In Vitro Drug Susceptibility of Bedaquiline, Delamanid, Linezolid, Clofazimine, Moxifloxacin, and Gatifloxacin against Extensively Drug-Resistant Tuberculosis in Beijing, China

ABSTRACT

Extensively drug-resistant tuberculosis (XDR-TB) is a deadly form of TB that can be incurable due to its extreme drug resistance. In this study, we aimed to explore the in vitro susceptibility to bedaquiline (BDQ), delamanid (DMD), linezolid (LZD), clofazimine (CLO), moxifloxacin (MFX), and gatifloxacin (GAT) of 90 XDR-TB strains isolated from patients in China. We also describe the genetic characteristics of XDR-TB isolates with acquired drug resistance. Resistance to MFX, GAT, LZD, CLO, DMD, and BDQ was found in 82 (91.1%), 76 (84.4%), 5 (5.6%), 5 (5.6%), 4 (4.4%), and 3 (3.3%) isolates among the XDR-TB strains, respectively. The most frequent mutations conferring fluoroquinolone resistance occurred in codon 94 of the gyrA gene (57.8%), and the strains with these mutations (69.2%) were associated with high-level MFX resistance compared to strains with mutations in codon 90 (25.0%) (P < 0.01). All 5 CLO-resistant isolates exhibited ≥4-fold upward shifts in the BDQ MIC, which were attributed to mutations of codons 53 (60.0%) and 157 (20.0%) in the Rv0678 gene. Additionally, mutation in codon 318 of the fbiC gene was identified as the sole mutation related to DMD resistance. In conclusion, our data demonstrate that the XDR-TB strains exhibit a strikingly high proportion of resistance to the current anti-TB drugs, whereas BDQ, DMD, LZD, and CLO exhibit excellent in vitro activity against XDR-TB in the National Clinical Center on TB of China. The extensive cross-resistance between OFX and later-generation fluoroquinolones indicates that MFX and GAT may have difficulty in producing the desired effect for XDR-TB patients.

INTRODUCTION

Tuberculosis (TB) remains a major cause of morbidity and mortality worldwide, with an estimated 10.4 million new cases and 1.4 million deaths in 2015 (1). Despite the decline in incidence and mortality of TB since the early 2000s, the emergence of drug-resistant TB, especially multidrug-resistant TB (MDR-TB), threatens our efforts to control and eliminate the disease (2). Notably, a more severe form of drug-resistant TB, named extensively drug-resistant TB (XDR-TB), was first reported from South Africa in 2006 (3). Compared with MDR-TB, these strains exhibit additional resistance to the most effective second-line therapeutic drugs commonly used to treat MDR-TB: fluoroquinolones and at least one of three injectable second-line drugs (4). Recently, XDR-TB has attracted more attention from the medical community, since poor survival and cure rates can be expected in this population with the severe degree of drug resistance (5). As a consequence, new anti-TB drugs and treatment regimens are urgently needed to tackle the growing global concern about virtually untreatable TB (2).

Although development of new drugs for TB is lengthy and expensive, a great achievement has been made in the global anti-TB drug pipeline in recent years (6). Several FDA-approved candidates exhibit promising anti-TB activity in vitro, as well as in clinical trials (7–10). Linezolid (LZD), an existing antibacterial compound, is increasingly used for patients with highly drug-resistant TB (11, 12). Numerous clinical studies have demonstrated that LZD is effective at achieving cure among patients with MDR and XDR pulmonary TB (9, 12). Similar to LZD, clofazimine (CLO), an old drug against mycobacteria, has shown good efficacy and low toxicity against drug-resistant TB in recent in vitro and in vivo trials (13, 14). In addition to these existing drugs, two new compounds, bedaquiline (BDQ) and delamanid (DMD), have gained importance and seem encouraging against drug-resistant TB (15, 16). Because each of the two drugs involves a mechanism of action that is different from that of other available drugs, promising efficacy against XDR-TB can be expected. In this study, we explored in vitro drug susceptibility of BDQ, DMD, LZD, CLO, moxifloxacin (MFX), and gatifloxacin (GAT) against clinical XDR-TB Mycobacterium tuberculosis isolates from TB patients seeking health care in the National Clinical Center on TB of China on the basis of MIC determination in liquid Middlebrook 7H9 medium. In addition, we describe the genetic characteristics of XDR-TB isolates with acquired drug resistance.

RESULTS

Drug susceptibility profiles of XDR-TB strains.

A total of 93 clinical XDR-TB strains were isolated from TB patients seeking health care in the National Clinical Center on TB of China, whereas 3 strains were excluded from the study due to subculture failure. Table 1 shows the drug susceptibility testing (DST) profiles of these 90 XDR-TB strains. Overall, the strains were all resistant to streptomycin (STR) (100.0%), ofloxacin (OFX) (100.0%), and amikacin (AMK) (100.0%), followed by levofloxacin (LVX) (94.4%; 85/90), ethambutol (EMB) (88.9%; 80/90), and capreomycin (CAP) (77.8%; 70/90). Notably, there were 61 strains (67.8%; 61/90) resistant to all the antituberculous drugs tested.

TABLE 1.

Drug resistance profiles of XDR-TB isolates used in the study

Drug	No. of resistant isolates	Percentage
STR	90	100.0
EMB	80	88.9
OFX	90	100.0
LVX	85	94.4
AMK	90	100.0
CAP	70	88.0
Total	90	100.0

We further analyzed the susceptibility of XDR-TB strains to MFX, GAT, LZD, CLO, BDQ, and DMD. As shown in Table 2, MFX had the highest resistance rate (91.1%; 82/90). In addition, resistance to GAT, LZD, CLO, and DMD was found in 76 (84.4%), 5 (5.6%), 5 (5.6%), and 4 (4.4%) isolates among the XDR-TB strains, respectively. The lowest resistance rate was observed for BDQ (3.3%; 3/90) when 0.25 mg/liter was set as the breakpoint value. Despite no significant difference being observed between the proportions of MFX- and GAT-resistant strains (P = 0.17), we found that the percentage of strains with high-level GAT resistance (25.6%; 23/90) was significantly lower than that of strains with high-level MFX resistance (46.7%; 42/90) among XDR-TB strains (P < 0.01).

TABLE 2.

MIC distribution of XDR-TB strains against MFX, GAT, LZD, CLO, BDQ, and DMD

Antimicrobial agent	No. of strains with MIC (mg/liter) of:											Breakpoint MIC for resistance (mg/liter)	No. (%) of resistant strains
	≤0.031	0.063	0.125	0.25	0.5	1	2	4	8	16	32
MFX	0	0	2	1	5	13	27	29	10	3	0	0.5	82 (91.1)
GAT	0	0	2	2	10	19	34	19	4	0	0	0.5	76 (84.4)
LZD	0	0	6	35	40	4	0	1	1	2	1	1.0	5 (5.6)
CLO	0	13	28	34	7	3	4	1	0	0	0	1.0	5 (5.6)
BDQ	81	2	3	1	3	0	0	0	0	0	0	0.25	3 (3.3)
DMD	86	0	0	0	1	0	0	0	0	0	3	0.125	4 (4.4)

Mutations conferring FQ resistance.

A total of 81 out of 90 (90.0%) XDR-TB strains carried a mutation located in the gyrA gene, while no nonsynonymous mutations were identified in the gyrB gene. The most frequent mutations conferring fluoroquinolone (FQ) resistance occurred in codon 94 (57.8%; 52/90), resulting in the amino acid substitution of Asp for Gly (45.6%; 41/90), Asn (4.4%; 4/90), Ala (3.3%; 3/90), Tyr (3.3%; 3/90), or His (1.1%; 1/90). In addition, the amino acid substitutions located in codon 90 (24.4%; 22/90) were the second most prevalent mutations among the XDR-TB isolates, followed by codon 91 (6.7%; 6/90) and codon 88 (1.1%; 1/90) (Table 3).

TABLE 3.

Mutations in the fluoroquinolone resistance-determining region (QRDR) of gyrA and fluoroquinolone MICs

Mutation	FQ	No. of isolates with MIC (mg/liter) of:									No. (%) of resistant isolates	No. (%) of high-level-resistant isolatesa
		0.13	0.25	0.5	1	2	4	8	16	Total
Gly88Ala	MFX		1							1	0 (0.0)	0 (0.0)
	GAT		1							1	0 (0.0)	0 (0.0)
Ala90Val	MFX	1		2	6	10	3			22	19 (86.4)	3 (13.6)
	GAT	1	1	3	8	8	1			22	17 (77.3)	1 (4.5)
Ser91Pro	MFX				2	2	1		1	6	6 (100.0)	2 (33.3)
	GAT				2	2	1	1		6	6 (100.0)	2 (33.3)
Asp94Gly	MFX				1	9	21	8	2	41	41 (100.0)	31 (75.6)
	GAT			1	5	17	15	3		41	40 (97.6)	18 (43.9)
Asp94Asn	MFX					1	3			4	4 (100.0)	3 (75.0)
	GAT				1	3				4	4 (100.0)	0 (0.0)
Asp94Ala	MFX				1	1		1		3	3 (100.0)	1 (33.3)
	GAT			1	1		1			3	2 (66.7)	1 (33.3)
Asp94Tyr	MFX					2	1			3	3 (100.0)	1 (33.3)
	GAT					3				3	3 (100.0)	0 (0.0)
Asp94His	MFX					1				1	1 (100.0)	0 (0.0)
	GAT					1				1	1 (100.0)	0 (0.0)
No mutation	MFX	1		3	3	1		1		9	5 (55.6)	1 (11.1)
	GAT	1		5	2		1			9	3 (33.3)	1 (11.1)

^aHigh-level resistant was defined as a MIC value higher than 2 mg/liter.

The distribution of FQ MICs according to different mutant types was further analyzed. As shown in Table 3, the proportion of MFX-resistant strains was not significantly higher than that of GAT-resistant strains among different mutation types (P > 0.05), whereas statistical analysis revealed that the percentage of strains with high-level resistance to FQs differed according to mutant type. Compared with the high-level MFX-resistant rate in strains with mutations in codon 90 (25.0%; 3/12; P < 0.01), the rate in strains with mutations in codon 94 (69.2%; 36/52) was significantly higher, while those in strains with mutations in codon 88 (0.0%; 0/1; P = 1.00) and codon 91 (33.3%; 2/6; P = 0.28) exhibited no statistical difference.

Mutations conferring cross-resistance between CLO and BDQ.

A total of 5 strains were identified as CLO resistant among XDR-TB strains. Interestingly, all these strains exhibited ≥4-fold upward shifts in BDQ MICs compared with the majority of XDR-TB strains with MICs lower than 0.031 mg/liter (Table 4). Further sequence analysis revealed that all these strains harbored no nucleotide substitution in the atpE gene, while amino acid substitutions due to mutations of codons 53 (60.0%; 3/5) and 157 (20.0%; 1/5) in the Rv0678 gene were observed in four CLO-resistant strains. In addition, the amino substitution Ser53Pro in the Rv0678 gene was also associated with BDQ resistance (66.6%; 2/3). In order to determine whether these mutations were specific for CLO resistance, the amplicons of the Rv0678 genes of the other 85 CLO-susceptible strains were further analyzed by DNA sequencing. However, no mutations in Rv0678 were identified in these strains. We also analyzed the nucleotide polymorphisms in the pepQ gene among the CLO-resistant isolates, and there were no genetic mutations found in the gene.

TABLE 4.

Correlation between MICs and Rv0678 mutations among CLO-resistant isolates

Mutation in Rv0678	No. of isolates	MIC (mg/liter)
		BDQ	CLO
Ser53Pro	2	0.5	2–4
Ser53Leu	1	0.25	2
Tyr157Asp	1	0.125	2
No mutation	1	0.5	2

Mutations conferring LZD and DMD resistance.

To investigate the mutations associated with LZD resistance, we examined 23S rRNA, rplC, and rplD among 5 LZD-resistant strains. Of the 5 strains, only 2 (40.0%) were found to contain the Cys154Arg allele in the rplC gene, while the other two genes seemed not to account for LZD resistance among these strains.

In addition, ddn, fgd1, fbiA, fbiB, and fbiC are candidate genes conferring DMD resistance in M. tuberculosis. Therefore, we analyzed the nucleotide sequences of these five candidates in four DMD-resistant isolates. The results are summarized in Table 5. All the DMD-resistant isolates harbored no mutations in ddn, fgd1, fbiA, and fbiB, while the mutations located in codon 318 of the fbiC gene were identified as the sole mutations related to DMD resistance. Further analysis of DMD-susceptible isolates confirmed that this novel mutation may be associated with DMD resistance in M. tuberculosis.

TABLE 5.

Mutations conferring LZD or DMD resistance among clinical XDR-TB isolates

Drug	Locus	Amino acid substitution	MIC (mg/liter)	No. of isolates (%)a
LZD	rplC	Cys154Arg	4–16	2 (40.0)
DMD	fbiC	Val318Ile	32	2 (50.0)

^aThe proportion represents the rate of isolates with mutation among all resistant isolates.

DISCUSSION

There is no doubt that XDR-TB is a deadly form of TB that can be incurable due to its extreme drug resistance (2). Recently, several promising candidates provided potential solutions for resolving this troublesome dilemma (6). In this study, our data reveal that XDR-TB exhibited a strikingly high proportion of resistance to the current anti-TB drugs in the National Clinical Center on TB of China. Almost 90% of XDR-TB strains tested were resistant to STR, EMB, OFX, LVX, and AMK, which makes it challenging to construct an effective regimen for XDR-TB patients. Hence, the treatment of XDR-TB with extremely broad-spectrum resistance has to rely on second-line anti-TB drugs from groups C and D on the basis of quality-assured in vitro DST results (17). However, most prefecture-level TB laboratories have well-documented competency only in performing first-line drug testing (18). As a consequence, XDR-TB patients would possibly be treated with an empirical regimen employing FQs and second-line injectable drugs, which will accelerate the emergence of further extremely drug-resistant TB. In view of the high rate of MDR- and XDR-TB in China, strengthening the laboratory capability for early detection of susceptibility of M. tuberculosis to second-line drugs is essential to guide the initiation of therapy for XDR-TB.

A recent meta-analysis by Jacobson et al. found that the use of later-generation FQs for the treatment of XDR-TB significantly improved treatment outcomes (19). Despite the decreased potency of OFX and LVX against XDR-TB, the higher levels of bactericidal activity of later-generation FQs may overcome in vivo low-level FQ resistance, thereby leading to improved XDR-TB treatment outcomes (19), which highlight a promising approach for the treatment of XDR-TB. Unfortunately, the high rates of MFX and GAT resistance among the XDR-TB isolates from our study indicate that these two antimicrobial agents may have difficulty producing the desired effect for XDR-TB patients. China has the world's highest growth rate of resistance because of antibiotic abuse (20). As highly effective broad-spectrum antibiotics, FQs have been the most misused antibiotics in the treatment of undiagnosed respiratory bacterial infection for more than 2 decades in China (21). It is worth mentioning that more than 90% of TB patients visited general hospitals to treat TB-related symptoms with FQs rather than first-line anti-TB drugs (22). In view of insufficient duration and weak efficacy of antibiotic monotherapy against M. tuberculosis, the increasing trend of FQ resistance in M. tuberculosis could be expected. Consistent with our hypothesis, a recent report by Pang and colleagues demonstrated that resistance to MFX increased significantly in China between 2000 and 2010 (23). The increasing emergence of FQ resistance highlights the urgent need to take action to fight FQ misuse in clinical practice.

Another interesting finding of our study is that BDQ, DMD, LZD, and CLO exhibit excellent in vitro activity against XDR-TB, and almost 95% of XDR-TB strains were susceptible in the National Clinical Center on TB of China. In spite of documenting the lowest resistance rate to BDQ in clinical XDR-TB strains, findings from previous studies have revealed that BDQ and CLO share the same efflux system, MmpS5-MmpL5, and exposure to CLO, therefore, may select for efflux-based resistance, resulting in cross-resistance between the two drugs (24). In accordance with previous observations, our data revealed that the amino acid substitutions in the Rv0678 gene are associated with CLO resistance, as well as increased BDQ MICs. On one hand, despite a small number of CLO-resistant isolates, our primary results indicate that the mutations in Rv0678 may serve as a promising predictor for CLO resistance in M. tuberculosis. On the other hand, previous treatment with CLO may create a high risk of acquired resistance of M. tuberculosis to BDQ. In addition, the rplC gene was identified in only 2 XDR-TB isolates (40.0%) as conferring LZD resistance, which is consistent with a molecular epidemiological study showing that less than 30% of LZD-resistant isolates from China harbored mutations in rplC and 23S rRNA (10). Further research is required to broaden our knowledge of the mechanisms of LZD resistance in M. tuberculosis, which is essential to develop a molecular diagnostic assay to monitor LZD resistance.

Adding to previous studies of mutations in the ddn and fgd1 genes, we identified a potential nonsynonymous substitution in fbiC conferring DMD resistance in M. tuberculosis isolates (25). As a member of the coenzyme F₄₂₀ biosynthesis pathway, FbiC catalyzes the transfer of the hydroxybenzyl group from 4-hydroxy-phenylpyruvate to pyrimidinedione (26). In a recent study by Haver et al., the greatest single nucleotide polymorphism (SNP) diversity was identified in fbiC among spontaneously generated mutant strains resistant to PA-824, another nitroimidazole similar to DMD, suggesting a potential role of fbiC in nitroimidazole resistance (26). Although we cannot entirely exclude the possibility that additional mechanisms may be involved in DMD resistance, we speculate that the mutation in fbiC serves as an important contributor to DMD resistance in M. tuberculosis. Notably, we found that two out of three strains with fbiC mutations had the same mutation in the gyrA gene. Despite lack of epidemiological data, our findings suggest that these two strains are related.

We acknowledge several limitations in our study design. First, although we have found several potential mutations conferring drug resistance on M. tuberculosis, further genetic evidence is urgently needed to confirm our findings, which will extend our knowledge of the mechanism of resistance to new antimicrobial agents in M. tuberculosis. Second, previous studies using multiple genotyping assays have demonstrated that recent transmission drives the MDR- and XDR-TB epidemic in several settings (27, 28). Therefore, it is interesting to identify genotypic clusters that reflect recent transmission among XDR-TB isolates in the present study. Unfortunately, the patients seeking health care in the Beijing Chest Hospital were from neighboring regions, and the strains may not be representative of XDR-TB circulating in Beijing. Nevertheless, this study provides hints for further application of new anti-TB drugs for the treatment of XDR-TB in China.

In conclusion, our data demonstrate that XDR-TB strains exhibit a strikingly high proportion of resistance to the current anti-TB drugs, whereas BDQ, DMD, LZD, and CLO exhibited excellent in vitro activity against XDR-TB in the National Clinical Center on TB of China. The extensive cross-resistance between OFX and later-generation FQs indicates that MFX and GAT may have difficulty in producing the desired effect for XDR-TB patients. The strains with mutations in codon 94 of the gyrA gene are more likely to be associated with high-level FQ resistance. In addition, the nonsynonymous mutation in the fbiC gene may confer DMD resistance on M. tuberculosis isolates. Further study will help us to more fully understand the potential role of F₄₂₀ biosynthesis in DMD resistance.

MATERIALS AND METHODS

Ethics statement.

The protocols applied in this study were approved by the Ethics Committee of the Beijing Chest Hospital, Capital Medical University, Beijing, China.

Bacterial strains and culture conditions.

XDR-TB strains were obtained from patients consecutively admitted to a medical service in Beijing Chest Hospital between June 2016 and December 2016. Each M. tuberculosis strain was isolated from a unique patient. XDR-TB isolates were selected by retrospectively reviewing in vitro drug susceptibility testing results performed in the National TB Clinical Laboratory. Drug susceptibility was determined using the absolute concentration method on Löwenstein-Jensen (L-J) medium containing the corresponding anti-TB drugs according to the guidelines of the World Health Organization (WHO) (29). In total, eight drugs were used to perform conventional DST, and the concentrations of anti-TB drugs were as follows: rifampin (RIF), 40 mg/ml; isoniazid (INH), 0.2 mg/ml; STR, 10 mg/ml; EMB, 2 mg/ml; CAP, 40 mg/ml; AMK, 30 mg/ml; OFX, 2 mg/ml; LVX, 2 mg/ml (27). XDR-TB was defined as resistance to RIF and INH and additional resistance to any fluoroquinolone and at least one of the injectable second-line drugs (30). All bacterial cells were stored in 7H9 broth containing 15% glycerol in a −80°C refrigerator. Prior to in vitro susceptibility testing, the strains were recovered on L-J medium for 4 weeks at 37°C.

MIC.

We used the alamarBlue assay to perform the MIC determination for XDR-TB identified by conventional drug susceptibility testing (31). Briefly, bacterial clones were harvested from the surfaces of L-J slants. After vigorous mixing on a vortex mixer for 1 min, the suspension was adjusted to a turbidity equivalent to 1.0 McFarland standard. The inoculum was further prepared by 1:20 dilution of 1.0 McFarland standard cell suspension with Middlebrook 7H9 broth containing 10% oleic acid-albumin-dextrose-catalase (OADC), and 100 μl of this inoculum was added to the wells of the 96-well plate containing the corresponding drugs. After 7 days of incubation, 70 μl of alamarBlue solution was added to each well and incubated for 24 h at 37°C. The color change was used to evaluate bacterial growth, and the MIC was defined as the lowest concentration of drug that prevented the color change from blue to pink. The concentrations of antibiotics in the test panel were 0.016 to 32 mg/liter for CLO, LZD, BDQ, and DMD. The MIC breakpoint concentrations were defined as 0.5 mg/liter for MFX and GAT, 1.0 mg/liter for LZD and CLO, 0.25 mg/liter for BDQ, and 0.125 mg/liter for DMD (10, 14, 21, 32, 33).

Genomic-DNA extraction.

Genomic DNA was extracted from freshly cultured bacteria as previously reported (34). The fresh bacteria were harvested from the surface of an L-J slant and then transferred into a microcentrifuge tube containing 500 μl Tris-EDTA (TE) buffer. After centrifugation at 13,000 rpm for 2 min, the supernatant was discarded, whereas the pellet was resuspended in 500 μl TE buffer and then heated in a 95°C water bath for 1 h. Following centrifugation of cellular debris, the crude DNA in the supernatant was used as the template for PCR amplification.

PCR amplification and DNA sequencing.

A total of 13 genes conferring drug resistance were analyzed using Sanger sequencing: gyrA and gyrB for fluoroquinolones; 23S rRNA, rplC, and rplD for LZD (10); atpE for BDQ; Rv0678 and pepQ for both BDQ and CLO (35); and ddn, fgd1, fbiA, fbiB, and fbiC for DMD. The primer pairs used in the study are listed in Table S1 in the supplemental material. The 50-μl reaction volume contained 25 μl of 2× PCR mixture (Genestar, China), 0.2 μM each primer, and 4 μl of template DNA. The PCR cycling conditions were as follows: an initial denaturation at 95°C for 5 min; 35 cycles of denaturation at 94°C for 1 min, annealing at 58°C for 1 min, and extension at 72°C for 1 min; and a final extension at 72°C for 5 min. The fragments were purified with a QIAquick gel extraction kit and provided to the Tsingke Company for DNA sequence analysis.

Statistical analysis.

The Pearson chi-square test or Fisher's exact test was used to compare proportions or rates. A P value of <0.05 was considered statistically significant. All statistical analysis was performed with SPSS version 17.0 software (SPSS Inc., Chicago, IL).