Intricate link between siderophore secretion and drug efflux in Mycobacterium tuberculosis

ABSTRACT

Drug-resistant Mycobacterium tuberculosis is a worldwide health-care problem rendering current tuberculosis (TB) drugs ineffective. Drug efflux is an important mechanism in bacterial drug resistance. The MmpL4 and MmpL5 transporters form functionally redundant complexes with their associated MmpS4 and MmpS5 proteins and constitute the inner membrane components of an essential siderophore secretion system of M. tuberculosis. Inactivating siderophore secretion is toxic for M. tuberculosis due to self-poisoning at low-iron conditions and leads to a strong virulence defect in mice. In this study, we show that M. tuberculosis mutants lacking components of the MmpS4-MmpL4 and MmpS5-MmpL5 systems are more susceptible to bedaquiline, clofazimine, and rifabutin, important drugs for treatment of drug-resistant TB. While genetic deletion experiments revealed similar functions of the MmpL4 and MmpL5 transporters in siderophore and drug secretion, complementation experiments indicated that the MmpS4-MmpL4 proteins alone are not sufficient to restore drug efflux in an M. tuberculosis mutant lacking both operons, in contrast to MmpS5-MmpL5. Importantly, an M. tuberculosis mutant lacking the recently discovered periplasmic Rv0455c protein, which is also essential for siderophore secretion, is more susceptible to the same drugs. These results reveal a promising target for the development of dual-function TB drugs, which might poison M. tuberculosis by blocking siderophore secretion and synergize with other drugs by impairing drug efflux.

INTRODUCTION

Tuberculosis (TB) is the second leading infectious killer after severe acute respiratory syndrome coronavirus 2 (1). An estimated 10 million people fell ill and approximately 1.5 million people died from tuberculosis in 2020 worldwide (1). While the standard treatment regimen of tuberculosis caused by drug-susceptible bacteria has frequent side effects and takes 6 months to complete, it is highly efficient with success rates of at least 85% if treatment guidelines are followed (1). However, treatment of multidrug-resistant tuberculosis, which comprises approximately 4% of all new and 20% of patients with previous treatments, takes up to 2 years, has more side effects, lower success rates (50%), and is prohibitively expensive (2, 3). Thus, new TB drug development and a better understanding of drug resistance mechanisms of Mycobacterium tuberculosis are urgently needed.

Mycobacteria including M. tuberculosis are resistant to many drugs and antibiotics, which are effective against most other bacteria (4). A major factor contributing to the intrinsic drug resistance of mycobacteria is the slow drug uptake by M. tuberculosis due to the outer membrane (OM) permeability barrier (5). However, synergistic contributions of other mechanisms are required to produce significant resistance levels as pointed out by Nikaido and co-workers (6). Drug-specific resistance mechanisms in mycobacteria include mutations of drug targets, of drug-activating, modifying and/or degrading enzymes and of regulatory proteins (7, 8). Drug efflux is an important and relatively unspecific mechanism of drug resistance in M. tuberculosis. While many inner membrane (IM) transporters have been implicated in drug efflux by M. tuberculosis, rigorous evidence by transport assays with appropriate controls is provided for only a few of these transporters (9). By contrast, a wealth of data indicates that MmpL5 (Rv0676c) is involved in drug efflux. MmpL5 belongs to the family of 13 MmpL transporters in M. tuberculosis with similarities to efflux pumps of the resistance-nodulation-division (RND) family and was initially identified in azole-resistant M. tuberculosis mutants with increased transcription of the mmpS5–mmpL5 genes and increased econazole efflux linked to mutations in the rv0678 gene (10). The rv0678 gene encodes a transcriptional repressor of the MarR family that binds to the intergenic regions between rv0678 and the divergent mmpS5-mmpL5 operon (10). Nontarget cross-resistance of M. tuberculosis to bedaquiline (11) and clofazimine (12) is caused by mutations in Rv0678 due to overexpression of mmpS5-mmpL5. Bedaquiline and clofazimine are two important drugs in the treatment of drug-resistant tuberculosis (13 – 15). The importance of the Rv0678 repressor and its regulation of mmpS5-mmpL5 transcription is underlined by Rv0678 mutations in clinical isolates of M. tuberculosis resistant to bedaquiline and clofazimine (16) and by the observation that more than 90% of bedaquiline-resistant M. tuberculosis strains selected in vitro had mutations in Rv0678 (17).

It is important to note that the physiological function of bacterial efflux pumps is not to transport antibiotics and/or TB drugs but rather to transport physiological substrates and/or confer resistance to natural substances in the environment and/or toxic compounds produced by the immune system (18). We showed previously that the mmpS4 and mmpS5 genes (Fig. 1A) are essential for siderophore secretion by M. tuberculosis (19, 20). Biochemical and genetic experiments showed that the MmpS4 and MmpL4 proteins interact, indicating that the MmpS4 and MmpL4 proteins form a membrane-spanning complex and export siderophores across the inner membrane (19). Lack of both the mmpS4 and mmpS5 genes reduced the bacterial burden in the lungs of infected mice by more than 20,000-fold and eliminated any visible lung pathology (19). This result was surprising since M. tuberculosis can utilize heme as an alternative iron source (20). However, the strong virulence defect was explained subsequently by the discovery that M. tuberculosis recycles siderophores and that disruption of siderophore export in the mmpS4/mmpS5 double deletion mutant leads to self-poisoning of M. tuberculosis through accumulation of active siderophores inside the bacterial cell (20).

FIG 1.

Construction of mmpL4 and mmpL5 single and operon deletion mutants of M. tuberculosis. (A) Schematic representation of the mmpS4-mmpL4 and mmpS5-mmpL5 loci in M. tuberculosis. The full and dotted black lines indicate the upstream and downstream regions used for the deletion of the mmpL genes and the complete operons, respectively. (B to D) Southern blots to validate gene deletions. wt: M. tuberculosis mc²6230. (B) Southern blot for the construction of M. tuberculosis ML2300 (ΔmmpL4::loxP, DCU) using the mmpL4 probe: expected fragment sizes are 2,800 bp (wt), 6,500 bp (DCO), and 4,000 bp (DCU). (C) Southern blot for the construction of M. tuberculosis ML2301 (ΔmmpL5::loxP, DCU) using the mmpL5 probe: expected fragment sizes are 2,430 bp (wt), 1,225 bp (DCO), and 2,990 bp (DCU). (D) Southern blot for the construction of M. tuberculosis ML2302 (ΔmmpL4::loxP, ΔmmpL5::loxP; DCU) using the mmpL5 probe: expected fragment sizes are: 2,430 bp (wt), 2,990 bp (ΔmmpL5::loxP, DCU), 2,430 bp (ΔmmpL4::loxP, DCU), and 2,990 bp (ΔmmpL4::loxP, ΔmmpL5::loxP; DCU). (E). Colony PCR to validate gene deletions in M. tuberculosis ML2302 (ΔmmpS4-mmpL4::loxP, ΔmmpS5-mmpL5::loxP; DCU) using primers for the mmpS4-mmpL4 locus (lanes 1 and 2) and primers for the mmpS5-mmpL5 locus (lanes 3 and 4). Lanes: 1 and 3, wt; 2 and 4, ML2302. DCO, double crossover; DCU, double crossover unmarked; wt, wild type.

In this study, we show that M. tuberculosis deletion mutants lacking the mmpL4 and mmpL5 genes are not only required for siderophore secretion as previously inferred from growth defects in the presence of (carboxy)mycobactins (21) and their association with the mmpS4 and mmpS5 genes (19, 20), but are also more susceptible to TB drugs including bedaquiline and clofazimine. While genetic deletion experiments revealed similar, redundant functions of the MmpL4 and MmpL5 transporters in siderophore secretion and drug efflux, complementation experiments indicated that the MmpS4-MmpL4 proteins alone are not sufficient to restore drug efflux in an M. tuberculosis mutant lacking both operons in contrast to the MmpS5-MmpL5 system. Importantly, a mutant lacking the recently discovered periplasmic Rv0455c protein, which is essential for siderophore secretion by M. tuberculosis (22), is also more susceptible to the same TB drugs. This study highlights the intricate link between siderophore secretion and drug efflux in M. tuberculosis and reveals a promising target for new drugs in TB chemotherapy.

RESULTS

Construction and characterization of M. tuberculosis mutants lacking components of the MmpS4-MmpL4 and/or MmpL5-MmpS5 efflux systems

To enable a comprehensive analysis of the role of each component of the MmpS4/MmpL4 and MmpS5/MmpL5 systems in siderophore secretion and their potential involvement in drug efflux with perhaps different drug specificities, we constructed M. tuberculosis strains lacking the individual mmpL4 and mmpL5 genes to complement the previous M. tuberculosis ΔmmpS4 and ΔmmpS5 mutants (19). The mmpL4 (rv0450c) and mmpL5 (rv0676c) genes (Fig. 1A) were replaced by a gfp ²⁺ _m hyg cassette flanked by loxP sites by allelic exchange to create in-frame gene deletions using the vectors pML3001 and pML3003, respectively (Table 1). Hygromycin-resistant clones with green fluorescence were selected and analyzed by Southern blots (Fig. 1B and C). One positive clone was selected for each gene deletion, and their gfp ²⁺ _m hyg cassettes were removed by Cre recombinase. The resulting unmarked deletion mutants were M. tuberculosis ML2300 (ΔmmpL4::loxP) and M. tuberculosis ML2301 (ΔmmpL5::loxP) (Table 1). Since we assumed that MmpL4 and MmpL5 have similar functions in siderophore secretion by M. tuberculosis as their corresponding mmpS4 and mmpS5 genes (19), we deleted the mmpL4 gene in the M. tuberculosis ΔmmpL5 strain by allelic exchange as described above (Fig. 1B and C). The removal of the gfp ²⁺ _m hyg cassette by Cre recombinase resulted in the unmarked deletion mutant M. tuberculosis ML2302 (ΔmmpL4::loxP; ΔmmpL5::loxP) (Table 1). We also constructed the operon deletion mutants M. tuberculosis ML1422 (ΔmmpS4-mmpL4::loxP) and ML2313 (ΔmmpS4-mmpL4::loxP;ΔmmpS5-mmpL5::loxP) (Table 1) using the same approach as described above to avoid potential interference of orphan proteins in the respective single-gene deletion mutants. The deletion of both operons was confirmed by PCR of the chromosomal DNA, which showed the expected fragment lengths at both loci for the parent M. tuberculosis strain and M. tuberculosis ML2313 (Fig. 1E). Sequencing of the resulting PCR products confirmed the in-frame deletions leaving the first three codons of mmpS4 and the last six codons of mmpL4, and the first six codons of mmpS5 and the last seven codons of mmpL5 in the genome of M. tuberculosis ML2313.

TABLE 1.

Role of the siderophore secretion systems in drug susceptibility by M. tuberculosis ^a

	M. tuberculosis strain (MIC90 µg/mL)
Drug	wt	ΔmmpS4/S5	ΔmmpL4/L5	Δrv0455c	Susceptibility factor
Isoniazid	0.062	0.125	0.062	n.d.	1
Cycloserine	6.25	6.25	6.25	n.d.	1
Rifampicin	0.06	0.03	0.06	0.02	2–3
Rifabutin	0.06	0.03	0.03	0.015	2–4
Kanamycin	0.5	0.5	0.25	2	0.25–2.0
Gentamycin	1	1	1	1	1
Clofazimine	2	0.25	0.5	0.25	4–8
Clarithromycin	16–32	16	16	n.d.	1
Bedaquiline	0.5	0.06	0.13	0.06	4–8
Delamanid	2	2	2	n.d.	1
Streptomycin	2	2	2	n.d.	1
Linezolid	4	8	4	n.d.	0.5–1.0
Ethionamide	80	80	80	n.d.	1
Oxifloxacin	1	0.5	1	1	0.5–1.0
Moxifloxacin	0.06	0.06	0.06	0.06	1
Econazole	16	4	16	4	1–4

^a The indicated M. tuberculosis strains were incubated with increasing concentrations of compounds, and their viability was determined using the microplate Alamar Blue assay. The minimal inhibitory concentrations reducing bacaterial growth by 90% (MIC₉₀) and the susceptibility factors of each strain are listed for each compound. n. d. = not determined.

The outer membrane lipid phthiocerol dimycocerosate (PDIM) plays an important role in outer membrane protein function (23) and in protection of M. tuberculosis from toxic compounds (24, 25) but is frequently lost during in repetitive in vitro cultures (26). Analysis of lipid extracts showed that the ΔmmpL4, ΔmmpL5, and ΔmmpL4/ΔmmpL5 mutants retained their ability to produce PDIM (Fig. S1). Since these strains were derived from the parent M. tuberculosis mc²6230 strain, its apparent, strongly reduced PDIM content is probably an experimental artifact. The PDIM levels of the previously constructed ΔmmpS5 and ΔmmpS4/mmpS5 mutants were also reduced and may have contributed to the virulence defect of the ΔmmpS4/mmpS5 mutant (19).

The MmpL4 and MmpL5 transporters perform essential functions in M. tuberculosis

We previously showed that the activity of MmpS4 or MmpS5 is required for growth of M. tuberculosis under low-iron conditions and in the presence of siderophores (19). Experiments in this study revealed a similar growth defect for the M. tuberculosis strain lacking both the mmpL4 and mmpL5 genes (Fig. 2). The growth defect of the M. tuberculosis ΔmmpL4/ΔmmpL5 mutant was enhanced in medium containing carboxymycobactin (cMBT) (Fig. 2B). Addition of siderophores to the medium causes a self-poisoning effect in M. tuberculosis siderophore secretion mutants (20 – 22) and thereby creates a toxic environment in contrast to medium with heme. Expression of the mmpS4-mmpL4 genes using the plasmid pML3601 (Table S2) completely restored growth of the M. tuberculosis ΔmmpL4/ΔmmpL5 mutant (Fig. 2D), demonstrating that the growth defect is indeed caused by the lack of the MmpL4 and/or MmpL5 transporters. This finding is consistent with the observation that the single ΔmmpL4 and ΔmmpL5 mutants do not have any growth defect. The observation that the growth defects of the M. tuberculosis ΔmmpS4/ΔmmpS5 and ΔmmpL4/ΔmmpL5 mutants are similar indicates that the MmpS4 and MmpS5 proteins are as important for growth of M. tuberculosis as their associated MmpL4 and MmpL5 efflux pumps.

FIG 2.

MmpL4 or MmpL5 are required for growth of M. tuberculosis under iron-limited conditions. Growth of M. tuberculosis mc²6230 (parent), ΔmmpL4, ΔmmpL5, ΔmmpL4/mmpL5, ΔmmpL4/mmpL5+mmpS4-mmpL4, ΔmmpS4/ΔmmpS5, and ΔmbtD::hyg was determined by measuring the optical density of the cultures at 600 nm (OD₆₀₀). The bacteria were grown in low-iron minimal medium with albumin and tyloxapol without additional iron source (A) and supplemented with 1-µM carboxymycobactin (B and D) or 10-µM hemin (C), respectively. Error bars represent standard deviations from the mean results of biological triplicates.

The MmpL4 or MmpL5 transporter proteins are required for siderophore secretion by M. tuberculosis

The self-poisoning phenotype of the M. tuberculosis ΔmmpL4/ΔmmpL5 mutant in the medium with carboxymycobactin is similar to that observed for other M. tuberculosis mutants with a defect in siderophore secretion such as the ΔmmpS4/ΔmmpS5 and the Δrv0455c mutants (20 – 22). This result, together with the association of the mmpS4 and mmpS5 with their corresponding mmpL4 and mmpL5 genes, indicates that the MmpL4 and MmpL5 proteins transport siderophores across the inner membrane of M. tuberculosis. To provide direct evidence for this hypothesis, we analyzed ¹⁴C-labeled siderophores extracted from the supernatants of different M. tuberculosis cultures. Thin-layer chromatography (TLC) revealed an almost complete lack of mycobactins and carboxymycobactins in the supernatants of the M. tuberculosis ΔmmpL4/ΔmmpL5 and ΔmmpS4-mmpL4/ΔmmpS5-mmpL5 mutants similar to that observed for the ΔmbtD mutant (Fig. 3), which is not capable of synthesizing siderophores (27). Complementation of the M. tuberculosis ΔmmpL4/ΔmmpL5 mutant with the mmpS4-mmpL4 genes fully restored siderophore production to wild-type (wt) level. The supernatant of a culture of the M. tuberculosis Δrv0455c mutant is almost completely devoid of siderophores similar to the ΔmmpS4/ΔmmpS5 mutant, as shown before (22). In conclusion, these results are consistent with the growth experiments shown above and provide direct evidence that the MmpL4 and MmpL5 efflux pumps are indeed required for siderophore secretion by M. tuberculosis. These results also show that the MmpS4/MmpL4 system is as capable of siderophore secretion as the MmpS5/MmpL5 system.

FIG 3.

MmpL4 or MmpL5 are required for siderophore secretion by M. tuberculosis. TLC of secreted siderophores extracted from cultures of wt M. tuberculosis mc²6230, ΔmmpL4 deletion mutant ML2300, ΔmmpL5 deletion mutant ML2301, mmpL4/mmpL5 deletion mutant ML2302, ΔmmpL4/mmpL5 deletion mutant complemented with mmpS4-mmpL4 (ML2310), ΔmmpS4-mmpL4/mmpS5-mmpL5 deletion mutant ML2311, and the siderophore biosynthetic mutant ΔmbtD::hyg (ML1424). Cultures were labeled with 7-[¹⁴C]-salicylic acid, which was run on the TLC as a control alongside ⁵⁵Fe-loaded cMBT and MBT. Lanes containing cell-associated extracts were loaded with 5,000 cpm, while media extracts were loaded with 7,500 cpm. MBT, mycobactin.

The MmpS4-MmpL4 and MmpS5-MmpL5 proteins have identical functions in siderophore secretion but distinct requirements for accessory proteins

Our experiments established that deletion of both the mmpS4 and mmpS5 or of both the mmpL4 and mmpL5 genes abrogates siderophore secretion in M. tuberculosis (Fig. 3), while deletion of the single genes does not have any detectable effect. In order to directly compare the function of the MmpS4/MmpL4 and MmpS5/MmpL5 systems, we constructed integration vectors constitutively expressing either the mmpS4-mmpL4 (pML4324, Table S2) or the mmpS5-mmpL5 (pML4329, Table S2) genes. First, we examined the expression levels of the encoded proteins in M. tuberculosis ML2313 lacking both operons (Table S1). Western blot showed that the previously produced antisera against purified proteins (19) specifically recognized the MmpS4 and MmpS5 in wt M. tuberculosis but did not detect any protein in the ML2313 strain as expected (Fig. 4). Expression of the mmpS4-mmpL4 and mmpS5-mmpL5 operons in ML2313 revealed slightly elevated MmpS4 and MmpS5 levels compared with the parent strain (Fig. 4). Higher expression levels in the recombinant strains are expected, considering the use of the strong p_smyc promotor (28). We then compared the ability of the MmpS4-MmpL4 and MmpS5-MmpL5 systems to restore siderophore secretion in different M. tuberculosis strains. To this end, we exploited the self-poisoning of siderophore secretion-deficient M. tuberculosis mutants in the presence of extracellular siderophores (20). All three M. tuberculosis mutants showed strong susceptibility to mycobactin (MBT) with minimal inhibitory concentrations (MICs) ranging from 0.03 to 0.13 µg/mL in Alamar Blue viability assays in contrast to wt M. tuberculosis, which grows well even in the presence of much higher MBT concentrations (Fig. 5A through C). Both expression vectors fully restored the viability of the mmpS4/mmpS5 and of the mmpL4/mmpL5 deletion mutants in the presence of high MBT concentrations demonstrating that these genes have similar functions in siderophore secretion consistent with the absence of phenotypes of the single-gene deletion mutants in experiments shown above and with previous complementation results with single genes (19). Surprisingly, the mmpS4-mmpL4 expression vector did not restore MBT tolerance in the ML2313 strain lacking both operons (Fig. 5C), despite the proven function of the MmpS4-MmpL4 exporter in siderophore secretion and despite higher expression levels compared with wt M. tuberculosis (Fig. 4A). This might be attributed to the lack of expression of an accessory gene such as rv0449c (Fig. 1A) when the mmpS4-mmpL4 genes are deleted together in contrast to deletion of the mmpL4 gene. Interestingly, the minimal inhibitory concentration of the mmpS4/mmpS5 deletion mutant was reproducibly slightly lower than that of the mmpL4/mmpL5 deletion mutant (Fig. 5A and B), indicating a more stringent requirement for the mmpS genes compared with the mmpL genes.

FIG 4.

MmpS4 and MmpS5 protein levels in complemented M. tuberculosis strains. Immunoblot analysis of the whole-cell lysate of M. tuberculosis mc²6230 (wt), M. tuberculosis ΔmmpS4-mmpL4/mmpS5-mmpL5 (ML2313), and ML2313 complemented with integrative vectors expressing the mmpS4-mmpL4 or mmpS5-mmpL5 genes. The whole-cell lysates were loaded on a 10% SDS-polyacrylamide gel, and the proteins were blotted on a polyvinylidene difluoride membrane. The proteins were detected with antisera specific for MmpS4 (A) and MmpS5 (B) generated as described previously (19). MW: molecular weight marker

FIG 5.

Efficacy of the mmpS4-mmpL4 and mmpS5-mmpL5 genes to detoxify bedaquiline of siderophore secretion-deficient M. tuberculosis mutants. M. tuberculosis mc²6230 (wt) and the strains lacking the mmpS4/mmpS5 genes (A and D), the mmpL4/mmpL5 genes (B and E), and the mmpS4L4/mmpS5L5 genes (C and F), and the respective strains complemented with the integrative vectors expressing the mmpS4-mmpL4 and mmpS5-mmpL5 operons were grown in minimal medium with heme and tyloxapol, and were incubated with increasing concentrations of mycobactin (A to C) or bedaquiline (D to F). The viability of the M. tuberculosis strains was determined by the microplate Alamar Blue assay. In each panel, the minimal inhibitory concentration MIC₉₀ is represented by a dotted line. Error bars represent standard errors of the mean values of biological triplicates.

The siderophore exporters MmpS4-MmpL4 and MmpS5-MmpL5 have similar functions in drug resistance of M. tuberculosis

Several reports revealed that inactivation of the repressor Rv0678 leads to overexpression of the mmpS5-mmpL5 operon and causes resistance of M. tuberculosis to bedaquiline and clofazimine (11, 12, 17, 29, 30). However, to our knowledge, no mutations were reported affecting the MmpS4-MmpL4 system, which appears to have similar functions in siderophore secretion as the MmpS5-MmpL5 system (Fig. 2 and 3). Hence, we examined whether the MmpS4-MmpL4 proteins are also capable in mediating drug resistance in M. tuberculosis. In addition, we aimed to define the substrate spectrum of these efflux systems. To this end, we determined the susceptibility to 16 selected drugs of M. tuberculosis mutants lacking different components of these efflux systems using the microplate Alamar Blue assay. None of the four M. tuberculosis mutants lacking the single mmpS4, mmpL4, mmpS5, and mmpL5 genes showed an increased susceptibility compared to the parent M. tuberculosis strain (Table S4). This observation is consistent with the results that either the MmpS4/MmpL4 or the MmpS5/MmpL5 need to be inactivated to affect siderophore secretion by M. tuberculosis (19, 20). The only exception was a twofold increased susceptibility to bedaquiline of both the mmpS5 and the mmpL5 mutants compared to the parent M. tuberculosis strain (Table S4). Interestingly, deletion of the mmpL4 or the mmpS4 genes increased the resistance of M. tuberculosis to bedaquiline two- to fourfold (Table S4). This counterintuitive result might be explained by a compensatory upregulation of the mmpS5-mmpL5 operon in an M. tuberculosis strain lacking a functional MmpS4/MmpL4 system. Both observations underline the important role of the MmpS5/MmpL5 efflux system for bedaquiline resistance in M. tuberculosis.

M. tuberculosis was four- to eightfold more susceptible to bedaquiline and clofazimine in the absence of both efflux systems in contrast to the mutants lacking only a single gene (Fig. 6A and B; Table 1). Importantly, rifampicin, one of the first-line TB drugs, and its analog rifabutin against M. tuberculosis were twofold more active in the absence of both efflux systems (Table 1; Fig. 6C). A similar observation was made for econazole, a fungicide with modest activity against M. tuberculosis. The latter result is also consistent with a previous report that overproduction of the MmpS5/MmpL5 efflux pump increased the resistance of M. tuberculosis to econazole (10). The absence of both efflux pumps did not alter the susceptibility of M. tuberculosis to any other tested drug (Table 1). Thus, the substrate spectrum of the MmpS4/MmpL4 and MmpS5/MmpL5 efflux pumps is limited to a few large and hydrophobic compounds with aromatic moieties and a branched structure (Fig. S2). In conclusion, these experiments identified the MmpS5-MmpL5 exporter as a potential source for clinically relevant resistance to rifampicin and its analogs in addition to bedaquiline and clofazimine.

FIG 6.

Siderophore-secretion M. tuberculosis mutants are more susceptible to TB drugs, antibiotics, and fungicides.M. tuberculosis mc²6230 (wt) and the strains lacking the mmpS4/mmpS5, mmpL4/mmpL5, and the rv0455c genes, respectively, were grown in minimal medium HdB medium containing tyloxapol and hemin and incubated with increasing concentrations of bedaquiline (A), clofazamine (B), rifabutin (C), and econazole (D). The viability of the M. tuberculosis strains was determined by the microplate Alamar Blue assay. In each panel, the minimal inhibitory concentration MIC₉₀ is represented by a dotted line. Error bars represent standard errors of the mean values of biological triplicates. HdB, Hartmans-de Bont.

The MmpS5-MmpL5 efflux pump is more efficient in detoxifying bedaquiline than MmpS4-MmpL4

To compare the efficacies of the MmpS4-MmpL4 and MmpS5-MmpL5 efflux pumps in detoxifying drugs, we used the microplate Alamar Blue assay to determine the bedaquiline susceptibilities of the siderophore secretion-deficient M. tuberculosis strains when complemented with either an mmpS4-mmpL4 or an mmpS5-mmpL5 expression vector. In these experiments the siderophore secretion-deficient strains were ~eightfold more susceptible to bedaquiline than wt M. tuberculosis (Fig. 6D through F). Surprisingly, the MmpS4-MmpL4 transporter only slightly increased the resistance of the ΔmmpS4/mmpS5 and ΔmmpL4/mmpL5 mutants to bedaquiline by twofold but did not reduce the bedaquiline susceptibility of M. tuberculosis strain lacking both operons (Fig. 6D through F). In contrast, expression of the mmpS5-mmpL5 genes increased the resistance of the M. tuberculosis ΔmmpS4/mmpS5 strain by more than 60-fold (Fig. 6D). This is a 30-fold higher MIC₉₀ compared to the same strain complemented with the mmpS4-mmpL4 genes, which does not reach wt levels of bedaquiline susceptibility (Fig. 6D). Similar results were obtained for the ΔmmpL4/mmpL5 and ΔmmpS4-L4/mmpS5 L5 mutants: expression of the mmpS5-mmpL5 genes strongly increased the resistance of these strains to bedaquiline. The mmpS5-mmpL5 genes also increased the resistance of the siderophore secretion-deficient strains to clofazimine to slightly higher levels than wt M. tuberculosis, in contrast to the mmpS4-mmpL4 genes, which did not complement any mutant (Fig. S4). We conclude that the MmpS5-MmpL5 efflux pump is much more efficient in detoxifying bedaquiline than MmpS4-MmpL4 in complementation experiments. These results might be explained by the differential requirement of accessory proteins for the activity of the MmpS4-MmpL4 system to secrete mycobactin versus drugs such as bedaquiline, while expression of the mmpS5-mmpL5 genes appears to be sufficient to complement all phenotypes of M. tuberculosis mutants deficient in siderophore secretion.

The periplasmic Rv0455c protein is required for siderophore secretion and resistance of M. tuberculosis to the substrates of the MmpS4-MmpL4 and MmpS5-MmpL5 efflux pumps

Since the periplasmic protein Rv0455c was recently shown to be essential for siderophore secretion by M. tuberculosis (22), we hypothesized that this protein might also be involved in drug efflux. To test this hypothesis, we determined the susceptibility of the M. tuberculosis rv0455c deletion mutant to selected TB drugs and antibiotics using the microplate Alamar Blue assay as described above. Indeed, the lack of Rv0455c sensitized M. tuberculosis to the same drugs as the lack of the MmpS4/MmpL4 and MmpS5/MmpL5 efflux pumps (Table 1; Fig. 6). Surprisingly, the phenotype is equal or stronger for the M. tuberculosis Δrv0455c mutant compared to both the ΔmmpS4-mmpS5 and the ΔmmpL4-mmpL5 double mutants (Fig. 6), indicating that Rv0455c is involved in drug efflux probably in a similar manner as its essential function in siderophore secretion by MmpS4/MmpL4 and MmpS5/MmpL5 (22).

Is the outer membrane permeability barrier of M. tuberculosis affected by the absence of mycobactin?

It is conceivable that the increased susceptibility of the M. tuberculosis mutants lacking the siderophore exporters MmpS4-MmpL4 and MmpS5-MmpL5 to TB drugs such as bedaquiline is caused by an indirect effect due to the lack of membrane-bound mycobactin in the outer membrane and not due to decreased efflux as deduced from the similarity of the MmpL transporters to efflux pumps of the RND family. To distinguish between these possibilities, we measured the accumulation of ethidium bromide, a hydrophobic molecule which permeates through lipid membranes, in different M. tuberculosis strains lacking siderophores. Internalized ethidium bromide can easily be measured due to a strong fluorescence increase upon intercalation into chromosomal DNA. Therefore, this assay is not sensitive to surface adsorption, which confounds uptake measurements of most other hydrophobic compounds (31). If loss of mycobactin in the outer membrane, which is the major permeability determinant in mycobacteria (5, 32), was responsible for the increased drug susceptibility of the M. tuberculosis mutants with impaired siderophore secretion, the ethidium bromide uptake should be faster in these mutants than in the parent M. tuberculosis strain. However, the kinetics of ethidium bromide uptake by all M. tuberculosis strains with a siderophore secretion defect are similar to that of the parent strain (Fig. 7), indicating that the strongly reduced mycobactin levels do not compromise the outer membrane permeability barrier in the siderophore secretion-deficient M. tuberculosis strains. Interestingly, the higher initial fluorescence of the siderophore biosynthesis mutant M. tuberculosis ΔmbtD indicates that the complete absence of mycobactin increases the permeability of ethidium bromide. The fact that the plateau levels of internalized ethidium bromide in both the M. tuberculosis ΔmbtD mutant and the parent strain are higher than that of the siderophore secretion mutants also indicates that ethidium bromide is not a substrate of the MmpS4-MmpL4 and MmpS5-MmpL5 transporters, since loss of a nonredundant efflux pump increases substrate accumulation, while overexpression of an efflux pump reduces substrate accumulation as shown previously for Rv0194 (31). The opposite phenotype as observed here for the M. tuberculosis mutants with a defective siderophore secretion system indicates that the loss of the MmpS4-MmpL4 and MmpS5-MmpL5 systems might increase the level of another efflux pump such as Rv0194, which is known to transport ethidium bromide (31).

FIG 7.

Ethidium bromide uptake by M. tuberculosis strains with defects in siderophore biosynthesis or secretion. The indicated M. tuberculosis strains were grown in minimal medium with hemin and were incubated with 20 μM ethidium bromide at 37°C, and the fluorescence was measured in relative fluorescence units (rfu) every minute at an excitation wavelength of 530 nm and an emission wavelength of 590 nm using a microplate reader.

In an alternative approach, we examined whether the complete absence of mycobactin alters the drug susceptibility of M. tuberculosis. To this end, we determined the susceptibility of the M. tuberculosis mbtD deletion mutant to bedaquiline and clofazimine using the microplate Alamar Blue assay as described above. We observed a minor increase in the susceptibility of the M. tuberculosis ΔmbtD mutant to bedaquiline, while the susceptibility to clofazimine was not altered (Fig. S5). As expected, mycobactin was not toxic for the ΔmbtD mutant at all measured concentrations (Fig. S5C). These results indicate that the increased susceptibility of the siderophore secretion-deficient mutants to bedaquiline and clofazimine is mainly due to the efflux activity of the MmpL4 and MmpL5 systems. These results are also consistent with the ethidium bromide uptake experiments, which revealed no difference between wt Mtb and the siderophore secretion-deficient mutants (Fig. 7). We conclude that bedaquiline and clofazimine are indeed substrates of the MmpS4-MmpL4 and MmpS5-MmpL5 efflux pumps.

DISCUSSION

The MmpS4-MmpL4 and MmpS5-MmpL5 efflux systems perform similar and essential functions in M. tuberculosis

Drug efflux is an important mechanism in bacterial drug resistance. While the 13 MmpL proteins constitute the largest drug efflux pump family of M. tuberculosis, none of the single-gene deletion mutants lacking MmpL proteins of the largest drug efflux pump family of M. tuberculosis showed a significant increase in drug susceptibility (33, 34). In this study, we provide evidence that this lack of phenotypes in drug efflux is caused by a functional redundancy of the MmpL4 and MmpL5 transporters and their associated MmpS4 and MmpS5 proteins, which form complexes with their respective MmpL transporter proteins (19). Interestingly, while the functions of the mmpS4-mmpL4 and mmpS5-mmpL5 genes are identical in their genetic context, complementation experiments revealed that the MmpS4-MmpL4 system alone is not capable of restoring wt levels of siderophore secretion and drug resistance in a mutant lacking both operons in contrast to the MmpS5-MmpL5 system (Fig. 5). This result indicated that the MmpS4-MmpL4 system may require an additional protein for its function, whose production might have been affected by the deletion of the entire operon. The lack of a promoter driving the expression of the rv0449c gene downstream of mmpL4 in the operon deletion mutant versus the single mmpL4 deletion mutant might be causing such a phenotype (Fig. 1A). Consistent with the similar functions of MmpL4 and MmpL5 as shown in this study, it should be noted that we did not observe any growth defect of our mmpL4 deletion mutant in contrast to a report (33). These results, combined with the sevenfold higher transcription levels of mmpL5 compared to mmpL4 in wild-type M. tuberculosis (35), might explain the dominant role of the MmpS5-MmpL5 efflux system in naturally selected bedaquiline- and clofazimine-resistant mutants of M. tuberculosis (11, 12). A similar phenomenon is observed in the clinical setting when multidrug-resistant M. tuberculosis strains acquired clofazimine resistance following bedaquiline treatment of TB patients (36, 37). It is worth noting that the absence of equivalent mutations affecting the MmpS4-MmpL4 system in clinical M. tuberculosis strains is not caused by an enhanced activity and/or different substrate specificity compared to the MmpS5-MmpL5 system. Based on the fact that wild-type M. tuberculosis strains are susceptible to all substrates of the MmpS4-MmpL4 and MmpS5-MmpL5 systems, we hypothesize that the natural protein levels of these efflux pumps do not provide sufficient protection. This might be due to the fact that the mmpS5-mmpL5 operon is downregulated in medium with high iron (38) and/or caused by saturation of these efflux systems with their natural substrates, the siderophores, under low-iron conditions (20), when mmpS5-mmpL5 appears to be upregulated (29).

The MmpS4-MmpL4 and MmpS5-MmpL5 efflux systems have a specific substrate set

Testing of a large set of 16 chemically diverse antibiotics and TB drugs revealed that the substrates of the MmpS4-MmpL4 and MmpS5-MmpL5 efflux systems are limited to large, nonlinear, and hydrophobic compounds with aromatic rings. This includes rifampicins and econazole, in addition to bedaquiline and clofazimine and the siderophores (Fig. S2). Our observation that M. tuberculosis is more susceptible to rifampicin and rifabutin in the absence of the MmpS4-MmpL4 and MmpS5-MmpL5 efflux pumps is consistent with the slightly increased resistance of clinical M. tuberculosis strains to rifampicin upon overexpression of the mmpS5-mmpL5 operon (29). Structural and mutational studies will help elucidate the molecular determinants of the substrate specificity of the MmpS4-MmpL4 and MmpS5-MmpL5 efflux pumps.

In this regard, it is interesting that the susceptibility of M. tuberculosis to the new TB drugs linezolid and delamanid is not affected by the absence of these efflux systems. The activity of linezolid, a new drug with excellent activity against M. tuberculosis in mice (39), is strongly enhanced in the presence of drug efflux pump inhibitors such as reserpine (29), indicating that linezolid is a substrate of drug efflux pumps (40). However, to our knowledge, no efflux pump has been identified that is capable of secretion of linezolid. Both linezolid and delamanid are hydrophobic, linear extended molecules in contrast to the hydrophobic, branched structure of MmpL4/MmpL5 substrates (Fig. S2). These results indicate that other drug efflux systems such as MmpL proteins which export lipids across the inner membrane could, in a redundant manner similar to MmpL4 and MmpL5, secrete lipid-like drugs such as linezolid and delamanid. Candidates are MmpL3, MmpL7, MmpL8, MmpL10, and MmpL11, which export mycolic acids (41, 42), phthiocerol dimycocerosate (43), sulfolipid SL-1 (44), monomeromycolyl diacylglycerol and mycolate ester wax (45), and acylated trehaloses (46).

The MmpS4-MmpL4 and MmpS5-MmpL5 efflux systems reveal an intricate link between siderophore secretion and drug resistance in M. tuberculosis

All mutations in this study which affect siderophore secretion also impair drug efflux by M. tuberculosis. Most importantly, this includes the periplasmic protein Rv0455c, which has recently been described by us as essential for siderophore secretion and for virulence of M. tuberculosis in mice (22). While the functional role of Rv0455c is still unknown, it was proposed that it might interact with the MmpS4-MmpL4 and MmpS5/MmpL5 efflux systems (22). It should be noted that we did not directly show drug efflux by MmpS4/MmpL4 and/or MmpS5/MmpL5. Considering that many MmpL proteins export lipids (34), it is conceivable that the MmpL4 and MmpL5 efflux systems also transport outer membrane lipids. For example, membrane-anchored mycobactin (47) could contribute to the permeability barrier of the outer membrane. However, uptake kinetics of the hydrophobic dye ethidium bromide are very similar for all siderophore secretion-deficient M. tuberculosis mutants compared to wt M. tuberculosis (Fig. 7), arguing against the possibility that the increased susceptibility of the M. tuberculosis mutants lacking MmpS4/MmpL4, MmpS5/MmpL5 or Rv0455c might be caused by a compromised outer membrane permeability barrier to hydrophobic drugs and not by impaired efflux. This conclusion is also consistent with the equal susceptibilities of all siderophore secretion-deficient mutants to the hydrophobic drugs delamanid and linezolid (Table 1).

Our finding that lack of Rv0455c enhances the susceptibility of M. tuberculosis to the same extent to the same drugs as the absence of the MmpS4/MmpS5 and MmpL4/MmpL5 proteins (Fig. 6) not only underlines the intricate link between siderophore secretion and drug efflux in M. tuberculosis but also validates Rv0455c as a drug target, whose inactivation was shown to reduce the virulence of M. tuberculosis in mice and enhance the efficacy of important TB drugs such as bedaquiline and rifampicin analogs against M. tuberculosis.

Model of siderophore and drug efflux by the MmpS4-MmpL4 and MmpS5-MmpL5 transporters

Our results show that the periplasmic Rv0455c protein is as important for drug efflux by M. tuberculosis as the MmpS4/MmpL4 and MmpS5/MmpL5 efflux pumps. Their shared, essential role in siderophore secretion (22) and the identical drug susceptibility profiles of the corresponding deletion mutants indicate that these proteins function in the same secretion pathway, which secretes siderophores and a selected group of hydrophobic compounds with aromatic rings. Therefore, we propose a model in which MmpS4/MmpS5 and MmpL4/MmpL5 form an inner membrane complex that translocates siderophores and hydrophobic drugs such as bedaquiline across the inner membrane of M. tuberculosis (Fig. 8). Periplasmic Rv0455c is exported by the Sec system and participates in a hitherto unknown way to this process (22). It is unclear whether MmpL4 and MmpL5 proteins directly transport drugs and siderophores from cytoplasm into the periplasm or whether it sweeps substrates exported by another transporter from the periplasm of M. tuberculosis in a similar manner as the AcrB-AcrA efflux pump in Escherichia coli. It is also unclear how mycobactins and carboxymycobactins are transported through the periplasm and across the outer membrane. M. tuberculosis does not have proteins with sequence similarities to TolC, the outer membrane secretion channel for enterobactin, a major siderophore of E. coli (48 – 50). It is likely that this protein forms a transmembrane channel in a similar manner as TolC, which connects to inner membrane drug efflux pumps such as AcrB and is involved not only in siderophore secretion but also in drug efflux in E. coli and other gram-negative bacteria (48 – 50). The challenge of identifying the outer membrane proteins involved in drug efflux might be functional redundancy as indicated by the lack of any candidate gene in a comprehensive TnSeq study of iron utilization in M. tuberculosis (21).

FIG 8.

Model of the function of the siderophore secretion system of M. tuberculosis in drug efflux. The RND-type efflux pumps MmpS4-MmpL4 and MmpS5-MmpL5 form complexes located in the mycobacterial IM. The Rv0455c preprotein is translocated across the IM via SecA1 secretion system. The mature, water-soluble Rv0455c protein is located in the periplasm and is essential for secretion of (carboxy)-mycobactin, possibly by interaction with the MmpL4/5-MmpS4/5 systems. The influx of the hydrophobic drugs such as bedaquiline, clofazimine, or econazole is not clear. The intracellular drugs, along with the de novo synthesized siderophores are exported by a process dependent on the MmpL4/S4 and MmpL5/S5 efflux systems (19 – 21). Transport of siderophores or drugs across the OM requires at least one hitherto unknown protein.

Conclusions

This study reveals the dual functions of two redundant but essential efflux systems in siderophore secretion and drug efflux by M. tuberculosis. Importantly, we showed that a single-gene inactivation is sufficient to block the efflux system constituted by MmpS4-MmpL4, MmpS5-MmpL5, and Rv0455c as demonstrated by the lack of siderophore secretion by the rv0455c deletion mutant (22). Thus, it might be feasible to inactivate this essential secretion pathway by a single drug and simultaneously to impair siderophore secretion and drug efflux by M. tuberculosis. This pathway is a promising target since inactivation of siderophore secretion in M. tuberculosis causes a 20,000-fold reduced virulence of M. tuberculosis in mice (19). A drug targeting the siderophore secretion system would directly kill M. tuberculosis by self-poisoning through the intrinsic synthesis and secretion of siderophores (20) and might sensitize M. tuberculosis to bedaquiline and rifampicins. Such a synergizing drug would be particularly beneficial in multidrug combination chemotherapies, which are standard in TB treatment regimens (51).

MATERIALS AND METHODS

Chemicals, enzymes, and DNA

Hygromycin B was purchased from Calbiochem. All other chemicals were purchased from Fisher or Sigma at the highest purity available. Enzymes for DNA restriction and modification were purchased from New England Biolabs. Isolation and modification of DNA were performed as described (52). Oligonucleotides were obtained from Integrated DNA Technologies (Table S1).

Bacterial strains and growth conditions

All bacterial strains used in this study are listed in Table S1. Escherichia coli DH5α was used for all cloning experiments and was routinely grown in LB medium at 37°C. M. tuberculosis mc²6230 strains were grown at 37°C in Middlebrook 7H9 liquid medium (Difco Laboratories) supplemented with 0.2% glycerol, 10% Middlebrook oleic acid, albumin, dextrose, catalase growth supplement (OADC), 0.2% casamino acids, 24-µg/mL pantothenate, and 0.02% tyloxapol (referred to herein as supplemented 7H9 medium). Hartmans-de Bont (HdB) minimal medium (53) is composed of 500-µM MgCl₂, 7-µM CaCl₂, 1-µM NaMoO₄, 2-µM CoCl₂, 6-µM MnCl₂, 7-µM ZnSO₄, 1-µM CuSO₄, 15-mM (NH₄)₂SO₄, 12-mM KH₂PO₄ (pH 6.8), and 1% (vol/vol) glycerol and was supplemented with 10% OADC , 0.2% casamino acids, 24-µg/mL pantothenate, and 0.02% tyloxapol. The iron concentration of supplemented HdB without OADC was <0.1 μM as determined by inductively coupled plasma mass spectrometry. Antibiotics were used at the following concentrations: hygromycin (200 µg/mL for E. coli and 50 µg/mL for mycobacteria) and kanamycin (50 µg/mL for E. coli and 30 µg/mL for mycobacteria).

Plasmid construction

E. coli strain DH5α was routinely used for plasmid construction and propagation. Plasmids used in this work are described in Table S2. Oligonucleotides are listed in Table S3. For gene deletions in M. tuberculosis, approximately 1,000 bp upstream and downstream of the gene of interest were amplified from chromosomal DNA of M. tuberculosis H37Rv by PCR and cloned into the deletion vector pML2424 flanking the loxP-psmyc-gfp_m ²⁺ -hyg-loxP cassette using the restriction sites SpeI/SwaI and PacI/NsiI, respectively. The deletion plasmids pML3001 and pML3003 were generated for deletion of the mmpL4 and mmpL5 genes, respectively, using the primer pairs mmpL4_Up_F/mmpL4_Up_R, mmpL4_Down_F/mmpL4_Down_R, and mmpL5_Up_F/mmpL5-Up-R, mmpL5_Down_F/mmpL5-Down-R as listed in Table S3.

Construction of gene deletion mutants of M. tuberculosis

Competent cells of M. tuberculosis mc²6230 were transformed with the deletion plasmids. Cultures of M. tuberculosis with the respective deletion plasmids were grown at 37°C on 7H9/OADC/Hyg medium with casamino acids and pantothenate and plated to select for double-crossover (DCO) mutants on 7H10/OADC/Hyg medium plus 2% sucrose at 40°C. DCO candidates were screened for the presence of of the red fluorescent protein tdTomato and green fluorescent protein (GFP). GFP-positive and tdTomato-negative DCO candidates were grown in liquid culture for approximately 7 days to prepare chromosomal DNA for Southern blots. Alternatively, colonies of positive clones were examined by colony PCR. The Cre recombinase expression vector pML2714 was used to excise the loxP-flanked gfp ²⁺ _m hyg cassette from the chromosomes of the selected DCO strains. Strains were cured of pML2714 by growth on 7H10/OADC/kan plates to identify GFP- and tdTomato-negative clones. These unmarked strains were examined again by colony PCR and/or Southern blots.

The unmarked mmpL4 and mmpL5 (ΔmmpL4::loxP, abbreviated ΔmmpL4, and ΔmmpL5::loxP, abbreviated ΔmmpL5) deletion mutants in M. tuberculosis mc²6230 were named ML2300 and ML2301 (Table 1), respectively. The double deletion mutant M. tuberculosis ML2302 (ΔmmpL4/ΔmmpL5) was obtained by electroporation of competent cells of M. tuberculosis ML2301 (ΔmmpL5) using the plasmid pML3001 as described above. This double mutant was complemented with the replicative vector pML3601 (mmpS4/mmpL4). The operon deletion mutant ML2311 (ΔmmpS4-mmpL4 and ΔmmpS5-mmpL5) was obtained by electroporation of the vector pML3425 (ΔmmpS5/L5) in M. tuberculosis ML1422 (ΔmmpS4L4::loxP). The PCR mmpS5 upstream region amplified 1026 bp and mmpL5 downstream region 991 bp.

Analysis of M. tuberculosis mutants by Southern blot and PCR of chromosomal DNA

Chromosomal DNA was extracted from the M. tuberculosis strains as described (54). Five-microgram genomic DNA was digested with the restriction enzymes PstI (M. tuberculosis ΔmmpL4) and SmaI (M. tuberculosis ΔmmpL5, ΔmmpL4/ΔmmpL5) and then separated on a 1% agarose gel. The gel was subsequently transferred to a positively charged nylon membrane (Hybond-N+) using standard protocols (52). The DNA was cross-linked to the membrane using a UV cross-linker (240,000 µJ) and prehybridized for 30 min at 42°C in a Dig-Easy hybridization solution (Roche). Probes were generated by PCR from M. tuberculosis genomic DNA using the primer pairs mmpL4_Down_R/mmpL4_Down_F and mmpL5_Down_F/mmpL5_Down_R (Table S3). The PCR fragments were labeled using the PCR DIG labeling mix (Roche). Hybridization of 250 ng of digoxigenin-labeled probes with the genomic DNA on the membrane was carried out at 55°C overnight. The membranes were washed and the hybridized digoxigenin-labeled probe was detected with a horseradish peroxidase-conjugated antidigoxigenin antibody following the protocol of the manufacturer (Roche). A gel imaging system and the software LabWorks (UVP) were used to visualize the luminescence of blots. The software Gimp version 2.0 was used to adjust the contrast of images if necessary. No parts of the images were altered separately.

Alternatively, chromosomal DNA of M. tuberculosis mc²6230 and of the M. tuberculosis ΔmmpS4-mmpL4/ΔmmpS5-mmpL5 mutant (Table S1) was amplified using the primer pairs mutS4/mutL4 mmpS4L4-Seq-F/mmpL4-down; mutS5/mutL5 mmpS5L5-seq-F/Gfp_qPCR_rev3v (Table S3). The PCR fragments were analyzed after electrophoresis of an agarose gel.

PDIM analysis

Extraction and analysis of apolar lipids of M. tuberculosis were carried out as described (26, 55). M. tuberculosis strains were grown in supplemented 7H9 medium at 37°C until midlog phase (~OD 0.8–1.0). Then, 0.1 µCi [1⁻¹⁴C] propionic acid (3.7 × 10³ Bq/mL) was added, and the cultures were incubated for two more days. Following centrifugation at 400 × g for 8 min, the cell pellets were resuspended in 700-µL methanol/0.3% NaCl (10:1, vol/vol) to which 700-µL petroleum ether was added. After vigorous mixing, the samples were centrifuged at 3,000× g for 5 min, and the upper petroleum ether phase containing the apolar lipid fraction was removed and transferred to a 2-mL screw cap tube. Another 700-µL petroleum ether was added to the cell pellets in the first tube, mixed, and centrifuged. Both extracts were pooled and were left under the hood until half of the petroleum ether was evaporated. Then, the lipid extracts were mixed with 700-µL chloroform, mixed, and left under the hood until the organic solvents evaporated completely. The lipids were then resuspended in 200-µL petroleum ether/diethyl ether (9:1, vol/vol) and 10-µL aliquots of each extract were spotted onto a thin-layer chromatography plate (250-µM silica gel 60 plates, EM Science). The TLC plate was run in petroleum ether/diethyl ether (9:1, vol/vol) for detection of PDIM. The TLC plates were visualized using a PhosphorImager (Storm 840, GE Healthcare) and imaged using the Typhoon Scanner software.

Siderophore production, purification, and analysis

Siderophores of M. tuberculosis were radioactively labeled as previously described with modifications (22). Precultures of the M. tuberculosis strains were grown in supplemented 7H9 medium with 20-µM FeCl₃ to an OD₆₀₀ of 1–2. To deplete intracellular iron stores, the strains were incubated in Hartmans-de Bont minimal medium (53) without added iron supplemented with 0.2% casamino acids, 24-µg/mL pantothenate, and 0.02% tyloxapol. The iron concentration of the HdB minimal medium was 13 µg/L, as determined by inductively coupled plasma mass spectrometry. The cells were washed in phosphate-buffered saline (PBS) and transferred into the supplemented HdB medium containing 10-µM hemin and 21-µM [7-¹⁴C]-salicylic acid (1 mCi/mL) for 11 days with shaking at 37°C. The cultures were centrifuged at 4,000 × g and supernatants were collected. The supernatants were incubated with 0.6-mM ferric chloride for 1 hour at room temperature to saturate siderophores with iron. The supernatants were then extracted twice with 5-mL CHCl₃. The organic fractions were pooled, evaporated in a vacuum (Vacufuge, Eppendorf), and resuspended in 200 µL CHCl₃. Extracts were then analyzed by thin-layer chromatography on a silica gel 60 plates (Sigma) using ethanol/cyclohexane/water/ethyl acetate/acetic acid (5.0:25.0:2.5:35.0:5.0) as a solvent. Plates were allowed to dry and then exposed to a storage phosphor screen for approximately 60 h. The screens were analyzed using a Storm Phosphor Imager (Molecular Dynamics). ⁵⁵Fe-labeled MBT and cMBT and ¹⁴C-salicylic acid were used as controls. The retention factors (Rf) for MBT (0.42) and cMBT (0.16) were similar to those previously reported (56).

Susceptibility of M. tuberculosis strains to mycobactin, TB drugs, and antibiotics

To determine the minimal inhibitory concentrations of antibiotics and TB drugs, the microplate Alamar Blue assay (MABA) was used as described previously (31). Briefly, M. tuberculosis mc²6230 and mutants were grown at 37°C in supplemented 7H9 medium until OD₆₀₀ of 1. The cultures were filtered using a 5-µM filter to remove clumps and were centrifuged at 3,000 × g for 5 min; the pellet was suspended in 1 mL of Hartmans-de Bont minimal medium (53) supplemented with 0.2% casamino acids, 24-µg/mL pantothenate, 20-µM hemin, and 0.02% tyloxapol. Then, the supernatants were removed and resuspended in 0.5 mL of HdB minimal medium with 0.02% tyloxapol. The cultures were then diluted in the same medium to an OD₆₀₀ of 0.05. Aliquots of 100 µL were added to the wells of a 96-well microplate containing 100 µL of each drug or HdB medium as a control. Dilutions of compounds were prepared in HdB medium and were serially twofold diluted from the following starting concentrations: mycobactin, 4 µg/mL; rifampicin, 4 µg/mL; rifabutin, 1 µg/mL; bedaquiline, 4 µg/mL; clofazimine, 4 µg/mL; kanamycin, 80 µg/mL; clarithromycin, 128 µg/mL; gentamycin, 80 µg/mL; isoniazid, 1 µg/mL; econazole, 64 µg/mL; and D-cycloserine, 50 µg/mL. After incubation of the microplates for 5 days at 37°C, resazurin was added to each well at a final concentration of ~90 µM. After an additional incubation of 6 h, the fluorescence of the metabolically converted resazurin dye was measured at 590 nm after excitation at 530 nm using a Synergy H1 microplate reader (BioTek). All experiments were done in triplicate. The minimal inhibitory concentrations (Table 1) were defined as the lowest concentrations of antibiotic or drug which reduced the viability of M. tuberculosis by at least 90%.

Analyis of cell lysates of M. tuberculosis by Western blots

The M. tuberculosis strains were grown in supplemented 7H9 medium to an OD₆₀₀ of 2. Cells were harvested by centrifugation, washed twice with PBS, and resuspended in 0.5 mL of PBS. Zirconia silica beads with a diamter of 0.1 mm were added to this sample and beaten for 30 s using a FastPrep machine. Then, the samples were kept on ice, and 50 µL of 10% SDS was added. This process was done three times. The lysed samples were centrifuged at 6,000 × g for 10 min at 4°C. The supernatant was collected; the protein concentration was estimated using a Bradford assay and then loaded on the 10% SDS-polyacrylamide gel. The proteins were transferred to a polyvinylidene difluoride (PVDF) membrane and detected in Western blots using 200-fold dilutions of the primary antisera specific for MmpS4 and MmpS5 and a secondary goat antirabbit antibody coupled with horse raddish peroxidase (#A6154, Sigma). The blots were developed using enhanced chemiluminescence Western blotting substrate (Pierce), and luminescence was visualized using a LabWorks (UVP, Inc.) imaging system.

Ethidium bromide uptake kinetics

The ethidium bromide uptake kinetics were measured as described (31). All strains were first grown to log phase in 10 mL of supplemented 7H9 medium. Cells were filtered through a 5 µM filter to obtain a single-cell suspension. Cells were then harvested at 3,000 × g for 5 min and resuspended to a final OD₆₀₀ of 0.5 in HdB medium with 0.2% casamino acids, 24-µg/mL pantothenate, 20-µM hemin, and 0.02% tyloxapol. For all strains, 200 µL of cell suspensions was added in triplicate in a 96-well plate, and ethidium bromide was added to a final concentration of 20 µM. The fluorescence was measured by excitation at 530 nm and emission at 590 nm at 1-min intervals using a Synergy H1 microplate reader (BioTek).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aac.01629-22.

Supplement. aac.01629-22-s0001.docx.

Figures S1 through S5 and Tables S1 through S4.

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