Pyrrolocin C and equisetin inhibit bacterial acetyl-CoA carboxylase

Antimicrobial resistance is a growing global health and economic concern. Current antimicrobial agents are becoming less effective against common bacterial infections. We previously identified pyrrolocins A and C, which showed activity against a variety of Gram-positive bacteria. Structurally similar compounds, known as pyrrolidinediones (e.g., TA-289, equisetin), also display antibacterial activity. However, the mechanism of action of these compounds against bacteria was undetermined. Here, we show that pyrrolocin C and equisetin inhibit bacterial acetyl-CoA carboxylase (ACC), the first step in fatty acid synthesis. We used transcriptomic data, metabolomic analysis, fatty acid rescue and acetate incorporation experiments to show that a major mechanism of action of the pyrrolidinediones is inhibition of fatty acid biosynthesis, identifying ACC as the probable molecular target. This hypothesis was further supported using purified proteins, demonstrating that biotin carboxylase is the inhibited component of ACC. There are few known antibiotics that target this pathway and, therefore, we believe that these compounds may provide the basis for alternatives to current antimicrobial therapy.


Introduction
Antimicrobial resistance is a growing public health concern worldwide [1,2]. Compounds derived from natural sources offer novel chemistry and may provide a solution to drug resistance. Most approved drugs currently available are derived from natural products or inspired by natural product pharmacophores. From the years 1981-2019, nearly 50% of small molecules approved by the FDA were natural products [3].
Fungal natural products from the class known as pyrrolidinediones are often antibacterial in the low micromolar range, yet their mechanisms of action are unknown. We previously identified pyrrolocin A (PYRA), which was isolated from a phylogenetically novel strain of endophytic fungus from Papua New Guinea (NRRL accession number 50135) and possessed moderate activity against the attenuated strain of Mycobacterium tuberculosis (Mtb), H37Ra [4]. A heterologous expresser was developed to improve production, yielding desmethyl isomers of pyrrolocin A: pyrrolocins B and C (PYRB and PYRC) [5]. In addition, pyrrolocins A-C were selectively active against additional Gram-positive bacteria including S. aureus, Streptococcus pneumonia, and Bacillus subtilis. Other pyrrolidinediones, including TA-289 and equisetin (EQI), also display antibacterial activity against a wide range of pathogens in a single-digit μg/mL range [6,7]. However, the mechanism of action of these compounds against bacteria was undetermined.
Here, we investigate the mechanism by which PYRC and EQI (Fig 1) exert toxicity in S. aureus, MRSA and additional drug resistant strains. We describe the initial transcriptomic and metabolomic approaches used to inform our hypothesis that PYRC and EQI inhibit fatty acid biosynthesis by targeting the bacterial acetyl-CoA carboxylase. Fatty acid rescue, acetate tracing and enzyme inhibition studies indicate that biotin carboxylase is catalytic domain inhibited by PYRC and EQI. There are few known antibiotics that target this pathway and, therefore, we believe that these compounds may provide interesting alternatives to current antimicrobial therapy.

PYRC and EQI broadly inhibit bacterial growth
PYRC and EQI showed activity against a broad panel of Gram-positive bacteria and mycobacteria, but were inactive against Gram-negative bacteria. Both compounds showed IC 90 s of 1 μg/mL against S. aureus, including methicillin-sensitive and methicillin-resistant strains ( Table 1 and  Because of this activity pattern, we suspected that drug penetration might be responsible for the lack of activity against Gram-negative bacteria. To test this, we measured the synergistic effect of EQI and PYRC with the outer membrane-permeabilizing agents polymyxin B and polymyxin B nonapapetide (Fig 2 and S2 Fig) [8]. When the membrane was permeabilized, EQI and PYRC both killed Escherichia coli. There was a significant potentiation effect (>90% inhibition) when EQI (32 μg/mL) was tested together with polymyxin B (0.125μg/mL) (Fig 2). Polymyxin B did not potentiate PYRC, but limited potentiation was observed using the nonapeptide. A maximum inhibitor effect of~30% was achieved when PYRC (128 μg/mL) was combined with polymyxin B nonapeptide (8 μg/mL) (S2 Fig). As we show below, this result is consistent with the relative potency of EQI and PYRC against the molecular target in E. coli.
Toxicity testing against a drug sensitive human cell line (CEMTART 1A2 [9]) yielded comparable therapeutic indices (TI) for the two compounds (Table 1 and  different mechanisms of drug action. As expected, the transcription component inhibitors rifampin and ciprofloxacin, formed a cluster when compared to the other drug treatments (circled in red; Fig 3). Gentamicin, a protein synthesis inhibitor, also formed a cluster spatially separate from all other treatments (circled in green; Fig 3). Isoniazid, para-aminosalicylic acid, and pyrazinamide did not form distinct clusters comparative to vehicle (DMSO, circled in blue; Fig 3) indicating that under our experimental conditions, these drugs did not elicit a consistent effect on Mtb. These drugs may require a longer incubation period in order to induce consistent effects on the transcriptome. The most striking feature that resulted from this analysis was that PYRC occupied an entirely separate region of the PCA plot from the known drugs, suggesting a unique transcriptomic signature (circled in cyan; Fig 3).
Pathway analysis of transcriptomic changes using the Database for Annotation, Visualization and Integrated Discovery (DAVID) [13] revealed that PYRC most impacted oxidative phosphorylation (S1 Table). We speculate that this might be mechanistically related to the reported activity of an EQI analog, TA-289, which causes aberrant mitochondrial morphology in yeast that is not due to reactive oxygen species, inhibition of cardiolipin synthesis, or direct effects on the electron transport chain [6].
Other transcription signatures that were significantly affected by PYRC treatment include (in order of decreasing enrichment score): ATP binding, pyrimidine metabolism, cation binding, and ribosomal protein processes (S1 Table). The meaning of these changes was difficult to discern. Ultimately, we attributed them to metabolic shifts and stress responses caused by PYRC. For example, transcripts of pathways in pyrimidine metabolism were upregulated. Protein synthesis inhibitors upregulate enzymes involved in pyrimidine metabolism, possibly to conserve nucleotides during translation inhibition [10,11]. Similarly, if energy metabolism were impeded by PYRC, induction of salvage pathways may be a compensatory response to reduced function of vital pathways, such as translation. However, one transcription process noted as a possible indicator of PYRC mechanism of action was cation binding, more specifically metal ion binding. Metal binding has been suggested as an important feature of EQI bioactivity [14][15][16].

EQI antibacterial activity not affected by metal ion concentration
Based upon the published data and transcriptome indicators, we explored metal homeostasis in EQI treated S. aureus. Tetramic acids are known to bind metal ions [16]. Upon proton loss, the tetramate anion has several resonance forms that promote bidentate interactions with metals [15]. In S. aureus metal homeostasis assessed by inductively-couple plasma-optical emission spectroscopy (ICP-OES), we observed EQI treatment to induce significant decreases in endogenous levels of iron, magnesium, and manganese, while causing an increase in copper levels ( Fig 4A). These findings might indicate that the drug sequesters metal or inhibits metal ion transport. We then attempted to rescue the bacteria from potential effects of such metal sequestration by supplementing the media with exogenous metal ions ( Fig 4B). We found no protection was conveyed against EQI (IC 90 ) at any of the test concentrations for magnesium or manganese. Furthermore, we found only moderate protection (~55%) at high, non-physiological, concentrations of iron. Metals did not increase PYRC toxicity. This result stood in contrast to previous work on an anti-TB drug with metal dependent-toxicity corresponding with transcription signatures of metal metabolism [10]. Since neither rescue nor toxic exacerbation could be achieved with exogenous metal ions, it seemed unlikely that metal sequestration or dysregulation of metal metabolism was responsible for S. aureus inhibition.

EQI decreases malonyl-CoA levels in S. aureus
A transposon S. aureus library [17] was screened for resistance to EQI, but no informative hits were observed. Metabolomic analysis of EQI-treated S. aureus lysates was therefore used to gain insight into mechanism of action. Several changes in key intracellular metabolites were observed. Tricarboxylic acid cycle (TCA) intermediates as well as amino acids were elevated (Table 2), which indicates a stress response in S. aureus due to addition of antibiotics [10,18,19]. Consistent with an increase in TCA intermediates, which is an NAD-consuming and NADH-generating cycle, there was a downward trend in NAD levels and upward trend in NADH levels in response to EQI. Critically, malonyl-CoA was significantly decreased following EQI treatment. Malonyl-CoA is the product of acetyl-CoA carboxylase, the first committed step in fatty acid biosynthesis. Also, NADP levels were decreased and NADPH levels were increased, suggesting reduced fatty acid biosynthesis. Taking these findings together, we proposed the hypothesis that EQI and PYRC inhibit fatty acid biosynthesis in S. aureus.

Exogenous fatty acids prevent growth inhibition by PYRC and EQI in S. aureus
In bacteria, fatty acids are generated through the fatty acid synthase (FAS) pathways (S3 Fig) [12]. S. aureus utilizes the type II FAS system [20][21][22]. Saturated branched-chain fatty acids (SBCFAs) constitute a major component of the S. aureus membrane. When challenged with compounds that inhibit fatty acid synthesis (e.g. andrimid, an inhibitor of ACC), S. aureus overcomes inhibition by incorporating exogenous SBCFAs [20,22]. This is attributed to availability of nonesterified acyl carrier proteins, which are otherwise depleted with FASII inhibitors in S. aureus [20][21][22]. SBCFAs can become incorporated into the membrane, thus maintaining membrane integrity [20][21][22]. In a study in which S. aureus was treated with fatty acid synthesis inhibitors, 50% of endogenous fatty acids were replaced following growth in media supplemented with oleic acid [22]. We found that SBCFAs (a15:0 and a17:0) and oleic acid (OA; 18:1Δ9) significantly rescued bacterial growth against PYRC and EQI treatments, consistent with the idea that PYRC and EQI target fatty acid biosynthesis ( Fig 5). When EQI or PYRC were tested for activity against an acetyl-CoA carboxylase (ACC) deficient S. aureus mutant (ΔaacD; generously provided by Dr. Rock); no effect was observed. This was because the mutant also requires addition of a15:0 and a17:0 or OA to grow, and these same fatty acids rescue S. aureus from EQI inhibition. Thus, neither EQI nor PYRC exhibit growth inhibition in a condition in which ACC is absent.

PYRC and EQI inhibit de novo fatty acid synthesis in S. aureus
De novo fatty acid synthesis is initiated by ACC [20]. ACC is a multi-subunit enzyme in S. aureus that converts acetyl-CoA to malonyl-CoA via the addition of an activated carboxyl group. We used radioisotope tracing to test whether PYRC and EQI would inhibit incorporation of acetate into fatty acids of treated S. aureus. Radiolabeled acetate is a well-established tool to trace de novo fatty acid synthesis [22][23][24]. Upon entering the cell, acetate is converted into acetyl-CoA by acetyl-CoA synthase and utilized by subsequent pathways, including fatty acid biosynthesis. Dose-dependent inhibition of [1-14 C]-acetate incorporation into  6). When the chloroform:methanol extractable-lipids were analyzed by LCMS, a complex mixture of modified fatty acids, lipids and lipid soluble metabolites was identified. Because of the complex nature of lipid metabolism, it was not possible to quantify changes in individual components. However, when the TLC plate films were simply analyzed by densitometry, it was apparent that the all radiolabeled components of the mixture (represented by TLC spots), were broadly and  proportionately diminished by EQI or PYRC in a dose dependent fashion, consistent with early inhibition of de novo fatty acid synthesis (Fig 7). Both normal phase and reverse phase TLC was used in order to take an unbiased approach in quantification of polar and nonpolar lipid fractions (Fig 6).

PYRC and EQI inhibit biotin carboxylase catalytic domain of ACC
Based upon the hypothesis that ACC was inhibited, we tested the actions of EQI and PYRC against recombinant E. coli ACC in vitro. ACC catalyzes the first committed and regulated step in fatty acid biosynthesis [30]. The reaction catalyzed by ACC involves two half-reactions and is shown in Scheme 1. Bacterial ACC is composed of three proteins: biotin carboxylase (BC), carboxyltransferase (CT), and biotin carboxyl carrier protein (BCCP). In the first halfreaction, BC catalyzes an ATP-dependent carboxylation of biotin, which is covalently attached to BCCP. In the second half-reaction, CT transfers the carboxyl group from biotin to acetyl-CoA to produce malonyl-CoA. Biotin carboxylase and carboxyltransferase retain their activity when isolated from the other components and can utilize free biotin as a substrate [30]. In order to determine which subunit of ACC is inhibited by PYRC and EQI, both subunits (BC and CT) were purified and assayed separately. PYRC and EQI inhibited biotin carboxylase by 36.6% and 84.5% respectively (Fig 8A). Control reactions showed neither PYRC nor EQI inhibited the coupling enzymes. Similarly, inhibition of CT was not observed by either PYRC or EQI (100μM each) (Fig 8B).

Discussion
EQI and related tetramic acid natural products have long been recognized for their wide activity against eukaryotic and Gram-positive bacterial cells. Multiple lines of evidence indicate that EQI and related compounds exert their effects on eukaryotic cells by affecting mitochondrial metabolism [5,6,24,26,27]. The compounds also bind to metals efficiently and may affect metal homeostasis or metalloproteins. Indeed, HIV integrase is targeted by EQI based upon the metal-binding property [28]. For these reasons, tetramic acids have been considered to be general toxins that may not be useful as drugs. In this work, we show instead that EQI and PYRC act against S. aureus specifically by inhibiting fatty acid biosynthesis. Moreover, our data strongly suggest that malonyl-CoA synthesis by the ACC enzyme is the direct target of EQI and PYRC We sought to investigate PYRC because we noted promising activity against M. tuberculosis that differed in mechanism from other TB drugs (Fig 3). Based upon the previous data on EQI, we initially explored energy metabolism and metal homeostasis. Indeed, transcriptome analysis brought up a number of transcripts involved in energy utilization and metal cation binding (S1 Table). However, none of the energy protein transcripts could readily be related to decreased ATP levels (which we observed to trend toward a decrease in the metabolome study; Table 2) nor to signature transcripts connected to the observed inhibitory activity [10,12]. In examining metal homeostasis by ICP-OES, we observed relatively large changes upon EQI treatment, but these were not tied to cell growth inhibition and could not be restored by addition of exogenous metals (Fig 4). Thus, we concluded that these effects were not likely central to PYRC or EQI mechanism of action.
We turned to metabolomics to further examine the effects of EQI on bacteria focusing on S. aureus, for which there is rich literature regarding its metabolism. When EQI was added at a sub-lethal dose to S. aureus, only a small number of metabolites showed a significant change. Among these, several were involved in the tricarboxylic acid cycle and the metabolic state of the cell, including a greatly increased amount of lactate. The increase in lactate could be explained by previous work on metal homeostasis in S. aureus, in which it was demonstrated that iron limitation led to an increase in fermentative metabolism [29].
Beyond central metabolism, the most profoundly decreased metabolite upon EQI treatment was malonyl-CoA, while acetyl-CoA was also modestly increased ( Table 2). These data led us to suspect that fatty acid synthesis inhibition might play a role in the biological activity of EQI. Moreover, an increased acetyl-CoA pool would feed into the TCA cycle, potentially explaining some of the metabolomics results. To determine whether fatty acid synthesis was indeed inhibited by EQI and PYRC, we examined [1-14 C]-acetate incorporation into growing fatty acids. Consistent with selective inhibition, de novo fatty acid synthesis was decreased in a dose-dependent manner that was similar to the MIC of EQI and PYRC. Furthermore, the attenuation of [1-14 C]-acetate incorporation was observed to impact the wide variety of lipid containing metabolites fractionated by TLC approximately evenly.
Fatty acid biosynthesis begins with the conversion of acetyl-CoA into malonyl CoA by the ACC enzyme complex [30][31][32][33][34]. Subsequently, in S. aureus and most bacteria (with the exception of mycobacteria and a few others) a type II fatty acid synthase (FAS) converts malonyl-CoA into fatty acids. Previously, Rock and co-workers demonstrated that fatty acid synthesis inhibition in S. aureus leads to an accumulation of malonyl-CoA, which is toxic to the cell [35]. However, inhibition of acetyl-CoA carboxylation by the ACC enzyme complex diminishes malonyl-CoA, consistent with our metabolomics result. Rock and colleagues restored growth of S. aureus in the presence of an ACC inhibitor by adding exogenous fatty acids, but the growth was not reestablished by fatty acids in the presence of FAS inhibitors [35]. We treated S. aureus with EQI or PYRC in the presence or absence of exogenous fatty acids (a15:0/ a17:0 and 18:1Δ9), demonstrating a recovery very similar to what was observed by Rock and colleagues in the presence of ACC inhibitor (Fig 5). These results implicated ACC as the target of the pyrrolidinedione acid metabolites, EQI and PYRC. Using purified E. coli ACC components, we demonstrated that the biotin carboxylation step of ACC catalysis was indeed inhibited by EQI and PYRC. Ongoing work is now focused on description of the inhibition kinetics using S. aureus ACC and determination of the crystal structure of EQI-and PYRC bound to S. aureus ACC.
ACC has emerged as an important drug target in both humans and in bacteria [11,36]. Several selective compounds exist that target only the eukaryotic ACC [36]. Among the drugs targeting bacterial ACC, several are natural products, including thailandamide B, andrimid, and moiramide B [36][37][38][39]. Mutational analysis of Salmonella enterica ACC resistant to thailandamide B revealed a series of residues in the beta protein adjacent to the interface between the alpha and beta subunits [37,38]. A crystal structure of moiramide B bound to the S. aureus ACC demonstrated that inhibition occurs at the interface between the alpha and beta subunits [39], which is nearby but not identical to the binding site proposed for thailandamide B.
Moiramide B and andrimid are 2,5-pyrrolidinedione-based natural products. It was shown that a beta-ketoamide / pyrrolidinedione was essential for activity; the beta-ketoamide was present in the enol tautomer, which bound in the same place that anion chemistry would take place during the catalytic cycle [14][15][16]. Similarly, tetramic acids are 2,4-pyrrolidinediones, with beta-ketoamides. The tautomerization of tetramic acids is essential to their bioactivity and metal binding activities and has been studied in detail [40,41]. For example, the pKa of tetramic acids is~2.5-4 [16]. Thus, we hypothesize that in tetramic acids, the pyrrolidinedione warhead could bind in the same place as in the moiramide B series. Alternatively, it is known that there is a zinc cation in the S. aureus ACC [39,42]; perhaps metal chelation is again important in this instance. Further research is required to determine the exact molecular interactions by which EQI and PYRC inhibit ACC.
It should be emphasized that the pyrrolidinedione moiety is not sufficient to impart inhibition of fatty acid biosynthesis. A series of moiramide derivatives was synthesized, in which required structural features were identified [11,26]. In the case of EQI and PYRC, very similar tetramic acid moieties are present. The methyl group of EQI is not essential, since it is absent in PYRC, and the methylated pyrrolocin A is nearly equipotent as an antibiotic to PYRC [4]. EQI and PYRC also have very similar 10-membered cyclic decalin subunits, as does TA-289 [6], which are decorated slightly differently with functional groups. Pyrrolocin B is much less active as an antibiotic, and its configuration at the decalin ring is cis, rather than trans as found in EQI and PYRC. Thus, the substituents attached to the tetramic acid moiety are highly important for specificity of action.
EQI and PYRC display approximately 1 log bacterial selectivity versus human cells in culture, which suggests that a safe therapeutic window may be achievable with this class of antibiotic. It is currently unclear what the molecular mechanism of action might be for EQI in eukaryotic cells. It is possible that the inhibition of fatty acid synthesis leads to mitochondrial dysfunction, as found in some known inhibitors [43]. Whether the target in eukaryotic cells is similar or much different, selective ACC inhibition suggests that these compounds may be modified to produce useful antibiotic agents. Moreover, recent data indicate that EQI can be formulated to act as a specific antibiotic agent in animals [44], and drug delivery of EQI as a topical application yielded reduced infection and increased wound healing [45]. Crucially, EQI and PYRC exhibit consistent activity against challenging, drug-resistant strains [46], including those resistant to methicillin, clindamycin, doxycycline and vancomycin. The novel mechanism of action may thus lead to utility in the ongoing battle against drug resistant infections.

Antimicrobial microdilution assay
Glycerol stocks of S. aureus, doxycycline/clindamycin-resistant S. aureus, and vancomycinresistant E. faecalis were thawed and streaked on Nutrient agar (NA) plates. Methicillin-resistant S. aureus, on the other hand, was streaked on Nutrient agar supplemented with 8μg/mL oxacillin. E. coli was streaked on Luria-Bertani agar plates. Plates were incubated at 37˚C for 8 -12h. A single colony from the NA plates was transferred into 7mL Mueller Hinton broth (MHB) (with additional 2% NaCl for MRSA)(LB broth for E. coli) and incubated for 6-8h at 30˚C, 150 rpm. The turbidity of the broth culture was then adjusted to match the 0.5 McFarland standard (1 x 10 8 cells/mL). The adjusted broth culture was then further diluted 200-fold and used as the final inoculum for the assay. 200 μL of the test organism was dispensed to each well of a 96 well plate. PYRC, EQI, positive control and 1% DMSO as solvent control were tested using a two-fold dilution scheme starting at 32μg/mL with 10 dilutions each. For E. coli, PYRC (2-128 μg/mL), EQI (2-128 μg/mL), polymyxin B (0.016-1 μg/mL) and polymyxin B nonapeptide (0.125-64 μg/mL) were tested singly, and in combination using a two-fold dilution scheme. Following an 18 -20h incubation period, ten microliters of MTT (5mg/mL) were added to wells and incubated for 2h. One hundred microliters of solubilizing reagent (50% DMF/20% SDS) was added to wells and incubated for 1 hour. The A 570 was measured using a Biotek-Synergy 2 Microplate Reader (Biotek;Winooski, VT).

Transcriptomic analysis of PYRC and prototypical anti-TB drugs
H37Ra was grown to log phase, measured by OD 600 . An aliquot of the bacterial culture was added to 7H9 media to achieve a final volume of 50 mL. The following drugs were added to the cultures, in triplicate, to result in the final concentrations (~10X minimum inhibitory concentration or IC90): rifampin (RIF; 12. . Differential expression analysis of drug compared to vehicle (DMSO) generated a gene list that were significantly upregulated or downregulated (foldchange: -2 � x � 2; p < 0.05). These gene lists were analyzed using functional clustering analysis in the National Institute of Allergy and Infectious Diseases DAVID website [13].

Total metal analysis of EQI-treated S. aureus by ICP-OES
An aliquot of an overnight S. aureus culture was added to LB broth, followed immediately with DMSO or EQI (4.36 μM). Cultures were incubated at 37˚C for 2 hours. Bacteria were pelleted and washed in 1X phosphate buffered saline (PBS). Bacterial pellets were digested in 70% nitric acid (trace metal grade) at 100˚C for 15 mins. Digestates were diluted with MilliQ water and submitted for ICP-OES analysis at the University of Utah Iron & Heme Core Facility. Total metal concentrations were adjusted based on final OD 600 .

Exogenous metal rescue of EQI-treated S. aureus
S. aureus cultures were incubated in DMSO or EQI (4.36 μM) in the presence or absence of exogenous metals at 37˚C for 2 hours. Metal solutions (MgCl, MnCl, & Fe(EDTA)) were added to reach the following concentrations: 1 μM, 10 μM, and 100 μM. Percent survival was determined by the MTT assay.

Metabolomic analysis of EQI in S. aureus
An aliquot of an overnight S. aureus culture was added to LB broth, followed immediately with DMSO or EQI (4.36 μM). Two sets of cultures were prepared in quintuplicate for DMSO and EQI (one set for the acidic preparation and one set for the basic preparation). Cultures were incubated at 37˚C for 2 hours. The final OD 600 was measured for each culture following the 2 hours. The bacteria were pelleted at 12000 x g, 4˚C for 10 min and washed with PBS. Bacterial pellets were reconstituted in cold 90% MeOH/1% formic acid (acidic preparation) or 90% MeOH/1% ammonium hydroxide (basic preparation). Preparations were bead milled on high for 2 min. Water was added to bring MeOH concentration to 50%. The preparations were pelleted at 12000 x g, 4˚C for 10 min. The supernatant was transferred and dried down en vacuo. Samples were submitted to the University of Utah Metabolomics Core Facility. Data was normalized to final OD 600 .
All LC-MS was performed using an Agilent 6550 QTOF fitted with an Agilent 1290 HPLC. A 100 x 2.1 mm, 3.5 μm, 100A SeQuantz zic-HILIC column (Merck Millipore, Darmstadt, Germany) was used for separation. Ten millimolar ammonium carbonate (pH 9) was used as the strong solvent with acetonitrile the weak solvent. The initial LC condition was held at 90% acetonitrile for one min followed by a 20-min ramp to 40% acetonitrile and a 2-min hold. The column was re-equilibrated over 10 min at 90% acetonitrile. Flow rate was 0.2mL/min. Data was recorded and translated to file format using MassHunter software (Agilent, Santa Clara CA).

[1-14 C] Acetate incorporation
An aliquot of an overnight culture of S. aureus was added to LB broth to yield an approximate initial OD 600 of 0.4. The culture was split for drug treatments with and without [1-14 C] acetate. [1-14 C] Acetate was added (4μCi/mL) and incubated at 37˚C for 2 hours. Duplicate cultures without [1-14 C] acetate was performed in parallel to normalize based on final OD 600 . Lipid fractions were extracted using a modified Bligh-Dyer method [25]. The bacteria were pelleted for 10 min at 12000 x g, 4˚C, washed with PBS, and pelleted again. The pellet was extracted by reconstituting in a chloroform to methanol mixture (1:2). Additional chloroform and water was added for a final ratio of chloroform:methanol:water of 2:2:1. The chloroform layer was transferred and evaporated. Chloroform volume was adjusted based on final OD 600 prior to loading onto normal or reverse-phase TLC plates. Hexane:acetone (25:4) developing solvent was used to develop normal-phase TLC plates and chloroform:methanol (3:2) was used to develop reverse-phase TLC plates. Carestream1Kodak1Biomax1 MR film was exposed to TLC plates for 7 days and then developed using a Mini-Medical 90 Processor (cat. # 9992305300-U; AFP Imaging, Elmsford, NY). Grayscale values of film images were analyzed in Fiji [47]).

LC-MS analysis of lipid fractions
A modified Bligh-Dyer method was used to extract the lipid fraction from bacteria cultures [25]. In brief, the bacteria were pelleted for 10 min at 12,000 × g, 4˚C, washed with PBS, and pelleted again. The pellet was extracted by re-constituting in a chloroform to methanol mixture (1:2) and then shaken for 16 hours at room temperature. Additional chloroform and water was added for a final ratio of chloroform:methanol: water of 2:2:1. The mixture was centrifuged to separate the methanol/water layer and chloroform layer. The chloroform layer was removed, mixed with 10 grams of silica gel, and dried en vacuo. The lipid fraction was fractionated further with the use of a silica gel flash column. Three column volumes of hexane: acetone (25:4) was passed through the column and individual fractions were collected, followed by 2 column volumes of hexane:acetone (50:50) and 1 column volume of hexane:acetone (10:90). Fractions were spotted on a normal-phase TLC plate and visualized by charring (10% CuSO4/ 8% phosphoric acid) to determine retention factor of contents. These fractions were dried down en vacuo and reconstituted in 100% isopropyl alcohol. The samples were submitted to University of Utah Mass Spectrometry Core for lipidomic analysis.

Protein purification and enzymatic assays
E. coli biotin carboxylase with a His-tag on the N-terminus was overexpressed and purified according to Blanchard et al. [48]. Carboxyltransferase with a His-tag on the N-terminus of the α-subunit was expressed and purified according to Blanchard & Waldrop [49].
The activity of biotin carboxylase was determined according to the method of Blanchard et al. [48]. The activity of carboxyltransferase was determined in the reverse non-physiological direction according to Blanchard & Waldrop [49].

Statistical analysis
Unpaired t tests were performed on each analyte identified by metabolomic analysis (in quintuplicate). For the fatty acid residue experiments, unpaired t tests were performed comparing conditions without fatty acid to conditions with fatty acids across 3 independent experiments run in triplicate. Significance was determined if the p-value was less than 0.05.  Table. PYRC treatment induces pathways enriched in energy metabolism in Mtb. (DOCX)