resistance, and rrl 2270/2271 mutation groups, respectively. Experiments were repeated >3 times for each isolate; error bars represent the standard error of the mean for each data set.

Efflux Pump Inhibition Assay

Previously, three highly expressed genes (MAB_2355c, MAB_1409c and MAB_1846) were found to encode efflux pumps capable of extruding CLA in mycobacteria (15–19). An efflux pump inhibition test was conducted to evaluate the role of these efflux pumps in intrinsic CLA resistance exhibited by M. abscessus. The addition of efflux pump inhibitors PAβN, CCCP and VP significantly decreased the MIC of CLA 47.4%, 52.6% and 63.2%, respectively, for the nonsusceptible isolates while there was no significant effect on the CLA sensitive isolates (Table 3). This decrease was more pronounced in resistant isolates that did not harbor a rrl 2270/2271 mutation. Although the level of efflux pump gene expression was significantly elevated, the inhibitory effect of efflux pump inhibitors was nearly absent in resistant isolates characterized by a rrl 2270/2271 mutation.

Table 3 Efflux Pump Inhibition Decreases the MIC of Clarithromycin for Non-Susceptible Isolatesa

Discussion

CLA remains the core antibiotic used to treat M. abscessus infections. Treatment options and outcomes differ significantly among patients infected with CLA-resistant versus -sensitive M. abscessus genotypes.³³ In a certain proportion of clinical M. abscessus isolates, resistance cannot be explained by the genotype,^12–15 alternate mechanisms exist. In this study, 194 isolates were obtained from M. abscessus lung infection cases in mainland China. In addition to CLA-target site mutations, efflux pumps played an important role in the intrinsic resistance of M. abscessus to CLA.

Almost all known mechanisms of intrinsic resistance to CLA involve changes in the CLA binding site due to mutations in rrl (i.e., 2270/2271) or adjacent sites.^11–15 The results reported provide additional support to indicate that the rrl genotype correctly predicts the CLA-resistant phenotype of most (181/194, 96.9%) clinical isolates. However, 6 (3.1%) intrinsic resistant isolates in this study exhibited an inconsistent, rrl mutation-negative, genotype. Recently, Lipworth and colleagues identified eight potential new resistance-conferring single nucleotide polymorphisms (SNPs) present only in the rrl and erm(41) genes of antibiotic-resistant, but not sensitive, isolates.¹⁴ None of the predicted SNPs or other resistance-conferring SNPs were found in rrl in the current study. Reportedly, mutations in ribosomal protein genes rplC, rplD and rplV are also associated with structural resistance to macrolides.³⁴ Mutations in these genes, however, were absent among all the CLA-resistant isolates collected in this study.

Efflux was an anticipated factor in the drug resistance exhibited by clinical M. abscessus isolates collected here. Multi-drug efflux pumps directly mediate the resistance of a number of mycobacteria species to CLA.^16–24 Reportedly, efflux inhibitors reduce the MIC of CLA for M. abscessus.^31,35 However, CLA-related efflux pumps have not been systematically demonstrated and validated in clinical, M. abscessus isolates specifically. In the present study, three putative efflux pumps (an ATP binding cassette efflux pump encoded by MAB_2355c, a tap like efflux pump encoded by MAB_1409c, and an ATP binding cassette efflux pump encoded by MAB_1846) contributed to CLA resistance exhibited by M. abscessus. The ATP binding cassette efflux pump encoded by MAB_2355c was the most notable. MAB_2355c is a homolog of genes found in: M. avium (MAV_3306), M. tuberculosis (Rv1473) and M. smegmatis (MSMEG_3140). Schmalstieg and colleagues first reported that exposure to azithromycin over a 3-day period led to a 56-fold increase in expression of MAV_3306, which encodes the ATP binding cassette efflux pump in M. avium.¹⁷ They envisioned that this increase was the first step in a general pathway to drug resistance, which eventually leads to high-level, chromosome mutation-related resistance. Macrolide resistance mediated by the ATP binding cassette efflux pump encoded by Rv1473 was further characterized in 2019.¹⁶ Rv1258c (MAB_1409c homolog) is associated with multiple drug efflux in M. tuberculosis.¹⁸ We also found this efflux pump was closely related to amikacin resistance in M. abscessus.³⁶ Recently, Vianna and colleagues reported the association between MAB_1409c overexpression and CLA resistance in M. abscessus, a finding consistent with that reported herein.³⁷ In contrast to their results, however, we found no correlation between MAB_3142c and CLA resistance; presumably, sample bias contributes to this difference. Purportedly, Rv2477c (MAB_1846 homolog) was involved in mycobacterial protein translation, and resistance to tetracyclines and macrolides in M. tuberculosis.²⁰ This finding is consistent with our results showing MAB_1846 overexpression is associated with macrolides (CLA) resistance. Notably, MAB_2355c, MAB_1409c and MAB_1846 encode factors that belong to the intrinsic macrolides resistance network which is positively regulated by whiB7. WhiB7 is a transcription factor that promotes intrinsic CLA resistance through self-activation and the activation of downstream genes involved in CLA efflux and the protection of ribosomes against antibiotic stress.¹⁶ Taken together, these findings are consistent with the results reported herein.

It is generally believed that the induction of efflux pumps is the first event leading to increased antibiotic resistance expressed by Mycobacterium, which usually exhibits low or moderate resistance prior to drug exposure.¹⁷ Similarly, most resistant isolates that lacked the rrl 2270/2271 mutation exhibited low or moderate resistance in the study reported here (ie, MIC ≤16mg/L, Supplementary Table 3), but exhibit relative higher expression of efflux pumps than do resistant isolates with rrl 2270/2271 mutation. Moreover, efflux pump inhibitors exerted a greater effect on resistant isolates that lack a rrl 2270/2271 mutation. As such, we speculate that a change in the target-site (e.g., a rrl 2270/2271 mutation) is the ultimate event that culminates in a high level of CLA resistance. Efflux pump overexpression, on the other hand, is a reversible, compensatory drug resistance mechanism making efflux pumps a potentially better target for clinical intervention. Whether efflux pump expression affects macrolide resistance induced by functioning erm(41) is currently unknown. Inasmuch as increases in both erm(41) and efflux pump expression reduce CLA sensitivity; however, we speculate that efflux pump expression can further enhance macrolide resistance dependent upon the degree of resistance induced by functional erm(41).

Conclusion

The present study identified three efflux pumps associated with the intrinsic resistance of M. abscessus to clarithromycin. Identification of these pumps adds to our current understanding of the factors that affect drug resistance, and suggests novel approaches to treating CLA-resistant M. abscessus infections.