Evolution toward extremely high imipenem resistance in Mycobacterium abscessus outbreak strains

ABSTRACT

Treatment of Mycobacterium abscessus pulmonary disease requires multiple antibiotics including intravenous β-lactams (e.g., imipenem). M. abscessus produces a β-lactamase (Bla_Mab) that inactivates β-lactam drugs but less efficiently carbapenems. Due to intrinsic and acquired resistance in M. abscessus and poor clinical outcomes, it is critical to understand the development of antibiotic resistance both within the host and in the setting of outbreaks. We compared serial longitudinally collected M. abscessus subsp. massiliense isolates from the index case of a cystic fibrosis center outbreak and four outbreak-related strains. We found strikingly high imipenem resistance in the later patient isolates, including the outbreak strain (MIC > 512 µg/mL). The phenomenon was recapitulated upon exposure of intracellular bacteria to imipenem. Addition of the β-lactamase inhibitor avibactam abrogated the resistant phenotype. Imipenem resistance was caused by an increase in β-lactamase activity and increased bla_Mab mRNA level. Concurrent increase in transcription of the preceding ppiA gene indicated upregulation of the entire operon in the resistant strains. Deletion of the porin mspA coincided with the first increase in MIC (from 8 to 32 µg/mL). A frameshift mutation in msp2 responsible for the rough colony morphology and a SNP in ATP-dependent helicase hrpA cooccurred with the second increase in MIC (from 32 to 256 µg/mL). Increased Bla_Mab expression and enzymatic activity may have been due to altered regulation of the ppiA-bla_Mab operon by the mutated HrpA alone or in combination with other genes described above. This work supports using carbapenem/β-lactamase inhibitor combinations for treating M. abscessus, particularly imipenem-resistant strains.

INTRODUCTION

Mycobacterium abscessus is a pathogenic, multidrug-resistant organism within the rapidly growing nontuberculous mycobacteria (NTM) that causes lung, soft tissue, and disseminated infections. Risk factors include preexisting lung conditions and immunodeficiency. Recent studies have shown that the incidence and prevalence of NTM infections are on the rise in the United States (1, 2). Lung disease accounts for the majority of NTM infections (3) in susceptible individuals such as those with cystic fibrosis (CF). According to the CF Foundation, 10% of CF patients presented with positive cultures for M. abscessus (4).

Notorious outbreaks of M. abscessus have occurred in CF Centers worldwide including Seattle, WA, USA (5), and Papworth, UK (6). Genomic analysis by Tettelin et al. showed high similarity among strains from the four patients involved in the Seattle outbreak but also a surprisingly high-level relatedness with isolates from geographically distant outbreaks (7). Genomic comparison of strains collected serially from CF patients and from different CF patients over time have allowed inferences on patient-to-patient transmission (6).

Recommended treatment for pulmonary infection by M. abscessus consists of an initial phase of IV and oral drugs including a macrolide, an aminoglycoside, a β-lactam (imipenem or cefoxitin), and a glycylcycline for 3 months followed by a continuation phase of four to five oral drugs for at least 14 months (1, 8, 9). Unfortunately, M. abscessus is naturally resistant to many antibiotics by different mechanisms (10, 11). Most strains display resistance to macrolides through the inducible erm41 gene (for M. abscessus subsp. abscessus and bolletii). In addition, strains may acquire macrolide resistance through specific mutations in the 23S rRNA gene.

The β-lactam imipenem is often used intravenously during the initial phase of treatment of M. abscessus. Several studies discussed the use of imipenem in combination with other drugs including β-lactamase inhibitors (12–14), other β-lactam drugs (15–17), and other antibiotics (18, 19) for the treatment of M. abscessus infections. Imipenem was strongly associated with treatment success for M. abscessus pulmonary disease in a meta-analysis study (adjusted odds ratio 2.65, 95% CI 1.36–5.10) (20). Imipenem covalently binds to penicillin-binding proteins (PBPs) and LD-transpeptidases, two distinct enzyme classes involved in peptidoglycan synthesis in M. abscessus (21). Among β -lactams, penicillins are most effective against PBPs while carbapenems (including imipenem) are most effective at inhibiting LD-transpeptidases (22–24).

Resistance to β-lactams is mediated by the production of a class A β-lactamase encoded by the bla_Mab gene (MAB_2875). Bla_Mab can efficiently hydrolyze penams (amoxicillin, ampicillin) and cephems in vitro. However, imipenem and cefoxitin, both part of the recommended treatment for M. abscessus, are hydrolyzed at a lower rate than other β-lactam drugs (25). β-lactamase inhibitors such as avibactam have shown promising results for M. abscessus when used in combination with imipenem (26, 27). Avibactam improved the efficacy of imipenem against M. abscessus in vitro as well as in macrophages and zebrafish embryos (28).

As part of the investigation of the Seattle outbreak, we obtained serial isolates of M. abscessus subsp. massiliense (hereafter referred to as M. massiliense) from the index case over 8 years, including the outbreak strain (Fig. 1). Recently, we described the development of clarithromycin and amikacin resistance in those organisms (29). Notwithstanding, we speculated that this organism could have developed resistance to additional antimicrobials, as part of its evolution within the patient, culminating in the outbreak strain.

Fig 1.

Timeline of serial isolates from patient 2B. Patient 2B had cystic fibrosis complicated by a long-term respiratory infection with M. abscessus subsp. massiliense. Isolates were obtained over an 8-year period until patient’s death. S, smooth colony morphology; R, rough colony morphology.

Using phenotypic and genomic analyses of serial isolates from the Seattle outbreak index patient and secondary cases, we describe the development of extremely high-level imipenem resistance and provide insights into the resistance mechanism.

RESULTS

Imipenem activity against patient 2B serial isolates

Serially collected isolates from patient 2B (Fig. 1) were screened for differences in susceptibility to imipenem. MICs of imipenem against 11 isolates and type strain M. massiliense CCUG48898^T were determined by the broth microdilution method in 7H9sB media (Table 1). The MIC of imipenem against isolate 2B1 to 2B4 was 8 µg/mL, comparable to that of CCUG48898^T (16 µg/mL). There was a fourfold increase in MIC against strains 2B5 and 2B6 (32 µg/mL). Remarkably, isolates 2B7 to 2B11 showed a dramatic increase in MIC beyond the limit of quantification of the assay (>512 µg/mL). This rise in MIC coincided with the shift from smooth to rough colony morphotype (first seen with strain 2B7). Based on these results, we selected three representative strains 2B1, 2B5, and 2B11 to further investigate the mechanism of resistance.

TABLE 1.

Minimal inhibitory concentrations of imipenem against 2B patient isolates and M. massiliense CCUG 48898T^a

Isolate	MIC of IPM (µg/mL)
2B-1	8
2B-2	8
2B-3	8
2B-4	8
2B-5	32
2B-6	32
2B-7	256
2B-8	>512
2B-9	>512
2B-10	>512
2B-11	>512
Mmass CCUG48898T	16

^a MICs were determined using the broth microdilution method in 7H9sB. MICs were read after 48 hours incubation at 37°C and are expressed in µg/mL. Values are the median of five independent replicates.

Imipenem activity against intracellular M. massiliense in THP1 macrophages

To evaluate antimicrobial resistance of our strains within the intracellular environment, THP1 cells differentiated into macrophages were infected with M. massiliense CCUG48898^T, 2B1, or 2B11 strains and treated with imipenem for 48 hours (30). Fold changes, expressed as colony forming unit (cfu) counts after 48 hours of treatment over initial cfu count, are shown for each condition in Fig. 2. Fold change values above 1 represent growth of intracellular bacteria, while those below 1 indicate bacterial killing. All three strains showed similar growth inside macrophages in the absence of imipenem (with growth fold change values of 50.5, 52.2, and 42.6 for strains CCUG48898^T, 2B1, and 2B11, respectively). When treated with imipenem at 32 µg/mL, the growth fold change of CCUG48898^T was 0.4, demonstrating intracellular killing. Strain 2B1 survived imipenem treatment showing a small increase in growth (1.5-fold change), even though the difference was not statistically significant. Remarkably, strain 2B11 growth increased by 7.8-fold in the presence of imipenem, highlighting the remarkably poor activity of the drug against intracellular 2B11. These data confirmed the higher level of resistance of strain 2B11 to imipenem compared with the patient’s initial strain 2B1 and reference strain CCUG48898^T in the context of infected macrophages.

Fig 2.

Activity of imipenem against intracellular M. massiliense strains in THP1 macrophages. PMA-differentiated THP1 cells were infected by different M. massiliense strains and then treated with imipenem at 32 ug/mL for 24 hours or left untreated (IPM 0). Cells were then lysed, and intracellular bacterial load was determined. Fold changes are calculated by dividing the cfu count at 48 hours by the initial bacterial load before treatment. Fold change > 1 corresponds to intracellular growth of the bacteria. Fold change < 1 reflects the killing of intracellular bacteria. *P < 0.05 and **P < 0.01, t-test analysis.

Comparison of β-lactamase enzymatic activity

Mycobacterium massiliense produces a class A β-lactamase, Bla_Mab, responsible for the hydrolysis of β-lactams, including imipenem. In order to investigate if the observed differences in MIC could be due to changes in β-lactamase activity, we measured the specific enzymatic activity in protein extracts from each strain with the chromogenic substrate nitrocefin. As seen in Fig. 3, strain 2B1 showed baseline β-lactamase activity (26.3 nmol/min/mg), which was comparable to that of M. massiliense CCUG48898^T control (35 nmol/min/mg). Interestingly, 2B5 β-lactamase activity was 2.5 times higher than that of 2B1 (64 nmol/min/mg, P < 0.05), while the enzymatic activity of 2B11 increased even further, to four times that of 2B1 (108.9 nmol/min/mg, P < 0.05). The higher β-lactamase activity, as measured using the chromogenic β-lactam nitrocefin, could provide a possible mechanism to the increase in imipenem MICs.

Fig 3.

β-Lactamase specific activity (SA) of 2B patient isolates and M. massiliense CCUG48898T. SA was determined by following the hydrolysis of nitrocefin by spectrophotometry (λ = 486 nm). Data shown are the mean and standard deviation of five independent replicates. *P < 0.05 and **P < 0.01, t-test analysis.

Comparison of transcription levels of the bla_Mab gene

We sought to determine whether the higher β-lactamase activity observed in isolate 2B11 was due to enhanced enzymatic activity or increased expression of Bla_Mab. We quantified the mRNA levels of bla_Mab and the reference housekeeping gene dnaK by quantitative reverse transcription-PCR (RT-qPCR) and normalized the values against those of the CCUG48898^T control strain [Relative Quantitation (RQ)]. Since bla_Mab was predicted to be part of an operon by BioCyc (31) and microbesonline databases (32, 33) (Fig. 4A), we also measured the mRNA level of the preceding gene, ppiA, with values calculated similarly as with bla_Mab. As shown in Fig. 4B, 2B1 and B5 strains showed baseline levels of bla_Mab mRNA similar to those of CCUG48898^T control (1.2 and 1.3, respectively). Interestingly, the bla_Mab transcription level of strain 2B11 was 2.6 times higher than those of 2B1, 2B5, and CCUG48898^T. Thus, the increase in bla_Mab mRNA levels in strain 2B11 correlated with the higher Bla_Mab activity described above. The transcription level of ppiA was significantly increased in 2B11 compared with 2B1 and 2B5 (10.84 vs 1.80 and 2.72, respectively; Fig. 4C). Concurrent increases in ppiA and bla_Mab transcription indicate upregulation of the entire operon in the resistant strains.

Fig 4.

Expression level of ppiA-bla_Mab operon in patient 2B isolates. (A) Schematic representation of ppiA-bla_Mab operon. The mRNA levels of bla_Mab (B) and ppiA (C) were determined by RT-qPCR using dnaK as housekeeping gene and calculated versus those of CCUG48898T control strain (RQ). Data shown are the mean and standard deviation of three independent replicates. *P < 0.05 and **P < 0.01 using t-test.

Inhibition of Bla_Mab activity by avibactam

Avibactam is a second-generation β-lactamase inhibitor from the diazabicyclooctanes (DBO) family that was shown to inhibit Bla_Mab in vitro (26). We determined the MIC of imipenem in combination with 4 µg/mL of avibactam. As shown in Table 2, addition of avibactam led to a significant decrease in MIC against all tested strains. MICs against 2B1 and CCUG48898^T decreased two- and fourfold, respectively, to 4 µg/mL. The MIC against 2B5 decreased by fourfold. Notably, avibactam led to a striking reduction in the MIC of strain 2B11 (from >512 to 32 µg/mL).

TABLE 2.

Minimal inhibitory concentrations of imipenem alone or in combination with avibactam against 2B patient isolates and M. massiliense CCUG48898T^a

Isolate	MIC of IPM (µg/mL)	MIC of IPM + avibactam (µg/mL)
2B-1	8	4
2B-5	32	8
2B-11	>512	32
Mmass CCUG48898	16	4

^a MICs were determined using the broth microdilution method in 7H9sB. Different concentrations of imipenem were tested in combination with 4 µg/mL of avibactam. MICs were read after 48 hours incubation at 37°C and are expressed in µg/mL. Values are the median of five independent replicates.

Imipenem activity against the Seattle CF outbreak strains

Strain 2B11 was identified to be part of a global dominant circulating clone of M. massiliense responsible for outbreaks in CF centers. We obtained four strains from other patients in the Seattle CF outbreak. Imipenem MICs were performed against strains G_2446, G_2258 G_2455, and G_2272, all of which were previously found to be closely related by genomic sequencing (7). As shown in Table 3, G_2272 and G_2446 showed imipenem MICs of 512 µg/mL and above, respectively. Imipenem MIC against G_2455 was 128 µg/mL. Surprisingly, MIC against G_2258 was only 8 µg/mL, comparable to that of the early isolates from the 2B patient. We performed further genomic comparisons between G_2258 and 2B11 and the other outbreak strains to identify genetic changes that could explain the variations in imipenem resistance.

TABLE 3.

Minimal inhibitory concentrations of imipenem against outbreak isolates collected in Seattle, USA^a

Isolate	MIC of IPM (µg/mL)
G_2258	8
G_2272	512
G_2446	>512
G_2455	128

^a MICs were determined using the broth microdilution method in 7H9sB. MICs were read after 48 hours incubation at 37°C and are expressed in µg/mL. Values are the median of five independent replicates.

Mutations concurring with the development of moderate and extremely high imipenem resistance

Following whole genome sequencing using Illumina platform reported previously, we performed long-read sequencing of strains 2B1, 2B5, 2B7, and 2B11 (29). The four complete, gap-free genome sequences were compared with genome sequences of other strains from the same outbreak (Table S3 for accession numbers) using Sybil (34, 35), and genomic regions spanning clusters of orthologous genes carrying changes between isolates were realigned using Muscle (36).

This analysis revealed several mutations that appeared concurrently with the development of imipenem resistance, shown in Table 4. A mutation in mspA (porin) coincided with the first increase in MIC in 2B5 (from 8 to 32 µg/mL). The second increase in MIC (from 32 to 256 µg/mL) was detected in 2B7, which was concurrent with a frameshift mutation in msp2, disrupting the glycopeptidolipid (GPL) biosynthesis pathway and changing the colony morphology from smooth to rough and a SNP in the ATP-dependent helicase hrpA. All three mutations persisted in 2B11 (MIC > 512 µg/mL), which caused the Seattle CF outbreak.

TABLE 4.

Mutations cooccurring with acquisition or loss of the resistance phenotype^a

Gene name	Description	GO biological process/function	MAB locus	Amino acids (in isolate)	In 2B isolates	Outbreak strains	Process	Potential explanations
mspA	porin	Porin activity	MAB_1080	224 (2B–1) 82 (2B–11)	ORF truncated from isolates 2B–5 to 2B–11	Identical to 2B11	Truncation deletion of a piece of sequence	Appears with the first increase in MIC (from 8 to 32 µg/mL)
mps2	Nonribosomal peptide synthetase	GPL production	MAB_4098c	2582 (2B–1) 267 and 2327 (2B–11)	Two ORFs in isolates 2B–7, 2B–9, and 2B–11	Identical to 2B11	Frameshift deletion of one A (G_2258 also has an additional deletion of a C further down, G_2246 also has an additional deletion of a G further down)	Appears with the second increase in MIC (from 32 to 256 µg/mL). Leads to the R morphotype.
hrpA	ATP-dependent RNA helicase	Nucleic acid metabolic process	MAB_0056c	1286	G->T in 2B7, 2B9, 2B11.	Identical to 2B11	N446K	Appears with the second increase in MIC (from 32 to 256 µg/mL)
ppiA	Probable peptidyl prolyl cis trans isomerase	Cis/trans isomerization of prolyl bonds	MAB_2874	307aa (2B1-2B11) 597aa (G_2258)	N.A.	Two ORFS fused in G_2258	Frameshift deletion of one C	Appears with the loss of imipenem resistance in outbreak strain G_2258. leads to the fusion of ppiA and BlaMab (MAB_2875) linked by random amino acids

^a Mutations were identified by using data from long-read sequencing data (this manuscript) combined with Illumina data (previously published). The appearance of the first three mutations coincides with resistance phenotype development. The four mutation coincides with loss of resistance in one of the outbreak strains. N.A. not applicable.

We also compared the genomes of the four additional strains from the Seattle outbreak. All four strains shared the same frameshift mutation in msp2 and SNP in hrpA genes. However, G_2258 and G_2246 presented a second mutation downstream in msp2 (deletion of C and G, respectively) that did not restore the frameshift of the first mutation.

Interestingly, outbreak strain G_2258, which displayed a low imipenem MIC (8 µg/mL) despite being related to the highly resistant strain 2B11, showed a single nucleotide deletion in ppiA (MAB_2874). In the M. abscessus genome, this gene precedes MAB_2875 encoding the β-lactamase, Bla_Mab, and is part of the same operon. This deletion caused a frameshift resulting in a fusion protein from both genes, albeit with a missense amino acid sequence. This can explain the loss of imipenem resistance in outbreak strain G_2258 (Table 3).

DISCUSSION

Using serial isolates from a CF patient over 8 years (Fig. 1), we were able to track and characterize the development of extraordinarily high-level imipenem resistance in M. massiliense from the first isolate 2B1 (MIC = 8 µg/mL) to isolate 2B5 (MIC = 32 µg/mL) through the last isolate 2B11 (MIC > 512 µg/mL). The patient was the index case of a well-documented CF center outbreak in Seattle, WA (5).

A review of M. abscessus susceptibility data from our laboratory in the past 2 years revealed that 62% and 32% of tested strains (n = 53) were intermediate (MIC of 8–16 µg/mL) and resistant (MIC ≥ 32 µg/mL) to imipenem, respectively. Multiple publications described the MIC90 of imipenem being at or above 64 µg/mL in clinical strains [Fröberg et al. used 1014 samples (37); Ying et al. used 46 samples (38), and Liu et al. used 114 samples (39)]. Of note, most clinical microbiology laboratories use commercial antimycobacterial susceptibility testing plates for MIC determinations. The highest imipenem concentration values tested in the commonly used RAPMYCO and RAPMYCO2 Sensititre panels are 64 and 32 µg/mL, respectively, which takes into consideration CLSI’s breakpoint of ≥32 µg/mL for imipenem resistance. Since this commercial panel only tests a limited range of MICs, there is scant information on the occurrence and frequency of high-level imipenem resistance in M. abscessus.

Using a host cell-mycobacteria infection model with the macrophage-differentiated THP1 cell line (30, 40), we showed that the high-level resistance phenotype was recapitulated upon exposure of intracellular bacteria to imipenem. While 32 µg/mL of imipenem showed killing activity against M. massiliense type strain CCUG48898^T after 48 hours, this effect was reduced for the first clinical strain 2B1, although not significantly (0.4 ± 0.1 vs 1.5 ± 0.9; Fig. 2). Remarkably, the last clinical strain 2B11 continued replicating within macrophages despite the antibiotic treatment (fold change of 7.8 ± 5.5) underscoring its high level of imipenem resistance.

Mycobacterium abscessus Bla_Mab displays a low catalytic efficiency against imipenem. In the study by Dubée et al., imipenem showed an MIC of 8 µg/mL against M. abscessus type strain ATCC 19977, which is in the susceptible category according to CLSI guidelines. The MIC decreases to 4 µg/mL upon deletion of the bla_Mab gene (strain M. abscessus Δbla_Mab) (26). We used a hydrolysis kinetic assay of the chromogenic β-lactam nitrocefin to assess specific β-lactamase enzymatic activity in our M. abscessus subsp. massiliense strains. The first strain 2B1 showed similar β-lactamase activity to that of the type strain control, CCUG48898^T. Notably, the β-lactamase activity of the last strain 2B11 was about four times higher (Fig. 3). This increased Bla_Mab activity could explain the high-level imipenem resistance observed in 2B11.

To further characterize the resistance phenotype, levels of bla_Mab mRNA transcripts were measured by RT-Q-PCR. Strain 2B11 showed 2.2 times higher levels of bla_Mab mRNA compared with 2B1 (Fig. 4), which correlated with the increased β-lactamase activity and the extremely high MIC observed in the resistance strain. In addition, it showed an increased expression of ppiA, a gene immediately upstream of bla_Mab and under the same promoter. Similar results were obtained for the first highly resistant strain, 2B7 (data not shown). We speculated whether the increase in β-lactamase mRNA and enzymatic activity alone could account for the extremely high imipenem MIC or was it multifactorial. To address this question, we assessed imipenem activity against the strains with and without the β-lactamase inhibitor avibactam.

Avibactam, a second-generation DBO β-lactamase inhibitor, has been previously shown to inhibit Bla_Mab (41) and rescue β-lactam activity in vitro, in intracellular infection models and in animal models (26, 42). Commercially, imipenem is available in combination with another DBO, relebactam. Imipenem/relebactam and imipenem/avibactam showed similar activity against M. abscessus ATCC19977 in vitro and in a macrophage model (40, 43). Avibactam reduced imipenem MIC against M. abscessus ATCC19977 to comparable levels of the bla_Mab gene-deleted strain (M. abscessus Δbla_Mab) and restored the ability of imipenem to kill intracellular mycobacteria. Similarly, in our study, the addition of avibactam significantly decreased the imipenem MIC of our highly resistant strains to levels closer to those of the early susceptible ones (Table 2). These results support the hypothesis that the higher transcription level of the bla_Mab gene leading to an increased amount of Bla_Mab is the main driver of the extraordinarily high imipenem MICs observed in our clinical strains. Our work further highlights the value of β-lactam/β-lactamase inhibitor combinations in treatment regimens for M. abscessus infections.

Genomic comparisons between the first isolate (2B1), the first resistant isolate (2B5), the first extremely highly resistant isolate (2B7), and the last isolate (2B11) were carried out to look for mutations or rearrangements that might explain these findings. We focused on those gene changes, which appeared in 2B5 and 2B7 and persisted throughout the later isolates. Mutations occurred in genes involved in cell envelope, transcriptional regulation, and lipid and amino acid metabolism (Table 4).

Sequencing of 2B isolates proved a large deletion of the mspA gene (MAB_1080) from 2B5 through 2B11, coinciding with the first (fourfold) increase in imipenem MIC observed in strains 2B5 and 2B6. It has been shown in M. smegmatis that MspA is the major porin protein (44). Previously, we described the effects of this deletion, which included a slower growth rate, explained by lower nutrient (e.g., glucose) intake. M. abscessus with a mutated mspA showed an increase in virulence and inflammation (29, 45). Deletion of porin genes from M. massiliense CCUG48898^T had no effect on the antibiotic susceptibility profile (45). In contrast, Danilchanka et al. reported higher quinolone and β-lactams resistance in M. smegmatis strains lacking mspA (46). Thus, the literature is equivocal about a possible role of Msp porins on β-lactams resistance. One possible explanation is that mutant strains in the described various studies (including this manuscript) differ in the type of msp gene mutations and the porin expression and activity.

Loss of GPL production through mutations in GPL pathway genes including mps2 (MAB_4098c) is responsible for the rough (R) colony morphology in M. abscessus (47, 48). We showed that the dramatic increase in imipenem MIC observed in strain 2B-7 through 2B11 coincided with the switch from a smooth to a rough morphotype. Different studies have tried to establish a link between colony morphotype and resistance to common antibiotics. Lavollay et al. (49) linked the R morphotype of clinical strains to higher imipenem and cefoxitin MICs. In contrast, Hershko et al. (50) used transposon technology to generate a rough GPL-defective strain and showed that the loss of MAB_4099c led to only a minor (one dilution) increase in the MIC of imipenem.

A single point mutation was found in hrpA (MAB_0056c) in 2B7 and conserved in all later isolates. hrpA encodes for an ATP RNA helicase containing a DEAH box domain (51). This protein has not been extensively studied in mycobacteria. However, in Neisseria meningitidis (52), it is involved in biofilm formation, while in E. coli, it is necessary for mRNA processing of the fimbrial operon (53). HrpA is also important in gene regulation of Borrelia burgdorferi at a post-transcriptional level and essential for virulence in mice (54, 55).

Using the database STRING, 10 proteins have been identified as interacting with HrpA, one of them being PpiA, a peptidyl polyl cis trans isomerase, which precedes the β-lactamase gene bla_Mab in an operon (data not shown). We showed in this study that transcription and expression of both ppiA and bla_Mab were increased in the highly resistant strains 2B7 and 2B11 (Fig. 4). We hypothesize that HrpA is involved in the regulation of ppiA at the RNA level, which subsequently affects Bla_Mab expression as part of the same operon. The hrpA mutation and subsequent N446K amino acid change observed in strains 2B7 and 2B11 could alter HrpA regulation of the ppiA-bla_Mab operon leading to a higher expression of Bla_Mab responsible for the extremely high resistance to imipenem.

Occurrence of extremely high-level resistance to imipenem is alarming, particularly in this case as the strains were clinical isolates from a CF patient that led to a serious outbreak. Transmission of a strain resistant to multiple antibiotics to patients who have not been treated with those drugs is a major concern (6).

Using serial isolates from a CF patient, we followed and dissected the development of extremely high-level resistance to imipenem appearing progressively over time in strains of M. massiliense that were involved in a CF outbreak. A partial deletion in the porin gene cooccurred with a first fourfold increase in MIC. A significant increase in transcription level and enzymatic activity of β-lactamase observed in the resistant strains accounted for the subsequent eightfold increase in MIC leading to a very high-resistance phenotype. While no mutations were found on the β-lactamase bla_Mab gene or its promoter, we hypothesize that increased Bla_Mab expression and enzymatic activity could be due to the observed mutation in hrpA alone or in combination with mutations in the additional genes described above. Our findings also re-affirm the importance of the addition of a β-lactamase inhibitor to the antimicrobial regimen for M. abscessus to prevent the β-lactamase from degrading imipenem. The imipenem/relebactam combination is available clinically and should be considered for the treatment of M abscessus, particularly for imipenem-resistant strains.

MATERIALS AND METHODS

Strains and media

Serial isolates of M. massiliense from the CF patient (referred to as patient 2B) and the Seattle outbreak strains were stored at –80°C. Information on the strains including name, date of isolation, and GenBank accession numbers is provided in Table S1 and a previous publication (29). M. abscessus subsp. massiliense CCUG48898^T was used as reference strain. The colony morphotype of each strain was assessed microscopically (100×) using 7H11 agar plates incubated at 37°C (with first evaluation after 3–5 days).

Strains were subcultured from frozen stocks onto 7H11 agar plates and incubated for 4 days at 37°C, 5% CO₂. Liquid media cultures were performed from single colonies in 7H9 supplemented with ADC and 1% glycerol (7H9sB) for 3 days at 37°C with shaking. Cultures were diluted to an optical density at 600 nm (OD600) = 0.05 in 7H9sB at 37°C and grown until exponential phase (OD600 ~0.7–1.0) prior to experiments.

Human leukemia monocytic cell line THP1 cells were grown from frozen stocks in RPMI supplemented with HEPES 10 mM, sodium pyruvate 1 mM, and 10% fetal bovine serum (RPMIsB) and incubated at 37°C, 5% CO₂. All cell experiments were performed using RPMIsB medium.

MIC testing

Minimal inhibitory concentrations were determined by the broth microdilution method. Bacteria grown in 7H9sB for 4 days were diluted to an OD600 = 0.05 and incubated for 24 hours at 37°C with shaking at 180 rpm. These cultures were used to inoculate 96-well round-bottom plates (~5.10⁵ cfu/mL) containing twofold dilutions of antibiotics in 7H9sB. Plates were examined for bacterial growth after 48 hours incubation at 37°C. The MIC was defined as the lowest drug concentration inhibiting visible growth. Experiments were performed in quintuplets, and data shown are the median in µg/mL. Imipenem was purchased from Sigma-Aldrich (St. Louis, MO). Antibiotic solutions were made fresh prior to each replicate.

Intracellular killing of M. massiliense in differentiated THP1 macrophages

THP-1 cells were differentiated into 12-well plates (5  ×  10⁵  cells per 1  mL well), with phorbol 12-myristate 13-acetate (PMA) at 20 ng/mL in RPMIsB for 24  hours at 37°C, 5% CO₂. Cells were then infected with M. massiliense CCUG48898^T, 2B1, or 2B11 strains at a multiplicity of infection (MOI) of 10:1 for 3  hours, at 37°C, 5% CO₂. Imipenem (32 µg/mL) was then added to the treatment wells and plates incubated for 48 hours. Medium with or without imipenem was renewed every 24 hours. Quantitative cultures of intracellular bacteria were performed by plating serial dilutions of macrophage lysates on lysogeny broth (LB) agar plates incubated for 4 days at 37°C. Results represent the means (±SD) of five independent experiments.

β-Lactamase enzymatic activity

The β-lactamase-specific activity of total protein extract of M. massiliense cultures was measured using the chromogenic β-lactam substrate nitrocefin. Briefly, strains were grown in 7H9sB until OD600 ~1. Five milliliters of cultures was spun down, and the bacterial pellet was resuspended in 1 mL PIPES 2 mM pH 6.8 containing zirconia/silica beads 0.1 mm in diameter (BioSpec Product). Bacteria were then mechanically disrupted using a bead beater homogenizer (PowerLyzer 24, Qiagen; two cycles of 45 s 5,000 rpm/5 minutes on ice). After an additional 5 minutes on ice, samples were spun down and supernatants containing total soluble proteins were transferred to new microtubes. Samples were used fresh or stored at – 20°C until use. Prior to experiments, samples were thawed and 10 µL of total protein extract was added to a 100-µM nitrocefin (Sigma-Aldrich) solution in PIPES 2 mM pH 6.8 and substrate hydrolysis was measured spectrophotometrically over 30 minutes (Cytation5, Biotek) at 486 nm at 20°C. In addition, protein concentration was quantified by the BCA assay (Pierce BCA Protein Assay Kit, Thermo Fisher). For an experiment to be considered valid, the linear phase, with constant velocity in the enzymatic reaction, had to span over a minimum of 15 measurements. Specific activity was calculated as the concentration of hydrolyzed nitrocefin produced per minute per milligram of total protein extract. Results represent the mean values expressed in nmol/min/mg of five independent experiments.

RNA extraction and quantification

RNA was extracted from each strain cultivated in 7H9sB at 37°C for 2 days using a Nucleospin Kit (Macherey-Nagel, Allentown, PA) following the manufacturer’s instructions with few modifications. Five milliliters of the same cultures used for protein extraction was spun down and resuspended in 350 µL of RA1 lysis buffer. Bacteria were then disrupted mechanically with zirconia/silica beads 0.1 mm in diameter (PowerLyzer 24 Qiagen) with two cycles of 45s 5,000rpm with 5 minutes on ice in between. After 5 minutes on ice, tubes were spun down and the supernatant was transferred to another tube and 3.5 µL of β-mercaptoethanol was added. RNA was obtained in RNase/DNase-free water, and the concentration was measured with NanoDrop (Thermo Fisher Scientific).

RT-Q-PCR was performed with the Power SYBR Green RNA-to-CT 1-Step Kit (Thermo Fisher Scientific). Primers were designed to target different genes of interests (e.g., ppiA and bla_Mab, dnaK) (Table S2).

Long-read sequencing

For genomic DNA (gDNA) extraction, strains 2B1, 2B5, 2B7, and 2B11 were cultured on 7H11 plates at 37°C for 4 days. Bacterial colonies were resuspended in 0.05 M Tris HCl–10 mM EDTA pH 8.0 containing 10% lysozyme and incubated at 37°C for 2 hours. Extraction of gDNA was performed using the Wizard Genomic DNA Kit (Promega). After quantification by NanoDrop, the gDNA samples were sent to Maryland Genomics core at the University of Maryland School of Medicine. Following quality assessment with the Agilent 5200 Fragment Analyzer System, gDNA samples were subjected to pooled barcoded large insert (≥20 kb) library construction with size selection for HiFi Multiplex sequencing on the PacBio Sequel II platform. Complete, gap-free genomes of the four strains were generated using the PacBio SMRT 11.0.0 Improved Phased Assembler. Genome sequences were submitted to NCBI for PGAP annotation and released with accession numbers listed in Table S1. Table S1 also includes accession numbers of previous sequencing data from these strains.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aac.00673-24.

Supplemental tables. aac.00673-24-s0001.pdf.

Table S1, sample description information; Table S2, RT-QPCR primer sequences; Table S3, NCBI accession numbers.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.