The Mycobacterial Efflux Pump EfpA Can Induce High Drug Tolerance to Many Antituberculosis Drugs, Including Moxifloxacin, in Mycobacterium smegmatis

ABSTRACT

Active efflux of drugs across the membrane is a major survival strategy of bacteria against many drugs. In this work, we characterize an efflux pump, EfpA, from the major facilitator superfamily, that is highly conserved among both slow-growing and fast-growing Mycobacterium species and has been found to be upregulated in many clinical isolates of Mycobacterium tuberculosis. The gene encoding EfpA from Mycobacterium smegmatis was overexpressed under the control of both a constitutive and an inducible promoter. The expression of the efpA gene under the control of both promoters resulted in >32-fold-increased drug tolerance of M. smegmatis cells to many first-line (rifampicin, isoniazid, and streptomycin) and second-line (amikacin) antituberculosis drugs. Notably, the drug tolerance of M. smegmatis cells to moxifloxacin increased by more than 180-fold when efpA was overexpressed. The increase in MICs correlated with the decreased uptake of drugs, including norfloxacin, moxifloxacin, and ethidium bromide, and the high MIC could be reversed in the presence of an efflux pump inhibitor. A correlation was observed between the MICs of drugs and the efflux pump expression level, suggesting that the latter could be modulated by varying the expression level of the efflux pump. The expression of high levels of efpA did not impact the fitness of the cells when supplemented with glucose. The efpA gene is conserved across both pathogenic and nonpathogenic mycobacteria. The efpA gene from Mycobacterium bovis BCG/M. tuberculosis, which is 80% identical to efpA from M. smegmatis, also led to decreased antimicrobial efficacy of many drugs, although the fold change was lower. When overexpressed in M. bovis BCG, 8-fold-higher drug tolerance to moxifloxacin was observed. This is the first report of an efflux pump from Mycobacterium species that leads to higher drug tolerance to moxifloxacin, a promising new drug for the treatment of tuberculosis.

INTRODUCTION

The occurrence of 4.8 million new drug-resistant tuberculosis (TB) cases is an alarming signal that the existing treatment regime may require change (1). The resilient nature of this microbe along with drug resistance are primary concerns and reasons for the continued emergence of TB as a global issue through decades (2). One of the primary mechanisms of resistance in Mycobacterium is mutations in drug target genes. However, many drug-resistant clinical strains have been reported in the past, which do not harbor mutations (3–6) in the known target genes. Increased drug resistance in Mycobacterium, without any classical mutations, points toward alternate drug resistance mechanisms that can also contribute to a substantially high level of resistance (7, 8). One possible mechanism of resistance is through the overexpression of efflux pumps. In fact, multiple efflux pumps have been observed to be differentially upregulated in many clinical strains, which suggests their contribution to drug resistance (9, 10).

Efflux pumps have been shown to play a significant role in both host-induced as well as drug-induced tolerance (11). Drug tolerance in Mycobacterium tuberculosis residing in macrophages has been attributed to macrophage-induced efflux pumps (12, 13). Low-level expression of efflux pumps has been thought to facilitate evolution toward higher resistance (14). For example, exposure to azithromycin led to the overexpression of many efflux pumps, including MAV_3306 and MAV_1406 of Mycobacterium avium, which have orthologs in M. tuberculosis and Mycobacterium leprae (14). Another study demonstrated that a lower level of efflux-mediated isoniazid resistance eventually resulted in mutations in the katG gene leading to higher isoniazid resistance (5). Thus, one can postulate that even low-level efflux-mediated resistance will subsequently be a reason for the development of drug resistance in Mycobacterium.

Multiple efflux pumps are present across the Mycobacterium genome. To understand their role, many of the efflux pumps have been deleted from the genome (15, 16). However, deletion alone is not sufficient to reveal their role in clinical isolates due to the presence of redundant efflux pumps in the genome (17). In the absence of drug pressure, efflux pumps are mostly transcriptionally silent due to the presence of a regulator (16, 18). Therefore, in drug-resistant clinical isolates, efflux pumps have been found to be mostly upregulated (10, 19, 20). Thus, their contribution to resistance can be assessed only when the pumps are overexpressed in the bacteria (17).

EfpA, a widely conserved efflux pump from the major facilitator superfamily (MFS) transporter family, was found to be upregulated in many resistant clinical isolates of M. tuberculosis, either constitutively or in response to drugs, including isoniazid (INH) and rifampicin (RIF) (20, 21). To determine the contribution of EfpA to intrinsic resistance, the efpA gene was previously deleted in Mycobacterium smegmatis (17), and the impact of the deletion on drug susceptibility was measured. A small change in the MICs of various drugs, including cationic dyes and fluoroquinolones (FQs), was observed. In this work, we study the effect of the overexpression of efpA on the resistance profile of M. smegmatis to elucidate its precise role in drug tolerance.

RESULTS

Conservation of efpA across Mycobacterium species.

EfpA is a 13-transmembrane efflux pump belonging to the MFS family of transporters. It is highly related to members of the QacA transporter family (22). The efflux pump protein was found to be widely conserved across the Mycobacterium species, with a homolog in more than 530 species with more than 70% identity and 90% query coverage. A phylogenetic tree was constructed with the top 61 species with >80% identity and 100% query coverage and is presented in Fig. 1. The phylogenetic tree shows conservation across both fast-growing (for example, M. smegmatis, Mycobacterium fortuitum, Mycobacterium mageritense, and Mycobacterium wolinskyi) and slow-growing (for example, M. tuberculosis, M. avium, and M. leprae) human pathogens.

FIG 1.

Phylogenetic tree based on efpA protein sequences with 80% identity and 100% query coverage.

Contribution of EfpA to efflux-mediated drug resistance.

To demonstrate the effect of the overexpression of efpA, we cloned the gene under the control of a constitutive promoter (P_myc1tetO) and expressed it extrachromosomally in wild-type (WT) M. smegmatis. The resultant strain is referred to as efpA_OE, whereas the cells expressing the empty vector are referred to as EV_OE. The expression of efpA from the plasmid was confirmed by measuring the transcript levels of efpA in EV_OE cells, which demonstrated a 7.4-fold increase in expression levels. The constitutive overexpression of efpA led to increased drug tolerance to multiple drugs, including FQs, first-line and second-line anti-TB drugs, and ethidium bromide (EtBr). Table 1 presents the MICs and the minimum bactericidal concentrations (MBCs) of the various drugs tested. The MICs of FQs increased severalfold in efpA_OE. While the MIC of norfloxacin (NOR) increased to 512 μg/ml from 2 μg/ml, there was a 320-fold increase in the MIC of ciprofloxacin (CIP) against the efpA_OE strain. Surprisingly, increased resistance was also observed toward the FQ moxifloxacin (MOXI), with the MIC increasing by 125-fold from 0.08 μg/ml to 10 μg/ml.

TABLE 1.

Susceptibility of efpA_OE to various FQs and first- and second-line drugs^a

Drug class	Drug	MIC (μg/ml)		MBC (μg/ml)		Fold change
		WT/EVOE	efpAOE	WT/EVOE	efpAOE
FQs	Norfloxacin	2	512	4	1,024	256
	Ofloxacin	0.6	32	1.2	64	53
	Ciprofloxacin	0.4	128	0.8	216	320
	Moxifloxacin	0.08	10	0.16	15	125
First-line anti-TB drugs	Rifampicin	32	128	64	216	4
	Isoniazidb	32	64	64	128	2
	Streptomycin	1	4	2	16	4
Second-line anti-TB drugs	Amikacin	0.5	4	1	16	16
Aromatic compounds	EtBr	7	14	14	28	2

^an = 3 × 3, where n represents the total number of replicates (number of days × biological replicates of that day).

^bThe MIC of isoniazid was estimated using the disk diffusion method. For the rest of the drugs, the broth microdilution method as well as the disk diffusion method were used. Similar results were obtained using both methods.

Overexpression of the efflux pump efpA led to multidrug resistance (MDR) against structurally unrelated compounds. A 2- to 4-fold increase in resistance was observed toward first-line anti-TB drugs, including rifampicin (from 32 μg/ml to 128 μg/ml), isoniazid (from 32 μg/ml to 64 μg/ml), and streptomycin (STP) (from 1 μg/ml to 4 μg/ml). Similarly, resistance to amikacin (AMIK), a second-line anti-TB drug, increased 16-fold, from 0.5 μg/ml to 4 μg/ml. EtBr, an aromatic compound, is a common substrate of most efflux pumps. We therefore tested the MIC of EtBr against efpA_OE. The MIC increased from 7 μg/ml against the WT to 14 μg/ml against the efpA_OE strain. The bactericidal and bacteriostatic concentrations have also been confirmed by measuring the CFU after treatment with each drug for 48 h (Fig. 2).

FIG 2.

Viability of WT and efpA_OE cells treated with norfloxacin (a), moxifloxacin (b), amikacin (c), and streptomycin (d). The CFU per milliliter were determined using the drop count method, 48 h after treatment with the drug, and compared with the CFU per milliliter with no drug. The concentration of the drugs is mentioned below the bars, where 0 depicts the condition without any drug, and solid colored bars represent the CFU per milliliter at 48 h at increasing concentrations of the drug. The P value is calculated with respect to the 48-h control CFU data. ***, 0.0001 < P < 0.001.

Consistent with the increased MIC, the intracellular drug concentrations of both norfloxacin and moxifloxacin were found to be 5- and 3-fold lower, respectively, in the efpA_OE cells than in the EV_OE control cells (Fig. 3a and b). Similarly, the accumulation of EtBr, a substrate for most efflux pumps, including efpA, was found to be negligible in the efpA_OE cells compared to the control cells (Fig. 3c). Furthermore, to determine if this decreased uptake is due to alteration of the membrane permeability of the efpA_OE cells, the uptake of 1-N-phenylnaphthylamine (NPN) was measured. As shown in Fig. 3d, the permeability of all cells was similar.

FIG 3.

Comparison of drug uptake in efpA_OE and EV_OE cells. (a and b) Intracellular concentrations estimated over time for norfloxacin (extracellular concentration of the drug used for treatment: 64 μg/ml) (a) and moxifloxacin (64 μg/ml) (b), as described in Materials and Methods. (c) Uptake of ethidium bromide (3 μg/ml) in efpA_OE and WT cells measured in terms of relative fluorescence units. (d) Uptake of NPN dye in WT and efpA_OE cells as a measure of membrane permeability. The endpoint NPN fluorescence is calculated based on equation 2. The concentrations of the drugs are mentioned in brackets. ***, P 0.0001 < P < 0.001.

The high MICs of various drugs can be reversed with the efflux pump inhibitor CCCP.

The efflux pump EfpA belongs to the MFS family of pumps that require a proton gradient for their functioning (23). Thus, compounds that inhibit the proton motive force in cells, such as carbonyl cyanide m-chlorophenylhydrazone (CCCP), are used as efflux pump inhibitors (EPIs) (24). To confirm that the increased MICs of the various drugs are due to increased expression of the pump, we investigated the effect of CCCP in combination with multiple drugs. In the presence of a subinhibitory concentration of CCCP, efpA_OE was found to be more susceptible to all the drugs (Table 2). While resistance to norfloxacin decreased by 32-fold, the MICs of streptomycin, ciprofloxacin, and amikacin decreased by 8- to 16-fold, and that of ofloxacin (OFLO) decreased by 6.4-fold, in the presence of CCCP. The MICs of other drugs, including rifampicin and EtBr, also decreased by 4-fold when CCCP was administered along with the drug. In contrast, the MICs of all drugs were reduced by 2-fold when the WT was treated with the drug in combination with CCCP (Table 2). Interestingly, the MICs of isoniazid and moxifloxacin were reduced by only 2-fold in the presence of CCCP. The decreased resistance in the presence of CCCP suggests that the activity of efpA is influenced in the presence of an EPI.

TABLE 2.

MICs of various drugs against efpA_OE decrease in the presence of CCCP^a

Drug class	Drug	MIC (μg/ml)
		Without CCCP		With CCCP (15 μg/ml)		Max fold decrease of MIC
		WT	efpAOE	WT	efpAOE	WT	efpAOE
FQs	Norfloxacin	2	512	1	16	2	32
	Ciprofloxacin	0.6	128	0.2	10	3	12.8
	Ofloxacin	0.4	32	0.3	5	1.3	6.4
	Moxifloxacin	0.08	10	0.04	5	2	2
First-line anti-TB drugs	Rifampicin	32	128	16	32	2	4
	Isoniazidb	32	64	16	32	2	2
	Streptomycin	1	4	0.5	0.5	2	8
Second-line anti-TB drugs	Amikacin	0.5	4	0.25	0.25	2	16
Aromatic compounds	EtBr	7	14	3.5	3.5	2	4

^an = 3 × 3, where n represents the total number of replicates (number of days × biological replicates of that day).

^bThe MIC of isoniazid was estimated using the disk diffusion method. For the rest of the drugs, the broth microdilution method as well as the disk diffusion method were used. Similar results were obtained using both methods.

Furthermore, we estimated the intracellular concentrations of norfloxacin, moxifloxacin, and EtBr in the presence of CCCP to evaluate the impact of EPIs. As shown in Fig. 4a and b, norfloxacin and moxifloxacin concentrations were higher in cells treated with a subinhibitory concentration of CCCP than in the absence of CCCP, in both WT and efpA_OE cells. Similarly, EtBr accumulation in the presence of CCCP (Fig. 4c) was higher in both WT and efpA_OE cells. EtBr accumulation was found to be restored to the levels of the WT in the presence of CCCP (25 μg/ml). This correlated with the decreased MICs of these drugs in the presence of CCCP.

FIG 4.

Effect of CCCP on drug uptake. Shown is increased accumulation of norfloxacin (4 μg/ml) (with CCCP [15 μg/ml]) (a), moxifloxacin (8 μg/ml and 16 μg/ml) (with CCCP [15 μg/ml]) (b), and EtBr (3 μg/ml) (c) in the presence of increasing concentrations of the efflux pump inhibitor CCCP (0 μg/ml to 25 μg/ml) after an hour. − refers to the absence of CCCP, and + refers to when CCCP was added along with the drug. ns, not significant; *, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.0001 < P < 0.001. Note that CCCP is used at subinhibitory concentrations. The MICs of CCCP are 25 μg/ml against the WT and >30 μg/ml against efpA_OE.

Tetracycline-controlled expression system for efpA in M. smegmatis.

Many MDR clinical strains of M. tuberculosis have been found to have higher transcript levels of the efpA gene, with a fold change of between 2 and 8 (5, 20, 21, 25). This correlates with the increased resistance observed in our work on overexpressing efpA under the control of a constitutive promoter. We next wanted to determine if the MICs of drugs can be modulated by varying the expression level of the pump. For this, we expressed efpA under the control of a tight tet-tunable expression system. We designed a construct, efpA_i, harboring efpA under the control of a tetracycline-inducible promoter, with a single copy of the gene integrated into the genome. In the absence of induction, the MICs were similar to that of the WT/EV_i for all the drugs (FQs, anti-TB drugs, and ethidium bromide). Nevertheless, upon inducing efpA_i cells with increasing concentrations of tetracycline over time, the MICs of various drugs increased. Figure 5 presents the MICs estimated using the broth microdilution method. There was a significant change in the level of drug resistance against first-line and second-line anti-TB drugs, FQs, as well as aromatic compounds. In the case of FQs, the MIC was observed to increase with the time of induction. While 15 min of induction led to an 8-fold increase in the MIC of norfloxacin, this increased to a 32-fold-higher MIC with 2 h of induction and a 128-fold-higher MIC by 5 h. The MIC of norfloxacin was measured to be 1,024 μg/ml against efpA_i cells induced for 14 h with tetracycline. A similar impact of induction was seen for the other drugs (Fig. 5). The same has been confirmed by CFU analysis for 0 h and 48 h for each of the drug treatments. Note that the MICs were similar when anhydrous tetracycline (ATc) was used as an inducer against efpA_OE for isoniazid, rifampicin, norfloxacin, and moxifloxacin.

FIG 5.

Effect of induction of efpA_i on the MICs of FQs and first- and second-line drugs. The inducer tetracycline was used at a concentration of 40 ng/ml for the time period indicated. Induction with anhydrous tetracycline instead of tetracycline resulted in similar MICs. The MIC of isoniazid was estimated using a disk diffusion assay. In the case of the drugs moxifloxacin, rifampicin, isoniazid, and norfloxacin, we used both inducers (tetracycline and anhydrous tetracycline) at the same concentrations, and similar results were estimated.

Corresponding to the increased MICs, the intracellular levels of three drugs, norfloxacin, moxifloxacin, and EtBr, were measured in cells induced for 14 h. The concentrations of all three drugs were found to be lower than those in the corresponding controls, WT and EV_i cells (Fig. 6a to c). The MICs of various drugs increased as the time of induction increased. We hypothesize that this increase is due to the increase in the efflux pump levels. Correspondingly, the mRNA expression levels of the efpA gene were measured after induction at various time points (Fig. 7a). Furthermore, the MICs of many drugs correlate with the mRNA levels of the efpA gene in efpA_i cells (Fig. 7b to f).

FIG 6.

Estimation of intracellular drug concentrations in efpA_i cells. (a and b) Cells were treated with norfloxacin (64 μg/ml) (a) and moxifloxacin (64 μg/ml) (b) for 2 h. Before treatment with the drug, the cells were induced with tetracycline (40 ng/ml) for 14 h. Uptake was compared to that in cells with the empty plasmid as a control. Uptake in efpA_i cells that had not been induced is also shown for comparison. (c) Estimation of EtBr accumulation over 1 h. Normalized EtBr is calculated based on equation 1. ***, 0.0001 < P < 0.001; ns, not significant.

FIG 7.

Correlation of MICs with mRNA levels. (a) Transcript levels of efpA in construct efpA_i after induction with tetracycline for the specified times. (b to f) MICs versus mRNA levels for efpA_i for norfloxacin (r = 0.83) (b), moxifloxacin (r = 0.86) (c), amikacin (r = 0.76) (d), streptomycin (r = 0.56) (e), and isoniazid (r = 0.86) (f) are plotted, and the correlation coefficient (r) was calculated. SigA is used as a housekeeping gene for normalization; fold changes are calculated with respect to the value at 0 h.

Effect of efpA expression on the fitness of bacteria.

To evaluate whether the overexpression of the efflux pump leads to any fitness deficit, we measured the growth kinetics of the constructs (efpA_OE and efpA_i) in comparison to the WT and empty plasmid (EV_OE and EV_i) strains. It was observed that efpA_OE, where efpA is under the control of the constitutive promoter, exhibited a growth kinetic profile similar to that of the WT. The expression of the pSTK vector also did not affect the growth of the cells (Fig. 8a). However, the construct under the control of the integration-inducible promoter exhibited compromised growth with a longer lag phase and an increased doubling time (Fig. 8b). This observed growth deficit could be due to the higher basal expression level of the efflux pump mRNA, which may lead to an increased requirement for energy during the exponential phase. Note that EV_i expression did not lead to a change in the growth kinetics of the bacteria. Thus, to overcome the potential energy deficit, we supplemented the growth medium with 10% (vol/vol) glucose. The addition of glucose as a supplement led to a restoration of the growth profile of the efpA_i construct to that of the WT/EV_i, under both induced and uninduced conditions (Fig. 8c). Furthermore, in order to rule out any possibilities that slow growth might have imparted the increased resistance of efpA_i cells, the MICs of two drugs, norfloxacin and rifampicin, were measured in the presence of 10% glucose. A similar resistance pattern was observed.

FIG 8.

Effect of overexpression of efpA on fitness. Fitness estimation was done through growth kinetics for constructs. (a) Comparison of growth kinetics of strains of M. smegmatis cells (WT, EV_OE, and efpA_OE). Note that the growth kinetics are measured in the presence of a selection marker (kanamycin [25 μg/ml]) for construct efpA_OE. (b) Comparison of growth kinetics of various inducible constructs in the presence of a selection marker (kanamycin [25 μg/ml]) and an inducer (tetracycline [40 ng/ml]). (c) Comparison of growth kinetics when medium is supplemented with 10% glucose. Kanamycin (25 μg/ml) was added to the medium during inoculation as a selection pressure, and tetracycline (40 ng/ml) was added as an inducer when cells reached an OD₆₀₀ of 0.5. (d) Percent viability of THP-1-derived macrophages infected with strains of M. smegmatis (WT and efpA_OE) compared with the respective noninfected macrophages for 24 h and 48 h. (e) Intracellular log₁₀ CFU counts for WT and efpA_OE cells at 0 h, 24 h, and 48 h of infection for both strains.

The survival of WT and efpA_OE cells inside THP-1-derived macrophages was also estimated (Fig. 8d). The intracellular cell counts of WT and efpA_OE cells were found to be similar immediately after infection. Furthermore, survival was similar to that of WT cells at the end of 48 h (Fig. 8e).

Effect of the efpA sequences from M. bovis BCG and M. tuberculosis H37Ra.

To further study the relationship of efpA expression and drug resistance in clinically relevant strains, we compared the efpA genes from M. bovis BCG and M. tuberculosis H37Ra and H37Rv with the sequence from M. smegmatis. The EfpA protein sequence from M. bovis/M. tuberculosis is 77% identical to that from M. smegmatis (Fig. 9). To understand the effect of this sequence variation, we overexpressed the efpA gene from M. bovis BCG/M. tuberculosis H37Ra in M. smegmatis. An 8-fold-higher level of resistance to isoniazid and a 4-fold increase in resistance to rifampicin and amikacin were observed. The MICs of other drugs, including ofloxacin, moxifloxacin, streptomycin, and EtBr (Table 3), also increased by 2-fold. Furthermore, we also overexpressed the efpA gene from M. bovis BCG in its native host. A drug susceptibility assay was performed, and we observed an 8-fold increase in resistance to both moxifloxacin and rifampicin. Resistance to isoniazid and amikacin was found to increase by 4-fold, whereas there was a 2-fold increase in the resistance of the efpA-bovis_OE strain to both ciprofloxacin and ofloxacin (Table 3).

FIG 9.

EfpA conservation across M. smegmatis (GenBank accession no. ABK73974.1), M. tuberculosis H37Rv (M. tb Rv) (accession no. P9WJY5.1), M. tuberculosis H37Ra (M. tb Ra) (accession no. WP_003414532.1), and M. bovis BCG (accession no. SIU01491.1). The efpA sequence is 100% identical in the M. bovis and M. tuberculosis H37Rv and H37Ra genomes.

TABLE 3.

Susceptibility of efpA-bovis_OE to various FQs and first- and second-line drugs^a

Drug class	Drug	efpA from M. bovis BCG/M. tuberculosis H37Ra in M. smegmatis			efpA from M. bovis BCG in M. bovis BCG
		MIC (μg/ml)		FC	MIC (μg/ml)		FC
		WT/EVOE	efpAROE		WT/EVOE	efpA-bovisOE
FQs	Norfloxacin	2	2	1	4	4	1
	Ciprofloxacin	0.6	0.6	1	0.24	0.48	2
	Ofloxacin	0.4	0.8	2	0.48	0.96	2
	Moxifloxacin	0.08	0.16	2	0.12	0.96	8
First-line anti-TB drugs	Rifampicin	32	128	4	0.03	0.24	8
	Isoniazidb	32	256	8	0.5	2	4
	Streptomycin	1	2	2	4	4	1
Second-line anti-TB drugs	Amikacin	0.5	2	4	1	4	4
Aromatic compounds	EtBr	7	14	2	1	1	1

^an = 3 × 3, where n represents the total number of replicates (number of days × biological replicates of that day). FC, fold change.

^bThe MIC of isoniazid was estimated using the disk diffusion method. For the rest of the drugs, the broth microdilution method as well as the disk diffusion method were used. Similar results were obtained using both methods.

DISCUSSION

Recent studies with clinical MDR strains of M. tuberculosis have revealed the overexpression of many efflux pumps, either constitutively or in response to drug treatment (20). EfpA, a highly conserved efflux pump, is one of the pumps found to be upregulated in clinical MDR strains (5). In this study, we find that the overexpression of efpA leads to increased drug tolerance to a wide variety of anti-TB drugs, from the first-line drugs rifampicin and isoniazid to second-line drugs, including FQs and amikacin. A significant change in this category was observed against fluoroquinolones, including moxifloxacin, a new addition to the anti-TB regime, where the MIC of the drug against M. smegmatis increased from 0.08 μg/ml to 15 μg/ml, representing a >180-fold increase. Although this study was done mostly in M. smegmatis, when the efpA gene from M. bovis BCG was overexpressed, an 8-fold increase in drug tolerance to moxifloxacin was seen, the highest change toward any of the FQs studied in this work.

Moxifloxacin belongs to the fourth-generation FQs and has activity against almost all anaerobes and aerobes (26, 27). In Mycobacterium, moxifloxacin can diffuse across the cell membrane and inactivate DNA gyrase by intercalating between the DNA during segregation, preventing the strands from religating (28). The drug has been found to be effective against ofloxacin-resistant MDR M. tuberculosis strains (29) and has also been shown to have higher efficacy than levofloxacin (30). Along with the first-line drugs, moxifloxacin has also been shown to reduce the treatment duration (29, 31, 32). Thus, the vulnerability of Mycobacterium to moxifloxacin makes this FQ stand out from the rest of its group. However, there is a need to preserve its use in order to limit the development of resistance against this drug (33).

Resistance to moxifloxacin treatment has been observed in several pathogenic organisms, including M. tuberculosis, primarily due to mutations in the gyrA and parC genes. In Streptococcus pneumoniae, ciprofloxacin and levofloxacin were used previously, which primarily target ParC of topoisomerase IV (34). However, with the rapid emergence of ciprofloxacin resistance in pneumococci, moxifloxacin was introduced into the treatment regime as it targets both GyrA and ParC (35). Subsequently, with the increased usage of moxifloxacin for treatment, a change in the resistance pattern was observed for S. pneumoniae with an increase in the resistant population with gyrA and parC double mutations (35, 36). Similarly, moxifloxacin-resistant strains have been reported in Clostridium difficile with mutations in the gyrA gene (37, 38). Similarly, mutations in the gyrA and gyrB genes have been identified in clinical strains of M. tuberculosis resistant to moxifloxacin, and many of them have been shown to confer resistance in different genetic backgrounds (39, 40).

In addition to mutations in gyrA genes, the overexpression of the NorB and NorC efflux pumps also leads to increased resistance to moxifloxacin in Staphylococcus aureus (41, 42). In a laboratory-generated mutant strain of S. aureus, increased resistance to moxifloxacin concomitant with reduced moxifloxacin accumulation was shown. Although the efflux pumps responsible for the higher level of resistance have not been identified, moxifloxacin’s susceptibility could be restored in the presence of EPIs (43), suggesting their role. In another report of urinary tract TB, out of 47 isolates of M. tuberculosis, 22 were resistant to moxifloxacin, where 3 strains showed significant resistance (MIC > 4 mg/liter) and had mutations in the gyrA or gyrB gene (44). The remaining low-level moxifloxacin-resistant strains responded to EPIs with a reduction in MICs by at least 4-fold in their presence. Similarly, in another work on Pseudomonas aeruginosa, a 22% reduction in the MIC of moxifloxacin was observed in the presence of EPIs (45).

Thus, the drug tolerance of both M. smegmatis and M. bovis BCG to moxifloxacin, through the overexpression of the efpA efflux pump, is of high significance. Previously, a 2-fold increase in resistance to moxifloxacin has been reported when an ABC transporter, Rv2686c-2688c, was overexpressed in M. smegmatis (40). We have also seen similar low-level resistance (∼2-fold increase) to moxifloxacin when we overexpressed other efflux pumps belonging to the same family (our unpublished data). Thus, this further signifies the importance of the correlation between efpA expression and moxifloxacin.

Efflux pumps are overexpressed in clinical strains generally through a mutation in the corresponding regulatory gene or a mutation in the promoter region (33, 46). However, not much is known about the regulation of efpA in either M. smegmatis or M. tuberculosis strains. Fold changes in the expression of efpA have been reported in the range of 2 to 8 (log₂ ratio of 1 to 3). We expressed the gene under the control of an inducible promoter to evaluate the relationship between efpA expression and the MICs of various drugs. The data suggest that resistance is proportional to gene expression, although the relationship is different for different drugs, as this will depend on the individual rates of uptake of each drug and pump specificity, etc. Interestingly, the expression of the pump did not cause any fitness defects, even at very high expression levels of the pump, when the culture was supplemented with glucose. Thus, strains with mutations that result in the overexpression of this pump should be selected during the evolution of resistance.

The pathogenic M. tuberculosis strains have around 267 drug transporters, compared to 713 drug transporters in M. smegmatis (47). The efflux pump efpA is a widely conserved pump across the Mycobacterium genus, with high percent identity. Nevertheless, the efpA protein from M. smegmatis is not identical to that from M. tuberculosis, with approximately 80% sequence identity. This sequence divergence seems to impact the extent of resistance provided by these pumps in the host. It would be interesting to determine the key residues in the efpA protein from M. smegmatis that are critical for the observed high-level resistance and whether mutations in these residues can impart higher-level resistance to the pathogenic mycobacteria.

In summary, the efpA efflux pump from the MFS family of transporters can impart high-level resistance to various anti-TB drugs, both first line and second line. A future detailed study of this pump from both slow- and fast-growing mycobacteria would open exciting prospects for the design of efflux pump inhibitors against this important transporter.

MATERIALS AND METHODS

Culture conditions.

M. smegmatis mc²155 (48) and M. smegmatis mc²155 harboring pSTK (overexpression plasmid) and pSTKiT (integration-inducible plasmid), along with empty plasmid controls used for the study, were cultured in Middlebrook 7H9 (M7H9) medium (HiMedia) supplemented with 10% (vol/vol) ADN (5 g albumin, 2 g glucose, and 0.85 g NaCl in 100 ml distilled water, filter sterilized prior to use), 0.44% glycerol (Sigma-Aldrich), and 0.15% Tween 80 (Sigma-Aldrich), with selection pressure (kanamycin) at 37°C at 180 rpm. ADN was initially filter sterilized before addition to the autoclaved medium. Tetracycline (HiMedia) and anhydrous tetracycline (ATc) (Sigma-Aldrich) were used as inducers for induction experiments in the case of pSTKiT (integration-inducible plasmid). Luria-Bertani (LB) medium (HiMedia) was used for growing the bacteria on an agar plate. For M. bovis BCG (48), experimental study strains were handled in a biosafety level 2 (BSL-2) facility. For liquid culture, all cells were grown in Middlebrook 7H9 broth (HiMedia) supplemented with 0.44% glycerol (Sigma-Aldrich), 0.15% Tween 80 (Sigma-Aldrich), and a 10% (vol/vol)ADN solution, with selection pressure (kanamycin). The cultures were maintained at 37°C with aeration at 200 rpm. For growth on solid media, cells were plated on Middlebrook 7H11 agar plates containing 10% (vol/vol) ADN and incubated at 37°C for at least 21 days.

Construction of efpA-overexpressing strains.

For cloning, the efpA gene was amplified from 500 to 600 ng of genomic DNA (gDNA) from the respective bacteria. For amplification from M. smegmatis, the primer sequences used were as follows: forward primer 5′-GGTAAGGTGACCGGTTATG-3′ and reverse primer 5′-GGTGTCAGAAACCTCTTGTCG-3′. For amplification from M. bovis BCG and M. tuberculosis (H37Ra), another set of primers was designed (forward primer 5′-AGGCGGCAAGCCTAATTCGCCG-3′ and reverse primer 5′-AGTCGGCCTGAGCGGCGTTC-3′). The PCR setup consisted of primers at a 0.5 mM final concentration, 1.5 U of DreamTaq DNA polymerase (Thermo Scientific), 1× reaction buffer, 0.2 mM deoxynucleoside triphosphates (dNTPs), and 1.5 mM MgCl₂ in a thermal cycler (HiMedia). This was followed by PCR cleanup. Two plasmids were used for this study: pSTK (constitutive expression) (Addgene plasmid 44560) and pSTKiT (inducible-integrating expression) (Addgene plasmid 44562) (a gift from Vinay Nandicoori, NII) (49). A two-step cloning procedure was used for cloning the gene of interest (efpA gene) under the control of both promoters (efpA_OE under the control of the constitutive promoter and efpA_i under the control of the integrative promoter). Restriction digestion was done using the specific restriction enzymes BamHI and HindIII (final concentration of 0.5 U) and 1× cut smart reaction buffer (New England BioLabs [NEB]) incubated in a water bath at 37°C for 2 h and 15 min, respectively, followed by heat inactivation for 15 to 20 min. The doubly digested plasmid and PCR products were ligated, and transformation was done by electroporation (0.1-cm cuvette) into M. smegmatis and M. bovis BCG, respectively.

MICs.

The MICs of rifampicin, isoniazid, norfloxacin (NOR), ofloxacin (OFLO), amikacin (AMIK), ciprofloxacin (CIP), moxifloxacin (MOXI), streptomycin (STP), ethidium bromide (EtBr), and carbonyl cyanide m-chlorophenylhydrazone (CCCP) were determined by the broth microdilution method. All the MIC experiments were performed according to Clinical and Laboratory Standards Institute (CLSI) guidelines (50). Kanamycin was added at concentrations of 50 μg/ml and 25 μg/ml for all efpA constructs in Escherichia coli and M. smegmatis, respectively. In the case of efpA_i, cells were induced by incubating them with tetracycline (40 ng/ml) for different time intervals varying from 15 min to 14 h. The MIC was estimated using the broth microdilution assay, postinduction, by subjecting the cells (∼10⁶ cells) to various concentrations of drugs. The MIC is defined as the concentration that inhibits growth after 48 h of incubation, as observed visually and by streaking 3 μl of cells onto LB plates. The minimum bactericidal concentration (MBC) is defined as the concentration that led to a >99% reduction in viability. CFU were estimated using the drop count method (51).

For disk diffusion assays, 20 μl of the respective cells was streaked evenly onto LB agar plates using cotton swabs (HiMedia). Furthermore, sterile disks (HiMedia) were placed onto the streaked agar, loaded with the required amount of drug, and incubated for 48 h at 37°C. The zones of inhibition (ZOIs) around the disks were recorded according to previously reported guidelines (52). The amount of loaded drug is proportional to the MIC for a given ZOI, diffusion constant, and time of diffusion as described previously by Kronvall (53). In order to estimate MICs, increasing concentrations of the drug were loaded onto the disk. The final drug amount that resulted in a ZOI of ≥10 mm was considered to be equivalent to the MIC (micrograms). The MIC was estimated using both of the above-defined methods, and similar results were observed.

EtBr accumulation assays.

EtBr accumulation assays were performed using a semiautomated fluorometric method (54). Briefly, mid-log-phase cells were centrifuged at 13,000 rpm for 3 min, and the pellet was washed and resuspended in phosphate-buffered saline (PBS) (pH 7.4). EtBr was added to the cellular suspension at a concentration of 3 μg/ml (less than MIC_1/2). The MICs of CCCP were 25 μg/ml and >30 μg/ml against the wild-type (WT) and efpA constructs, respectively. Fluorescence data were acquired every minute for 60 min. To determine if EtBr binds to the cell membrane, cells were incubated with EtBr at 4°C, assuming negligible diffusion into the cell at this temperature, for an hour, and the EtBr fluorescence was measured. This obtained fluorescence was negligible compared to that obtained at 37°C. Normalized EtBr accumulation was calculated using the following equation (RFU is relative fluorescence units):

$$\text{Normalized EtBr uptake}=\frac{{\left(\frac{{\text{RFU}}_{\text{60\hspace{0.17em}min}}-{\text{RFU}}_{\text{0\hspace{0.17em}min}}}{\text{CFU}}\right)}_{\text{condition}}}{{\left(\frac{{\text{RFU}}_{\text{60\hspace{0.17em}min}}-{\text{RFU}}_{\text{0\hspace{0.17em}min}}}{\text{CFU}}\right)}_{\text{control}}}$$ (1)

To study the effect of the efflux inhibitor on EtBr accumulation, CCCP was added at the desired subinhibitory concentration. The MICs of CCCP were 25 μg/ml and >30 μg/ml against the WT and efpA constructs, respectively. Fluorescence was measured in a microplate reader (Spectra Max multimode M5; Molecular Devices) at 37°C using 530 nm as the excitation (Ex) wavelength and 585 nm as the emission (Em) wavelength.

Estimation of membrane permeability.

The extent of membrane permeabilization was assessed using 1-N-phenylnaphthylamine (NPN) dye (HiMedia) (55), using a modified protocol in our laboratory for Mycobacterium (56). Cells were grown until the optical density at 600 nm (OD₆₀₀) reached 0.5, and the cells were then centrifuged and resuspended in 5 mM HEPES buffer (HiMedia). One hundred microliters of the sample and 100 μl of 50 μM NPN in HEPES were mixed in a 96-well plate. Fluorescence was measured immediately after addition (excitation at 355 nm and emission at 402 nm). CFU were also quantified for each sample. Fluorescence was normalized to the corresponding CFU data for neglecting the effect of different cell viabilities (56). Normalized NPN uptake was calculated as follows:

$$\text{Normalized NPN uptake}=\frac{\text{RFU}}{\text{CFU}}$$ (2)

Norfloxacin and moxifloxacin uptake.

Cells at an OD₆₀₀ of 0.5 (M. smegmatis) were concentrated to an OD₆₀₀ of 5 and resuspended in M7H9 broth with 0.44% glycerol and 0.15% Tween 80. The cells from the WT, EV_OE, and efpA_OE strains of M. smegmatis were further incubated with 64 μg/ml of the drug (norfloxacin/moxifloxacin) for 4 h at 37°C with shaking at 200 rpm. Similarly, cells of the WT, EV_i, and efpA_i strains of M. smegmatis were further incubated for 2 h at 37°C with shaking at 200 rpm with 64 μg/ml of the drug (norfloxacin/moxifloxacin). After incubation, cells were centrifuged at 8,000 rpm at 4°C for 10 min, followed by washing with chilled PBS. The pellet was resuspended, lysed by adding 0.1 M glycine-HCl (pH 3), and kept for 12 h at 37°C with shaking at 200 rpm. Samples were centrifuged, the supernatant was used for fluorescence measurement at wavelengths of 281 nm (Ex) and 440 nm (Em), and the pellet was dried for dry cell weight (DCW) measurements. A standard fluorescence curve of various drug (norfloxacin/moxifloxacin) concentrations in glycine-HCl was plotted for the final estimation of the concentration from the fluorescence of the supernatant. Final results were plotted in micrograms of drug (norfloxacin/moxifloxacin) accumulated intracellularly per milligram of DCW. For estimation by high-performance liquid chromatography (HPLC), treated samples were centrifuged, and the supernatant was filtered using a 0.2-μm polytetrafluoroethylene (PTFE) membrane filter. The concentration of norfloxacin was estimated using HPLC, where the mobile phase used was a mixture of methanol (0.05 M dihydrogen sodium phosphate buffer [pH 3.5], trifluoroacetic acid [TFA])-acetonitrile-water (20:20:60) at a flow rate of 1.5 ml/min and a retention time of 2 min (57). The moxifloxacin concentration was estimated using two methods. In the HPLC-based method, the mobile phase used was a mixture of methanol-water (0.4% TFA [trifluoroacetic acid; pH 3.5], OPA [o-phtalaldehyde]) at a flow rate of 2 ml/min and a retention time of 5 min (58). Both estimations were done using a C₁₈ reverse-phase column (Agilent 1260 infinity, quaternary liquid chromatography VL). For the estimation of moxifloxacin using Spectromax (Molecular Devices), the fluorescence and absorbance of the supernatant were measured using wavelengths of 294 nm (Ex) and 560 nm (Em).

Expression studies.

Real-time PCR (quantitative PCR [qPCR]) was performed to study the transcript levels of efpA in both the WT and constructs. RNA was isolated (59) using the TRIzol method (Sigma-Aldrich), followed by cDNA synthesis using reverse transcriptase (RT) enzyme (Thermo Scientific). The reaction mixture for cDNA synthesis contained 4 μg RNA template, 1 U RT enzyme, 1× RT buffer, 2 mM dNTPs, 0.2 μg random primers, and water to make the final volume up to 40 μl. After the completion of the reaction cycle, comprising 25°C for 10 min, 50°C for 1 h, and a final extension step at 70°C for 10 min, the mixture was diluted to obtain a final cDNA concentration of 100 ng/μl. The transcript levels of efpA were measured using 100 ng of the cDNA template with 0.5 μM (each) efpA-specific primers (forward primer 5′-TCGGATTCATCCCGTTCGTG-3′ and reverse primer 5′-GTGCAGTGTCGAACCGTAGA-3′) with 1× SYBR green using Bio-Rad qPCR. The 2^−ΔΔCT method was used for analysis where SigA rRNA was used as the housekeeping gene, followed by normalization with the empty vector.

Phylogenetic tree analysis.

A phylogenetic tree based on the efpA protein sequence with ≥80% percent identity and 100% query coverage hits shows the relationship between different Mycobacterium strains. The efpA sequence from M. smegmatis was used for the identification of homologs using BLASTp (60). The phylogenetic tree was constructed using the neighbor-joining method based on the method of Naruya Saitou and Masatoshi Nei (61). Evolutionary analyses were conducted in MEGA5.2 (62).

Intramacrophage survival of mycobacteria.

THP-1 cells were seeded at a cell density of 3 × 10⁵ cells/ml in a 24-well plate and treated with 25 ng/ml of phorbol 12-myristate acetate (PMA) for 24 h at 37°C with 5% CO₂. Following PMA treatment, the differentiated THP-1 cells were washed and stabilized in fresh RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) (complete RPMI 1640 medium) for 24 h under the same conditions of temperature and CO₂. Macrophage cells differentiated from THP-1 cells were cocultured with M. smegmatis cells (WT and efpA_OE) at an OD₆₀₀ of 0.5 with a multiplicity of infection (MOI) of 10:1 for 3 h. After 3 h, extracellular bacteria were removed by washing with Ca-Mg-supplemented PBS. Furthermore, cells were incubated with 200 μg ml⁻¹ amikacin-containing RPMI 1640 medium (HiMedia) supplemented with 10% FBS (HiMedia) for 4 h for the elimination of extracellular bacteria (63). For the removal of amikacin, cells were washed twice with Ca-Mg-supplemented PBS, and fresh RPMI 1640 medium supplemented with 10% FBS was added to each well. This time point is considered 0 h. Fresh complete RPMI 1640 medium containing 1 μg/ml amikacin was added to the wells, and cultures were monitored for 48 h. Intracellular bacteria were estimated by the drop count method after lysing the macrophages with 0.05% SDS (63). The differentiated THP-1 cell count was determined from a representative well after trypsinization with a trypsin-EDTA solution (HiMedia) and cell counting using the trypan blue assay.

Statistical analysis.

All the experiments were performed on multiple days (minimum of 3 days) with a minimum of three biological replicates each day. Data are presented as means ± standard deviations. Furthermore, significance between two treatment groups was quantified using Student’s t test with equal variances (*, 0.01 < P < 0.05; **, 0.001 < P < 0.01; ***, 0.0001 < P < 0.001; ns, not significant).