M. smegmatis mc²155 was used both as the wild-type and parental strain for siderophore mutants. ∆mbtD and ∆fxbA strains were generated using the p0004S allelic-exchange system, as described (Larsen et al., 2007). Experiments with virulent M. tuberculosis were performed using strain Erdman TMCC 107 gifted by Dr. Anne Lenaerts (Colorado State University).

M. smegmatis and M. tuberculosis were grown in Dubos media consisting of 2 g asparagine, 0.5 g casitone, 1 g KH₂PO₄, 2.5 g Na₂HPO₄, 10 mg MgSO₄·6H₂O, 0.5 mg CaCl₂, 0.1 mg ZnSO₄, 0.1 mg CuSO₄, 7.5 g glucose, 5 g bovine serum albumin fraction 5, 1.7 g NaCl and 2.5 mL 20% Tween-80 in 1 liter of diH₂O unless otherwise stated; media was adjusted to pH 6.64 with HCl. Standard Dubos medium was supplemented with 50 mg/L ferric ammonium citrate equating to a total iron content of approximately 190 μM. In the absence of this supplement, residual iron levels were approximately 2 μM, as determined using a bathophenanthraline disulfonate iron detection assay (Nilsson et al., 2002). Avirulent M. tuberculosis ML1600 (∆mbtD, ∆RD1 ∆panCD), gifted by Dr. Michael Neiderweis (University of Alabama, Birmingham), was grown in Middlebrook 7H9 prepared from powdered base (Sigma Aldrich) with 0.2% casitone, 24 μg/mL pantothenic acid, 20 μM porcine hemin and 0.05% Tween-80, as previously described (Wells et al., 2013). All bacterial cultures were grown at 37°C with stirring in 125 mL Erlenmeyer flasks prior to experiments. All chemicals were obtained from Sigma Aldrich unless otherwise specified.

M. smegmatis was diluted from freezer stocks to OD₆₀₀ 0.0006 and grown for 19 h in Dubos medium. Cultures were then diluted to OD₆₀₀ 0.03, dispensed in 1.5 mL aliquots into 14 mL snap cap tubes which were loosely capped and incubated for 90 min at 37°C with shaking at a 30° angle. Drugs were added to the tubes at indicated concentrations and the tubes were returned to the incubator. Tubes were assayed for CFU at indicated timepoints via plating in triplicate on Dubos media plates containing 4 g/L activated charcoal. M. smegmatis experiments were conducted over 6 days to observe the full extent of CFZ lethal activity.

M. tuberculosis Erdman was settled at 37°C for 4 days and avirulent ∆mbtD M. tuberculosis strain ML1600 for 6 days at 37°C in 10 mL of media before being brought to 50 mL. Strains were incubated with stirring at 37°C for 6 days with periodic monitoring and dilution to keep cultures between OD₆₀₀ 0.05 and 0.5. For killing assays, cultures were diluted to OD₆₀₀ 0.05 and dispensed in 2 mL aliquots to 24 mL borosilicate tubes containing sterile magnetic stir bars; ML1600 cultures were diluted to OD₆₀₀ 0.1. Erdman cultures were then incubated with stirring at 37°C with 5% CO₂ for 150 min and without CO₂ for ML1600. Drug or vehicle was introduced to tubes as indicated and tubes were returned to incubator. Cultures were assayed for CFU at indicated timepoints via plating.

Filter-sterilized aliquots of the antibiotics CFZ and kanamycin were stored at -80°C and norfloxacin at 4°C before use. Chlorpromazine was prepared fresh prior to each use. Cell-associated mycobactin J (MBTJ) isolated from M. avium subsp. Paratuberculosis (Schwartz and De Voss, 2001) and obtained from Allied Monitor was dried down in 200 μg aliquots and resuspended in ethanol prior to use. Salicyl-AMS was synthesized as previously described (Somu et al., 2006).

M. smegmatis and M. tuberculosis were cultured with or without CFZ in the indicated medium under the aforementioned growth conditions for 24 h with 0.1 μCi/mL 7–¹⁴C salicylic acid (American Radiolabeled Chemicals). Cell pellets were extracted for cell-associated MBT, as previously described (Wells et al., 2013). Cell pellets were boiled in the presence of 4% SDS for 2 h then quantified using the BCA protein assay. Extracts were normalized to 100 mg total protein for M. smegmatis, 50 mg for M. tuberculosis. Samples were separated via thin-layer chromatography on 0.25 mm glass-backed silica plates using 2:3:3 petroleum ether:n-butanol:ethyl acetate solvent system (De Voss et al., 2000) which were then exposed to a Kodak Imaging screen (K-screen). Radiation was visualized and quantified using a Typhoon FLA 7000 molecular imager.

CFZ at a concentration of 100 μM was added to an aqueous buffer solution consisting of 1 g KH₂PO₄ and 2.5 g Na₂HPO₄ in 1 liter of water, pH adjusted to 6.64. Increasing concentrations of ferric ammonium citrate were introduced, with or without the addition of 50 mM sodium ascorbate in 14 mL snap cap tubes. Samples were incubated with shaking for 30 min after which samples were centrifuged, the buffer was decanted, and the tube washed with 1 mL DMSO to remove precipitate. The wash was read at OD₄₅₄ and compared to a standard curve consisting of known quantities of CFZ in DMSO.

M. tuberculosis H37Rv was grown in the Rapid Anaerobic Dormancy (RAD) model as previously described (Leistikow et al., 2010) using DTA broth supplemented with either 50 or 250 mg/L of ferric ammonium citrate, equating to 190 μM and 950 μM total iron, respectively. Cultures were grown in the RAD model for 8 days to achieve complete O₂ depletion. Cultures were then transferred to an anaerobic environmental chamber (Shell lab, Cornelius, OR, USA), unsealed and challenged with either 106 μM CFZ or a DMSO vehicle control. Cultures were resealed, removed from the anaerobic chamber and incubated at 37°C. Cultures were assayed for the accumulation of Fe²⁺ via a colorimetric assay using the bidentate chelating agent ferene (3-(2-pyridyl)-5,6-bis(2-(5-furyl sulfonic acid))-1,2,4-triazine) which forms a stable complex with Fe²⁺ with a peak absorbance at 593 nm (Eskelinen et al., 1983). Fe²⁺ accumulation was assayed at 5, 24, 48, 120, and 192 h following the addition of CFZ. At indicated timepoints, cultures were transferred to the anaerobic chamber. Cultures were unsealed, 1.0 mL aliquots were collected and centrifuged for 5 min at 3,000 rpm to pellet out cell material. Supernatant was removed and supplemented with ferene at a final concentration of 50 μM. Samples were incubated for 5 min at room temperature to allow for complete color development. Samples were removed from the anaerobic chamber and assayed for Fe²⁺ by measuring the absorbance at 593 nm. Total sample Fe²⁺ was inferred by plotting against a standard curve prepared using aqueous solutions of FeSO₄.

RNA was isolated from M. tuberculosis as previously described (Rodriguez et al., 2002) from cultures challenged with either CFZ or DMSO. RNA was sequenced using an Illumina NovaSeq6000 platform with a minimum of 60 million reads per sample. Sequencing reads were mapped to M. tuberculosis Erdman (NC_AP012340) using CLC genomics workbench. To normalize raw gene count by transcript length and sequence depth, the gene expression values were calculated as reads per kilobase per million. Differential transcript expression between the control and experimental samples was calculated using generalized linear model to correct for library size variations between samples.

M. smegmatis cultures were pre-grown for 19 h in 2 μM iron Dubos medium supplemented with 1 μCi/mL ⁵⁵FeCl₃ (American Radiolabeled Chemicals). Bacilli were washed before outgrowth, resuspended in either 2 μM or 190 μM iron Dubos medium and treated with CFZ or DMSO as indicated by the assay. Samples were harvested at indicated timepoints with 1 mL per sample spun down for 3 min twice. 500 μL of supernatant was analyzed using a Beckman LS 6500 liquid scintillation counter on the ³H channel for 8 min.

Expression profiling data from this study are available at the ArrayExpress at EMBL-EBI¹ under accession number E-MTAB-9350.

To determine how M. tuberculosis adapts to CFZ exposure we performed RNAseq transcriptional profiling. Cultures were subjected to either 20 μM CFZ or DMSO vehicle control for 24 h before being harvested for RNA isolation and RNAseq analysis. Exposure of M. tuberculosis to 20 μM CFZ resulted in several days of growth before the onset of killing to the limit of detection by 6 days (Figure 1B). Alignment of reads to a reference genome (Miyoshi-Akiyama et al., 2012) demonstrated a significant (p < 0.05) increase of 2-fold or more in 353 genes (Supplementary Table S1) and a 2-fold or more decrease in 178 genes (Supplementary Table S2). A prominent change observed during CFZ treatment was the substantially increased expression of genes in the IdeR regulon involved in iron homeostasis (Table 1). IdeR differentially regulates iron-associated genes based on the binding of intracellular Fe²⁺ and ensures that iron storage proteins are expressed when iron is replete and the systems necessary for de novo siderophore production and iron acquisition are expressed when iron is scarce (Gold et al., 2001; Rodriguez et al., 2002).

The two MBT biosynthetic operons comprising mbtA-mbtK and mbtL-mbtN were induced from 11- to 59-fold during CFZ exposure (Table 1), as were irtA and irtB associated with internalizing MBT-bound iron (Rodriguez and Smith, 2006). Components of the MmpS4/L4 and ESX-3 MBT transport systems (Wells et al., 2013; Siegrist et al., 2009), which fall under IdeR control, were also upregulated, albeit by a lesser degree than observed for the mbt and irt operons, with induction ranging from approximately 2- to 8-fold over DMSO vehicle control. The iron storage bacterioferritin genes, bfrA and bfrB, are induced by IdeR under high iron conditions and were downregulated by 3.5- and 4.8-fold by CFZ, respectively.

The impact of CFZ on the IdeR regulon and other iron-regulated genes, as defined by Rodriguez et al. (2002), is represented in a volcano plot with genes grouped by both their dependence on IdeR and whether iron increases or decreases expression (Figure 2A). Expression analysis revealed a pattern of iron-associated gene expression consistent with a low iron response for both IdeR-dependent and independent genes. The production of MBT during conditions used for expression profiling was demonstrated using ¹⁴C-salicylate MBT labeling (Figure 2B). The DMSO vehicle control culture produced no observable MBT as would be expected for iron replete conditions, but CFZ exposure resulted in a visible accumulation of cell-associated MBT (De Voss et al., 2000), validating the CFZ-associated increase in MBT-associated gene expression.

We next investigated the role of environmental iron availability on CFZ killing. M. smegmatis and M. tuberculosis were grown with either the standard 50 mg/L ferric ammonium citrate supplement (190 μM total iron) or 0 ferric ammonium citrate (2 μM total iron) referred to as high and low iron, respectively. High levels of MBT were observed from cell extracts of M. smegmatis grown under low iron but absent from cultures grown at high iron, confirming iron limiting and replete conditions (Supplementary Figure S1). M. smegmatis cultures were treated with 50 μM CFZ and M. tuberculosis cultures with 20 μM CFZ. Results indicated a stark impact of low iron on CFZ killing in M. smegmatis (Figure 1A). High iron cultures exhibited a one-day period of growth of 1.7-log₁₀ that transitioned to a rapid killing phase after 3 days and terminated at the limit of detection (250 CFU mL⁻¹) after 6 days, constituting a 6.0-log₁₀ reduction in CFU from the 2-day high. Meanwhile low iron cultures experienced a relatively slow decline in CFU, dropping only 2.7-log₁₀ over the same period. The differential killing by CFZ of M. tuberculosis, at high and low iron was less pronounced than in M. smegmatis (Figure 1B), but a significant difference was observed.

To determine why iron limitation reduced CFZ killing, we assessed the role of individual siderophores. M. smegmatis mutants were generated to specifically deprive the bacteria of each class of siderophore: deletion of the polyketide synthase-encoding mbtD prevents production of both MBT and cMBT while deletion of the putative formyltransferase-encoding fxbA prevents the production of exochelin (Zhu et al., 1998; Fiss et al., 1994; Quadri et al., 1998; McMahon et al., 2012). Deletion of mbtA or mbtB would have been sufficient to block MBT and cMBT production (Lun et al., 2013; De Voss et al., 2000); however, ∆mbtD was chosen in order to phenocopy the MBT mutant in avirulent M. tuberculosis ML1600, used in this study (Jones et al., 2014). Mycobactin and exochelin minus strains as well as wild-type M. smegmatis were then challenged with 50 μM CFZ at low iron (Figure 3A). Inhibition of MBT production led to increased killing at day 2 and no detectable viable M. smegmatis after 5 days exposure. Inhibition of exochelin production did not increase killing but resulted in a more static effect of CFZ relative to that observed in the wild-type. The increased killing of ∆mbtD cultures by CFZ was complemented by the addition of 5 μM exogenous MBTJ, serving to confirm the impact of MBT on CFZ activity. The impact of MBT on M. smegmatis CFZ killing was further interrogated with the MbtA inhibitor Salicyl-AMS (Lun et al., 2013). Exposure of wild-type M. smegmatis to both CFZ and 16 μM Salicyl-AMS increased killing that was comparable to the impact of the mbtD deletion (Figures 3A,C). Addition of 5 μM MBTJ to the ∆mbtD strain prevented CFZ killing as it did when added to the Salicyl-AMS treated culture. Wild-type ∆mbtD and ∆fxbA M. smegmatis strains displayed no difference in killing by CFZ under high iron conditions (Figure 3B). Use of ¹⁴C salicylate to measure MBT production indicated that MBT was present in conditions with low CFZ activity and absent from conditions that exhibited high activity (Figure 3E). Interestingly, wild-type M. smegmatis produced MBT during CFZ exposure but less than low iron cultures alone (Supplementary Figure S1). Therefore, CFZ did not strongly induce M. smegmatis MBT synthesis as it did in M. tuberculosis (Figure 2B), likely due to the ability to use exochelin in M. smegmatis.

With regards to M. tuberculosis, CFZ challenge of ML1600 ∆mbtD demonstrated a statistically significant difference for the first 6 days between cultures treated with CFZ or with CFZ plus 5 μM MBTJ. However, the survival advantage observed in MBTJ supplemented cultures was not significant by day 8 and all culture viability fell below the limit of detection by day 12 (Figure 3D).

We next investigated the response of the M. smegmatis siderophore mutants to other antimicrobials to determine if the effect of MBT on CFZ activity was specific to CFZ. Wild-type, ∆mbtD, and ∆fxbA cultures were grown with low iron and monitored for CFU during challenge with 25 μM norfloxacin, 300 nM kanamycin, or 200 μM chlorpromazine (Figures 4A–D). Statistically significant differences were observed at 3 and 12 h in control cultures without antibiotic but 24-h growth indicated that strains were within one doubling of each another. Killing by norfloxacin, kanamycin, or chlorpromazine was remarkably similar between strains. Slight statistical significance was observed at 3 h for all drugs and additionally at 24 h for chlorpromazine, but these differences did not alter the overall trends observed for each agent.

Wild-type M. smegmatis cultures grown in high iron and challenged with CFZ did not demonstrate substantially reduced killing when 5 μM MBTJ was added concurrently with CFZ or 2 days post CFZ addition (Figure 5A). Only the day 4 timepoint showed a significant difference between the three conditions; however, all three cultures reached the limit of detection by day 6, indicating that the difference observed was minimal. ∆mbtD cultures grown in low iron displayed rescue by MBTJ only when added concurrently with drug; MBTJ addition after 2 days of CFZ exposure even increased killing over CFZ alone by approximately two-fold at days 3 and 4 (Figure 5B). Therefore, MBTJ addition did not rescue CFZ killing in high iron conditions, nor did it complement the ∆mbtD cultures when given 2 days after CFZ exposure began. As such, the presence of MBT alone was not sufficient to prevent CFZ killing.

We next sought to investigate the broader impact of CFZ on intracellular iron levels given the importance of MBT in CFZ killing. Intracellular iron efflux was detected and quantified by preloading M. smegmatis with ⁵⁵FeCl₃ and challenging with CFZ. Substantial iron loss in the presence of CFZ was observed within 6 h in cultures grown in low iron medium that would be expressing MBT (Figure 6A) while high iron cultures lost iron at a far slower rate (Figure 6B). Low iron loss occurred even though high iron conditions resulted in far greater CFZ killing than low iron conditions. Plating controls confirmed CFZ did not result in death during the timepoints tested, making the presence of supernatant ⁵⁵Fe from lysis unlikely (Supplementary Figure S2).

It has been demonstrated via fluorescence emission spectroscopy that CFZ is capable of interacting with a number of divalent transition metal cations including biologically relevant elements such as Cu²⁺, Zn²⁺, and Mn²⁺ (Mohd et al., 2011). In aerobic conditions, extracellular iron exists primarily in the oxidized Fe³⁺ form. We investigated the impact of increasing Fe³⁺ on CFZ solubility in buffer both alone and with addition of 50 mM sodium ascorbate as a reducing agent (Figures 7A,B). The reducing agent alone did not significantly improve CFZ solubility. Buffer alone and buffer plus sodium ascorbate each enabled less than 10% of CFZ to be maintained in solution after 30 min. However, increasing amounts of iron in the presence of sodium ascorbate led to improved solubility, while Fe³⁺ alone did not improve solubility over buffer alone until at least 100 μM was added. The combination of iron and a reducing environment shifted CFZ from less than 10 to 82% soluble in aqueous solution as a function of iron concentration indicating an interaction of Fe²⁺ with CFZ or reduced CFZ with Fe³⁺.

Given the apparent interaction between Fe²⁺ and CFZ, we next measured the impact of CFZ on the oxidation state of iron. Anaerobic M. tuberculosis culture conditions were used to prevent the rapid reoxidation of Fe²⁺ in the presence of oxygen. Reduced iron was assayed in the presence of 190 μM iron and a 5-fold excess (950 μM) of iron supplement using the Fe²⁺-specific chelator, ferene (Eskelinen et al., 1983). CFZ exposure led to a considerable increase in the rate of Fe²⁺ accumulation (Figure 8A). Accumulation of Fe²⁺ at a faster rate and to higher levels was observed at 950 μM iron compared to 190 μM iron; although, both CFZ-exposed conditions generated significantly more Fe²⁺ than controls.

We also investigated the role of iron availability on anaerobic CFZ activity. M. tuberculosis cultures grown in either 2 μM, 190 μM or 950 μM iron (5-times base iron supplement) for 8 days in an anaerobic M. tuberculosis dormancy model (Leistikow et al., 2010) were treated with 106 μM CFZ and plated for survival (Figure 8B). In a reversal of the role of iron concentration in aerobic conditions, higher levels of media iron displayed lower CFZ killing with an 8 day drop in CFU of 1.7-log₁₀ at 950 μM iron, 2.1-log₁₀ at 190 μM iron, and 2.7-log₁₀ at 2 μM iron. DMSO control cultures were not reduced in viability over the 8 days of anaerobiosis and were not differentially affected by iron availability. These findings demonstrated that iron reduction correlated with decreased CFZ killing in anaerobic conditions and further indicate an interaction between CFZ activity and iron homeostasis.