Mycofactocin Is Associated with Ethanol Metabolism in Mycobacteria

Tuberculosis is caused by Mycobacterium tuberculosis, and the increasing emergence of multidrug-resistant strains renders current treatment options ineffective. Although new antimycobacterial drugs are urgently required, their successful development often relies on complete understanding of the metabolic pathways—e.g., cholesterol assimilation—that are critical for persistence and for pathogenesis of M. tuberculosis. In this regard, mycofactocin (MFT) function appears to be important because its biosynthesis genes are predicted to be essential for M. tuberculosis in vitro growth in cholesterol. In determining the metabolic basis of this genetic requirement, our results unexpectedly revealed the essential function of MFT in ethanol metabolism. The metabolic dysfunction thereof was found to affect the mycobacterial growth in cholesterol which is solubilized by ethanol. This knowledge is fundamental in recognizing the bona fide function of MFT, which likely resembles the pyrroloquinoline quinone-dependent ethanol oxidation in acetic acid bacteria exploited for industrial production of vinegar.

RiPPs share many structural and enzymatic similarities with nonribosomal peptides (NRPs), except for the polyketide-derived moieties that are ubiquitous in NRPs (10). Exochelin and mycobactin are notable mycobacterial NRPs which are vital for iron acquisition and pathogenesis of M. tuberculosis (11,12). Unlike NRPs, the significance of RiPPs such as MFT in mycobacteria is still elusive, although MFT is predicted to function as a cofactor, similar to PQQ, for its nicotinoprotein redox partners (3,13). PQQ is a redox cofactor that supports the functions of several quinoproteins, including ethanol (EtOH) and aldehyde dehydrogenases catalyzing the oxidation of ethanol to acetic acid with acetaldehyde as a reaction intermediate (14)(15)(16)(17)(18). Besides the possibility of acting as an analog of PQQ and catalyzing ethanol oxidation in mycobacteria, MFT could have other cellular functions: (i) products encoded by mft genes have been suggested to be part of the electron transport chain, which is vital for M. tuberculosis survival under both replicating and nonreplicating conditions (19); (ii) independent saturated transposon mutagenesis studies (20,21) have predicted the essentiality of most of the mft genes for in vitro growth in cholesterol-solubilized in ethanol-a carbon source required for host adaptation of M. tuberculosis during infection (22)(23)(24). While data from the studies mentioned above strongly imply the role of MFT in M. tuberculosis pathogenesis, the precise nature of the biochemical function of MFT in M. tuberculosis metabolism, particularly in cholesterol catabolism, remains unclear. Accordingly, exploring the conditions under which mft genes are essential for in vitro growth of M. tuberculosis in cholesterol could uncover the bona fide function(s) of MFT. Consequently, insights into the MFT-associated pathway(s) for potential therapeutic intervention could be provided and could light shed on the functions of MFT homologs conserved in many other bacterial species.
In this study, using genetic and transcriptomic methods, we revealed that the essentiality of mft genes for growth in cholesterol is dependent on the presence of ethanol in M. smegmatis, M. marinum, and M. tuberculosis. We showed further that disruption of mft genes affects the mycobacterial growth in ethanol. These findings have implications for respiration and redox regulation during mycobacterial growth/ adaptation in ethanol.

RESULTS
MFT gene essentiality for M. smegmatis growth in cholesterol is ethanol dependent. To determine the essentiality of mft genes for the growth of M. smegmatis in cholesterol, single in-frame deletion mutants of all mft genes (mft mutants) were constructed and characterized (see Fig. S1 in the supplemental material; see also  Fig. 2A). In contrast, the growth rates of all strains in glycerol-supplemented medium were comparable (Fig. S2A). Given the key role of MftC in the initial steps of MFT biosynthesis (3,(5)(6)(7)(8)(9) and the nearidentical phenotypes of mft mutants, the ΔmftC strain was chosen for further analyses. Genetic complementation of strain ΔmftC with a full-length copy of mftC at the attB site, referred to as strain ΔmftC-Comp, reversed the growth defect, albeit with a delay, in cholesterol:EtOH. Thus, disruption of any of the mft genes resulted in impaired growth of M. smegmatis in cholesterol:EtOH. Subsequently, to interrogate whether the growth limitation of strain ΔmftC was ethanol dependent, the growth of strain ΔmftC   (20,21). Consequently, we assessed whether ethanol would also affect strain ΔmftC growth in the presence of another carbon source such as propionate, acetate, or glycerol. Supplementation of ethanol had no effect on strain ΔmftC growth in propionate (0.5% [wt/vol]) or glycerol (0.2% [vol/vol]) but severely delayed strain ΔmftC growth in acetate (0.5% [wt/vol]) in comparison with the growth of the wild-type strain ( Fig. S2D to G). Thus, we conclude that the effect of ethanol on strain ΔmftC growth depends on the carbon source and that MFT and ethanol metabolism in M. smegmatis are linked.
The MFT gene cluster is required for ethanol assimilation in M. smegmatis. Similar to M. marinum (25), M. smegmatis can utilize ethanol as the sole growth substrate ( Fig. 3A; see also Fig. S2C). To directly establish the role of MFT in ethanol assimilation, the growth rates of mft mutants and the wild-type strain in ethanol were compared. Strikingly, the mft mutants did not grow in ethanol (except for strain ΔmftE, which presented delayed growth), whereas the wild-type strain as well as the ΔmftC-Comp strains did grow in ethanol. In summary, the function of each mft gene affects ethanol assimilation.
The inability of strain ΔmftC to grow in ethanol could be attributed to exacerbated toxicity, because high concentrations of ethanol disrupt cell wall and membrane integrity (26,27). However, ethanol toxicity was not observed as a growth-limiting factor in our study. The growth of strain ΔmftC remained unrestored even at the lowest concentration of ethanol tested ( Fig. S3A and B), and there was no ethanol-dependent potentiation with respect to the effect of cell wall-damaging agents such as sodium dodecyl sulfate (SDS) and lysozyme (Fig. S3C). To interrogate the functionality of MFT in catabolism of related substrates, the growth rates of the wild-type strain and ΔmftC mutant in other primary alcohols were compared. M. smegmatis grew in all primary alcohols examined (C2 to C6), except in methanol, but with different growth rates under the conditions tested. In contrast, disruption of mftC eliminated growth of M. smegmatis ( Fig. 3B; see also Fig. S3D to I), thereby strongly establishing the essential role of MFT in primary alcohol assimilation.
MFT is associated with ethanol oxidation in M. smegmatis. PQQ-dependent ethanol assimilation involves alcohol and aldehyde dehydrogenases to oxidize ethanol to acetic acid, a redox reaction, with acetaldehyde as a reaction intermediate (14)(15)(16)(17)(18). Given the similarities between MFT and PQQ (3)-and the inability of the respective biosynthetic mutants to grow in ethanol (17)-it has been proposed that MFT acts as a redox cofactor to enable the function of alcohol dehydrogenase(s) (3,13). Accordingly, we assessed whether supplementation of acetaldehyde would rescue strain ΔmftC growth in ethanol (probably by compensating for inactive MFT-dependent alcohol dehydrogenase). Indeed, acetaldehyde (0.001% [vol/vol]) supplementation partially restored the growth of strain ΔmftC in ethanol (Fig. 3C) as well as in cholesterol: EtOH (Fig. 3D). Neither the mutant nor the wild-type strain grew in medium containing only acetaldehyde (data not shown). Intriguingly, a similar acetaldehyde-dependent growth correction has been reported to eliminate an ethanol-induced lag phase of Saccharomyces cerevisiae via NAD ϩ replenishment by NADH oxidation (28)(29)(30)(31). Therefore, disruption of mft genes probably causes redox imbalance due to dysregulation/ dysfunction of MFT-associated alcohol dehydrogenase in M. smegmatis.
Ethanol alters the expression of genes associated with respiration and redox homeostasis in M. smegmatis. Microarray-based global gene expression analysis was performed to identify ethanol-inducible genes and to determine the cause of the growth retardation of strain ΔmftC in ethanol and cholesterol:EtOH ( Fig. 4A; see also Table S2). Following ethanol treatment, genes associated with M. smegmatis respiration (type I NADH:menaquinone oxidoreductases [nuo gene cluster]), acetate metabolism, ectoine, B 12 -dependent glycerol dehydrogenase, and propane-diol metabolism, as well as genes encoding several oxidoreductases, were upregulated in both the wild-type strain and ΔmftC mutant. In contrast, genes associated with ATP synthase subunits (atpB and atpE) and Sdh2 (sdhA, sdhB, sdhC, and sdhD), genes associated with nitrogen metabolism (ureases and ammonium transporter), and genes rplL and MSMEG_1364 (50S ribosomal protein) were markedly downregulated in both strains after ethanol treatment. In particular, the gene expression profiles of the ethanol-treated and cholesterol:EtOH-treated strains were similar, suggesting that ethanol is preferred in the hierarchy to induce specific transcriptional responses.
With regard to the ethanol-induced transcriptional response, whiB3 (MSMEG_1831) and MSMEG_3644 (MerR family) were upregulated in strain ΔmftC in comparison with the wild-type strain. Expression of MSMEG_6242, annotated as a glycerol/alcohol dehydrogenase, resulted in higher levels of accumulation in the wild-type strain than in the mutant strains during ethanol treatment ( Fig. 4A; see also Table S2). Assessment of this transcriptional modulation by quantitative reverse transcription-PCR (qRT-PCR)based gene expression analysis revealed that the level of expression of MSMEG_6242 was significantly higher in strain ΔmftC-both untreated and treated-than in the wild-type strain (Fig. 4B). Taking the data together, transcriptional profiling indicated that ethanol affects respiration and redox regulation and that the putative alcohol dehydrogenase MSMEG_6242, requires MFT for its function.
MSMEG_6242 is a putative MFT-associated alcohol dehydrogenase. To validate the role of MSMEG_6242 in ethanol assimilation in M. smegmatis, a deletion mutant was constructed (Fig. S1) and phenotyped. Strikingly, the ΔMSMEG_6242 mutant failed to grow in ethanol (Fig. 4C) as well as in cholesterol:EtOH (Fig. S4). However, this growth impairment was corrected, as in the case of strain ΔmftC, by the supplementation of acetaldehyde, thus confirming the role of MSMEG_6242 in ethanol assimilation.
In addition to MSMEG_6242, other oxidoreductase/dehydrogenase genes (MS-MEG_1410, MSMEG_2687, and MSMEG_5866) have been predicted to be probable MFT-associated nicotinoproteins, given their co-occurrence with mft genes in many bacterial genomes (3,13). The role of these dehydrogenases in ethanol assimilation was analyzed in a genetic approach. While repeated attempts to isolate a viable MS-MEG_2687 mutant strain were unsuccessful, mutant strains lacking MSMEG_5866 or MSMEG_1410 were readily isolated (Fig. S1). Both of those mutant strains-unlike the ΔMSMEG_6242 and ΔmftC mutants-showed wild-type-like growth in ethanol and cholesterol:EtOH ( Fig. 4C; see also Fig. S4), suggesting their dispensability or redundancy for ethanol assimilation. Given that the mutant's growth phenotype can be used as a criterion to establish functional association-along with prior bioinformatics predictions (3)-MSMEG_6242 can be regarded as a putative MFT-dependent alcohol dehydrogenase gene required for M. smegmatis growth in ethanol.
Note that no direct homolog of MSMEG_6242 is present in either M. marinum and M. tuberculosis; however, only the latter grows poorly in ethanol (22,25), suggesting species-specific variations in MFT-associated growth and metabolism in ethanol. To interrogate the relevance of MFT for the growth of M. marinum in the presence of ethanol and cholesterol:EtOH, we disrupted mftD as a representative of the mft genes (Table S1). Strikingly, the growth of M. marinum ΔmftD (ΔmftD Mm ) was impaired in minimal medium containing ethanol (0.5% [vol/vol]) or cholesterol: EtOH, thus emphasizing the significance of MFT in ethanol assimilation as well as establishing the functional existence of an MFT-associated but nonorthologous alternative for MSMEG_6242 in M. marinum (Fig. 4D).
Ethanol limits growth of mftC mutant of M. tuberculosis in cholesterol. Unlike M. smegmatis or M. marinum, M. tuberculosis has been found to grow poorly in ethanol (22). To validate the function of MFT in M. tuberculosis, an in-frame deletion mutant lacking mftC, here referred to as ⌬mftC Mtb , was constructed (Fig. S1). As expected, strain ΔmftC Mtb exhibited a partial growth defect in medium containing cholesterol as the sole carbon source, and a genetically complemented derivative, strain ΔmftC Mtb -Comp, showed restored growth in cholesterol:EtOH (Fig. 5A). Notably, the growth of strain ΔmftC Mtb was not severely retarded in cholesterol:DMSO (Fig. 5B). Because the M. tuberculosis wild-type strain presented poor growth in ethanol, growth deficiency, if any, in the mutant was undetectable (Fig. S5A). However, similar to the results seen with M. smegmatis, acetaldehyde supplementation restored the growth of strain ΔmftC Mtb in cholesterol:EtOH to a level comparable to that seen with the wild-type strain (Fig. 5A). Moreover, growth of strain ΔmftC Mtb was unaffected in glycerol, or in acetate, but was partly affected in propionate under conditions of ethanol repletion (Fig. S5B to D). These results suggest that the function of mftC is conserved between M. smegmatis and M. tuberculosis, regardless of their differential ability to grow in ethanol.
Ethanol treatment does not impact redox and bioenergetics parameters in the mftC mutant. Transcriptional analysis indicated that ethanol treatment affects the respiration and redox status of the mftC mutant. We speculated that the cytosolic NADH/NAD ϩ ratio-which can reliably indicate redox status-is altered upon ethanol treatment. Indeed, the NADH/NAD ϩ ratio was increased in the ethanol-treated wildtype and ⌬mftC Mtb strains (Fig. 6A) in comparison with the untreated culture. Intriguingly, there was no mutant-specific defect in recycling of NADH and NAD ϩ . In addition, there were no differences in membrane potential (Fig. 6B) and in susceptibility to cell wall-damaging agents (Fig. 6C) in both the wild-type and ⌬mftC Mtb strains following 24 h of ethanol treatment ( Fig. 6B and C). Furthermore, ethanol treatment did not enhance the susceptibility of the wild-type and mutant strains to agents inducing redox stress (1 mM cumene hydroperoxide [CHP] or tert-butyl hydroperoxide [tBHP]), suggesting that MFT is not directly associated with cellular oxidative stress responses (Fig. 6D). Likewise, there were no differences between the wild-type and ⌬mftC Mtb strains with respect to the oxygen consumption rate (OCR), as measured by the use of  Fig. 6E and F). Most notably, similar findings were also obtained with M. smegmatis strains (Fig. S6). Although transcriptional evidence suggested a role for MFT in respiration/redox regulation, none of these results were confirmatory. Thus, MFT may be involved in a noncanonical redox mechanism in mycobacteria.

DISCUSSION
Our results establish the significance of MFT in mycobacterial metabolism of primary alcohols complementing a range of biological functions documented for RiPPs. So far, PQQ-dependent ethanol oxidation by acetic acid bacteria, which is exploited for the industrial production of vinegar, has been well characterized (14)(15)(16)(17)(18). Intriguingly, the M. smegmatis genome (but not the M. tuberculosis or M. marinum genomes) comprises genes putatively involved in biosynthesis of PQQ and its dependent alcohol dehydrogenases (MSMEG_3721 to MSMEG_3727) (3). However, the established genetic requirement of MFT for the growth of M. smegmatis in ethanol excludes the possibility of overlapping functions of PQQ-dependent reactions under the in vitro conditions examined in the present study.
While MFT biosynthesis genes are conserved, the numbers of predicted MFTassociated nicotinoproteins encoded in mycobacterial genomes differ greatly and could impact the ethanol assimilatory capacity-complete or partial-of each mycobacterial species. For instance, the lack of a homolog of MSMEG_6242 in M. tuberculosis could account for its poor growth in ethanol. Yet M. marinum, lacking a direct homolog of MSMEG_6242, was found to exhibit sustained growth in ethanol, probably owing to the functionality of other MFT-associated dehydrogenases. Moreover, the proposition of a functional relationship between MFT and MSMEG_6242 is solely based on the mutant's shared phenotypes and on prior bioinformatics prediction (3) and thus requires further biochemical confirmation. Nevertheless, the established role of MSMEG_6242 in ethanol assimilation adds to the functional diversity of this enzyme in methylotrophy (32,33) and in glycerol metabolism and as a target for posttranscriptional/posttranslational regulation and S-mycothiolation (34)(35)(36). Therefore, MSMEG_6242, and probably also MFT, is likely to function at the crossroads of regulation, metabolism, and growth in M. smegmatis. In contrast, ethanol metabolismand related MFT function-appears to be irrelevant for M. tuberculosis, perhaps because the status of availability of these volatile compounds with hydroxy groups during infection is ambiguous. Nonetheless, ethanol, as a poor growth substrate for M. tuberculosis, has been demonstrated to alter the cellular state, with NADH accumulation corresponding to the increase in expression of DosR regulon genes (37), mimicking ascorbate treatment (38) and hypoxia adaptation (39). Accordingly, an inability to support active growth is not a strong criterion to predict the impact of a substrate on cellular respiration. Therefore, it is tempting to speculate that a similar poor growth substrate, particularly under nutrient/oxygen-deprived conditions (40,41), could contribute to respiratory flexibility and, in turn, prolong the survival of M. tuberculosis.
Mycobacterial respiration under hostile conditions requires utilization of different electron donors and acceptors, the diversity of which is not completely known (19,41). To this end, the proposed function of MFT as an electron carrier involved in redox exchange with its nicotinoprotein partners gains significance (3). The present study demonstrated that ethanol-induced transcriptional signatures exhibit strong associations between MFT function and redox homeostasis/respiration. Notably, loss of function of mftC resulted in multifold increases in expression of the DosR regulon genes in ethanol-treated M. tuberculosis. The DosR regulon expression is triggered by the redox sensor DosS in response to a reduced menaquinone pool of the respiratory chain (38,42). Furthermore, MftC contains at least two iron-sulfur clusters with an intrinsic ability to regulate the redox state of bacterial cells (3,5,6,43). This notion is further supported by the identification of mft genes as part of a transcriptional network, which are differentially regulated in response to (or in compensation for) the loss of function of whiB4, which recognizes the intracellular redox signals in M. tuberculosis (44). Nonetheless, we noted that the M. tuberculosis strain lacking mftC function was neither hypersusceptible to redox stressors nor impaired in recycling nicotinamide cofactors in comparison with the wild-type strain in the presence of ethanol. One possible reason for this lack of differences in NADH/NAD ϩ recycling could be that the MFT-associated nicotinoprotein probably does not rely on cofactor exchange but instead uses MFT as an external redox partner and a nonexchangeable NAD for multiple turnover, as suggested earlier (3,13). Alternatively, the functions of MFT are likely analogous to those of PQQ, which acts as a redox couple (PQQ/PQQH 2 ) during alcohol oxidation in quinoprotein dehydrogenase (14,18,(45)(46)(47). This PQQ-dependent enzyme is known to donate electrons to c-type cytochromes or directly to ubiquinone instead of using NAD ϩ or NADP ϩ as an electron acceptor (48). Therefore, understanding the nature of MFT-associated redox chemistry (with its nicotinoprotein partners) and of its impact on mycobacterial cellular metabolism is crucial and the subject of ongoing investigation.
Intriguingly, we found that disruption of mft genes profoundly impaired the ability of mycobacterial species to assimilate cholesterol in the presence of ethanol. Griffin et al. (20) pioneered the identification of essential genes for cholesterol catabolism by comparing mutant pools grown under glycerol:EtOH and cholesterol:EtOH conditions. Here we found that, remarkably, ethanol selectively compromised the ability of strain ΔmftC to assimilate cholesterol, but not glycerol, which likely influenced the essentiality prediction of MFT genes. Similarly, Nazarova et al. (21) demonstrated that disruption of mft genes alleviated the growth defect of M. tuberculosis strain Δicl1::hyg in the presence of cholesterol:EtOH and a carbohydrate. We propose that, under such conditions, ethanol-derived suboptimal cholesterol catabolism (by partial degradation of the carbon) in strain ΔmftC reduces propionyl-coenzyme A (propionyl-CoA) flux and consequently methylcitrate toxicity, enabling the growth of strain Δicl1::hyg. Although the effect of ethanol was obvious with additional results obtained from M. smegmatis and M. marinum, the possibility of an impact of other constituents of media on growth cannot be precluded. Nevertheless, in addition to the results seen with cholesterol, the rate of growth of M. smegmatis in acetate was also restricted by ethanol and was reminiscent of catabolite repression. This effect was, however, less pronounced in M. tuberculosis, implying species-specific differences in potential mechanisms of ethanolinduced inhibition of cholesterol metabolism. The mechanisms underlying ethanolinduced growth stasis/impairment are multifactorial, as observed in other bacterial systems: ethanol (i) impairs the regeneration of NADH/NAD ϩ (29-31); (ii) affects the uptake of acetate (16,(49)(50)(51); and (iii) suppresses the genes required for acetate assimilation and metabolic turnover of acetyl-CoA (52)(53)(54). Although these factors may still be relevant in the case of strain ΔmftC growth impairment in acetate or cholesterol: EtOH, our transcriptomic analysis suggests that the extent of ethanol-induced growth impairment is probably owed to compounded and yet species-specific effects of redox imbalance and altered bioenergetics.
Our genetic analysis adds to the expanding knowledge about MFT biosynthesis (3,(5)(6)(7)(8)(9) in the context of ethanol metabolism. However, the inferred functional discordance of the data resulting from analyses of the mftE gene was unexpected, despite its well-recognized significance in MFT maturation (7)(8)(9). Nevertheless, the growth permissiveness of strain ΔmftE in ethanol could have been the result of residual functions of an end metabolite(s) lacking an MftE-mediated modification(s) or of other compensatory gene functions. Complementing the ongoing efforts (5)(6)(7)(8)(9) in elucidating the significance of mftE and other mft genes in the structural/functional diversity of MftA-derived metabolites, future studies should directly investigate nonpathogenic M. smegmatis strain as a more appropriate natural host with fully functional associated metabolic pathways.
In conclusion, we identified a previously unknown function of MFT, a peptidederived cofactor, in ethanol assimilation in mycobacteria. MFT is most likely associated with several other biochemical processes, because it is predicted to act as a redox cofactor for other dehydrogenases whose functions were shown here to be unrelated to ethanol metabolism. Although the role of MFT in M. tuberculosis survival, persistence, and pathogenesis cannot be reliably predicted, the complete loss of MFT biosynthesis in M. leprae (3) and species-specific variations in genes encoding MFT-associated dehydrogenases suggest, at the least, a selective advantage for the retention of MFT in mycobacterial species to inhabit diverse ecological niches.

MATERIALS AND METHODS
Bacteria, media, and growth conditions. Strains used in this study are listed in Table S1 in the supplemental material. Escherichia coli DH5␣ was propagated in Luria-Bertani (LB) medium at 37°C or on LB agar plates containing no or appropriate antibiotics. Mycobacterial strains were propagated in Middlebrook 7H9 broth (Difco) supplemented with albumin-dextrose-catalase enrichment (Becton, Dickinson), 0.2% glycerol, and 0.05% Tyloxapol at 37°C with agitation/rotation (100 rpm) or on Middlebrook 7H11 agar containing 10% (vol/vol) oleic acid-albumin-dextrose-catalase enrichment (Becton, Dickinson) at 37°C. All other chemicals were from Sigma-Aldrich/Fisher Scientific unless otherwise specified.

Construction of mutants.
Allelic exchange mutants were constructed by two-step selection using the p2NIL/pGOAL system (55). Bacterial genotyping was done by PCR and Southern blotting. Genetic complementation studies used integration plasmid pMCpAINT (56). The details of PCR oligonucleotides and suicide plasmids are provided in Table S1.
RNA extraction and microarray analysis. Bacterial strains were treated with 1% ethanol or cholesterol:EtOH (0.01% cholesterol-1% ethanol) for 4 h (M. smegmatis) or 24 h (M. tuberculosis). RNA was extracted with TRIzol reagent (Invitrogen). Cell lysis was achieved through three cycles of bead beating in a FastPrep-24 instrument (MP) at 4.5 m/s for 30 s. RNA was prepared using a Turbo DNA-free kit (Ambion), and RNA concentrations and quality were determined with a NanoDrop ND-1000 spectrophotometer and an Agilent 2100 Bioanalyzer, respectively. Quantitative PCR (qPCR) was performed with Fast SYBR green Master Mix kit (Applied Biosystems).
Microarray experiments were performed as dual-color hybridizations on custom M. smegmatis 4-by-44K (Agilent-017184/expression) and on custom 8-by-60K M. bovis BCG-M. tuberculosis H37Rv M. tuberculosis V3 (Agilent-035148/multipack) arrays. To compensate for dye-specific effects, dye-reversal color-swap hybridizations were applied. Total RNA labeling and amplification were performed using a Quick Amp gene expression labeling kit (Agilent Technologies) and a 5:1 mixture of FullSpectrum MultiStart T7 primer (BioCat GmbH) with oligo(dT)-T7 primer (Agilent Technologies). After precipitation, purification, and quantification, 0.75-g volumes of both Cy-labeled cRNAs were fragmented and hybridized to Agilent-035148 or 1.25 g of each labeled cRNA was used with Agilent-017184. Scanning of microarrays was performed with either 3-m resolution and 20 bits (8 by 60K) or an extended dynamic range and 5-m resolution and 16 bits (4 by 44K) using a G2565CA high-resolution laser microarray scanner (Agilent Technologies). Microarray image data were processed using Image Analysis/Feature Extraction G2567AA v. A.11.5.1.1 software and default settings and the GE2_1105_Oct12 extraction protocol. The extracted MAGE-ML files were analyzed with Rosetta Resolver Biosoftware, Build 7.2.2 SP1.31 (Rosetta Biosoftware). Ratio profiles comprising single hybridizations were combined in an error-weighted fashion to create ratio experiments. A 1.5-fold change expression cutoff value for ratio experiments was applied together with dye-swapped ratio profiles, rendering the microarray analysis data highly significant (P Ͻ 0.01), robust, and reproducible (57). Alternatively, the extracted raw data files (.txt) were analyzed using R and the associated BioConductor limma R package (58).
Measurement of NAD/NADH levels and membrane potential. Bacterial cells were subjected to ethanol treatment for 4 h (M. smegmatis) or 24 h (M. tuberculosis). The NAD and NADH concentrations were determined using a FluroNAD/NADH detection kit (Cell Technology). The membrane potential of M. smegmatis strains was determined after incubation with DiCO2 fluorescent dye (3 mM) for 30 min and was further analyzed on a BD fluorescence-activated cell sorter (FACS) LSRII instrument, whereas the membrane potential of M. tuberculosis strains was measured as previously reported (59) with some modifications. Samples were transferred into clear-bottom 96-well microtiter plates and centrifuged at 3,500 ϫ g for 10 min. DiOC2 (15 M) was then added to each well and incubated for 20 min at room temperature, followed by fixation with 4% formaldehyde for an additional 60 min. Subsequently, samples were washed with fresh 7H9 medium to remove extracellular dye. A SpectraMax M5 spectrofluorimeter (Molecular Devices) was used to measure green fluorescence (488-nm wavelength/530-nm wavelength) and shifts to red fluorescence (488-nm wavelength/610-nm wavelength). Membrane potential was determined as the ratio of red fluorescence to green fluorescence. Untreated and carbonyl cyanide m-chlorophenyl hydrazine (CCCP)-treated cells were included as controls. Each condition was measured in triplicate, and each experiment was performed at least twice. Extracellular flux analysis. OCR values were measured as described earlier (60), with some modifications, using unbuffered (pH 7.35) minimal medium (M. smegmatis) and Middlebrook 7H9 medium (M. tuberculosis) free from any carbon sources.
For the M. smegmatis experiments, a culture in the mid-log phase of growth (optical density at 600 nm [OD 600 ] of 0.8) was starved for 6 h in carbon source-free minimal medium containing 0.01% tyloxapol. Then, 2 ϫ 10 5 to 4 ϫ 10 5 cells/well were added to a Cell-Tak (Corning)-coated XF 96 cell culture microplate and were analyzed using an Agilent Seahorse XFe96 analyzer. OCR values were measured for about 21 min before the addition of preloaded ethanol (1%) or of medium through injection ports of the sensor cartridge. Finally, 5 mM CCCP was added to the wells and the measurement was continued for another 21 min.
For the M. tuberculosis experiments, a culture in the log phase of growth (OD 600 of 0.6 to 0.8) was starved for 18 to 24 h in carbon source-free 7H9 medium containing 0.01% tyloxapol. Subsequently, about 5 ϫ 10 5 cells/well were added to a Cell-Tak-coated XFp miniplate and analyzed using an Agilent Seahorse XFp analyzer. The assay conditions were similar to those described for M. smegmatis.
Statistical analysis. GraphPad Prism 7.03 software was used for statistical analyses. P values of less than 0.05 were considered statistically significant.
Data availability. Microarray data are available from the NCBI GEO database under accession no. GSE121398.