Drug Degradation Caused by mce3R Mutations Confers Contezolid (MRX-I) Resistance in Mycobacterium tuberculosis

ABSTRACT

Contezolid (MRX-I), a safer antibiotic of the oxazolidinone class, is a promising new antibiotic with potent activity against Mycobacterium tuberculosis (MTB) both in vitro and in vivo. To identify resistance mechanisms of contezolid in MTB, we isolated several in vitro spontaneous contezolid-resistant MTB mutants, which exhibited 16-fold increases in the MIC of contezolid compared with the parent strain but were still unexpectedly susceptible to linezolid. Whole-genome sequencing revealed that most of the contezolid-resistant mutants bore mutations in the mce3R gene, which encodes a transcriptional repressor. The mutations in mce3R led to markedly increased expression of a monooxygenase encoding gene Rv1936. We then characterized Rv1936 as a putative flavin-dependent monooxygenase that catalyzes the degradation of contezolid into its inactive 2,3-dihydropyridin-4-one (DHPO) ring-opened metabolites, thereby conferring drug resistance. While contezolid is an attractive drug candidate with potent antimycobacterial activity and low toxicity, the occurrence of mutations in Mce3R should be considered when designing combination therapy using contezolid for treating tuberculosis.

INTRODUCTION

Tuberculosis (TB), caused by Mycobacterium tuberculosis (MTB), afflicts about 10 million patients and causes approximately 1.5 million deaths every year (1). Although the global treatment success rate is 86% for drug-susceptible TB, the success rates decrease to 59% for multidrug-resistant (MDR) (1), 47% for preextensively drug-resistant (pre-XDR) (2), and 39% for XDR-TB cases (3). The introduction of the oxazolidinone linezolid into tuberculosis treatment regimens has effectively improved treatment success rates for MDR/XDR MTB strains (4–6), but around 48 to 87% of patients on prolonged linezolid therapy experience adverse toxic effects, including mainly reversible hematologic problems and irreversible neuropathies (4–9). To address this toxicity, several new oxazolidinone agents have been rationally designed to reduce the myelosuppression and monoamine oxidase inhibition (MAOI) that are associated with linezolid toxicities (10).

Contezolid (MRX-I) is a novel potent oxazolidinone developed by Shanghai MicuRx Pharmaceutical Co., Ltd., and was recently approved by the National Medical Products Administration (NMPA) of China for the treatment of complicated skin and soft tissue infections (cSSTI) (11–13). Nonclinical and clinical studies have confirmed that, compared with linezolid, contezolid exhibits significantly reduced myelosuppression and MAOI while maintaining equivalent or even higher antibacterial activity both in vitro and in vivo (10, 14–20). Contezolid also has excellent antitubercular activity against both susceptible and resistant MTB, including MDR/XDR strains (MIC₅₀ = 0.5 μg/mL, MIC₉₀ = 1 μg/mL) (21), and its lower toxicity might allow contezolid to supplant linezolid in the long-term treatment regimens required to cure MDR/XDR-TB. However, to fully appreciate the potential of contezolid as an anti-TB agent, it would be advantageous to understand the frequency, location, and molecular mechanisms of mutations associated with contezolid resistance in MTB. To understand the development of contezolid resistance in MTB, the present study in vitro selected contezolid-resistant mutants, determined their resistance profiles, identified the mutational targets, and explored the underlying mechanisms conferring the resistance.

RESULTS

In vitro selection of contezolid-resistant MTB mutants.

Ten independent cultures of the pan-susceptible wild-type (WT) H37Rv strain were grown to log phase, and cells from each 3-mL culture were equally spread onto 7H10-OADC plates containing either 2.5 or 5 μg/mL contezolid, while the MIC of the WT strain was 1 μg/mL (Table 1). As a result, similar numbers of colonies have appeared on plates with the two different selective concentrations mentioned above. From the total of about 2.1 × 10¹⁰ CFU plated, there arose almost 6,000 resistant colonies, yielding a spontaneous contezolid resistance frequency of approximately 3 × 10⁻⁷.

TABLE 1.

Mutations identified in 106 contezolid-resistant strains of M. tuberculosis

Mutation type	No. of strains	Percentage
SNPs in mce3Ra	59	55.6
Nonsynonymous mutations	47	44.3
Nonsense mutations	12	11.3
Indels in mce3R	40	37.8
Insertion mutations	11	10.4
Deletion mutations	27	25.5
Multiple variants in mce3R	2	1.9
Unknown	7	6.6
Total	106	100

^aSNP, single nucleotide polymorphism.

WGS identified mce3R as the most frequently mutated gene.

A total of 120 resistant colonies, representing equally the 10 original cultures, were replated onto 7H10 agar plates containing 2.5 μg/mL contezolid and confirmed to be resistant. The genomic DNA was successfully extracted from 108 of the 120 resistant isolates and subjected to whole-genome sequencing (WGS). Two isolates with sequencing depths less than 15 were excluded from further analysis. The WGS data revealed that 93.4% (99 of 106) of the resistant isolates had mutations in mce3R (Rv1963c), which encodes a TetR family transcriptional regulator, while the remaining 6.6% (7 of 106) harbored no confirmable mutations. The mutations in mce3R were scattered throughout the gene with no mutagenic “hot spot” regions (Fig. 1A). A total of 35 nonsynonymous mutations were found in 47 resistant strains and 6 nonsense mutations were found in 12 strains. In addition, 10 different insertions were present in 11 resistant strains, and 14 different deletions were found in 27 resistant strains (Table 1). Although most insertions and deletions involved a single nucleotide causing a frameshift, there were 5 resistant strains that contained either a 98-bp (417_527del) or a 111-bp (583_680del) deletion within the mce3R gene. Additionally, in two resistant strains, the WGS data showed multiple mutations within the 339 to 466-base region of the mce3R gene, which were confirmed by Sanger sequencing (Fig. 1B).

FIG 1.

The genetic variants in contezolid-resistant M. tuberculosis. (A) Distribution of mce3R mutations associated with contezolid resistance. The full length of mce3R gene is 1,221 bp, encoding a putative protein of 406 amino acids. Single nucleotide polymorphisms (SNPs) are showed in gray with nonsense mutations in bold, with numbers that correspond to the amino acid positions. Insertion and deletion mutations are shown in indigo and purple, respectively, with numbers indicating base positions. (B) Two contezolid-resistant strains harbored multiple mutations within the 339 to 446 base region of the mce3R gene. SNPs, insertions, and deletions are indicated by red, blue, and purple boxes, respectively. The “Query” and “Sbjct” sequences belong to contezolid-resistant strains and the wild-type H37Rv strain, respectively. The numbers indicate the base positions of the mce3R gene. (C) Distribution of hitchhiking mutations identified in contezolid-resistant strains by whole-genome sequencing (WGS) analysis, among 10 independent selection cultures indicated with different colors.

By exploring the entire genomes of the 106 contezolid-resistant mutants, we also identified 20 other mutations in 17 isolates, but 90% of these mutations (18 of 20) were found in strains that also contained mce3R mutations. Of the 20 other mutations, 12 were intergenic or synonymous and predicted to have no impact on the function of their encoded proteins, and most mutations (18 of 20 mutations) were found in a resistant strain obtained from just one of the 10 original cultures used for mutant selection. Therefore, it was presumed that these 20 mutations fixed in MTB were forms of genetic hitchhiking unrelated to contezolid resistance.

Determination of contezolid resistance target and its cross-resistance with linezolid.

To confirm that the mce3R mutations were responsible for the contezolid resistance, we cloned the entire WT mce3R gene into high-copy-number nonintegrative plasmid pSMT3-LxEGFP and complemented one of the mce3R mutants with a large mce3R deletion (417_527del) using the constructed plasmid. The complemented mce3R deletion strain recovered its susceptibility to contezolid with the same MIC of 1 μg/mL as the WT H37Rv parent strain (Table 2), proving that the mce3R mutations were responsible for contezolid resistance.

TABLE 2.

MIC values for contezolid and linezolid in M. tuberculosis strains analyzed in this study

Strain ID	Characteristic	MIC (μg/mL)
		Contezolid	Linezolid
MTB-260	mce3R_R15G mutant	16	1
MTB-243	mce3R_Q6stop mutant	16	1
MTB-224	mce3R_838_839insT mutant	16	1
MTB-197	mce3R_264delG mutant	16	1
MTB-21	rrl_G2270T mutant	16	16
MTB-33	rrl_G2814T mutant	16	32
MTB-12	rplC_T460C mutant	16	16
H37Rv	WT strain	1	1
△mce3R	mce3R_417_527del mutant	16	1
△mce3R O/E mce3R	mce3R_417_527del mutant complemented with a WT copy of mce3R	0.5 to 1	1
H37Rv O/E Rv1936	MTB overexpressing Rv1936	16	1
H37Rv O/E Rv1938	MTB overexpressing Rv1938	1	1
H37Rv O/E Rv1936–Rv1941	MTB overexpressing Rv1936–Rv1941 operon	16	1

We then evaluated the effect of the mce3R mutations on both contezolid and linezolid resistance. The MTB strains harboring different mce3R mutations (R15G, Q6stop, 838_839insT, 264delG, and 417_527del) uniformly exhibited a 16-fold increase in contezolid resistance (MIC = 16 μg/mL) but remained susceptible to linezolid (MIC = 1 μg/mL) (Table 2). In contrast, our previously isolated linezolid-resistant MTB strains, harboring canonical oxazolidinone resistance mutations in rrl or rplC (22), exhibited cross-resistance to contezolid, with MICs (16 μg/mL) equal to that of the mce3R mutants (Table 2). These results suggested that mutations in the mce3R gene confer resistance specifically to contezolid and do not constitute a generalized oxazolidinone resistance mechanism.

Overexpression of Rv1936 induced by Mce3R mutations confers contezolid resistance.

As Mce3R is a TetR transcriptional repressor that negatively regulated genes involved in lipid metabolism, redox reactions and stress resistance (23–25), we reasoned that genes in its regulon might be involved in the molecular mechanism of contezolid resistance in MTB. Previous proteomic results directed our attention to two genes located in the Rv1936–Rv1941 operon, which was negatively regulated by Mce3R, in that their coding products, namely, a monooxygenase encoded by Rv1936 and an epoxide hydrolase EphB encoded by Rv1938, have been found to be dramatically more abundant in the mce3R knockout mutant than in the WT H37Rv strain (23). Similar enzymes have been shown to be generally involved in the detoxification and catabolism of xenobiotics in microorganisms (26, 27). To better understand their association with contezolid resistance, we constructed strains containing plasmids overexpressing Rv1936, Rv1938, or the Rv1936–Rv1941 operon. Real-time fluorescent quantitative PCR (RT-qPCR) confirmed that Rv1936 and Rv1938 genes were markedly coupregulated in our mce3R mutants and the strain containing plasmid with the Rv1936–Rv1941 operon, as well as individually upregulated in their corresponding overexpressing strains, whereas they were expressed at the WT level in the mce3R mutant complemented with a WT copy of mce3R (Fig. 2). We then tested their contezolid susceptibilities and found that MTB strains containing plasmids overexpressing Rv1936 or the Rv1936–Rv1941 operon had a MIC of 16 μg/mL, showing a level of resistance as high as that displayed by the mce3R mutants, whereas the strain overexpressing only Rv1938 had a MIC of 1 μg/mL to contezolid, remaining unchanged from the WT strain regarding susceptibility (Table 2). These findings suggest that the mutations in mce3R abrogate its negative regulatory function, leading to overexpression of the monooxygenase encoded by Rv1936, which then confers contezolid resistance in MTB.

FIG 2.

Gene transcript level detected by real-time fluorescent quantitative PCR (RT-qPCR). The figure shows the change in transcriptional expression, relative to the wild-type (WT) H37Rv, of the Rv1936 and Rv1938 genes in five in vitro-isolated mce3R mutants, the mce3R_417_527del mutant complemented with a WT copy of mce3R gene (△mce3R O/E mce3R), and three MTB strains with plasmids overexpressing (O/E) the indicated genes.

Rv1936 degrades contezolid in resistant MTB stains.

We next sought to relate the overexpression of Rv1936 to the mechanism of contezolid resistance in MTB. As Rv1936 is nonessential for the in vitro growth of MTB (28), which was unlikely to be the target of contezolid, we hypothesized that Rv1936 inactivates contezolid through drug degradation or modification. Sequence alignment analysis found that Rv1936 shares ~98% sequence homology with the flavin-dependent oxidoreductase MelF in Mycobacterium marinum (Fig. 3A), indicating a reliable enzymatic activity of this MTB monooxgenase (29). Moreover, given that contezolid in humans is mainly metabolized by the liver enzyme flavin-containing monooxygenase 5 (FMO5) (30), we reasoned that Rv1936 might act as a putative FMO that degrades contezolid to confer resistance in MTB.

FIG 3.

MTB Rv1936-promoted contezolid degradation. (A) Alignment of the amino acid sequences of the MTB Rv1936 and the flavin-dependent oxidoreductase MelF in M. marinum. Highlighted amino acids represent identity between sequences. The numbers indicate the amino acid positions of Rv1936 or MelF. (B) Chromatogram representation of the parent contezolid and its two principal degradation products MRX445-1 and MRX401 detected by ultraperformance liquid chromatography (UPLC)/triple time-of-flight mass spectrometry (TOF MS), with their corresponding m/z values indicated. The retention time and mass are confirmed as described before (30). (C) Comparative quantification of contezolid, MRX445-1, and MRX401 based on their chromatographic peak area in the WT H37Rv, in the three contezolid-resistant mce3R mutants (L266P, 755_756insC, and 417_527del), in △mce3R complemented with a constitutive expression vector for mce3R (△mce3R O/E mce3R), and in the WT H37Rv that constitutively expresses Rv1936 (H37Rv O/E Rv1936). All strains were treated with 10 μM contezolid for 15 h, with the exception of the WT H37Rv being treated with 2 μM contezolid to avoid drug inhibition effect. (D) Schematic representation of oxidation of contezolid by Rv1936 and subsequent hydrolysis of its 2,3-dihydropyridin-4-one (DHPO) ring-expanding enol lactone, the oxidative intermediate product indicated in the dotted box, into two principal DHPO ring-opened metabolites MRX445-1 and MRX401. Chemical groups indicating the key DHPO ring-opening process of contezolid degradation are highlighted in red. The structures of contezolid and MRX445-1 were confirmed with authentic molecules, and the proposed structure of MRX401 is consistent with available high-resolution mass spectrometry data and its biotransformation pathway. FMO, flavin-containing monooxygenase.

To test the Rv1936-promoted contezolid degradation hypothesis, we grew MTB cells for 15 h in the presence of contezolid and used ultraperformance liquid chromatography (UPLC)/triple time-of-flight mass spectrometry (TOF MS) to identify the parent drug, as well as its metabolites, in the coincubation mixtures. Overall, the most abundant components were the parent drug contezolid (C₁₈H₁₅N₄O₄F₃) and its two 2,3-dihydropyridin-4-one (DHPO) ring-opened metabolites: MRX445-1 (C₁₈H₁₉N₄O₆F₃) and MRX401 (C₁₆H₁₅N₄O₅F₃) (Fig. 3B and C). In detail, compared with results obtained from contezolid incubated with contezolid-susceptible MTB strains (i.e., WT H37Rv and the mce3R complemented mutant strain), we observed a dramatic reduction on the amount of the parent drug contezolid in the incubation samples of contezolid treated with contezolid-resistant MTB strains (i.e., mce3R_755_756insC mutant, mce3R_417_527del mutant, mce3R_L266P mutant, and the strain overexpressing the Rv1936 gene) (Fig. 3C). Meanwhile, the amount of two DHPO ring-opened metabolites MRX445-1 and MRX401, as a consequence of contezolid degradation, increased in the resistant mixtures relative to the susceptible ones (Fig. 3C). Our findings suggested that the overexpression of the putative FMO Rv1936 promotes the degradation of contezolid, presumably by oxidizing the DHPO ring-expanding process for subsequent hydrolytic degradation (Fig. 3D), thereby conferring drug resistance in MTB.

DISCUSSION

Contezolid is a new oxazolidinone antibiotic that retains the potent antitubercular activity of linezolid but is less toxic and could potentially replace linezolid in the treatment of drug-resistant TB (10, 18–21). This study sought to understand the development of resistance to contezolid by in vitro isolating and characterizing spontaneous contezolid-resistant MTB mutants. Surprisingly, 93.4% of the contezolid-resistant strains had mutations in mce3R, a TetR-like transcriptional repressor. A wide variety of mutations were found throughout the mce3R gene, suggesting that any loss-of-function mutations would confer resistance. Consistently, complementation of mutant with a WT copy of mce3R restored contezolid susceptibility. The inactivation of Mce3R in these mutant strains led to increased expression of Rv1936, which appeared to code for the critical functional enzyme that degraded contezolid for degradation, thereby reducing the level of contezolid and conferring resistance in MTB.

In humans, contezolid is principally metabolized into MRX445-1 and MRX459 via the Baeyer-Villiger oxidation of DHPO ring, which is specifically catalyzed by the liver FMO5, followed by the hydrolytic DHPO ring-opening process (30). In the current study, we have discovered that MTB exploits a similar mechanism to counter the antitubercular activity of contezolid by degrading the parent drug into its DHPO ring-opened metabolites MRX445-1 and MRX401. Based on the established literature of structure-activity relationship (SAR) data for oxazolidinones, metabolites of contezolid are not expected to exhibit potent antibacterial activity (18). Further MIC tests have confirmed that the two human metabolites MRX445-1 and MRX459 showed no antibacterial activities against Gram-positive pathogens, including Staphylococcus aureus, Enterococcus faecium, Enterococcus faecalis, and Streptococcus pneumoniae (31). Given that MRX401 is the dealkylated metabolite derived from MRX445-1, it should also exhibit no antibacterial activity.

Mutations in mce3R or overexpression of Rv1936 contributed to contezolid resistance in MTB, however, did not confer resistance to linezolid, because the chemical structure of linezolid lacks the DHPO ring targeted by the putative FMO Rv1936. Through searching on the NCBI BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) and the mycobrowser database (https://mycobrowser.epfl.ch/releases), we found only orthologues of the MTB Mce3R and Rv1936 in Mycobacterium bovis, Mycobacterium canettii, Mycobacterium marinum, Mycobacterium ulcerans, and Mycobacterium vanbaalenii. Thus, it is possible that the degradation of contezolid as a mechanism of resistance may be limited in bacteria with a suitable monooxygenase whose expression is controlled by a nonessential negative regulator, such as mce3R. While the drug-degradation mechanism of contezolid resistance appears to be unique to a few mycobacteria, degradation of antibiotics is a common mechanism through which pathogens develop drug resistance. One of the most prominent examples is the enzymatic degradation of β-lactam antibiotics through the acquisition of β-lactamases that hydrolyze the β-lactam ring (32). Although MTB possesses strong, constitutive β-lactamase activity that renders it intrinsically resistant to β-lactam antibiotics (33–37), the current study provides another example of the acquired degradation mechanism conferring drug resistance in MTB.

Given that the contezolid resistance could be achieved by overexpressing Rv1936 alone in the WT H37Rv rather than overexpressing Rv1938 alone, we reasoned that oxidation of the DHPO ring catalyzed by monooxygenase Rv1936 is pivotal for the degradation of contezolid, which appears to be the rate-limiting step, and MTB should have other intrinsically abundant epoxide hydrolase(s) rather than Rv1938 to catalyze the subsequent hydrolysis of the oxidized DHPO ring. Preliminary studies suggested that the MTB genome encodes six other putative monooxygenases (i.e., Rv0892, Rv0565c, Rv3854c, Rv1393c, Rv3049c, and Rv3083) capable of performing Baeyer-Villiger oxidations and seven putative epoxide hydrolases (i.e., Rv3617, Rv1938, Rv1124, Rv2214c, Rv3670, Rv0134, and Rv2740, corresponding to Eph A to G) (38, 39). Further enzymatic assays have confirmed the oxidizing activity of Rv3854c, Rv3049c, and Rv3083 (38, 40), as well as the hydrolyzing activity of Rv3617, Rv1938, Rv1124, Rv0134, and Rv2740 (40–43). In particular, Rv3854c (EthA) and Rv3083 (MymA) have been confirmed to similarly function as the activating enzymes of the anti-TB prodrug ethionamide, whereas they individually activate different sulfur-containing compounds, suggesting the overlap in enzymatic function and substrate specificity of these monooxygenases in MTB (44). It remains for further study to identify whether the six monooxygenases in MTB mentioned above could catalyze the oxidation of DHPO ring and determine which epoxide hydrolase(s) participated in the subsequent hydrolytic ring-opening process.

Although contezolid has been used clinically for the treatment of cSSTI, very little is known about the potential mechanisms of its clinical resistance. The only published data on contezolid resistance found mutations in the canonical oxazolidinone gene targets rrl and rplC in spontaneous in vitro-isolated contezolid-resistant strains of S. aureus (45). In the present study, however, none of our 106 in vitro-isolated resistant MTB strains harbored mutations in rrl or rplC, even though we showed that MTB strains with mutations in these genes are resistant to contezolid. In subsequent work, we isolated additional contezolid-resistant strains and found mutations in rrl or rplC. The predominance of mce3R mutations in contezolid-resistant strains is likely because these inactivating mutations can occur anywhere in the gene and are found at an estimated frequency of approximately 3 × 10⁻⁷, which is nearly one log higher than the previously reported mutation frequencies in rrl and rplC of 2 × 10⁻⁸ to 5 × 10⁻⁹ (46, 47). Moreover, such mutational bias raises the possibility that mutations in the nonessential gene mce3R might display lower fitness cost on MTB strains than in the essential genes rrl and rplC. Interestingly, there remained seven in vitro-isolated contezolid-resistant MTB strains harboring mutations neither in the mechanistic rrl and rplC targets nor in our newly identified nonmechanistic mce3R target, suggesting that there may be other alternative mutational targets and mechanisms for contezolid resistance in MTB.

Another intriguing observation is that the contezolid-resistance-conferring mutations in mce3R were highly polymorphic and included 35 nonsynonymous mutations, 6 nonsense mutations, and 24 frameshift mutations scattered throughout the whole mce3R gene rather than in a “hot spot” region. Similar polymorphic patterns have been observed for mutations in pncA that confer resistance to pyrazinamide and in Rv0678 that confer resistance to bedaquiline. Studies based on in silico structure analysis illustrated that, in addition to frameshift mutations, mutations can inactivate the protein function by impairing protein conformational flexibility, folding, stability, or catalytic activity (48–50). Further investigations are warranted to elucidate the effect of the contezolid-resistance-conferring mutations on Mce3R protein function. Although the frequency of mce3R mutations, 3 × 10⁻⁷, is relatively high, this may not discourage the use of contezolid against tuberculosis as the two critical first-line anti-TB drugs isoniazid and ethambutol have comparable even higher resistance frequencies of 1.78 × 10⁻⁶ and 1 × 10⁻⁷, respectively (51, 52). MTB resistance occurs rapidly if any single anti-TB drug is used in monotherapy; however, the clinical successful induction and maintenance therapy for TB is always based on strategy with a multidrug therapy approach that could minimize the risk of acquiring drug resistance and achieve the maximum therapeutic efficacy. In addition, our study was based on in vitro selection of resistant isolates derived from the WT H37Rv strain, whereas the mutational location and frequency of in vivo-developed contezolid resistance in clinical MTB strains may be different.

In conclusion, this study identified the mce3R gene as a novel target for mutations conferring resistance to contezolid but not linezolid in MTB. The mutations in mce3R lead to upregulated expression of Rv1936, a putative FMO that initiates the degradation of contezolid into inactive metabolites. Our results provide fundamental insight into the in vitro-raised contezolid resistance in MTB and suggest that the occurrence of mutations in mce3R gene should be considered when designing combination therapy using contezolid for treating tuberculosis.

MATERIALS AND METHODS

Bacterial cultures.

Escherichia coli TOP10 and BL21 Star (DE3) strains used for cloning were grown in LB with or without 1.5% agar. MTB strains were grown in Middlebrook 7H9 broth supplemented with 0.2% (vol/vol) glycerol, 0.05% (vol/vol) tyloxapol, and 10% (vol/vol) oleic acid-albumin-dextrose-catalase (OADC) or on Middlebrook 7H10 agar plates supplemented with 0.5% (vol/vol) glycerol and 10% (vol/vol) OADC. All bacteria were grown aerobically at 37°C with constant agitation. When indicated, kanamycin and hygromycin were added at 25 and 50 μg/mL, respectively. Contezolid was used for mutant selection and MIC determination at concentrations ranging from 2.5 to 64 μg/mL.

Spontaneous contezolid-resistant mutant selection.

For each 10 independent log-phase cultures of laboratory reference strain H37Rv, 3-mL aliquots were concentrated and equally spread onto three 7H10-OADC plates containing either 2.5 or 5 μg/mL contezolid (Shanghai MicuRx Pharmaceutical Co., Ltd.), corresponding to 2.5 or 5 times the MIC of 1 μg/mL contezolid for H37Rv, respectively. The CFU were calculated by serially diluting the original cultures 10⁻⁴- to 10⁻⁵-fold and then triplicate plating aliquots onto antibiotic-free 7H10-OADC plates. After incubation at 37°C for 4 weeks, colonies appearing on the contezolid-containing plates were replated onto 7H10-OADC plates containing 2.5 μg/mL contezolid to confirm their resistance. The frequency of spontaneous contezolid resistance was estimated as the ratio between the number of resistant colonies obtained and the total CFU plated.

Whole-genome sequencing and data analysis.

The genomic DNA of 108 contezolid-resistant colonies and the parental H37Rv was extracted with the cetyltrimethylammonium bromide-lysozyme method and used for the Illumina construction of 300-bp-fragment length genomic libraries for paired-end sequencing on an Illumina HiSeq 2500 instrument. A previously validated pipeline was used for mapping the sequencing reads to the reference genome (53). In brief, the Sickle tool was used for trimming and sequencing reads with Phred base quality scores above 20 and lengths longer than 30 bp were kept for analysis. Mutations were called with at least two reads in both forward and reverse directions, using MTB H37Rv strain (NC_000962.2) as a reference. Sequencing reads were mapped to the reference genome using Bowtie 2 (version 2.2.9) and SAMtools (version 1.3.1) was used for single nucleotide polymorphism (SNP) calling with mapping quality greater than 30. Fixed mutations (frequency ≥ 75%) were identified using VarScan (v2.3.9) with at least 10 supporting reads with the strand bias filter option on. SNPs in repetitive regions of the genome (e.g., proline-proline-glutamate (PPE)/proline-glutamate (PE)-polymorphic guanine-cytosine-rich sequence (PGRS) family genes, phage sequences, insertion, or mobile genetic elements) were excluded. Small indels were identified by VarScan (version 2.3.9) using the mpileup2indel function.

MIC determination.

MIC values were determined in Middlebrook 7H9-OADC broth in 96-well plates using a standard microdilution method. Serial 2-fold dilutions of antibiotics were prepared in 96-well plates. Early-log phase (optical density at 600 nm [OD₆₀₀] = 0.5–1.0) MTB cultures were diluted to ~1 × 10⁵ CFU/mL (OD ≈ 0.001), and then equal volumes (100 μL) of diluted cultures were added to each well to yield a final testing volume of 200 μL. The plates were incubated at 37°C for 2 to 3 weeks, and then MICs were determined visually as the minimum drug concentration preventing bacterial growth. Contezolid and linezolid were generously provided by Shanghai MicuRx Pharmaceutical Co., Ltd., and solubilized in dimethyl sulfoxide (DMSO) as a stock solution of 5 mg/mL. For drug susceptibility determination, these stocks were diluted in 7H9-OADC broth to final concentrations of 64, 32, 16, 8, 4, 2, 1, and 0.5 μg/mL. To exclude the bacteriostatic effect of the solvent, DMSO was diluted as well for drug-free control.

Plasmid construction and transformation.

The high-copy-number nonintegrative plasmid pSMT3-LxEGFP was used for the genetic overexpression and complementation. Briefly, the gene containing DNA fragments were PCR-amplified from MTB H37Rv with the primers shown in Table S1 and then inserted into plasmid pSMT3-LxEGFP restricted with BamHI and HindIII (Thermo Scientific) using T4 DNA ligase. The recombinant plasmids containing Rv1936 and Rv1938 genes and the Rv1936–Rv1941 operon were then independently electroporated into the transformation competent WT H37Rv. As for mce3R mutant complementation, the recombinant plasmid containing a WT copy of mce3R was electroporated into the transformation competent mce3R_417_527del mutant, which has an 111-bp deletion in mce3R.

RNA extraction and quantitative PCR (qPCR).

Cultures were grown to early-log phase (OD₆₀₀ = 0.5 to 1.0) and harvested at 4°C. Cell pellets were resuspended in 1 mL TRIzol reagent (Invitrogen) and subjected to bead-beating with 0.1-mm silica beads at maximum speed for 30 s three times, with cooling on ice between shaking. After centrifugation at 18,000 × g for 1 min at 4°C, the supernatant was transferred to a phase lock gel tube (TIANGEN) containing 400 μL chloroform, inverted vigorously for 30 s, and centrifuged at 18,000 × g for 15 min. The upper aqueous phase (about 500 μL) was then transferred to a new tube, and the RNA was precipitated with 450 μL isopropanol for 30 min at room temperature. RNA was washed sequentially with 80% and pure ethanol, and the pellet was then dissolved in 50 μL RNase-free water. The total RNA yield was quantified with a Nanodrop (Thermo Scientific). cDNA was synthesized using the PrimeScript RT reagent kit (TaKaRa) following the manufacturer’s instructions. qPCRs were performed with iQ SYBR green supermixture kit (TaKaRa) in the CFX96TM real-time PCR system (Bio-Rad). The sequences of all primers are listed in Table S1.

Coincubation sample collection and metabolite profiling.

The contezolid stock solution was added to 4 mL of stationary-phase H37Rv cultures at final concentrations of 2 and 10 μM (final DSMO concentrations, 0.025 and 0.125%) for susceptible and resistant strains, respectively. The mixtures were incubated at 37°C for 15 h, heated to 80°C for 30 min to kill the MTB, and then subjected to bead-beating with 0.1-mm silica beads at maximum speed three times for 30 s each with cooling on ice between shakes. After centrifugation at 11,000 × g for 3 min at 4°C, the supernatants were directly analyzed by UPLC/triple TOF MS to identify contezolid and its metabolites as previously described (30). In brief, 10 μL of supernatant was injected into the Acquity UPLC HSS T3 column (100 × 2.1-mm inner diameter, 1.8 μm) eluted with a linear gradient. The mobile phase consisting of 0.05% formic acid in water and methanol with the flow rate maintained at 0.45 mL/min. Contezolid and its metabolites were detected using a triple TOF 5600+ MS/MS system (AB Sciex, Concord, Ontario, Canada) in the positive electrospray ionization mode, with the mass range set at m/z 100 to 1000. Information-dependent acquisition (IDA) was used to trigger the acquisition of MS/MS spectra for contezolid and its metabolites. A real-time multiple mass defect filter was used for the IDA criteria.

Quantification and statistical analysis.

Graphics and data analyses were performed in Prism version 7.0 software (GraphPad) and SnapGene version 4.3.6 software (GSL Biotech LLC).

Data availability.

The WGS data were deposited in the Genome Sequence Archive (https://bigd.big.ac.cn/gsa) under accession number CRA007448.