Proteomic and Metabolomic Analyses of a Tea-Tree Oil-Selected Staphylococcus aureus Small Colony Variant

Tea tree oil (TTO) is hypothesized to kill bacteria by indiscriminately denaturing membrane and protein structures. A Staphylococcus aureus small colony variant (SCV) selected with TTO (SH1000-TTORS-1) demonstrated slowed growth, reduced susceptibility to TTO, a diminutive cell size, and a thinned cell wall. Utilizing a proteomics and metabolomics approach, we have now revealed that the TTO-selected SCV mutant demonstrated defective fatty acid synthesis, an alteration in the expression of genes and metabolites associated with central metabolism, the induction of a general stress response, and a reduction of proteins critical for active growth and translation. SH1000-TTORS-1 also demonstrated an increase in amino acid accumulation and a decrease in sugar content. The reduction in glycolytic pathway proteins and sugar levels indicated that carbon flow through glycolysis and gluconeogenesis is reduced in SH1000-TTORS-1. The increase in amino acid accumulation coincides with the reduced production of translation-specific proteins and the induction of proteins associated with the stringent response. The decrease in sugar content likely deactivates catabolite repression and the increased amino acid pool observed in SH1000-TTORS-1 represents a potential energy and carbon source which could maintain carbon flow though the tricarboxylic acid (TCA) cycle. It is noteworthy that processes that contribute to the production of the TTO targets (proteins and membrane) are reduced in SH1000-TTORS-1. This is one of a few studies describing a mechanism that bacteria utilize to withstand the action of an antiseptic which is thought to inactivate multiple cellular targets.


Introduction
Staphylococcus aureus is a leading cause of hospital-acquired infections that demonstrates a propensity for acquiring resistance to antimicrobials [1]. One mechanism that enhances the ability of S. aureus to cause infections and tolerate antimicrobial challenge is the ability to produce small colony variants (SCVs) [2]. S. aureus SCVs are typically characterized by slow growth, reduced electron transport chain activity, decreased virulence factor production, reduced susceptibility to antimicrobials, and altered metabolism [3,4]. A number of transcriptional profiling and proteomic analysis studies have revealed that S. aureus SCVs display an altered expression of genes and proteins involved with both glycolysis and the tricarboxylic acid (TCA) cycle [5][6][7][8][9].

Locus ID Gene Symbol Function
Fold-Change in Gene Expression in SH1000-TTORS-1
The fatty acid biosynthesis inhibitor triclosan selected SCVs that harbored mutations in one or more genes encoding FabI and FabD, a teicoplanin resistance-associated protein, and a NADH-dependent flavin oxidoreductase [42]. Triclosan-selected SCVs were also reported to produce higher concentrations of FabF, as well as FabI and FabH (β-ketoacyl-ACP synthase); however, triclosan-selected SCVs demonstrated decreased concentrations of FabD [42].
Uncharacterized proteins SAOUHSC_01348 and SAOUHSC_2604 were increased in relative concentrations in SH1000-TTORS-1. SAOUHSC_01348 exhibited 50% amino acid identity with the Bacillus subtilis acyl-CoA thioesterase YneP, that is suspected to be involved with fatty acid degradation [43]. It has been reported however that under laboratory conditions, S. aureus cannot degrade fatty acids [44]. SAOUHSC_02604 is an uncharacterized oxidoreductase that has a conserved FabI superfamily domain and demonstrates 26% protein identity along the entire length of the S. aureus enoyl-ACP-reductase FabI, which is required for fatty acid biosynthesis [45].
Growth of the triclosan-selected SCVs on media containing the fatty acid supplement Tween 80 led to increased colony size, which demonstrated that the SCV mutants were deficient in fatty acid biosynthesis [42].

Proteins Associated with Stress and Stringent Response Are Altered in SH1000-TTORS-1
The general stress response RNA polymerase sigma factor SigB is intimately involved with the general stress response of S. aureus, and plays an essential role in the formation of SCVs [47,48]. RsbU is a phosphatase that positively regulates SigB activity by releasing SigB from the anti-sigma factor RsbW [49]. The relative concentrations of both SigB and RsbU were both increased in SH1000-TTORS-1 (Table 1). In addition, a number of proteins produced by genes controlled by SigB were also altered in SH1000-TTORS-1. The genes that produce Cap5O (UDP-N-acetyl-D-mannosaminuronic acid dehydrogenase) and CidC are positively regulated by SigB [50], Cap50 was only detected in SH1000-TTORS-1, and CidC was increased in SH1000-TTORS-1 ( Table 1). The genes that produce PycA, Hel (5 -nucleotidase), and SAOUHSC_02820 are negatively regulated by SigB [50], and the relative concentration of the proteins produced from these genes were decreased in SH1000-TTORS-1 ( Table 1).
The stringent response in S. aureus is activated when nutrients such as amino acids are limited, which results in the accumulation of the alarmone guanosine 3 , 5 -bisdiphosphate and decreased expression of genes encoding products involved with protein biosynthesis [51]. RelA is an alarmone synthetase required for the stringent response that also demonstrates guanosine 3 , 5 -bisdiphosphate hydrolysis activity [51]. S. aureus has two additional (p)ppGpp synthetases, RelP and RelQ, that both lack the RelA C-terminal hydrolyzing domain [52]. The relative concentration of RelQ was increased in SH1000-TTORS-1 compared to SH1000 (Table 1). This finding suggests that a component that controls the stringent response in increased in SH1000-TTORS-1, supporting the suggestion that the stringent response is partially activated in this mutant.
Additional proteins previously reported to be reduced during the stringent response [53] were also reduced in SH1000-TTORS-1 compared to SH1000. These proteins included: PdhD, GpmI, SAOUHSC_02574 (an uncharacterized NAD/NADP octopine/nopaline dehydrogenase alpha-helical domain), TrxB (a thioredoxin reductase), and TypA (a GTP-binding protein) ( Table 1). The protein SAOUHSC_02665 which encodes an uncharacterized general stress protein induced during a stringent response [53] was also increased in SH1000-TTORS-1 ( Table 1). The altered expression of these proteins could be linked to the increased expression of RelQ in SH1000-TTORS-1.

Additional Proteins Altered in SH1000-TTORS-1
The relative concentration of a number of proteins involved with translation were decreased in SH1000-TTORS-1. These proteins included: PheT (phenylalanyl-tRNA synthetase subunit beta), TrpS (trytophanol tRNA synthetase), MetG (methionyl-tRNA synthetase), and RplY (50S ribosomal protein L25/general stress protein) ( Table 1). MetG is responsible for the production of methionyl-tRNA which when formylated can be used to initiate translation [54]. The methionine aminopeptidase protein (Map) is then responsible for cleaving the N-terminal methionine following protein production [54], and the relative concentration of Map was also reduced in SH1000-TTORS-1 (Table 1).

Metabolomics Analysis
A total of 105 metabolites were identified in both strains: 16 amines and polyamines, 29 amino acids, 38 polar organic acids, and 22 sugars (Table S2). Out of these metabolites, 35 metabolites were found to be significantly altered in SH1000-TTORS-1 compared to SH1000, three metabolites were only identified in SH1000, and four metabolites were only identified in SH1000-TTORS-1 (Table 3; Table 4).
Eleven amino acids were found in higher relative concentrations (Table 3) and 10 sugars were found in lower concentrations (Table 4) in SH1000-TTORS-1 compared to SH1000.
Asparagine and aspartic acid are increased in SH1000-TTORS-1 and during the degradation of these amino acids they can enter the TCA cycle after being turned into oxaloacetate [56] (Figure 2), and aspartic acid is one of eight amino acids that is initially degraded by S. aureus during growth in media without a preferred carbon and energy source [56]. The relative concentration of RocD which is involved with ornithine and proline biosynthesis [57] was increased in SH1000-TTORS-1 (Table 1, Figure 2), and both ornithine and proline are increased in SH1000-TTORS-1 (Table 3, Figure 2). The increased proline accumulation could be attributed to the increased relative concentration of the proline/choline/glycine betaine transporter OpuD [58] in SH1000-TTORS-1 (Table 1). Ornithine is synthesized from arginine during the urea cycle and is a precursor needed in the synthesis of glutamate and proline (Figure 2) [57]. Both proline and ornithine can be metabolized into glutamate which can then enter the TCA cycle as α-ketogluterate (Figure 2).
The concentration of glucose 6-phosphate, fructose 6-phosphate, and fructose 1,6-bisphosphate were decreased in SH1000-TTORS-1 compared to SH1000 (Figure 2, Table 3) indicating that gluconeogenesis or sugar accumulation is reduced in this mutant. The concentrations of TCA cycle intermediates fumarate and succinate were both increased in SH1000-TTORS-1 which occurs in conjunction with increased SucD expression (Table 1; Table 3).
Lactate is synthesized from pyruvate in the absence of oxygen via lactate dehydrogenase [59] and lactate concentrations were decreased in SH1000-TTORS-1, yet there was no difference in the D-lactate dehydrogenase levels in SH1000-TTORS-1 and SH1000 (Table S1). In conjunction with the reduced lactate level in SH1000-TTORS-1, we note that the relative concentration of a protein (Lqo) which is responsible for metabolizing lactate into pyruvate, was also decreased (Figure 2) [60].
The only sugar found to be increased in SH1000-TTORS-1 and not detected in SH1000 was N-acetylglucosamine (Table 3) which is a precursor for peptidoglycan biosynthesis [61].
The reduction in PurA (adenylosuccinate synthetase) observed in SH1000-TTORS-1 (Table 1), that occurs with a concomitant reduction in adenosine and guanosine, was not even detected in this mutant ( Table 3).

Discussion
SH1000-TTORS-1 harbors a mutation in acpP [30], which leads to the production of an ACP with an amino acid alteration (A34D) that is 2 amino acids away from the highly conserved S36 that is modified with a 4 -phosphopantetheine moiety that acts as the site for fatty acid attachment during fatty acid elongation [34]. We now report that SH1000-TTORS-1 exhibits a number of other alterations that further supports the hypothesis that fatty acid biosynthesis in this mutant is defective (Figure 1). SH1000-TTORS-1 demonstrated the altered expression of genes required for the fatty acid and lipid biosynthesis (plsX, fabZ, and acpP) and the control of this process (fapR). SH1000-TTORS-1 also expressed an increased quantity of CoaBC and reduced quantity of PanE, which implied that the production of holo-ACP is altered in this mutant. Furthermore, the relative concentration of FabD (malonyl-CoA-ACP transacylase) and FabF (3-oxoacyl-synthase) ( Figure 1) and two uncharacterized proteins (SAOUHSC_01348, SAOUHSC_02604) possibly involved with fatty acid biosynthesis and degradation, were all increased in SH1000-TTORS-1. Lastly, similar to triclosan-selected SCVs [42], growth of SH1000-TTORS-1 with the fatty acid supplement tween-80 led to an increase in colony size. We previously reported that the overall fatty acid content of SH1000 and SH1000-TTORS-1 were similar [30], so we propose that fatty acid biosynthesis is defective and slowed in SH1000-TTORS-1.
Similar to previously characterized S. aureus SCVs [5][6][7][8][9], SH1000-TTORS-1 also demonstrated alterations in central metabolism. The relative concentration of GapA1 and Gpm1 which are central to glycolysis and gluconeogenesis were reduced in SH1000-TTORS-1. Six proteins involved in pyruvate metabolism were also reduced in SH1000-TTORS-1. The defect in fatty acid biosynthesis observed in SH1000-TTORS-1 probably also contributes to altered acetyl-CoA metabolism. Collectively, this data implies that the flow of pyruvate and acetyl-CoA into the TCA cycle is altered in SH1000-TTORS-1. At the same time, the relative concentration of CidC and ackA and pta expression were increased in SH1000-TTORS-1, which indicated an alteration in acetate metabolism. ATP production via the Pta/AckA pathway is thought to be important for growth when glucose is not abundant [56]. Collectively, these findings indicate that pyruvate, acetyl-CoA and acetate metabolism are skewed within SH1000-TTORS-1. Surprisingly, metabolomic analysis revealed that the concentrations of pyruvic acid in both SH1000 and SH1000-TTORS-1, were not significantly different (Table S2).
In SH1000-TTORS-1 SucD, succinate, and fumarate, were all increased in SH1000-TTORS-1 indicating increased flow of carbon in this part of the TCA cycle in SH1000-TTORS-1. During carbon catabolite repression (CCR) in bacteria, the presence of preferred carbon sources leads to the repression of genes whose products are involved in the catabolism of non-preferred carbon sources [62]. S. aureus can readily degrade a number of amino acids and it was suggested that glutamate and amino acids that generate glutamate, particularly proline, serve as the major carbon source in defined media lacking a preferred carbon source [56]. We propose that the reduction in sugar content in SH1000-TTORS-1 alleviates carbon catabolite repression (CCR) allowing for the catabolism of amino acids (ornithine, proline, asparagine and aspartate), which are provided from the increased amino acid pool observed in SH1000-TTORS-1, to produce TCA intermediates and energy ( Figure 2).
It has been previously reported that the production of ribosomal proteins can be altered in S. aureus SCVs [8] and an S. aureus SCV mutant harboring a relA mutation demonstrated increased expression of TCA cycle enzymes [16]. Protein synthesis and the expression of tRNA synthetases are also reduced during the stringent response [51]. SH1000-TTORS-1 demonstrated reduced production of proteins essential to translation, including tRNA-amino acid synthetases, increased RelQ production and the altered production of proteins affected by the stringent response. The increase in free amino acid pools in SH1000-TTORS-1 therefore likely results from downturned protein production and the activation of an aberrant stringent response.
The increased production of SigB and RsbU, and the altered expression of SigB-controlled proteins demonstrates that the general stress response is activated in SH1000-TTORS-1. Considering the altered metabolic state of SH1000-TTORS-1 overall, it is not surprising that SigB and RsbU are increased in an effort to deal with the stress this mutant is experiencing and it should be noted that SigB is required for SCV formation [47,48].
The thinned cell wall [30], increased expression of MurZ, increased N-acetylglucosamine content, and alteration in cell wall antimicrobial susceptibility reported in this study supports the notion that peptidoglycan biosynthesis is also compromised in SH1000-TTORS-1. Furthermore, the reduction in DNA polymerase III, the major DNA replicative complex of the cell, indicates that DNA replication is reduced in SH1000-TTORS-1. This reduction in adenosine and guanosine is likely linked to a reduced DNA replication and ATP production brought on by the slowed growth and the altered metabolism observed in this mutant.
Gradient plate analysis was performed with MHB overnight cultures and Mueller Hinton agar (MHA) as previously described [64], and confluent growth along the gradient was measured in mm ± standard deviation (n = 3). Colony size determination was carried out as previously described [29] by diluting overnight MHB cultures and plating on MHA and MHA containing 0.1% (v/v) Tween 80. Random colonies were then measured for each strain grown on both media using a caliper in mm ± standard deviation (n = 10).

Protein Extraction
Total proteins were isolated from S. aureus strains SH1000 and SH1000-TTORS-1 in triplicate following the procedure of Wolff et. al. 2008 [65]. Briefly, overnight cultures were used to inoculate 250 mL of LB (initial OD 580 = 0.01) which were then incubated (37 • C, 200 rpm) until the cultures reached an OD 580 = 0.7. The cells were then pelleted (8000× g, 10 min, 4 • C) and stored at −80 • C. These cell pellets were then thawed and washed twice with TBS buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4) and re-suspended in 1 mL of lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, pH = 7.4). These suspensions were then loaded into a 2-mL cryovial tube containing 1 g of 0.1 mm glass beads (Biospec Products, Bartlesville OK) and the cells were lysed by agitation using a Mini-Beadbeater (Biospec Products) (5 × 50 sec @ 4200 rpm) with 2-min resting periods on ice in between agitation cycles. The cell lysates were then pelleted (8000× g, 10 min, 4 • C) twice to remove whole cells and glass beads, and sodium azide was added (final concentration of 0.02% v/v) before samples were stored at −80 • C. Protein concentrations were determined using the Bradford method [66] using bovine serum albumin (Bio-rad, Hercules CA) as a standard.

Mass Spectrometry Analysis
Each protein sample was analyzed using two replicate LC-MS/MS analyses on a hybrid LTQ-Orbitrap mass spectrometer (LC-MS/MS) (Thermo Fisher Scientific, Waltham, MA) as previously described [67]. Proteins were identified by using the Andromeda application with MaxQuant [68,69] to search the MS data against a database of S. aureus proteins downloaded from Uniprot (S. aureus strain NCTC 8325, ID 9306). Alterations in protein expression were quantified on the basis of peptide peak intensities, via the LFQ algorithm [70] embedded in MaxQuant v. 1.5.3.8. Statistically significant differences in protein levels were calculated via the t-test algorithms embedded in Perseus 1.5.3.2 ( [71,72]; Max Planck Institute of Biochemistry). A two-fold alteration in protein concentration and a Student's t-test p-value of < 0.05 was used as the threshold criteria for significant differences in protein expression (Table 1).

Metabolite Extraction and Analysis
Metabolites were extracted from both SH1000 and SH1000-TTORS-1 in triplicate as previously described [73]. Briefly, overnight cultures were used to inoculate 125 mL of LB (initial OD 580 = 0.01) and incubated (37 • C, 200 rpm) until an OD 580 = 0.07 was reached. Cells from each sample were then collected by centrifugation (10,000× g, 5 min, 4 • C) and washed once with 1× PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH = 7.4) and re-pelleted (10,000× g, 5 min, 4 • C). Metabolic quenching was achieved by adding 500 µL of cold methanol (−20 • C) and all samples were stored at −80 • C. Metabolite analysis was carried out by the Roy J. Carver Biotechnology Center, University of Illinois at Urbana-Champaign, Urbana, IL, USA as previously described [73]. Metabolite relative concentrations were normalized using 100 mg of dry cell weight and statistically significant alterations in metabolite levels were based upon a Student's t-test p-value < 0.05 and reported as the average ± standard error of the mean for all three biological replicates.

RNA Extraction and RT-PCR
Initially, exponentially-grown 50-mL LB cultures (see above) were pelleted (8000 rpm, 4 • C, 10 min), and re-suspended in 1 mL of TRIzol (Ambion-Thermo Fisher Scientific, Waltham, MA). Cell lysis was completed using a single cycle of bead beating as described above. The resulting supernatant was then extracted with chloroform (1/5 the total volume) with agitation, and then centrifuged (12,000× g, 4 • C, 15 min). The aqueous layer was then placed in a fresh microfuge tube and the RNA was then precipitated with ice cold isopropanol (2.5 × the total volume) on ice, followed by centrifugation (12,000× g, 4 • C, 20 min). The resulting RNA pellet was then washed with 70% ethanol, re-pelleted (7500× g, 4 • C, 10 min) and air-dried. All RNA samples were treated with the DNA-free Kit (Ambion-Thermo Fischer Scientific, Waltham, MA) and cDNA samples were produced utilizing SuperScript III Reverse Transcriptase per the manufacturer's instructions (Invitrogen-Thermo Scientific, Waltham, MA).
cDNA samples were then interrogated by real-time PCR using the LightCycler 96 Real-Time PCR system (Roche, Indianapolis, IN) and FastStart Universal SYBR Green Master (ROX) (Roche). Gene-specific primers used in the RT-PCR analysis are found in Table S3. Critical threshold values were normalized using 16S rRNA and expression values were calculated using the 2 −∆∆Ct method [74].

Conclusions
Overall this study provides additional insight into a mechanism that a bacterial pathogen utilizes to thwart the action of a popular "natural" antiseptic that inactivates multiple targets. We propose that the slowed or reduced production of the general targets of TTO (proteins and membranes) and altered metabolism allow SH1000-TTORS-1 to withstand the antimicrobial action of TTO better than the parent strain. This reduction in major biosynthetic pathway activity, stress response activation, and overall alteration in metabolism also likely contributes to the reduced growth and small colony phenotype observed in SH1000-TTORS-1.

Conflicts of Interest:
The authors declare no conflicts of interest.