Accumulation of Succinyl Coenzyme A Perturbs the Methicillin-Resistant Staphylococcus aureus (MRSA) Succinylome and Is Associated with Increased Susceptibility to Beta-Lactam Antibiotics

ABSTRACT Penicillin binding protein 2a (PBP2a)-dependent resistance to β-lactam antibiotics in methicillin-resistant Staphylococcus aureus (MRSA) is regulated by the activity of the tricarboxylic acid (TCA) cycle via a poorly understood mechanism. We report that mutations in sucC and sucD, but not other TCA cycle enzymes, negatively impact β-lactam resistance without changing PBP2a expression. Increased intracellular levels of succinyl coenzyme A (succinyl-CoA) in the sucC mutant significantly perturbed lysine succinylation in the MRSA proteome. Suppressor mutations in sucA or sucB, responsible for succinyl-CoA biosynthesis, reversed sucC mutant phenotypes. The major autolysin (Atl) was the most succinylated protein in the proteome, and increased Atl succinylation in the sucC mutant was associated with loss of autolytic activity. Although PBP2a and PBP2 were also among the most succinylated proteins in the MRSA proteome, peptidoglycan architecture and cross-linking were unchanged in the sucC mutant. These data reveal that perturbation of the MRSA succinylome impacts two interconnected cell wall phenotypes, leading to repression of autolytic activity and increased susceptibility to β-lactam antibiotics.

antibiotics and reduces autolytic activity via perturbation of global lysine succinylation in MRSA.

RESULTS
TCA cycle genes sucC and sucD control resistance to b-lactam antibiotics in MRSA. A screen of the Nebraska Transposon Mutant Library (NTML) (28) revealed that the TCA cycle mutants, NE569 (sucC::Tn) and NE1770 (sucD::Tn) were more susceptible to cefoxitin (Fig. 1A and B) and oxacillin ( Fig. 1C; Table 1) compared to wild-type JE2. Other mutations in TCA cycle genes did not affect susceptibility to oxacillin (Table 1), specifically implicating sucCD-encoded succinyl-CoA synthetase in b-lactam resistance. The oxacillin MICs of the sucC and sucD mutants in Mueller-Hinton agar (MHA) were 2 to 4 mg/ml, compared to 32 to 64 mg/ml for JE2, as measured using M.I.C.Evaluator (Fig. 1C) and agar dilution assays (Table 1). Both mutants also exhibited reduced growth in Mueller-Hinton broth (MHB) in the absence of antibiotic (Fig. 1D). Phage 80a-mediated backcross of the sucC::Tn allele into JE2, USA300 FPR3757 (4), and clinical MRSA strain DAR173 (29,30) was accompanied by increased susceptibility to oxacillin ( Fig. 1E to G; Table 1). In addition, both NE569 and NE1770 were successfully complemented by the sucCD genes carried on plasmid pLI50 ( Table 1). The sucC and sucD genes are separated by only 21 bp, suggesting that they are organized in an operon (Fig. 1H) and that the transposon insertion in sucC is likely to have a polar effect on sucD expression. Consistent with this, NE569 was complemented by psucCD but not by plasmids carrying sucC (psucC) or sucD (psucD) alone (Table 1 and Fig. 1H). NE1770 was complemented by psucCD and partially complemented by psucD, but not by psucC (Table 1 and Fig. 1H).
HoR mutants of NE569 (sucC) have mutations in relA and relQ. Expression of the HoR phenotype is dependent on activation of the (p)ppGpp-mediated stringent response (31)(32)(33)(34). SucCD, succinyl-CoA synthetase, produces GTP, which is a substrate for (p)ppGpp synthesis by the RelA/SpoT homolog (RSH), RelP, and RelQ, raising the possibility that the NE569 (sucC) or NE1770 (sucD) mutations may negatively affect (p)ppGpp production. To investigate the possible relationship between the impact of sucCD mutations and the stringent response, the ability of JE2, NE569, and NE1770 to produce HoR mutants on brain heart infusion (BHI) agar supplemented with oxacillin 100 mg/ml was compared. Two stable NE569 HoR mutants, designated sucC HoR1 and sucC HoR2 with oxacillin MICs .256 mg/ml (see Fig. S1A in the supplemental material), were genome sequenced. sucC HoR1 had a single nucleotide deletion, frameshift mutation in relA (Table 2; see also Fig. S1B), while sucC HoR2 had a nonsynonymous point mutation, resulting in an A 178 V substitution in RelQ (Table 2; see also Fig. S1C). The N-terminus of RelA contains (p)ppGpp hydrolase and synthase domains, while mutations in the C-terminal domain deregulate synthase activity (35)(36)(37). The NTML relA mutant, NE1714, contains a transposon insertion in the C-terminal domain and is associated with increased b-lactam resistance, presumably due to increased (p)ppGpp synthase activity (Table 1). To investigate the impact of the relA::Tn mutation on b-lactam susceptibility in NE569, a relA sucC double mutant was constructed. First, the erythromycin resistance (Erm r ) marker in NE569 was swapped for a kanamycin resistance (Kan r ) marker, as described in Materials and Methods, to generate sucC::Tn-Kan r . The sucC::Kan r allele was then transduced into NE1714 using phage 80a. Similar to the sucC HoR mutants, the oxacillin MIC of the relA sucC double mutant was also .256 mg/ ml (Table 1). RelQ contains a (p)ppGpp synthase domain only and has previously been implicated in adaptation to vancomycin-and ampicillin-induced cell wall stress (38). A relQ mutant is not available in the NTML, indicating that this gene may be essential and that the RelQ A 178 V substitution in sucC HoR2 may enhance ppGpp synthase activity to promote a HoR phenotype. Given that previous studies have demonstrated the role of ppGpp synthase mutations in the HoR phenotype (5,38,39), these data suggest that increased b-lactam susceptibility in the sucCD mutants is not related to impaired GTP/ppGpp production. Mutations in sucA and sucB reverse NE569 (sucC) mutant phenotypes. Fastergrowing, more-pigmented suppressor mutants were readily observed among colonies of NE569 grown on MHA ( Fig. 2A). Three of these suppressor mutants were isolated and comparison of their genomic DNA sequences to NE569 identified mutations in either sucA or sucB ( Table 2). The sucA gene, which encodes 2-oxoglutarate dehydrogenase E1 subunit, is found in a two-gene operon with sucB, which encodes dihydrolipoyl succinyltransferase E2 subunit. Together, SucAB catalyze the synthesis of succinyl-CoA, the substrate for SucCD complex (Fig. 1A). A single nucleotide variant (SNV) leading to a Ser 9 STOP codon in sucA was present in sucC suppressor 1, sucC suppressor 2 had a SNV leading to a Ile 361 Thr substitution in SucB, and sucC suppressor 3 had a 46-bp deletion in sucB. The possible significance of additional SNVs in the tatD and hflX genes of sucC suppressors 1 and 2, respectively, is unknown. However, all three sucC suppressor mutants exhibited wild-type levels of oxacillin resistance (Table 1 and Fig. 2B), implicating the sucA and sucB mutations in this phenotype. The sucC suppressor 1 mutant, which was chosen for more detailed analysis, also exhibited wild-type growth in MHB (Fig. 2C). Similarly, in chemically defined media supplemented with glucose (CDMG), the NE569 growth defect was reversed by the sucA suppressor mutation (Fig. 2D). However, both NE569 and sucC suppressor 1 were unable to grow in chemically defined media lacking glucose (CDM) (Fig. 2E), which is consistent with previous results showing growth in CDM is dependent on an intact TCA cycle (9). Extension of these experiments to NTML mutants revealed no change in oxacillin susceptibility in NE547 (sucA) and NE1391 (sucB) ( Table 1). Using phage 80a to disrupt sucA in the sucC mutant restored wild-type colony morphology, oxacillin resistance, and growth ( Fig. 3A to C). For control purposes, a sucC sdhA double mutant was also constructed. The succinate dehydrogenase complex SdhAB catalyzes the conversion of the SucCD product, succinate, to fumarate (Fig. 1A). Mutations in sdhA or sdhB did not impact susceptibility to oxacillin (Table 1) and the sucC sdhA double mutant exhibited the same colony morphology, oxacillin resistance, and growth characteristics as the sucC mutant (Fig. 3A, B, and D). Taken together, these data demonstrate that mutations in sucA or sucB overcome the b-lactam susceptibility and growth defects of the sucC mutant.
Succinyl-CoA is significantly increased in the sucC mutant. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was used to quantify the accumulation of intracellular metabolites from the TCA cycle and the pyruvate node of glycolysis in NE569 (sucC), NE569 psucCD, and sucC suppressor strain 1 collected from late- exponential-phase cultures grown aerobically in MHB. Consistent with the predicted impact of a sucC mutation, succinyl-CoA levels were significantly increased in NE569 (Fig. 4). Furthermore, accumulation of succinyl-CoA in the sucC mutant was accompanied by a concomitant decrease in the levels of succinate (Fig. 4). Succinyl-CoA was reduced to wild-type levels in sucC suppressor 1 and the complemented sucC mutant, implicating accumulation of this metabolite in sucC-dependent modulation of b-lactam resistance. The levels of a-ketoglutarate were also significantly increased in sucC suppressor 1 compared to JE2, NE569, and NE569 psucCD (Fig. 4), supporting the conclusion that the SucA Ser 9 STOP mutation in this strain has impaired a-ketoglutarate dehydrogenase activity. Mutation of sucC also impacted the glycolytic pathway as evidenced by significantly reduced phosphoenol pyruvate (PEP) and increased levels of pyruvate (Fig. 4). The levels of acetyl-CoA were also significantly reduced in the sucC mutant ( Fig. 4), as were citrate and isocitrate, albeit not significantly, further demonstrating the impact of this mutation on TCA cycle homeostasis.
Mutation of sucC significantly impacts the global MRSA proteome. Similar to acetyl-CoA, succinyl-CoA can react with the NH 3 1 group located on the lysine side chain of proteins in general resulting in the succinylation of this residue in a pH-and concentration-dependent manner (21,40). This PTM has been shown to occur extensively in prokaryotes (22,23,(40)(41)(42)(43), and a recent study demonstrated that a murine succinate dehydrogenase mutation altered succinyl-lysine distribution in chromatin (44). Here, the accumulation of succinyl-CoA measured in the sucC mutant NE569 prompted us to investigate the global distribution of succinyl-lysines by analyzing the bacterial proteome and succinylome using LC-MS/MS.
For these experiments, cells were collected from JE2 and NE569 cultures grown to exponential and stationary phase as described in Materials and Methods. The global proteome analysis (the workflow is summarized in Fig. S2A in the supplemental material) quantified approximately 90% (n = 2,283) of the 2,607 predicted S. aureus proteins. After principal-component analysis and batch correction, JE2 early exponential and exponential-phase samples were found to be largely the same but significantly different from JE2 stationary-phase samples (see Fig. S2B). Compared to JE2, 381 proteins were significantly upregulated in NE569 during exponential growth (see Fig. S2C) and 330 proteins were upregulated in stationary phase (see Fig. S2D), including proteins involved in PG biosynthesis, cell wall organization, and cell division. Significantly downregulated proteins in NE569 during exponential-phase (307 proteins) and stationaryphase (187 proteins) growth included virulence regulators (e.g., SaeS, SaeR, SarR, SarZ, and AgrB), components of type VII secretion systems, hemolysins, leukotoxins, and immune evasion/inactivation proteins (e.g., Spa, Sbi, SraP, and PsmA1) (see Fig. S2C and D), suggesting that virulence of the sucC mutant may be attenuated.
Global patterns of protein succinylation were increased in the sucC mutant. In total 9,545 peptides with succinylated lysine residues (P , 0.05, A-Score . 13) (45,46) were identified. Peptides that could not be quantified in .75% of samples were disregarded, leaving 5,762 succinylated-lysine peptides derived from 1,000 unique proteins. The abundance of approximately 58% of the succinylated peptides (3,340) was significantly modulated across the conditions tested (ANOVA [analysis of variance] adjusted P , 0.05) (Fig. 5A). Hierarchical clustering revealed that while most of the changes in the succinylome were growth phase dependent, succinylated peptides in clusters 1 and 2 were more abundant in NE569 compared to JE2 in the stationary phase and were associated with "penicillin binding," "cytolysis," "cell wall," "amidase activity," "transferase activity transferring acyl-groups," "response to oxidative stress," and "metalloendopeptidase activity" (Fig. 5B).
The median number of succinylated peptides per protein was 3, and 670 proteins had ,5 succinylated peptides. Intriguingly, the top 40 most succinylated proteins (Fig. 5C) had .20 succinyl peptides each, which represented .20% of all succinylated peptides. Consistent with a recent study in S. epidermidis (23), numerous proteins involved in glycolysis and the TCA cycle were highly succinylated. The major autolysin (Atl) was the most succinylated protein in JE2 and NE569, with the mecA-encoded FIG 4 Mutation of sucC alters the central metabolism in S. aureus. JE2, NE569 (sucC), sucC comp (NE569 psucCD), and sucC supp (sucC suppressor 1, which has a SucA Ser 9 STOP mutation) strains were grown aerobically in MHB. The cells were harvested in the exponential phase (6 h), and intracellular metabolites associated with the pyruvate node and TCA cycle were analyzed by LC-MS/MS. The metabolites included in the analysis are indicated in the TCA cycle pathway, along with their associated enzymes. The slit arrow indicates a predicted break in the TCA cycle due to mutation of sucC in NE569. n = 3; cps, count per second. The image was created using Biorender.com. penicillin binding protein 2a (PBP2a) and PBP2 also among the 40 most succinylated proteins (Fig. 5C). Atl had 102 succinylated peptides quantified, mapping to 82 succinyl-lysine sites, of which 12 showed significantly higher levels of succinylation in the sucC mutant in the stationary phase (Student t test, P , 0.05) (Fig. 5D).
Mutation of sucC does not affect mecA transcription, PBP2a expression, or peptidoglycan structure and cross-linking. LightCycler RT-qPCR analysis revealed that the relative expression of mecA was not significantly affected in NE569 compared to JE2 in BHI media or in BHI media supplemented with oxacillin 0.5 mg/ml (Fig. 6A). Western blotting also revealed similar PBP2a levels in JE2, NE569, and NE569 psucCD and NE569 (sucC) grown to exponential phase in BHI or BHI supplemented with 0.5 mg/ml oxacillin. Experiments were repeated at least three times, and standard deviations are shown. Student t test analysis revealed no significant differences between either strain or in the presence of absence of oxacillin. (B) Western blotting of PBP2a in JE2, NE569 (sucC), NE569 psucCD, and MSSA strain 8325-4 (negative control). Total protein was extracted from cells collected during the exponential phase of growth in MHB plus 2% NaCl supplemented with 0.5 mg/ml oxacillin, with the exception of 8325-4, which was grown without oxacillin. A portion (8 mg) of total protein was separated on a 7.5% Trisglycine gel, transferred to a polyvinylidene difluoride membrane, and probed with anti-PBP2a antibody (1:1,000 dilution), followed by horseradish peroxidase-conjugated protein G (1:2,000 dilution) and (Continued on next page) Campbell et al. ® grown in MHB with 2% NaCl at 35°C supplemented with 0.5 mg/ml oxacillin (Fig. 6B). MSSA strain 8325-4 was included as a mecA-negative control.
Quantitative PG compositional analysis was performed using UPLC analysis of muramidase-digested muropeptide fragments extracted from exponential-or stationaryphase cultures of JE2, NE569, and sucC suppressor 1 grown in MHB or MHB supplemented with oxacillin 3 mg/ml or 32 mg/ml. The PG profiles of all three strains were similar under all growth conditions tested (see Fig. S3 in the supplemental material). Supplementation of MHB with oxacillin was associated with significant changes in muropeptide oligomerization and reduced cross-linking, but these effects were the same in all strains (Fig. 6C). Interestingly, although PBP2a is the second most succinylated protein in the S. aureus proteome (Fig. 5C) and contains 40 succinylated lysine residues, only the lysine residue K 47 , which is not located near to the enzyme active site or dimerization domains, showed increased succinylation in the sucC mutant (MassIVE ID MSV000086976 and MassIVE ID MSV000086971), perhaps making it unlikely that this PTM influences the transpeptidase activity of this enzyme in NE569.
Mutation of sucC does not impact susceptibility to the lipoteichoic acid synthase inhibitor Congo red or the alanylation inhibitor D-cycloserine. Experiments comparing the susceptibility of JE2 and NE569 to Congo red revealed no differences as evidenced by similar CFU counts, although the morphology of the sucC mutant was drastically changed (see Fig. S4). Similarly, the D-cycloserine (DCS) MIC for JE2 (32 mg/ ml) was not significantly different from NE569, NE569 psucCD, or sucC suppressor 1, all of which had DCS MICs of 16 to 32 mg/ml. Congo red is a selective inhibitor of lipoteichoic acid synthase (LtaS) activity (47), while DCS blocks the D-alanine racemase and D-alanine ligase enzymes required for the production of D-alanine (48,49), an important precursor for PG, wall teichoic acid (WTA), and lipoteichoic acid (LTA) biosynthesis. Our data already showed that PG structure was unaffected by the sucC mutation (Fig. 6C), and these observations further suggest that altered expression or stability of WTA and LTA may not be involved in the reduced b-lactam susceptibility of NE569.
Autolytic activity is impaired in the sucC mutant. The 12 lysine residues in Atl that exhibited increased levels of succinylation in the sucC mutant were evenly distributed throughout the protein, including within the amidase and glucosaminidase domains (Fig. 5D). To investigate whether altered succinylation of Atl impacted its enzymatic activity, Triton X-100-induced autolysis was compared in JE2, NE569, NE569 psucCD, sucC suppressor 1, the sucC sucA double mutant, and the NE460 (atl) mutant. Autolytic activity was strikingly reduced in the sucC mutant compared to JE2, and this phenotype was complemented to wild-type levels in NE569 psucCD and reversed in the sucC/sucA and the sucA Ser 9 STOP suppressor mutant strains (Fig. 7). Indeed, Triton X-100-induced autolysis was similar in NE569 and NE460 (Fig. 7). These data suggest that the increased succinylation of the 12 lysine residues in Atl is associated with blocked autolytic activity by interfering with proteolytic cleavage of Atl or the activity of the amidase or glucosaminidase PG hydrolases. Although mutation of atl in several MRSA strains, including the HoR strain COL, was previously associated with reduced resistance to methicillin (50), the oxacillin MIC of NE460 (32 mg/ml) was similar to the parent strain JE2. Mutation of Sle1, which like Atl is required for daughter cell separation after cell division, has previously been reported to reduce autolytic activity (51) and increase MRSA susceptibility to oxacillin (52). Succinyl-PTM of the Sle1 K 200 residue was significantly increased in NE569 compared to JE2 (Student t test, P , 0.05) (Fig. 7B). (sucC), and sucC suppressor 1 grown to exponential phase in MHB or MHB supplemented with oxacillin at 3 or 32 mg/ml. PG analysis from NE569 (sucC) is shown only at 3mg/ml oxacillin because 32mg/ml exceeds its MIC. Each profile shown is a representative of three biological replicates. Significant differences were determined using a Student t test (**, P , 0.01; ***, P , 0.001; ****, P , 0.0001).
sucCD-Dependent Control of b-Lactam Resistance in MRSA ® Overall, our data reveal that accumulation of succinyl-CoA in the sucC mutant negatively impacts two interconnected cell wall-associated phenotypes: autolysis and susceptibility to b-lactam antibiotics.

DISCUSSION
The TCA cycle is centrally involved in the production of biosynthetic precursors, reducing potential and energy. Here, we report that mutations in sucC and sucD genes encoding the a and b subunits of succinyl-CoA synthetase, which catalyzes the conversion of succinyl-CoA to succinate, significantly increased susceptibility to b-lactams. The sucC and sucD mutants grew as smaller, less-pigmented colonies on MHA and exhibited impaired growth in MHB. Genetically blocking the production of succinyl-CoA in the sucC mutant by mutating sucA or sucB reversed the growth and b-lactam susceptibility phenotypes. In contrast, mutation of sdhA in the sucC mutant had no phenotypic impact. Succinyl-CoA levels were significantly increased in the sucC mutant and were restored to wild-type levels by psucCD complementation or mutation of sucA.
The accumulation of succinyl-CoA in the sucC mutant perturbed global protein succinylation, which is an important PTM previously described in several pathogens (21,22,43), including S. epidermidis (23). Although several PBPs, including mecA-encoded PBP2a, were among the proteins with the highest number of succinyl-lysines, PG architecture and cross-linking were unchanged in the sucC mutant, even under oxacillin stress when mecA expression is increased. The absence of structural changes in PG also indicates that the accumulation of succinyl-CoA does not impact b-lactam resistance via altered biosynthesis of lysine, which is an important component of the cell wall (53). Succinyl-CoA is used in the biosynthesis of lysine from aspartate and in Corynebacterium glutamicum disruption of the sucCD locus, and the resulting accumulation of succinyl-CoA was accompanied by overproduction of lysine (54). It is difficult to envisage how increased lysine accumulation would reduce b-lactam resistance, and in any event the unchanged PG structure in the sucC mutant does not implicate lysine biosynthesis in this phenotype. Succinyl-CoA synthetase activity generates GTP, which is a substrate for RelA, RelP, and RelQ enzymes that produce the stringent response alarmone (p)ppGpp. (p)ppGpp plays a central role in the control of MRSA b-lactam resistance (31)(32)(33)(34). However, the ability of the sucC mutant to produce stable HoRs with mutations in relA or relQ suggests that intracellular GTP is not limited in the sucC and sucD mutants or associated with reduced b-lactam resistance. Metabolomic analysis further revealed significantly reduced levels of acetyl-CoA in the sucC mutant, raising the additional possibility that the acetylome may also have been perturbed. In this context, changes in PG acetylation have also been implicated in autolysis and antibiotic resistance (55), and future comparison of the acetylome and succinylome in sucCD mutants to identify proteins that are modified by both PTMs may provide further insights into how these mutations impact resistance.
The reduced pigmentation of the sucCD mutants may be a consequence of altered production of the S. aureus carotenoid staphyloxanthin, composed of a glucose residue esterified with a 30-carbon carboxylic acid chain and a 15-carbon fatty acid (56). In S. aureus most fatty acids are odd-numbered branched-chain fatty acids (57). The b-oxidation of odd-numbered fatty acids generates acetyl-CoA and propionyl-CoA, the latter of which can be converted to succinyl-CoA. Interestingly, staphyloxanthin-derived lipids interact with flotillin to form functional membrane microdomains required for oligomerization and activity of PBP2a (58). In contrast to the reduced pigmentation, the proteomic analysis revealed increased levels of staphyloxanthin biosynthetic enzymes in the sucC mutant (MassIVE ID MSV000086976 and MassIVE ID MSV000086971), perhaps reflecting efforts by the sucC mutant to compensate for reduced pigmentation.
The discovery that Atl was the most succinylated protein in the MRSA proteome and that 12 of the 82 succinyl-lysines were significantly more succinylated in the sucC mutant suggested a potential connection to reduced b-lactam resistance. Fisher and Mobashery recently proposed a model in which the bactericidal activity of b-lactams is the result of deregulated Atl activity at the cell division septum (59). Our data showing that autolytic activity was significantly reduced in the sucC mutant and that the oxacillin MIC of the atl transposon mutant NE460 was unchanged are not consistent with this possibility. Furthermore, previous work in our laboratory linked increased autolytic activity with increased b-lactam resistance. Specifically, atl transcription was activated in a HoR mutant of USA300 LAC (oxacillin MIC . 256 mg/ml), which exhibited significantly increased autolytic activity, and growth of USA300 LAC in sub-MIC oxacillin was also associated with significantly increased autolysis (60). Thus, while it seems clear that autolysis and b-lactam susceptibility are interconnected, the precise mechanistic interactions between these two phenotypes needs to be elucidated further.
Processing of the Atl proprotein, produces a signal peptide, a propeptide, a N-acetylmuramoyl-L-alanine amidase (AM) enzyme, and a C-terminally located endo-b-Nacetylglucosaminidase enzyme (GL) (61). The region between the AM and GL catalytic domains contains three repeat regions (R1 to R3) with GW-dipeptide motifs required to target Atl proprotein to the equatorial ring on the cell surface during cell division (62). None of these 12 lysine residues exhibiting increased succinylation are in the AM catalytic domain, 5 are in the GL catalytic domain, 2 are in the propeptide region, and the remaining 5 are in the R1 and R2 regions. The single lysine residue (K 200 ) of Sle1 that is more succinylated in the sucC mutant is located in a LysM cell wall hydrolase domain and may be important for activity of the enzyme.
Susceptibility to the LtaS inhibitor Congo red was unchanged in the sucC mutant, indicating that increased b-lactam susceptibility may not be associated with impaired expression or stability of WTA or LTA. Similarly, susceptibility to the alanylation inhibitor DCS was the same in the wild-type and sucC mutant. Given that D-alanine is an important component of PG, this observation is consistent with the absence of any changes in PG structure in the sucC mutant. Furthermore, because D-alanine is an important component of WTA and LTA, unchanged susceptibility to DCS does not point to roles for these cell envelope glycopolymers in sucC-dependent b-lactam susceptibility. Nevertheless, given that almost 58% of all quantifiable succinylated peptides were significantly changed in the sucC mutant, we propose a model in which perturbation of the succinylome likely modulates the activity of multiple enzymes, including Atl and Sle1, that collectively control growth and interconnected cell envelope characteristics such as autolysis and susceptibility to b-lactam antibiotics (Fig. 8). The U.S. Food and Drug Administration-approved anticancer drug streptozotocin specifically targets succinyl-CoA synthetase in human cells to limit proliferation (63) and is used primarily to treat tumors that cannot be surgically removed. The findings described here may open the door to the possibility of sensitizing MRSA to b-lactam antibiotics using compounds that specifically target succinyl-CoA synthetase or protein succinylation generally within the cell.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. Escherichia coli strains were cultured in Luria-Bertani (LB) broth or LB agar. S. aureus strains were grown in Mueller-Hinton broth (MHB), Mueller-Hinton agar (MHA), tryptic soy broth (TSB), brain heart infusion (BHI), tryptic soy agar (TSA), chemically defined medium (CDM), and CDM supplemented with glucose (CDMG) and, where indicated, supplemented with erythromycin (Erm) at 10 mg/ml, chloramphenicol (Cm) at 10 mg/ml, ampicillin (Amp) at 50 mg/ml, or kanamycin (Km) at 75 mg/ml. For growth experiments in MHB, 25 or 50 ml of MHB in 250-ml flasks were used for a 10:1 or 5:1 flask/volume ratio, respectively. Overnight cultures in MHB were used to inoculate the media at a starting optical density at 600 nm (OD 600 ) of 0.01, and flasks were incubated at 37°C shaking at 200 rpm. For CDM and CDMG growth experiments, overnight cultures were grown in TSB, washed once in 5 ml of phosphate-buffered saline (PBS), and used to inoculate 25 ml of CDMG or CDM in a 250-ml flask to a starting OD 600 of 0.05. Flasks were incubated at 37°C with shaking at 200 rpm. For all growth experiments, CFU were enumerated in serially diluted 20-ml aliquots removed from flask cultures. At least three biological replicates were performed for each strain, and average data are presented. For experiments to compare growth in MH and in MH supplemented with 5% glucose, cultures were grown at 37°C in 250-ml flasks with 50 ml of media and shaking at 200 rpm.
Cefoxitin disk diffusion assays and MIC measurements. Cefoxitin disk diffusion susceptibility testing was performed in accordance with Clinical and Laboratory Standards Institute (CLSI) guidelines (64). MIC measurements by broth microdilution or agar dilution were performed in accordance with CLSI methods for dilution susceptibility testing of staphylococci (65). Disk diffusion and MIC results were interpreted using CLSI standard M100, and strains were classified as susceptible or resistant (66). For oxacillin MIC measurement, M.I.C.Evaluator (Oxoid) strips were used in accordance with manufacturer guidelines. Strains were grown at 37°C on MHA for 24 h, 5 to 6 colonies were resuspended in 0.85% saline and adjusted to a 0.5 McFarland standard. The suspension was swabbed evenly three times across the surface of an MHA 2% NaCl plate (4-mm agar depth). An M.I.C.Evaluator strip was applied, followed by incubation for 24 h at 35°C. Three independent measurements were performed for each strain. Genomic DNA extraction and whole-genome sequencing. Genomic DNA extraction, sequencing by MicrobesNG (http://www.microbesng.uk), and analysis were performed as described previously (67). Wild-type JE2 reads were aligned against the published USA300 FPR3757 genome sequence (RefSeq accession number NC_007793.1) and assembled into a contig. USA300 FPR3757 annotation and base numbering were transferred onto the wild-type JE2 sequence. The reads for sucC suppressor strains and sucC HoR strains were then mapped onto the assembled JE2 sequence. High-frequency (70%) and goodquality base changes were identified using the CLC Genomics Workbench software package (version 20.0.2). Sequence data for NE569 (sucC), sucC HoR1, sucC HoR2, and sucC suppressors 1, 2, and 3 are available from the European Nucleotide Archive (project PRJEB43960, accession numbers ERS6142066 to ERS6142071).
Genetic manipulation of S. aureus. Phage 80a was used to transduce the transposon insertion from NE569 into JE2, DAR173, and USA300 to ensure that background mutations were not responsible for the antibiotic resistance phenotype, as described previously (67). Transposon insertions were verified using PCR amplification of target loci.
A 1,680-bp fragment encompassing the sucC gene and a 2,610-bp fragment, including both sucC and sucD, were PCR amplified from JE2 genomic DNA using the primers INF_sucC_F/INF_sucC_R and INF_sucCD_F/INF_sucCD_R, respectively (see Table S2 in the supplemental material) and cloned into the E. coli-Staphylococcus shuttle vector, pLI50, using Clontech infusion cloning kit 2. The psucCD plasmid was digested with HpaI and SpeI and treated with T4 DNA polymerase (Roche) and deoxynucleoside triphosphates to create a blunt-end fragment that was religated using T4 ligase (Roche). The resulting plasmid, designated psucD, contained a 627-bp deletion at the 59 end of the sucC gene, with only the sucD gene remaining intact. Recombinant plasmids were first transformed into cold-competent E. coli HST08 (supplied with the Clontech infusion cloning kit) before being transformed by electroporation into the restriction-deficient S. aureus strain RN4220 and finally into NE569 and NE1770.
The sucC sucA, sucC sdhA, and sucC relA double mutants were generated by first exchanging the sucC::Tn containing an Erm r cassette in NE569 for a transposon containing a Kan r cassette, generating strain sucC::Tn-Kan r . This was performed by allelic exchange, using the plasmid pKAN and a method described previously (68). Genomic DNA extraction and PCR using the primers suc_F and sucC_R was carried out to verify the allelic exchange process.
To construct the sucC relA double mutant, the sucC::Tn-Kan r was transduced into NE1714 and transductants were selected on TSA with 75 mg/ml kanamycin. To construct sucC sucA and sucC sdhA double mutants, the sucA::Tn allele from NE547 and the sdhA::Tn allele of NE626 were each transduced into sucC::Tn-Kan r using phage 80a, and transductants were selected on TSA with 10 mg/ml erythromycin. PCR was used to verify the transposon insertions in the double mutants using primers for sucC, relA, sucA, and sdhA (Table S2).
RNA purification and real-time RT-PCR. Cultures were grown in BHI media to midexponential phase. Harvested cells were pelleted and immediately stored at -20°C in RNAlater (Ambion) to ensure maintenance of RNA integrity. RNA was extracted according to the manufacturer's guidelines using an RNA Mini-Extraction kit (Sigma). RNA integrity was examined visually by agarose gel electrophoresis and RNA concentration was determined using a Qubit Fluorometer 4 (Qiagen). Quantitative reverse transcription-PCR (RT-qPCR) was used to measure mecA transcription on the Roche LightCycler 480 instrument using a LightCycler 480 SYBR green kit (Roche) with the primers mecA1_Fwd and mecA1_Rev (see Table S2), as described previously (49). The gyrB gene amplified with the primers gyrB_Fwd and gyrB_Rev (see Table S2) served as an internal standard. Each RT-qPCR experiment was performed three times, and average data are presented, including standard deviations.
PBP2a Western blot analysis. Overnight MHB cultures were used to inoculate 25 ml of MHB plus 2% NaCl, with or without 0.5 mg/ml oxacillin, to a starting OD 600 of 0.05, followed by incubation at 35°C (with 200-rpm shaking) until an OD 600 of 0.8 was reached. The cells were then pelleted and resuspended in PBS to an OD 600 of 10, and PBP2a Western blot analyses were performed as described previously (67). Three independent experiments were performed and a representative blot presented.
Isolation of sucC suppressor mutants. For experiments comparing the growth of JE2 and the sucC mutant NE569, aliquots removed from 25-ml MHB cultures grown in 250-ml flasks were serially diluted, and the CFU were enumerated on MHA. Suppressor mutants of NE569 were readily identified on MHA due to their larger colony size and deeper pigmentation, which contrasted with the smaller, pale colony morphology of parental NE569. Three suppressor mutants (1, 2, and 3) were chosen for further analysis. DNA extraction and PCR verification of the Bursa aurealis transposon using the primers sucC_F and sucC_R (see Table S2) was performed to verify the presence of the sucC transposon insertion. Comparative genome sequence analysis was used to identify the suppressor mutations.
Oxacillin resistance population analysis. Population analysis profiles (PAPs) was generated as described previously (69). Overnight cultures were grown in BHI, adjusted to an OD 600 of 1, and 10-fold serially diluted from 10 21 to 10 27 , and a 20 ml aliquot of each dilution was plated onto a series of BHI agar plates supplemented with oxacillin at 0. 25, 0.5, 1, 2, 4, 8, 16, 32, 64, or 128 mg/ml. The CFU were enumerated after overnight incubation at 37°C, and the results were expressed as the CFU/ml at each oxacillin concentration. Average data from three independent experiments are presented.
Isolation of homogeneously resistant mutants. Overnight cultures of NE569 were grown in BHI, adjusted to an OD 600 of 1, serially diluted, plated onto BHI agar supplemented with 100 mg/ml oxacillin, and incubated at 37°C to isolate HoR mutants. The CFU were enumerated on BHI agar to calculate the rate of HoR mutant production. HoR mutants were passaged for 14 days in antibiotic-free BHI broth to identify stable mutants that were then verified using PAPs and oxacillin MIC measurements. The experiments were performed twice, and the results of a representative experiment are presented.
sucCD-Dependent Control of b-Lactam Resistance in MRSA ® LC-MS/MS metabolite analysis of NE569 (sucC), NE569 psucCD, and sucC suppressor 1. For LC-MS/MS analysis, samples were prepared as previously described (70). Briefly, strains were inoculated in MHB to an OD 600 of 0.06 and grown aerobically (250 rpm, 37°C) for 6 h. Culture volumes corresponding to OD 600 of 10 were harvested and rapidly filtered through a membrane (0.45 mm; Millipore). The cells on the membrane were washed twice with 5 ml of cold saline and immediately quenched in ice-cold 60% ethanol containing 2 mM Br-ATP as an internal control. The cells were mechanically disrupted using a bead homogenizer set to oscillate for three cycles (30 s) of 6,800 rpm with a 10-s pause between each cycle. Cell debris was separated by centrifugation at 13,000 rpm. The supernatant containing intracellular metabolites were lyophilized and then stored at 280°C.
A triple-quadrupole-ion trap hybrid mass spectrometry (Sciex) connected with a Waters ultraperformance liquid chromatography I-class (UPLC) system was used for the metabolite analysis. The chromatographic separation was performed by ionic liquid chromatography using a Column XSELECT HSS XP (150 mm Â 2.1 mm inner diameter; 2.5-mm particle size), and a binary solvent system with a flow rate of 0.250 ml/min was used for chromatographic separation. Mobile phase A was composed of 10 mM tributylamine, 10 mM acetic acid, 5% methanol, and 2% 2-propanol; mobile phase B was 100% methanol. The column was maintained at 40°C, and the autosampler was maintained at 7°C. The A/B solvent ratio was maintained at 90/10 for 1 min, followed by a gradual increase in B to 65% for 10 min. Solvent B was increased to 90% over next 1 min, which was maintained for 4 min. The gradient was again reduced to 90/10 (A/B) within 0.5 min and was equilibrated for 5.5 min before the next run. A QTRAP 65001 mass spectrometry system (Sciex) operated in negative-ion mode was used for targeted quantitation in multiple reaction monitoring (MRM) mode. MRM details for each analyte are listed in Table S3 in the supplemental material. The electrospray ionization parameters were optimized for a 0.25-ml/min flow rate and were as follows: an electrospray ion voltage of 24,400 V, a source temperature of 400°C, a curtain gas of 40, and gas 1 and 2 of 40 and 45 lb/in 2 , respectively. Analyzer parameters were optimized for each compound using manual tuning.
Proteomics sample preparation and analysis. Cells were collected from four biological replicates of JE2 and NE569 (sucC) flask cultures grown to exponential and stationary phases in MHB. Due to different growth characteristics and CFU/ml in wild-type and sucC mutant cultures (see Fig. S1), JE2 and NE569 cells were collected after 3 and 4 h, respectively, for exponential growth and after 10 and 12 h, respectively, for stationary-phase growth. As an added control four biological replicates of JE2 cells were also collected after 3 h (early exponential phase). Cell pellets were collected promptly from culture samples and snap-frozen in liquid nitrogen.
The cell pellets from all 20 samples were lysed, denatured, and digested with trypsin before being labeled with tandem mass tags (TMT) using TMT11plex, as described previously (71). Because there were 20 samples and only 11 channels in the TMT11plex kit, an equal concentration of peptides from all the cell pellets was mixed to generate a reference sample used for normalization purposes. Two samples from each cell pellet were labeled with different isobaric labels and placed in each of the two TMT sets, the channel 131C was reserved for the reference sample. The peptide samples were cleaned-up by C 18 solid-phase extraction (SPE). An immunoaffinity purification was performed using the PTMScan acetyl-lysine motif (Ac-K) kit (Cell Signaling Technology, Danvers, MA) to bind acetyl peptides, thereby enriching succinyl-lysine peptides in the unbound fraction, which was cleaned-up using a C 18 SPE and fractionated (into 12 fractions) using high-pH reverse-phase chromatography as described previously (72). Peptides were analyzed by reverse-phase separation (C 1 8) coupled with a QExactive HF-X mass spectrometer. The instrument .raw files generated were deposited and are available at MassIVE data repository (MassIVE ID MSV000086976 and MassIVE ID MSV000086971). Raw MS data were searched with MS-GF1 (73) against UniProt/Swiss-Prot S. aureus database (UP000001939) bovine trypsin and human keratin sequences. Methionine oxidation and succinylation were set as dynamic modifications, and cysteine alkylation and TMT labeling of N termini and lysines were set as static modifications. The identified spectra were filtered based on their MS-GF1 scores, resulting in a false discovery rate of ,1%. For quantitative analysis, the TMT reporter ion intensities were extracted with MASIC (74). Intensities were normalized to reference channel intensity. Analytes were considered for quantification when reporters were observed for at least three-quarters of the samples per given condition. The subsequent data were processed using the package RomicsProcessor (http://doi.org/10.5281/zenodo.3956544). Briefly, the data were log 2 transformed, median centered, and batch corrected using the sva ComBat method (75), before missing values were imputed using random values drawn from a normal distribution downshifted by 2 standard deviations with a width of 0.5 standard deviation similar to the Perseus imputation method (76). ANOVA and Student t tests were used to determine significance. The DAVID modified Fisher exact test (EASE score) (77) was used to measure significant enrichment of Gene Ontology (GO) terms in proteins whose abundance was altered by the sucC mutation.
Peptidoglycan analysis. Wild-type JE2, NE569, and sucC suppressor 1 were grown in MHB and in MHB supplemented with oxacillin at 3 or 32 mg/ml. For each strain and growth condition tested, independent quadruplicate 50-ml cultures were grown in flasks at 37°C with 200 rpm shaking to an OD 600 of 0.5, and cell pellets were collected promptly and snap-frozen in liquid nitrogen. PG was extracted from the samples as described previously (49,78). This analysis of wild-type JE2 PG was also used as a part of a separate study (67) and is reported again here for comparison to NE569 and sucC suppressor 1. Mass spectrometry (MS) was performed on a Waters XevoG2-XS QTof mass spectrometer. Structural characterization of muropeptides was determined based on their MS data and MS/MS fragmentation pattern, matched with the PG composition and structure reported previously (79)(80)(81)(82).
Autolytic activity assays. Samples (200 ml) from overnight cultures were inoculated into 20 ml of TSB, grown at 37°C (200 rpm) to an OD 600 of 0.5, washed with 20 ml of cold PBS, resuspended in 5 ml of cold PBS, and then adjusted to an OD 600 of 1. Then, 1 ml of the cell suspension was transferred to a cuvette, and Triton X-100 was added at a final concentration of 0.1% (vol/vol). The initial OD 600 was recorded before incubation at 37°C with shaking (200 rpm). Thereafter, the OD 600 was recorded every 30 min for 4 h, and autolytic activity is expressed as a percentage of the initial OD 600 . NE460 (atl::Tn) was used as a control, and at least three biological replicates were performed for each strain.
Data availability. Proteomic .raw files generated during this study are available at MassIVE data repository (MassIVE ID MSV000086976 and MassIVE ID MSV000086971). The UniProt/Swiss-Prot S. aureus database (UP000001939) was used as a reference for the proteomic analysis.
Whole-genome sequence data are available from the European Nucleotide Archive (project PRJEB43960, accession numbers ERS6142066 to ERS6142071). The SAUSA300_FRP3757 (TaxID:451515) reference genome sequence is available from the National Center for Biotechnology Information.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.   The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.