Genetic factors affecting storage and utilization of lipids during dormancy in Mycobacterium tuberculosis

In vitro, carbon starvation has been used to model dormancy in Mtb (41, 42), where the bacterium can survive months of carbon starvation with minimal loss of viability in comparison to other bacterial species (Fig. S1). To better understand which Mtb genes are required to survive dormancy, we screened a transposon mutant library of >40,000 mutants created with the phiMycoMarT7 element (43),(44) for mutants that were impaired for survival during carbon starvation (39, 41). We subjected the transposon mutant pool in vitro to carbon starving for 5 weeks followed by a 5-day resuscitation period in growth media and repeated three cycles of growth, starvation, and resuscitation over a total of 4.5 months (Fig. 1A). The cycling had the advantage that differences in the numbers between survivor and non- or impaired survivor mutants would amplify. To avoid bottlenecks created by cycling the transposon mutant pool, we ensured a passage of more than 10⁸ cells during each step of the experiment (on average >1,000× mutants per TA site) and determined the copy number for each transposon mutant before and after carbon starvation using transposon insertion sequencing (Tn-seq) (34, 42, 45, 46). We identified mutants significantly underrepresented in the output pool compared to the input pool using the previously developed conditional essential genes identifier con-ARTIST (47). Mutants of interest were defined as having a significance value of P < 0.03 (Mann-Whitney U test) and a log 2-fold-change [log₂(FC)] between input and output pools of <−3.5. This methodology was highly reproducible in two replicates of the experiment (r² = 0.78, Fig. S2; Table S1).

We identified 102 genes and 6 intergenic regions that were significantly underrepresented in the output pool compared to the input pool, indicating that these mutations resulted in a high fitness cost to the bacterium during carbon starvation and/or resuscitation (Fig. S3; Table S2). Gene set enrichment analysis (GSEA) revealed that mutations that conferred the most substantial fitness cost were those in genes involved in oxidative phosphorylation [nuo genes, sdh, cydA; normalized enrichment score (NES) = −2.2], lipid transporters (mce genes, NES = −1.5), the dormancy regulator sigB (Fig. 1B; Fig. S4), other transporters, and several genes involved in different metabolic pathways. In addition, the stringent response regulator relA was among the top hits, thus corroborating the validity of our results (48, 49). Of note, the toxin-anti-toxin systems that are often referred to as being crucial for establishing persistent or dormant populations were not among the hits, which is perhaps attributable to functional redundancy among the 88 toxin-anti-toxin systems in Mtb (50).

The identification of the importance of mce lipid transporters for survival during carbon starvation raised the hypothesis that lipids could serve as a vital energy source during dormancy. Among the four mce gene clusters present in Mtb (51, 52), we identified five genes in the mce1 cluster, one in the mce4 cluster, and the uncharacterized orphaned mce gene omamC as being important for survival during carbon starvation (30) (Fig. 1B; Fig. S5). Since the Tn-seq screen did not distinguish between genes required for survival during carbon starvation and genes required for resuscitation, we performed a transcriptomic analysis comparing a 5-week carbon-starved wild-type culture with a culture that had been subsequently resuscitated (Table S3). Among the seven mce genes identified in the Tn-seq data set, omamC had mRNA levels uniquely enriched during carbon starvation compared to growth in rich media [differential expression analysis (DESeq2), log₂(FC) = 1.7, Fig. 1C]. As expected, the stress regulators sigB and relA were also upregulated during starvation [log₂(FC) = 2.6 and 4.3, respectively, Fig. S6]. Although the absence of changes in the transcript levels for other mce genes does not exclude their involvement in carbon starvation, we focused on understanding how the poorly characterized, orphaned mce gene omamC affects Mtb survival during carbon starvation.

We constructed a clean deletion of the omamC gene in the Mtb strain H37Rv (ΔomamC). We confirmed that, indeed, deletion of omamC results in a moderate but significant reduction in Mtb’s ability to survive carbon starvation (Fig. 2A). These findings are similar to previous observations with major dormancy regulators like relA (52). Meanwhile, constitutive overexpression of omamC, in either the wild-type or ΔomamC background, rendered Mtb more resilient to carbon starvation stress (two-tailed non-parametric t-test P_ΔomamC = 0.035, P_ΔomamC− P_omamC+ = 0.009, Fig. 2A), an observation also supported by a calcein-based viability assay (Fig. S7). Perturbing omamC during carbon starvation and resuscitation on solid agar also affected colony size, with deletion of omamC resulting in the formation of smaller colonies, while overexpression (omamC+) resulted in larger colonies (Fig. 2B). While OmamC-dependent differences in colony size on solid agar could arise from differences in growth rate, no omamC-dependent differences in liquid axenic culture were observed (two-tailed non-parametric t-test, P = 0.01, Fig. S8A). However, omamC+ was slightly impaired in its ability to actively replicate inside J774 macrophages compared to wild-type Mtb (Fig. S8B).

Since in vitro carbon starvation models have been shown to induce a drug-tolerant state that is refractory to killing by the majority of anti-tubercular agents, with the exception of rifampicin (10), we examined whether perturbations in the ability to survive carbon starvation might affect the efficacy of rifampicin. Overexpression of omamC increased tolerance toward rifampicin as reflected in an increase of the minimal bactericidal concentration and a decreased kill rate of pre-starved Mtb (two-tailed non-parametric t-test, P = 0.0178; Fig. 2C) (50).

We hypothesized that omamC could be involved in lipid metabolism, given its modest similarity with two other mce-associated genes, rv0177 and rv1972 (Fig. S9). While Mce-associated proteins are poorly characterized (53), their interaction with Mce membrane complexes (35, 54) suggests that they could play some role in conferring lipid substrate specificity. We thus considered two possible roles for OmamC that might contribute to better survival during carbon starvation: (i) OmamC could increase the uptake or biosynthesis of (certain) fatty acids during logarithmic growth in nutritional replete environments so that Mtb has increased stores of accessible fatty acids going into carbon starvation, and/or (ii) OmamC could affect fatty acid oxidation during carbon starvation to increase fatty acid utilization and thus improve the bacterium’s survival of carbon starvation. In this carbon starvation model, since fatty acids are not available from the environment through nutrient supplementation, Mtb would only be able to utilize its own stored lipids or scavenge lipids from deceased siblings.

To explore whether OmamC might play a role in fatty acid storage in times of carbon availability and utilization during periods of carbon starvation, we compared the incorporation of [1-¹⁴C] acetate into the lipids of mutants in which omamC was deleted or overexpressed (ΔomamC and omamC+, respectively). We grew Mtb in the presence of [1-¹⁴C] acetate 1 day (starting at day −1) before we subjected cultures to carbon starvation and measured the concentration of fatty and mycolic acids using thin layer chromatography (TLC) at day 1 and 20. On day 1 after exposure to [1-¹⁴C] acetate in rich media, omamC+ had twice the amount of ¹⁴C-labeled fatty acids and 1.5 times the amount of ¹⁴C-labeled mycolic acids compared to wild type (normalized to cell number, Fig. 3), suggesting that OmamC plays a role in increasing uptake of [1-¹⁴C] acetate and its incorporation into fatty and mycolic acids during periods of exposure to a rich environment. While we did not observe statistically significant alterations in ¹⁴C-labeled incorporation into fatty acids in ΔomamC, we did observe subtle trends that are consistent with this role for OmamC. The lack of a more robust phenotype for ΔomamC could be attributed to some functional redundancies between mam genes, with OmamC’s role in these phenotypes being more pronounced following overexpression rather than deletion.

After 20 days of carbon starvation, the picture had reversed, and omamC+ had a modest but statistically significant lower amount of ¹⁴C-labeled fatty and mycolic acids compared to wild type (omamC+ day 1 vs day 20: P_{fatty acids} = 0.01; P_{mycolic acids} = 0.002), while ΔomamC had significantly more ¹⁴C-labeled mycolic acid compared to wild type (ΔomamC day 1 vs day 20: P_{mycolic acids} = 0.003, Fig. 3A). These findings are consistent with a role for OmamC in enhancing fatty and mycolic acid utilization.

Apart from alterations in the absolute concentration of fatty and mycolic acids during carbon starvation, the composition of lipids was also affected by OmamC. Using liquid chromatography-mass spectrometric (LC-MS) analysis, compared with wild type, we found that during starvation, omamC+ had a relatively lower concentration of lipids containing unsaturated fatty acids (Fig. 3B; Tables S4 and S5). This observation was also reflected in the pool of free fatty acids for omamC+, where unsaturated fatty acids were underrepresented and saturated fatty acids overrepresented compared to wild type (Fig. 3C). Meanwhile, the opposite result was observed for ΔomamC, with a relative increase in lipids with unsaturated fatty acids compared to wild type (Fig. 3B and C).

These data show that prior to dormancy entry, in a setting of rich media, overexpression of omamC results in an enrichment of fatty and mycolic acids during nutrient-rich periods, which are then able to be utilized during periods of starvation, with preferential utilization of unsaturated fatty acids. In contrast, while we did not observe alterations in the ΔomamC’s ability to store fatty or mycolic acids during periods of growth in rich media, ΔomamC was unable to utilize available mycolic acids and unsaturated fatty acids as effectively as wild type or OmamC+ during periods of starvation. Taken together, these data suggest that OmamC plays a role in lipid storage during growth in rich media and lipid utilization during starvation.

Finally, to examine OmamC’s role in fatty acid oxidation and the utilization of fatty acids for central metabolism during starvation, we used LC-MS to measure the incorporation of [all-¹³C]-oleic acid into succinate, a key metabolite in the TCA cycle. We starved cells for 3 weeks, then added [all-¹³C]-oleate to cells in order to compare the incorporation of ¹³C into succinate in wild type, ΔomamC, and omamC+. In all cases, we observed an initial burst of ¹³C incorporation into newly produced succinate 10 hours after the addition of [all-¹³C]-oleic acid. However, we measured a 3-to-14-fold increase in isotopically labeled succinate (two or four ¹³C atoms incorporated per molecule succinate, Fig. 3D) in omamC+ compared to wild type or ΔomamC. These findings were similarly supported by the finding that in response to supplementation with oleic acid as a sole carbon source, carbon-starved omamC+ metabolically responded, as measured with the redox-sensitive dye resazurin, more rapidly than wild type or ΔomamC after 2 weeks of starvation (Fig. 3E). In general, adding fatty acids with various degrees of saturation also triggered a faster response of omamC+ (Fig. S10). After 7 weeks of carbon starvation, however, there was little difference in the ability of these strains to metabolize oleic acids or other fatty acids (Fig. 3E; Fig. S10).

The omamC gene is in an operon encoding putative stress regulators (rv1363c; Fig. S13). This location contrasts with that of other mce-associated genes, which are usually encoded in gene clusters associated with Mce transporters and are known to be important for shifting lipid composition, cholesterol metabolism (mce4 cluster), and infection. Given its atypical genomic association, we explored what other pathways might be impacted by overexpression of omamC. We compared the expression profiles of omamC+ to a set of five strains, including wild type and four control strains overexpressing genes with no fitness consequence during carbon starvation (Fig. S12). Differential expression analysis (Tables S6 and S7) revealed that only eight genes were upregulated in omamC+ compared to the five reference strains after 3 weeks of carbon starvation [log₂(FC) >1.2]. OmamC transcripts themselves were eight times enriched. Interestingly, the transposon mutants corresponding to all eight genes had also shown a fitness defect in the initial Tn-seq experiment (Table S2) and were either related to the stress regulon sigma D (rv0516c, rv3413c, rv3414c), molybdopterin metabolism/regulation (rv1404, rv3205c, rv3206c, rv3355c) or ribosome dimerization (rv3241c, Fig. 4). Seven of the eight genes (except rv3355c) showed enriched transcript levels during starvation, which decreased during resuscitation in the wild-type background, a pattern similar to that observed for omamC expression (Fig. S13).

We were interested if any of these genes when overexpressed results in similar expression profiles to omamC+ using a principal component analysis (Table S8) to define components of a more general transcriptome supportive of carbon starvation survival. Indeed, the profile of the α-α sigD (rv0516c) overexpressing strain (α-α sigD+) clustered with omamC+, while all other expression profiles clustered with wild type (Fig. 4B). Surprisingly, the sigD (rv3414c) overexpression strain (sigD+) fell in the cluster with wild type, suggesting that the effects of α-α sigD (rv0516c) overexpression might not be mediated by the sigD regulon. In addition to the similarity between the transcriptomes of omamC and α-α sigD+, α-α sigD+ also phenocopied omamC+ in its enhanced ability to survive carbon starvation compared to wild type or strains overexpressing sigD or α-sigD, where overexpression of the latter two genes was detrimental to survival (Fig. 4C).

Since the expression profiles of carbon-starved α-α sigD+ and omamC+ correlated well (R = 0.79, both transcriptomes were normalized to carbon-starved wt using DESeq2, Fig. 4D; Table S8) and both displayed an enhanced ability to survive carbon starvation, we speculated that genes that are upregulated in these two strains might be important during dormancy. The top 30 most upregulated genes included three members of the sigD regulon [dnaK, an hsp70 co-factor, and a thiol peroxidase (Fig. 4E; Fig. S14; Tables S9 and S10)] (55), five genes involved in stress responses, and eight genes involved in cell wall maintenance and cell division, including fbpC and ftsZ. The remaining half of the 30 most upregulated genes were either conserved hypothetical proteins, possible or probable transcriptional regulatory proteins, or were otherwise poorly annotated. Upregulation of fbpC was intriguing as it encodes one of the mycolyltransferases Ag85 that catalyzes the transfer of mycolic acids between trehaloses, consistent with our earlier finding that omamC overexpression results in increased consumption of fatty and mycolic acids during starvation. This finding suggests that there may be significant changes to the mycomembrane (Fig. 3) and mycolic acid metabolism, and its transcriptional upregulation may represent a compensatory mechanism for dwindling mycolic acid concentrations. The upregulation of ftsZ, involved in cell division, was also interesting, given the increased exit rate from dormancy in omamC+. We chose 5 from the top 30 highest expressed genes and constructed overexpression plasmids (rv0251c, rv0519c, rv1158c, rv3084). However, none conferred a survival advantage on their own during carbon starvation in the manner of omamC+ and α-α sigD+, suggesting that either other genes are involved, or the concerted action of several genes under the impact of omamC+ and α-α sigD+ may be required to increase fitness (Fig. S15).

Having identified a role for OmamC in lipid metabolism and survival during starvation, we sought to link the role of fatty acid utilization and survival during carbon starvation. We treated carbon-starved Mtb with tetrahydrolipstatin (THL), a non-specific lipid esterase inhibitor, including lipases and mycolyltransferases ( Fig. S16) (56). We reasoned this exposure would inhibit the release of free fatty acids for β-oxidation during starvation, thus hindering ATP production through lipid metabolism. THL treatment, indeed, resulted in Mtb killing (55–57), dramatically decreasing Mtb survival during carbon starvation at both early (1 week) and late (7 weeks) time points during starvation. Bacteria died within the course of a few days after THL exposure (Fig. 5). Of note, this killing was in an OmamC-dependent manner at the earlier time point as the expression of higher levels of OmamC allowed Mtb to delay the lethal effects of lipase inhibition. We did not observe the positive benefit of omamC overexpression when we added THL late, after 40 days, with all strains dying with similar kinetics. Overexpression of omamC also did not affect the minimal inhibitory concentration of THL in rich media ( Fig. S17). These data show that THL-dependent interference with lipid metabolism, potentially through inhibition of lipid consumption, is detrimental during dormancy and that the activity of OmamC can contribute to better survival in the early stages of dormancy.

Mtb H37Rv and derivative strains were grown at 37°C in Middlebrook 7H9 (Difco) liquid culture medium supplemented with 10% oleic acid-albumin-dextrose-catalase (OADC), 0.5% glycerol, and 0.05% tyloxapol (referred to as growth in rich media in the text) or on Middlebrook 7H10 (Difco) solid culture medium supplemented with 10% OADC (BD Biosciences) and 0.5% glycerol. For carbon starvation, liquid cultures were grown to an OD₆₀₀ of 0.8–1, washed once in carbon starvation media (Middlebrook 7H9 supplemented with 0.05% tyloxapol), and subsequently incubated in carbon starvation media at 37°C and 100 rpm. All liquid cultures had an overhead space two to four times of the culture volume for adequate aeration. Plasmid constructs carrying a P_myc1tetO-inducible gene (e.g., omamC) were not induced with anhydrotetracycline (atc), unless otherwise indicated, as promoter leakage was sufficient to induce transcription.

For construction of overexpression plasmids, sequences were PCR amplified from genomic DNA of H37Rv and cloned into the vector pUVtetORm (73) at PacI and PstI (NEB) restriction sites. The ΔomamC deletion was created by recombineering as previously described (74) with few modifications. Briefly, 1 kB of flanking sequence immediately upstream of omamC was amplified with primers Rv1363ca-fw and Rv1363ca-rev and cloned upstream of the hygromycin resistance cassette hyg in the plasmid pJG1100 using SfbI, PmeI. One kilobytes of flanking sequence immediately downstream of omamC were amplified with primers Rv1363cb-fw and Rv1363cb-rev, and cloned downstream of the hyg cassette in the resulting plasmid using the PacI and AscI sites (75). The flanking regions of omamC were then amplified from this plasmid, along with the hygromycin cassette using primers Rv1363ca-fw and Rv1363cb-rev, and the resulting PCR product was electroporated into H37Rv harboring the plasmid-encoded Che9c RecET recombination system (pNitET-SacB-kan) (76), which facilitates replacement of genome-encoded omamC with the hygromycin cassette by allelic exchange. The mutant candidates were plated on selective media (7H9 + OADC agar containing 50 µg/mL hygromycin), and resulting colonies were tested for the insertion of the cassette using control PCR primers reading from the hygromycin cassette to a genomic region just outside the flanking regions used for the cloning. A complete list of strains, plasmids, and primers that were used in the study are listed in Tables S11 and S12.

Construction of the transposon library in Mtb has been described in detail (77). Briefly, M. smegmatis Mc²155 was infected at 30°C with a TM4 phage derivative carrying the conditional transposon vector phiMycoMarT7 to generate high titers of phage. TM4 is virulent at permissive temperatures (30°C) but is unable to replicate at 37°C. For the generation of the transposon pool, 10 mL of a mid-log phase Mtb (H37Rv) culture was washed with Phage buffer (50 mM Tris-HCl pH7.4, 10 mM MgSO₄, 2 mM CaCl₂, 150 mM NaCl) and subsequently incubated in Phage buffer for 24 hours at 37°C, concentrated 10-fold (OD₆₀₀ = 10), and infected with the phage-carrying phiMycoMarT7 at an MOI of 10 for 3 hours at 37°C. The mix of Mtb and Phage was then cultured for 24 hours in 7H9 OADC and then pelleted and plated to 7H10 agar supplemented with 100 µg/mL kanamycin.

A volume of 10 mL dense (OD₆₀₀ >1) Mtb culture was harvested at 3,000 rcf (Allegra X15R Centrifuge, Beckman) and resuspended in 450 µL TE-Glucose containing 10 mg lysozyme and incubated for 1 hour. To further lyse the cells, 100 µL 10% SDS was added and gently mixed. An amount of 10 mg proteinase K was then added and incubated for 30 minutes at 55°C. Finally, 200 µL 5 M NaCl and 160 µL centrimide were added and incubated at 65°C for 10 minutes. To extract the DNA, an equal volume of chloroform:isoamyl-alcohol (24:1) was added to the lysed cells, mixed, and phases were separated by centrifugation >5,000 rcf. The aqueous phase containing the DNA was precipitated using an equal volume of isopropanol, and the resulting DNA pellet was solubilized in water.

One microgram of DNA was then sheared (Covaris sonicator E220) resulting in fragments of approximately 500 bp. The sheared DNA was cleaned using AMPure XP beads (Beckman Coulter), end repaired (END-It, Epicentre) and adenosine residues were attached by Taq polymerase using 20 mM dATP at 72°C for 45 minutes (Qiagen) to facilitate adaptor ligation. The single-stranded adapters 5′-TACCACGACCA-NH₂ and 5′-ATGATGGCCGGTGGATTTGTGNNANNANNNTGGTCGTGGTAT-3′ were denatured at 95°C and then slowly annealed using a 1% ramp down to 20°C (2 hours). This double-stranded adapter was then ligated to the A-tailed DNA fragments (T4 DNA ligase) (NEB). Adapter-ligated fragments were then purified using AMPure XP beads. Using the selective primer pairs 5′-TATGATGGGCGGTGGATTTGTG-3′ and 5′-TAATACGACTCACTATAGGGTCTAGAG-3, the region between the transposon internal T7 promoter and the adaptor was amplified using Phusion high-fidelity polymerase (NEB). To multiplex different samples for Illumina sequencing on MiSeq (Illumina), these fragments were uniquely barcoded using the primer mixes listed in Table S2. Sequences were trimmed and mapped using the software Bowtie2 (http://bowtie-bio.sourceforge.net/bowtie2/manual.shtml) and the H37Rv reference sequence (NC_000962). For the further analysis, we used the con-ARTIST script [MATLAB (Mathworks)] (47).

Genes identified from the Tn-seq analysis having a significance value of P < 0.03 (Mann-Whitney U test) between input and output pools (Tables S4 and S5) were analyzed in the gene set enrichment analysis using KEGG gene ontology terms. Default parameters of the software GSEA (GSEA v4.0.1, Broad Institute and University of California) were used to determine the normalized enrichment score. For a better visualization, the enriched gene sets were further combined into four broader customized categories metabolism, respiration, stress/signaling, and transporter subsets. Inclusion gene set size was set very lenient to 2 as the input data set size was small (Fig. S4; Table S13).

Mtb cultures were pelleted and resuspended in TRIzol (Thermo Fisher Scientific) and frozen at −80°C to preserve RNA. In general, for exponentially growing cultures (OD₆₀₀ = 0.6–1.0), 1 mL of culture was harvested for RNA isolation. For starved cultures whose mRNA levels were substantially lower and which also contained different degrees of dead cells, we harvested 10 mL of culture. The Pellet-TRIzol mix was then bead beaten (0.1 mm beads zirconia/silica) (Biospec) three times for 1 minute with 1-minute intervals on ice to cool the sample. The lysed sample was then applied to a direct-zol RNA Miniprep column according to supplier’s protocol (Zymo Research). A range of 0.1–0.5 ug total RNA was used for library construction, described previously in detail (78). Briefly, samples were DNase treated and fragmented for barcoded adaptor ligation and sample identification. Multiple samples were then pooled, and rRNA was depleted (Ribo-Zero) (Illumina). The sequences of the adapters were then used for cDNA synthesis. After degradation of the RNA strand and a second-adaptor ligation to be able to further multiplex the libraries, we amplified these fragments with index primers for Illumina sequencing (Next-Seq 500, 800 M paired-end reads). Analysis was performed using DESeq 2.

Mtb H37Rv and derivative strains were grown to an OD₆₀₀ of 0.8 in Middlebrook 7H9 10% OADC, 0.5% glycerol, and 0.05% tyloxapol. Then, [1-¹⁴C]-acetate (Perkin Elmer) was added at a concentration of 1 uC/mL and incubated for a further 12 hours. Then cultures were harvested and washed in starvation media before being resuspended in starvation media and incubated with shaking at 37°C. One day after the switch to starvation media, we sampled at t = 0 and after 20 days (t = 20). The relatively early time points were chosen to avoid measuring dead bacteria that occur later in starvation. Samples were spun down and washed in H₂O and resuspended in H₂O and heat killed at 80°C overnight. Preparation of the samples for thin layer chromatography has been previously described (79) with minor modifications. Briefly, saponification was performed by adding 40% tetrabutylammonium hydroxide to the samples and incubating for 20 hours at 100°C (vol/vol 1:1). Methylene chloride (vol/vol 1:1) and methyl iodide (vol/vol 1:40) were then added to the cooled mixture methylation and slowly rotated for 1 hour. After phase separation, the upper aqueous phase was discarded. This was repeated after adding 3 N HCl (vol/vol 1:2) and again repeated after adding H₂O (vol/vol 1:2). Residual water was removed by NaSO₄. The lipophilic phase was dried and resolved in methylene chloride and applied to a Silica Gel 60 F₂₅₄ plate (Millipore Sigma). Separation was achieved using solvent (vol/vol) 95:5 hexane/ethyl acetate in a TLC chamber.

Mtb was kept in starvation media (Middlebrook 7H9 supplemented with 0.05% tyloxapol) for 3 weeks as described above with an OD₆₀₀ = 1 at start of starvation. A volume of 3 mL of the starved culture was then pelleted at 10,000 rpm at room temperature (Eppendorf Centrifuge 5430) and resuspended in 500 µL isopropanol for lipid analysis or methanol for free fatty acids liquid-liquid extraction and 20% chloroform to ensure killing of the bacteria. To break up the hardy cell wall, silicate beads were added to the suspension, and bacteria were lysed using a Mini-Beadbeater (BioSpec). Silica beads were pelleted by centrifugation (10,000 rpm) at room temperature (Eppendorf Centrifuge 5430), and the isopropanol chloroform fraction from the resulting supernatant was allowed to evaporate to prepare the sample for LC-MS analysis. Polar and non-polar lipids were profiled using LC-MS system comprising Shimadzu Nexera X2 U-HPLC (Shimadzu Corp.) coupled to an Q Exactive HF orbitrap mass spectrometer (Thermo Fisher Scientific). Dried samples were resuspended in 100 µL of (95:5 vol/vol) isopropanol/water containing 1,2-didodecanoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids) as an internal standard and centrifuged (10 minutes, 9,000 × g, 4°C). Supernatants (10 µL) were injected onto a 100 × 2.1 mm, 1.7 µm ACQUITY BEH C8 column (Waters). The column was eluted isocratically at 450 µL/min with 80% mobile phase A (95:5:0.1 vol/vol/vol 10 mM ammonium acetate/methanol/formic acid) for 1 minute followed by a linear gradient to 80% mobile phase B (99.9:0.1 vol/vol methanol/formic acid) over 2 minutes, a linear gradient to 100% mobile phase B over 7 minutes, then 3 minutes at 100% mobile phase B. MS analyses were carried out using electrospray ionization in the positive ion mode using full-scan analysis over 220–1,100 m/z at 120,000 resolution and 3 Hz data acquisition rate. Other MS settings were sheath gas 50, in source CID 5 eV, sweep gas 5, spray voltage 3 kV, capillary temperature 300°C, S-lens RF 60, heater temperature 300°C, microscans 1, automatic gain control target 1e6, and maximum ion time 125 ms. Raw data were processed using Progenesis QI (nonlinear dynamics) for peak detection and integration of both lipids of known identify and unknowns. Lipid identities were determined based on comparison to reference plasma membrane extracts and are denoted by total number of carbons in the lipid acyl chain(s) and total number of double bonds in the lipid acyl chain(s). Free fatty acids were profiled using a Shimadzu Nexera X2 U-HPLC (Shimadzu Corp.) coupled to a Q Exactive hybrid quadrupole orbitrap mass spectrometer (Thermo Fisher Scientific). Dried samples were resuspended in 120 µL of 3:1 methanol water containing 15R-15-methyl-PGA₂, 15R-15-methyl-PGF_2alpha, 15S-15-methyl-PGD₂, 15S-15-methyl-PGE₁, and 15S-15-methyl-PGE₂ (Cayman Chemical Co.) internal standards and centrifuged (10 minutes, 9,000 × g, 4°C). Supernatants (10 µL) were injected onto a 150 × 2 mm ACQUITY BEH C18 column (Waters). The column was eluted isocratically at a flow rate of 450 µL/min with 20% mobile phase A (0.1% formic acid in water) for 3 minutes followed by a linear gradient to 100% mobile phase B (acetonitrile with 0.1% acetic acid) over 12 minutes and then isocratic elution using 100% mobile phase B for 3 minutes. MS analyses were carried out in the negative ion mode using electrospray ionization, fullscan MS acquisition over 70–850 m/z, and a resolution setting of 70,000. Other MS settings were sheath gas 45, sweep gas 10, spray voltage −3.5 kV, capillary temperature 320°C, S-lens RF 60, heater temperature 300°C, microscans 1, automatic gain control target 1e6, and maximum ion time 250 ms. Raw data were processed using Progenesis QI (nonlinear dynamics) for peak detection and integration. Identities of free fatty acids were confirmed using authentic reference standards.

Mtb H37Rv strains were starved at an OD₆₀₀ of 1 for 2 or 7 weeks using the standard starvation media before it was diluted 10-fold in fresh starvation media containing 0.8 mg/mL oleic and 0.1 mg/mL resazurin. Absorbance was measured at 570 and 600 nm. Reduction grade for each time point was calculated using the following equation:

where ε is the molar extinction coefficient for resazurin depending on the wavelength and its oxidation state, A is the absorbance at a given wavelength, t₀ is the initial measurement, and t_x is the measurement at a given time x. The values for the extinction coefficient are the following: εOX_{_600nm} = 117.216, εOX_{_570nm} = 80.586, εRED_{_570nm} = 155.677, and εRED_{_600nm} = 14.652 [Biosystems, (80)]. The experiments were performed in sextuplicate in black 96 well plates with clear flat bottom (Corning) in a final volume of 100 µL per well. The microtiter plates were incubated in a closed container with a damp towel at 37°C.

Infections were carried out as previously described (81). Wild-type Mtb H37Rv, H37RvΔomamC, or H37Rv harboring the omamC overexpression plasmid was used to infect J774 macrophages (ATCC). Mtb strains used were grown to mid-log phase, washed in PBS, resuspended in PBS, and subjected to a low-speed spin to pellet clumps. J774 were infected at the indicated MOI, allowing 4 hours for phagocytosis. Cells were then washed once with PBS, and media were added back to washed, infected cells. Infected J774 were incubated for 3 days, then macrophages were lysed with Triton X-100 (0.5%), and surviving bacteria were enumerated by colony-forming units. Cell lines were verified to be free of mycoplasma contamination using the ATCC Universal Mycoplasma Detection Kit.

Bacterial cultures were exposed to carbon starvation as described in Bacterial strains and culture conditions. In this case, bacteria were kept in 2 mL starvation media in a tightly sealed 30 mL flask gently shaken at 100 rpm. To investigate the effects of lipase inactivation, 30 µg/mL tetrahydrolipstatin (Orlistat, Sigma) was added to the media early on starting at week 1 and re-applied at weeks 2 and 3. For the effects in the late stage of carbon starvation, THL was added at weeks 7, 8, and 9. Each week, the cultures were sampled, and dilutions plated on Middlebrook 7H10 (Difco) solid culture medium supplemented with 10% OADC (BD Biosciences) and 0.5% glycerol.

Inhibition of mycolyltransferase activity by THL in M. tuberculosis H37Rv was examined using an optimized quencher-trehalose-fluorophore (N-QTF) (82). Bacteria were grown to OD₆₀₀ = 0.2, and mycolyl transferase activity was measured by adding N-QTF (2.5 µM) in a final volume of 50 µL per well (n = 3). Relative fluorescence units were measured at ex/em 485/525 and monitored for 80 hours at 37°C. Each triplicate per strain is shown together with a linear fit (sampling frequency 1 /hour, all R2 >0.98, rates: wt = 13.4 h⁻¹, corrected for cell number).