Identification of a metabolic disposal route for the oncometabolite S-(2-succino)cysteine in Bacillus subtilis

Cellular thiols such as cysteine spontaneously and readily react with the respiratory intermediate fumarate, resulting in the formation of stable S-(2-succino)-adducts. Fumarate-mediated succination of thiols increases in certain tumors and in response to glucotoxicity associated with diabetes. Therefore, S-(2-succino)-adducts such as S-(2-succino)cysteine (2SC) are considered oncometabolites and biomarkers for human disease. No disposal routes for S-(2-succino)-compounds have been reported prior to this study. Here, we show that Bacillus subtilis metabolizes 2SC to cysteine using a pathway encoded by the yxe operon. The first step is N-acetylation of 2SC followed by an oxygenation that we propose results in the release of oxaloacetate and N-acetylcysteine, which is deacetylated to give cysteine. Knockouts of the genes predicted to mediate each step in the pathway lose the ability to grow on 2SC as the sulfur source and accumulate the expected upstream metabolite(s). We further show that N-acetylation of 2SC relieves toxicity. This is the first demonstration of a metabolic disposal route for any S-(2-succino)-compound, paving the way toward the identification of corresponding pathways in other species.

The respiratory intermediate fumarate, an electrophilic ␣,␤unsaturated dicarboxylate, reacts spontaneously and readily with soft nucleophiles such as sulfhydryl groups in a process known as succination (1,2). Fumarate and the sulfhydryl group of cysteine undergo succination to form the stable compound S-(2-succino)cysteine (2SC), 3 which exists as two diastereomers due to the chirality of the S- (2-succino) bond; this reaction can occur with free cysteine as well as with cysteinyl residues of proteins and small-molecule thiols such as GSH (1, 2) (Fig. 1). Succination of proteins and GSH is biologically significant due to their high cellular concentrations and because their cysteinyl residues can be significantly more acidic (and hence nucleophilic) than free cysteine (1,2). Cysteine residues located in enzyme active sites are likely to exist as strongly nucleophilic thiolate anions (3), making thiol enzymes particularly prone to succination (4). Several proteins with functional cysteine residues are known to be susceptible to succination that results in impaired functionality, including glyceraldehyde-3-phosphate dehydrogenase (5,6), aconitase (7), actin and tubulin (8,9), and chaperone proteins (9). At least 182 succinated proteins have been detected in humans (10). Succination is thus a common and well-described post-translational chemical modification of proteins.
The degree of succination of biological thiols is directly related to fumarate concentration (1). Accordingly, 2SC can serve as a biomarker for abnormalities in aerobic respiration that cause fumarate buildup. In humans, germline mutations that inactivate the citric acid cycle enzyme fumarate hydratase cause fumarate accumulation and increase the succination of cellular thiols; such mutations predispose to hereditary leiomyomatosis and renal cell cancer syndrome (11)(12)(13). 2SC is considered an oncometabolite because of its accumulation in certain cancers (14). An increase in succination due to fumarate accumulation is also observed in adipose tissue under the hyperglycemic conditions associated with type 2 diabetes and obesity (15)(16)(17). Succination of proteins has many deleterious effects (18 -21). Additionally, succination of GSH can lead to persistent oxidative stress and cellular senescence (22). The harmful effects of succination suggest that this process is more than just a symptom of mitochondrial stress or dysfunction in chronic diseases but that succination actually contributes to the pathogenesis of disease complications. In support of this, recent studies suggest that succination is the mechanistic link between mitochondrial and endoplasmic reticulum complications in diabetes (23).
Much research has focused on targets of succination and the role of succination in disease but not on the metabolism of 2SC after it forms. No metabolic disposal route for 2SC has yet been demonstrated. Here, we present biochemical and genetic analyses showing that Bacillus subtilis has a breakdown pathway that is specific for 2SC. These analyses implicate the yxe operon in 2SC breakdown, show that the pathway begins with acetylation and ends with deacetylation, and confirm that 2SC is toxic.

B. subtilis uses 2SC as a sulfur source
We first tested whether Escherichia coli or B. subtilis can utilize 2SC as the sulfur source. 2SC served as an excellent sulfur source for B. subtilis but not E. coli (Fig. S1). A cysteine auxotroph of B. subtilis (⌬yrhA ⌬cysK) (24) also grew well on 2SC (Fig. S2), demonstrating that B. subtilis can metabolize 2SC to cysteine.

A high-throughput screen implicates the yxe operon in 2SC utilization
To identify the gene(s) required to metabolize 2SC, we used a high-throughput screen with a B. subtilis library containing ϳ7.3 ϫ 10 4 unique magellan6x transposon insertions (i.e. Tnseq) (25). The initial transposon-mutant population was grown for six generations on defined medium containing 2SC, sulfate, or methionine as the sulfur source. Afterward, the transposon insertion sites were sequenced, allowing us to map the location and relative frequency of each transposon insertion for each experiment. If a gene is important for growth on one particular sulfur source, then the relative frequency of transposon insertions located within that gene will decrease in the population grown on that sulfur source relative to the others. Different transposon insertions can result in varying degrees of gene silencing, so a gene may still contain transposons even if it is essential under a certain growth condition.
Sulfate-or methionine-grown populations served as both controls and comparisons for the 2SC-grown population. As expected, nearly every gene involved in cysteine and methionine synthesis was essential when grown on sulfate but not when grown on methionine (Fig. S3). Because B. subtilis can metabolize 2SC to cysteine (Fig. S2), growing cells on 2SC is functionally equivalent to growing cells on cysteine. Accordingly, genes involved in the conversion of cysteine to methionine were essential when grown on 2SC (Fig. S3). Observing the expected trends in the different sulfur-source populations validated the screening procedure. The only genes that significantly affected fitness in the 2SC-grown population but not in either sulfate-or methionine-grown populations were genes of the yxe operon (Fig. 2), particularly yxeK and yxeL.

The yxe operon and its distribution
The yxe operon consists of seven genes that encode a predicted FAD-dependent monooxygenase, yxeK; an acetyltransferase, yxeL; three components of an ATP-transporter complex, yxeMNO; an N-acetylcysteine deacetylase, yxeP; and a putative dehydratase, yxeQ. No specific role for the yxe operon has been ascribed, but there is evidence that the acetyltransferase yxeL and N-acetylcysteine deacetylase yxeP may participate in the degradation of S-methylcysteine, and they have been named snaB and sndB, respectively (26). Here, we retain yxe designations. The yxe operon is widely distributed in bacteria, occurring in firmicutes and ␣-, ␤-, and ␥-proteobacteria (Fig.  S4). E. coli lacks both the yxe operon and the ability to use 2SC as a sulfur source (Fig. S2). A priori, it is reasonable to predict that the yxe operon plays a role in 2SC metabolism.

Analysis of mutants confirms a role for yxe genes in 2SC metabolism
Because Tn-seq indicated that yxe operon genes play a role in 2SC metabolism, we obtained a BKE knockout mutant (27) for each of the yxe operon genes and tested the mutants' ability to grow on 2SC as the only sulfur source. Deletants of the monooxygenase yxeK and the acetyltransferase yxeL were severely limited in their ability to grow on 2SC with only a slight trace of growth observed for either mutant (Fig. 3A). In contrast, no obvious growth defect was observed for the other yxe gene deletants (Fig. 3A).
The acetyltransferase YxeL has a homolog in B. subtilis, SnaA, that is reported to acetylate certain S-alkylcysteine adducts such as S-methylcysteine (26). An snaA deletant grew well on 2SC; however, a mutant lacking both acetyltransferase genes yxeL and snaA showed no sign of growth on 2SC (Fig. 3B), indicating that an acetylation step is critical for 2SC metabolism and that snaA can partially compensate for yxeL deficiency. Besides yxeP, B. subtilis has two other functionally redundant N-acetylcysteine deacetylase genes, sndA and sndC (26). As was observed for the yxeP deletant, mutants of sndA and sndC grew

Deconstruction of the oncometabolite S-(2-succino)cysteine
well on 2SC. However, a mutant lacking all three deacetylase genes did not grow on 2SC (Fig. 3B). These results suggest that a pathway to metabolize 2SC to cysteine involves acetylation and deacetylation steps, similar to a proposed pathway for the degradation of other S-alkylcysteine adducts (26).
Three monooxygenase genes, cmoO, cmoI, and cmoJ, have all been reported to be essential for the degradation of S-methylcysteine and other S-alkylcysteine adducts (26). Mutants of each of these genes grew well on 2SC (Fig. 3C), indicating that, apart from the acetylation and deacetylation steps, the breakdown of 2SC requires distinct genes from that of S-methylcysteine.

YxeL is a highly efficient 2SC N-acetyltransferase
Tn-seq and growth experiments indicate that the acetyltransferase gene yxeL is important for 2SC metabolism, but growth tests show that snaA may partially compensate for yxeL. To test their relative ability to acetylate 2SC, we purified heterologously expressed YxeL and SnaA proteins and assayed their 2SC N-acetyltransferase activity. YxeL could acetylate essentially all the 2SC present in the assay mixture, indicating that it acts on both diastereomers of 2SC, which should be present in equimolar amounts (1, 2). The K m for YxeL is in the low micro-molar range, which is more than 50-fold lower than that for SnaA ( Table 1). The turnover number for YxeL is more than 250-fold higher than that for SnaA, making YxeL about 13,000fold more catalytically efficient (Table 1). This fits with the Tnseq and growth data that indicate that yxeL is far more important than snaA for 2SC metabolism. These kinetic parameters are consistent with 2SC being a physiological substrate for YxeL.

YxeL alleviates 2SC toxicity in B. subtilis by N-acetylation
YxeL is a highly efficient 2SC N-acetyltransferase and catalyzes a critical step in 2SC metabolism. Acetylation of toxic or unwanted compounds is a common strategy used by bacteria (28), raising the possibility that YxeL alleviates 2SC toxicity. To assess this, we tested whether 2SC inhibits growth of a B. subtilis yxeL mutant. Wildtype (WT) or yxeL deletant cells were overlaid on a plate containing sulfate as the sulfur source, and 2SC was added to the center of the plate. 2SC caused a zone of growth inhibition to occur on yxeL deletant but not WT B. subtilis, whereas N-acetyl-S-(2-succino)cysteine (N-acetyl-2SC) did not inhibit growth of either strain (Fig. 4). These data indicate that 2SC is a toxic compound, at least in B. subtilis at high exogenous concentrations, and that N-acetylation relieves toxicity.

Metabolite profiling confirms a breakdown pathway
Our data indicate that the first step in 2SC breakdown is N-acetylation and that oxygenation and deacetylation steps are also required. A reasonable scenario involves acetylation of 2SC followed by oxygenation of N-acetyl-2SC and formation of N-acetylcysteine (NAC), which is deacetylated to give cysteine, analogous to the proposed breakdown pathway of other S-alkylcysteine compounds (26). If this proposed pathway for 2SC deconstruction is correct, then mutants should accumulate their respective upstream pathway intermediates.
The first step in the pathway is acetylation of 2SC. Thus, we predicted that acetyltransferase (e.g. yxeL) mutants would accumulate 2SC. We profiled B. subtilis cells grown on medium containing fumarate as the carbon source in an attempt to increase the spontaneous succination of thiols by fumarate, thereby increasing 2SC formation. 2SC was detected in both WT and snaA deletant cells, but yxeL deletants accumulated around 70-fold more 2SC (Fig. 5A). There was no significant difference in 2SC levels between the yxeL deletant and yxeL snaA double mutant cells (Fig. 5A). These data confirm that the yxeL gene product is the main enzyme controlling 2SC levels in vivo.
We proposed that YxeK is a monooxygenase that acts on N-acetyl-2SC. Thus, yxeK mutants should accumulate N-acetyl-2SC. We profiled B. subtilis cells grown on medium , and various monooxygenase mutants (C) were grown overnight in LB medium with the appropriate antibiotic; washed twice with water; and diluted to optical density 10 0 , 10 Ϫ1 , 10 Ϫ2 , and 10 Ϫ3 , and 10 l was spotted on medium with either 1 mM 2SC or 2 mM sulfate as the sulfur source. Incubation was for 16 h at 37°C.

Deconstruction of the oncometabolite S-(2-succino)cysteine
containing glucose as the carbon source but were unable to detect N-acetyl-2SC in WT or yxeK deletant cells (Fig. 5B). In an attempt to force carbon flux through the 2SC breakdown pathway, we added 2SC to the culture either 10 or 60 min prior to harvesting the cells. When 2SC was added to the culture 10 min prior to harvesting, yxeK deletant cells showed a massive accumulation of N-acetyl-2SC to around 8 mol mg Ϫ1 of protein (Fig. 5B). By 60 min after 2SC addition, N-acetyl-2SC was still readily detectable, but the levels decreased about 25-fold compared with 10 min after 2SC addition (Fig. 5B). There was also a small but significant increase in NAC at 60 min after 2SC addition (Fig. 5B). These data indicate that the yxeK gene product plays a major role in the breakdown of 2SC by acting on N-acetyl-2SC, likely giving rise to NAC. Because the levels of N-acetyl-2SC decreased and NAC slightly increased between 10 and 60 min after 2SC addition to the mutant, there may be one or more functionally redundant mechanisms that can partially compensate for the loss of YxeK. YxeP is predicted to be an NAC deacetylase that completes the 2SC breakdown pathway. Because growth tests indicated that B. subtilis has three functionally redundant NAC deacetylase genes (Fig. 3B), we profiled the triple mutant strain (⌬yxeP ⌬sndAC). Cells were grown in the same manner as the yxeK deletants. There was no increase in NAC 10 min after addition of 2SC, but there was a significant increase in N-acetyl-2SC (Fig. 5C). At 60 min after 2SC addition, NAC levels increased about 60-fold, and N-acetyl-2SC levels increased about 10-fold compared with the values obtained at 10 min after 2SC addition (Fig. 5C). These data show that the levels of both upstream intermediates, N-acetyl-2SC and NAC, increase in the NAC deacetylase triple mutant and provide evidence in support of the proposed pathway.

Discussion
Our results provide strong biochemical and genetic evidence that genes of the yxe operon are primarily responsible for 2SC breakdown in B. subtilis. The initial breakdown step is N-acetylation of 2SC, a reaction that YxeL performs remarkably well ( Table 1). The catalytic efficiency (K cat /K m ) of YxeL is ϳ350fold higher than that of bacterial serine acetyltransferase, which is another small-molecule acetyltransferase involved in cysteine metabolism (29). This high efficiency might be necessary because of the toxicity of 2SC (Fig. 4) and thus the need to avoid its accumulation. Our data show that the concentration of 2SC is maintained below 5 M in B. subtilis cells (assuming a cell volume of 4.6 fl (30)) unless yxeL is disrupted, even when grown under conditions that should favor an increase in succination (Fig. 5A). In addition to initiating the salvage of cysteine from 2SC, YxeL may also prevent the accumulation of a toxic metabolite by converting it to a more benign form. If that were the case, then yxeL would fit the definition of a metabolite damagepreemption gene (31).
The next step in 2SC breakdown, carried out by YxeK, is likely oxygenation of the succinyl moiety of N-acetyl-2SC, leading to the release of NAC. One possible mechanism for this step is that YxeK hydroxylates the 2-position of the succinyl moiety of N-acetyl-2SC, causing a spontaneous elimination reaction of the resulting hemithioketal that generates oxaloacetate and NAC (Fig. 5D). Although other mechanisms are formally possible, the intermediate described above is analogous to the hemithioacetal intermediate of the glyoxalase I reaction that is formed from the reversible reaction of methyglyoxal with GSH (32), so it is supported by a well-defined precedent. We were unable to detect oxaloacetate in any bacterial sample; thus, we cannot determine whether it accumulated in the cysteine deacetylase triple mutant (⌬yxeP ⌬sndAC) as our proposed YxeK reaction schemes suggest it might (Fig. 5D). In any case, further biochemical studies are needed to elucidate how YxeK forms NAC from N-acetyl-2SC.
The last step in the pathway is deacetylation of NAC to give cysteine. Our results show that deacetylation can occur by YxeP or other functionally redundant NAC deacetylases (Figs. 3B and 5C). Growth assays indicate that yxeM, yxeN, and yxeO, which are predicted to encode the three subunits of an ABCtype transporter, are not essential for 2SC utilization in B. subtilis (Fig. 3A). Still, there is reason to suspect a 2SC transport function for these genes. Our Tn-seq results show that transposon insertions in these genes slightly decrease fitness when grown on 2SC-containing medium (Fig. 2). Furthermore, B. subtilis has at least one other ABC-type transporter that has been shown to transport sulfur-containing compounds, including S-alkyl-derivatives of cysteine (33-36). yxeMNO genes may encode a 2SC transporter that is not essential for growth on 2SC in B. subtilis because of functional redundancy. Similarly, yxeQ is not essential for 2SC utilization, but this could also be due to functional redundancy. It is also possible that yxeQ allows the yxe operon to participate in the disposal of compounds other than 2SC.

Deconstruction of the oncometabolite S-(2-succino)cysteine
The 2SC breakdown pathway involves acetylation, oxygenation-initiated deconstruction, and deacetylation, which is similar to the pathway proposed for the metabolism of S-methylcysteine, and the operons that are primarily used to metabolize each compound (the yxe operon and ytmItcyJKLMNytmOytn IJrbfKytnLM operon) are similar in content as noted previously (26). This is likely the reason for the high degree of functional redundancy observed between enzymes of each operon in the breakdown of 2SC and S-methylcysteine. The deacetylases (and probably ABC-type transporters) perform essentially the same task and can efficiently moonlight in the breakdown of either compound, whereas the acetyltransferases and monooxygenases perform more specialized tasks and have a limited ability to moonlight. Based on our evidence that the yxe operon is predominantly responsible for 2SC breakdown, we propose renaming it as the S-(2-succino)cysteine metabolism (scm) operon.
To our knowledge, this is the first demonstration of a metabolic route for 2SC breakdown. It was previously reported that succinated GSH is a substrate of GSH reductase (37). However, these results are almost certainly artifactual because (a) dimethyl fumarate was used instead of fumarate to succinate GSH, (b) a mixture of dimethyl fumarate and GSH was assayed instead of purified substrate, and (c) the reaction GSH reduc-tase catalyzes, reduction of a disulfide bond, is different from the hydrolysis reaction needed to break the thioether bond of succinated GSH. Furthermore, we were unable to replicate this finding in vitro with GSH reductase enzyme assays. The expected enzyme activity was detected with GSH disulfide as a substrate, but we could not detect any activity with purified S-(2-succino)GSH, 2SC, or N-acetyl-2SC as substrates (Table  S1).
Although it is well-established that succination of cellular thiols can have deleterious effects in vivo (e.g. by inactivating enzymes with functional cysteine residues (18 -21)), this is the first report that 2SC itself is toxic, at least to B. subtilis. Breakdown of 2SC may be an important way to deal with this unavoidable and toxic product of metabolism, more so than as a means to recycle a damaged metabolite. Acetylation appears to be a common strategy for prokaryotes to manage toxic and/or reactive compounds. Bacteria often have specific acetyltransferases that prevent the toxic buildup of sugars such as lactose (38) and maltose (39,40), and acetylation of antibiotics is a common mechanism of bacterial resistance (41). 2SC disposal substantiates the emerging paradigm that cells often handle toxic metabolites by pathways that begin with acetylation and end with deacetylation in direct analogy to the blockingdeblocking strategies used in organic chemistry (26). Asterisks denote levels that are significantly different (*, p Ͻ 0.05; ***, p Ͻ 0.001; t test). †, not detectable (detection limit, 0.1 nmol mg Ϫ1 protein). D, proposed breakdown pathway for 2SC in B. subtilis. A possible reaction mechanism for YxeK as discussed in the text is shown. Solid or dotted lines indicate enzymecatalyzed or spontaneous reactions, respectively.

Deconstruction of the oncometabolite S-(2-succino)cysteine
The yxe operon occurs in widely taxonomically diverse bacteria (Fig. S3), but the total number of species containing this operon is quite small. Of a representative set of over 1600 diverse bacterial genomes (see "Experimental procedures"), the yxe operon (defined as containing the monooxygenase and at least two of the acetyltransferase, deacetylase, putative dehydratase, and ABC transporter subunits) occurs in only 21. Homologs of the monooxygenase gene yxeK, which is the most distinct gene in the pathway and perhaps an indicator for the ability to breakdown 2SC, are found outside of the context of the yxe operon but are still limited to Bacteria, Archaea, and Ascomycota fungi. Because succination of thiols is likely to occur nearly ubiquitously, this raises the possibility that other distinct 2SC breakdown pathways exist. It is also possible that most organisms do not metabolize 2SC but instead simply excrete it. If this were the case, then having a breakdown pathway would enable the utilization of 2SC that has been released to the environment.
Humans lack homologs of the yxe genes, and it is so far unknown whether they have a pathway to metabolize 2SC. Our work provides an example of how such a pathway could operate. Considering the importance of succination to the progression of diseases such as diabetes (15)(16)(17)42) and, in general, of the role spontaneous modifications of macromolecules is thought to play in aging (43), the identification of a mammalian system to metabolize 2SC, or otherwise deal with 2SC accumulation, could provide important medical insights.

Chemicals
2SC was made by combining 40 mmol of cysteine with 45 mmol of sodium fumarate in 50 ml of water, adjusting pH to 8.0 with NaOH, and incubating at 22°C for 35 h with stirring. The mixture was lyophilized, and the resulting crystals were stored at Ϫ20°C. 2SC was further purified by HPLC using a Hypersil GOLD 250 ϫ 4.6-mm C 18 column (Fisher Scientific) with 0.1% formic acid as the mobile phase (flow rate, 1.0 ml min Ϫ1 ). Fractions containing 2SC (detected by absorbance at 210 nm) were collected, and successive runs were pooled, lyophilized, and stored at Ϫ20°C. S-(2-Succino)GSH and N-acetyl-2SC were synthesized and purified in the same manner except that 8 mmol of GSH and 9 mmol of sodium fumarate were combined in 15 ml of water or 8 mmol of N-acetyl-L-cysteine and 9 mmol of sodium fumarate were combined in 12 ml of water, respectively. All other chemicals were from Sigma-Aldrich.

Bacterial strains and culture conditions
All strains are listed in Table S2. All BKE mutant loci were back-crossed into WT B. subtilis 168 by transforming 5 g of isolated DNA into WT cells (45) and selecting for recombinants on LB agar plates containing 1 g ml Ϫ1 erythromycin and 25 g ml Ϫ1 lincomycin. The presence of the BKE cassette in the expected locus was verified (Fig. S5). The B. subtilis transposon library used for Tn-seq experiments was described previously (25). For growth assays and preparing Tn-seq and metabolite profiling samples, strains were grown in ED minimal medium (8 mM K 2 HPO 4 , 4.4 mM KH 2 PO 4 , 30 mM NH 4 Cl, 2 mM MgSO 4 , 0.6 mM MgCl 2 , 27 mM glucose, 0.3 mM Na 3 -citrate, 0.25 mM L-tryptophan, 0.1 mM FeCl 3 , 50 M CaCl 2 , 5 M MnCl 2 , 12 M ZnCl 2 , 2.5 M CuCl 2 , 2.5 M CoCl 2 , and 2.5 M Na 2 MoO 4 ). To prepare medium with 1 mM 2SC or methionine as the sulfur source (or for medium with no sulfur), MgSO 4 was replaced with MgCl 2 at the same magnesium concentration (2 mM). Solid medium was prepared by adding low-melt agarose to a final concentration of 0.8% (w/v). Growth assays were performed as indicated. To prepare samples for Tn-seq, 0.4 ml of the B. subtilis transposon library (stored at Ϫ80°C at an optical density (600 nm) of 1.0) was combined with 7.6 ml of LB medium and incubated at 37°C with shaking for 1.5 h until optical density reached 0.38. The library was then washed twice in ED minimal medium (without sulfur) and used to inoculate 20 ml of ED medium or ED medium containing either 2 mM sulfate, 1 mM Met, or 1 mM 2SC to an optical density of 0.02. Cultures were incubated at 37°C with shaking for 9 -10 h until optical density reached 1.3 (six doublings) at which point the cells were harvested by centrifugation (8000 ϫ g, 10 min), frozen in liquid N 2 , and stored at Ϫ80°C. For preparing metabolite profiling samples, glucose was replaced with 30 mM sodium fumarate (when indicated). Overnight cultures were used to inoculate fresh medium to an optical density of 0.05, and cultures were grown at 37°C with shaking for 6 -8 h until optical density reached 1.5 Ϯ 0.1 at which point an equivalent of 2 ml at optical density of 1.0 was harvested by centrifugation (21,000 ϫ g, 30 s), frozen in liquid N 2 , and stored at Ϫ80°C. When indicated, 2SC was added to a final concentration of 0.5 mM either 10 min or 1 h before harvesting.

Tn-seq analysis
Tn-seq was performed essentially as described previously (25,46). DNA was isolated from the harvested cell pellets using the GeneJET Genomic DNA Purification kit (Thermo Scientific, Waltham, MA). 6 g of DNA was treated with the restriction enzyme MmeI (New England Biolabs, Ipswich, MA) for 3 h at 37°C, then 10 units of calf intestine phosphatase (New England Biolabs) was added, and incubation was continued for 1 h. The DNA was then extracted and precipitated (25) and dissolved in 27.5 l of 2 mM Tris-Cl, pH 8.5. To the DNA was added 2 l of annealed DNA adapter (200 mM each of oCJ25 and oCJ26 oligonucleotides prepared as described (25)), 3.5 l of 10ϫ T4 DNA ligase buffer, and 1.5 l (600 units) of T4 DNA ligase (New England Biolabs). The ligation mixture was incubated for 16 h at 16°C, and DNA was isolated with the GeneJET Genomic DNA Purification kit. 5 l of adapter-ligated DNA was used in PCRs with the oCJ23 primer and either primer oCJ22-2 (sulfate-grown sample), oCJ22-4 (methionine-grown sample), or oCJ22-6 (2SC-grown sample) as described previously (25). PCR products were electrophoresed on 2% agarose

Deconstruction of the oncometabolite S-(2-succino)cysteine
gels, the ϳ128-bp band of interest was excised, and the DNA was recovered in 15 l of 10 mM Tris-Cl, pH 8.5, with a DNA Gel Extraction Mini kit (Qiagen, Hilden, Germany). Sequencing was performed at the University of Florida NextGen DNA Sequencing facility (University of Florida Interdisciplinary Center for Biotechnology Research). Using the oCJ24 oligonucleotide to obtain the primary sequence and oCJ27 oligonucleotide to obtain the barcode sequences, we first used a MiSeq system (Illumina, San Diego, CA) for 1 ϫ 50 cycles to determine how much to load for optimum clustering. Sequencing was then performed for 50 cycles with a NextSeq 500 (Illumina) using a 1 ϫ 75 high-throughput v2 cartridge. Custom primers (25) were added to the standard Illumina primers to allow for use of PhiX (10%). Approximately 320 million "passing-filter" reads were generated per run. The data were trimmed with Illumina software to remove everything except the 16 -17-bp region corresponding to the transposon-genome junctions. Transposon insertion sites were mapped in the B. subtilis reference genome sequence (NC_000964.3) using 16-mer sequence tags. k-mer profiles (16-mer) were generated for sequence data sets from each of the three treatments, SO 4 , Met, and 2SC, using the count function of JELLYFISH (47). The k-mer count profiles were then queried with the reference genome sequence using the query function of JELLYFISH to determine relative frequencies and genome coordinates of insertions for each treatment. Insertion coordinates were then overlaid with the genome annotation (AL009126.gff) to identify genes that were differentially targeted in the three data sets.

Enzyme assays
Purified SnaA and SnaB proteins were assayed for acetyltransferase activity against 2SC using Ellman's reagent as described previously (48,49). Assays (50-l total volume) contained 50 mM Tris-Cl, pH 7.5, 0.4 mM acetyl-CoA, and between 8.0 and 0.001 mM 2SC and were started by adding either 8.8 g of SnaA or 4.0 ng of SnaB. After a 10-min incubation at 37°C, 50 l of a solution containing 50 mM Tris-Cl, pH 7.5, and 6 M guanidine HCl was added to stop the reaction. 50 l of Ellman's reagent (50 mM Tris-Cl, pH 7.5, 0.2 mM 5,5Ј-dithiobis(2-nitrobenzoic acid), and 1 mM EDTA) was added, and the mixture was incubated at 22°C for 10 min before measuring optical density at 412 nm. Assays were replicated at least three times. An extinction coefficient of 14,150 cm Ϫ1 M Ϫ1 for 3-carboxy-4-nitrobenzenethiol was used to calculate the amount of CoA formed. Kinetic parameters were calculated by fitting data to the Michaelis-Menten equation using GraphPad Prism Software (version 4.00 for Windows, GraphPad Software, San Diego, CA).

Metabolic profiling
Extractions were performed essentially as described (50). Briefly, 0.5 ml of ice-cold 80:20 (v/v) acetonitrile/water was added to each sample and shaken at 4°C for 15 min on an orbital shaker (Torrey Pines Scientific) at maximum speed and centrifuged at 14,000 ϫ g, and 0.5 ml of supernatant was transferred to a clean microcentrifuge tube. The extraction process was repeated to give a total cell extract volume of 1 ml. Aliquots (0.2 ml) of the cell extract were dried under vacuum at room temperature and frozen at Ϫ20°C. For metabolic profiling, samples were resuspended in 0.05 ml of acetonitrile/water (4:1, v/v) containing internal standards (e.g. 5 g/ml Val-Tyr-Val). 5 l was injected onto a hydrophobic interaction LC column (Waters Acquity UPLC BEH Amide, 150-mm length ϫ 2.1-mm inner diameter, 1.7-m particle size) with guard column (Acquity VanGuard BEH Amide precolumn, 5-mm length ϫ 2.1-mm inner diameter, 1.7-m particle size) held at 45°C on a Thermo Fisher Vanquish Focused ultra HPLC system. Mobile phase A was LC-MS-grade water with 10 mM ammonium formate and 0.125% formic acid. Mobile phase B was LC-MSgrade 95:5 (v/v) acetonitrile/water with 10 mM ammonium formate and 0.125% formic acid. Total run time per sample was 17 min with a flow rate of 0.4 ml/min. The gradient started at 100% B for 2 min; was brought to 70% B by 7.7 min, 40% B by 9.5 min, and back to 100% B by 12.75 min; and then was held at 100% B until 17 min. Mass spectra were recorded on a Thermo Fisher Q-Exactive HF mass spectrometer in positive mode between 60 and 900 m/z. Data were collected at resolutions of 60,000 for MS1 and 15,000 for MS/MS. MS/MS information was collected under data-dependent conditions for the top four ions per MS1 scan with one m/z isolation window and normalized collision Deconstruction of the oncometabolite S-(2-succino)cysteine energy of 20, 30, and 40. Automatic gain control target was set at one million ions for MS1 and 150,000 ions for MS/MS. Capillary temperature was set to 300°C, spray voltage was 3 kV, and auxiliary gas flow rate was set to 20. S- cysteine, N-acetyl-2SC, and N-acetylcysteine standards were used to identify and quantify these metabolites based on accurate mass, retention time, and MS/MS matching. Peak height was used for quantification. For each standard, an MS/MS spectrum was uploaded to the Mass Bank of North America (MONA) database. Open source software MS-DIAL (51) was used for peak deconvolution, peak picking, alignment, and MS/MS matching.