Proteomic Analysis of Ketogulonicigenium vulgare under Glutathione Reveals High Demand for Thiamin Transport and Antioxidant Protection

Ketogulonicigenium vulgare, though grows poorly when mono-cultured, has been widely used in the industrial production of the precursor of vitamin C with the coculture of Bacillus megaterium. Various efforts have been made to clarify the synergic pattern of this artificial microbial community and to improve the growth and production ability of K. vulgare, but there is still no sound explanation. In previous research, we found that the addition of reduced glutathione into K. vulgare monoculture could significantly improve its growth and productivity. By performing SEM and TEM, we observed that after adding GSH into K. vulgare monoculture, cells became about 4–6 folds elongated, and formed intracytoplasmic membranes (ICM). To explore the molecular mechanism and provide insights into the investigation of the synergic pattern of the co-culture system, we conducted a comparative iTRAQ-2-D-LC-MS/MS-based proteomic analysis of K. vulgare grown under reduced glutathione. Principal component analysis of proteomic data showed that after the addition of glutathione, proteins for thiamin/thiamin pyrophosphate (TPP) transport, glutathione transport and the maintenance of membrane integrity, together with several membrane-bound dehydrogenases had significant up-regulation. Besides, several proteins participating in the pentose phosphate pathway and tricarboxylic acid cycle were also up-regulated. Additionally, proteins combating intracellular reactive oxygen species were also up-regulated, which similarly occurred in K. vulgare when the co-cultured B. megaterium cells lysed from our former research results. This study reveals the demand for transmembrane transport of substrates, especially thiamin, and the demand for antioxidant protection of K. vulgare.


Introduction
L-Ascorbic acid (L-AA), also known as vitamin C, has been widely used in pharmaceutical, food and cosmetic industries mainly for its antioxidant property [1,2]. Seven-step Reichstein method has been employed to produce vitamin C since 1934 [3]. However, high energy cost and chemical pollution have made this chemical-based method less attractive. In China, a two-step microbial fermentation process was developed [4] and has been successfully used for industrial production of vitamin C [5]. The fermentation process consists of one step of conversion from Dsorbitol to L-sorbose by Gluconobacter oxydans [6], followed by another step of converting L-sorbose to 2-keto-gulonic acid (2-KGA), the precursor of vitamin C, by a mixed culture system of Ketogulonicigenium vulgare and Bacillus megaterium [6,7,8]. In the mixed culture, the 2-KGA producing bacteria K. vulgare which grows poorly when cultivated alone even on rich media [9], can grow better and carry out the conversion with the coculture of B. megaterium with high efficiency [8]. In spite of obvious advantages over monoculture of K. vulgare, the mixed culture has relatively higher cost for fermentation and contamination issue caused by the B. megaterium spore [10].
Previously, Leduc et al added 100 mg/L of reduced glutathione (GSH) into the K. vulgare LMP P-20356 monoculture [11], but no growth improvement of K. vulgare was observed. However, we have recently found that the addition of 1 g/L GSH into the K. vulgare monoculture could considerably enhance cell growth and improve 2-KGA production [12]. The discrepancy of our result and that of Leduc's may be due to different concentrations of GSH used, and could also be due to the species differences between two studies. In addition, we also found that K. vulgare cells elongated after GSH treatment. K. vulgare, formerly classified in Gluconobacter oxydans [13], belongs to the a-Proteobacteria. It was reported that G. oxydans when confronted with certain stresses could form intracytoplasmic membranes (ICM) at the end of exponential phase [14]. Accompanying the ICM formation, cells became longer and the activities of several membrane-bound dehydrogenases including glycerol [14], sorbitol and other polyol dehydrogenases [15] were improved significantly. Improved 2-KGA production suggested probable enhanced activities of two membrane-bound dehydro-genases (sorbose dehydrogenase and sorbosone dehydrogenase) in K. vulgare responsible for the conversion of L-sorbose to 2-KGA. Thus, it could be speculated that similar ICM formation might also occurred in K. vulgare grown under GSH treatment. After our transmission electron microscopy observation, this speculation was confirmed. Proteomic analysis has been successfully applied in our laboratory to study cellular responses of Saccharomyces cerevisiae to external stimuli [16,17]. Considering its accuracy in quantification and relative lower protein amount requirement, iTRAQ (isotopic Tags for Relative and Absolute Quantification)-coupled 2-D LC-MS/MS technique was used in this study to identify and quantify differential proteomes in K. vulgare cells grown with and without GSH, with the aims to reveal the molecular mechanism of GSH's enhancing effects, to study the similarities and differences between the mechanisms of GSH and Bacillus megaterium on K. vulgare and further in order to provide insights into the study of the synergism of the co-culture system, and ultimately to find possible targets of gene modification in K. vulgare for further improving cell growth and 2-KGA production. The results showed that significant changes in the proteome of K. vulgare occurred at 15 h after GSH addition. After principal component analysis (PCA), a series of membrane related proteins involved in thiamin/thiamin pyrophosphate (TPP) transport, oligopeptide transport, outer membrane integrity maintenance, and L-sorbose metabolism were found to contribute the most to the distinguishment of the GSH addition group at 15 h. Several proteins in pentose phosphate pathway (PPP) and tricarboxylic acid (TCA) cycle also showed upregulation. Besides, some proteins combating against intracellular reactive oxygen species (ROS) were also up-regulated, which resembled our former research results when the B. megaterium cells lysed in the mix-culture system [18], suggesting a reducing environment should be important for the growth and 2-KGA production of K. vulgare, and this demand could be satisfied either by the addition of B. megaterium, or by the addition of reduced GSH.

Results and Discussion
Physiological Profile and Morphology of K. vulgare Treated with and without GSH A comparative physiological study of K. vulgare with and without GSH addition was used to determine the influence of GSH on K. vulgare ( Figure 1). Compared with the control group, K. vulgare with 1 g/L GSH addition could achieve 3.6-fold of growth and 5.6-fold of 2-KGA production at 70 h of the fermentation. The extracellular GSH content was shown in Figure 1C, suggesting that GSH in the environment was completely depleted after 36 h. The cells were harvested at 1 h and 15 h after adding GSH, which represented the beginning and middle phases of GSH depletion, respectively. Scanning electron microscopy (SEM) was used to determine the morphology change of K. vulgare (Figure 2), and after measuring cell size in 20 replicates, we obtained the following results: the size of the K. vulgare cell without GSH addition was about (0.660.2)6(0.560.1) mm (Figure 2A), and at 24 h after GSH addition, the cell size became (2.460.7)6(0.660.1) mm ( Figure 2B). Interestingly, when the elongated cells resulting from GSH treatment were transferred to agar plate, the cells were able to change back to their original size ( Figure 2C), indicating the variation in morphology was reversible, largely depending on the environmental conditions. Similar change in cell morphology also occurred to G. oxydans as described previously [14,15,19]. It was reported that among the non-photosynthetic gram-negative prokaryotes, complex ICM were formed most notably in four physiological groups: Azotobacter, Gluconobacter, the obligate methane and methanol oxidizers, and the ammonium of nitrite oxidizers [15]. Studies with G. oxydans showed that viable cells became two to four times longer as a result of the differentiation at the end of exponential growth by forming dense regions containing accumulations of ICM and ribosomes at the end of each cell [14]. It was also revealed that several membrane-bound dehydrogenases for glycerol, sorbitol and other polyols acquired improved activities probably as the ICM were extensions of plasma membranes [14]. In our study, the cells had a 4-6 folds of cell elongation and 5.6 folds of 2-KGA producion by the membrane-bound sorbose/sorbosone dehydrogenase. Considering the homology between K. vulgare and G. oxydans, similar cell differentiation and ICM forming under GSH stimulation are quite possible. In order to confirm our speculation, transmission electron microscopy (TEM) observation was conducted, and the resluts were shown in Figure 3. From Figure 3B, we can see that after adding GSH into K. vulgare monoculture, the cells formed ICM as we conjectured.

Protein Function Analysis
Proteomic analysis was employed to seek further clues on GSH's effects. A total of 319 proteins were identified from three biological replicates by 2-D nano-LC-MS/MS, and among them 265 proteins were quantified. The quantified proteins were listed in Table S1. Phylogenetic classification of the identified proteins was performed by searching the Clusters of Orthologous Groups of proteins (http://www.ncbi.nlm.nih.gov/COG) by COGnitor. As shown in Figure 4, the three major protein categories were proteins for translation, ribosomal structure and biogenesis (22%), followed by amino acid transport and metabolism (17%), and energy production and conversion (11%). Proteins involved in posttranslational modification, protein turnover, chaperones, transcription and other 13 categories of functions were also identified. In our previous study on the metabolic cooperation of K. vulgare and B. megaterium by spatially cultivating them on a solid agar plate [20], we have found that K. vulgare had the ability of secreting amino acids to the environment, suggesting K. vulgare itself may have a high-efficiency amino acid transport system. In our former proteomic analysis of the mix-culture system [18], the proportion of proteins for amino acid transport and metabolism in K. vulgare accounted for 22%, higher than that in the monoculture with GSH, suggesting in the mix culture B. megaterium might helped amino acid transport of K. vulgare to a larger extent than GSH. Proteins with just 4% proportion for carbohydrate transport and metabolism was probably correlated with the deficient TCA cycle and poor PPP in K. vulgare, while in the mix-culture system, the proportion increased to 8% [18], revealing the significant assistance of B. megaterium on the central carbon metabolism of K. vulgare.
Among the 41 proteins participating in translation, ribosomal structure and biogenesis, 31 were up-regulated for over 2 folds in K. vulgare with GSH addition at 15 h, compared with the control group, consistent with previous report that the differentiation of G. oxydans caused the accumulation of ribosomes at the poles of each cell [14]. With the increase in expression of ribosome proteins, more mRNA could be translated into amino acids, then polypeptide chain, and ultimately protein.

PCA Analysis of the K. vulgare Proteins
In this study, we applied PCA to the quantified proteins, and the result was shown in Figure 5. In the score plot in Figure 5A, the datasets from 115/114, 116/114, and 117/114, which represented GSH-1 h, Kv-15 h and GSH-15 h compared with Kv-1 h, respectively, could be discriminated as circled by the three ellipses.
Especially, the 117/114 dataset was conspicuously away from the other two datasets, which indicated significant protein variations had occurred in the GSH treated K. vulgare at 15 h. While the 115/ 114 and 116/114 groups didn't show much distinction, suggesting that K. vulgare didn't have much variation during its monoculture process, and yet no significant changes occurred in protein level at 1 h after the addition of GSH. The loading plot in Figure 5B showed that thiamin/thiamin pyrophosphate-binding protein (ThiB), YkoF-related family protein, bacterial extracellular solute-binding proteins, family 5 Middle family protein, protein TolR, sorbose/sorbosone dehydrogenase (SSDH), resolvase, ribosomal protein, integration host factor were important proteins for the discrimination of the 117/114 group from the other two groups.
ThiB, which belongs to the ThiBPQ transport complex for thiamin and TPP, is used as a marker of the complex [21]. Besides, the YkoF protein is also involved in thiamin transport [22]. These two proteins were the most up-regulated proteins in K. vulgare in response to GSH at 15 h compared with the control (thiB, 7.4 folds; YkoF, 5.3 folds), and their relative abundances were shown in Figure 6A, suggesting these two proteins possibly have close relationship with the GSH stimulation. The similarities of these two proteins in their functions led us to examine the function of thiamin (vitamin B 1 ). After its uptake by cells, thiamin undergoes pyrophosphorylation yielding thiamin pyrophosphate (TPP), which is the active substance [23]. TPP participates in carbohy-drate metabolism as an important coenzyme for pyruvate dehydrogenase, alpha-ketoglutarate dehydrogenase and transketolase [24], and thus is important to TCA cycle and PPP by affecting their key enzymes. The relative abundances of these three enzymes were also shown in Figure 6A, which all showed upregulation in K. vulgare grown with GSH at 15 h. K. vulgare was reported to have a deficient TCA cycle, and mainly depended on PPP to carry out central carbon metabolism [25]. While in K. vulgare with GSH addition, five important enzymes in TCA cycle got more than 2-fold up-regulation, and three enzymes in PPP got more than 1.7-fold up-regulation ( Figure 6B). In addition, increased ATP synthesis could be observed from Figure 6C, in which four subunits of ATP synthase were up-regulated to about 2-3 folds at 15 h under the stimulation of GSH. These results suggested that the addition of GSH might facilitate the thiamin/ TPP transport and then improve the activities of important enzymes in TCA cycle and PPP. In this way, more ATP could be generated and in return be used to supply energy for GSH transport itself and also thiamin/TPP transport, forming a positive circulation.
In order to validate the conjecture of the importance of thiamin transport to K. vulgare, we performed a negative validation by adding inhibitor of thiamin transport-amprolium, which competed with thiamin in being transported as an analogue [26,27]. It had been used in E. coli as a thiamin analogue to explore the molecular recognition characteristics of the TPP riboswitch [27]. In our study, a final concentration of 1 mg/mL of amprolium was added to K. vulgare monoculture and K. vulgare with GSH addition, respectively. The growth and 2-KGA production of K. vulgare under different treatments were shown in Figure 7, from which we could see that after a culture of 24 h, amprolium slightly lowered the growth and 2-KGA production of mono-cultured K. vulgare, while in K. vulgare with both amprolium and GSH, the growth and 2-KGA produciton decreased to a considerable larger extent.  These results suggested that thiamin played an essential role in the growth and 2-KGA production of K. vulgare, and that GSH could enhance the transport process regardless of thiamin or its analogue.
Considering the importance of thiamin transport process, there is a possibility that adding thiamin into the K. vulgare monoculture might improve cell growth. In our research, we added thiamin with a final concentration of 0.2 mg/mL. While from Figure S1, we didn't see much difference in growth and 2-KGA production of K. vulgare after adding thiamin. This phenomenon was probably due to the fact that active transport process was not only substrate restricted, but also closely related with indispensable trasnporter and energy support. As a poor growing bacterium, K. vulgare has defects in both upstream protein synthesis and downstream metabolism including energy production. Thus, in order to improve its gowth and production ability, a global regulation should take place. The response of K. vulgare to glutathione was such a global regulation.
Bacterial extracellular solute-binding proteins, participating in active transport of solutes across the cytoplasmic membrane, generally correlated with the nature of the solute bound, are grouped into eight families of clusters. The family 5 mainly consists of periplasmic oligopeptide-binding protein (oppA), dipeptidebinding protein, nickel-binding proteins, and heme-binding lipoprotein. It was reported in Bacillus subtilis and Lactococcus lactis that the oligopeptide transport system (Opp) was encoded by the gene oppDFBCA [28,29]. This transport system typically transports peptides of 3 to 5 amino acids, independently of peptide sequence [28]. According to genome sequencing results, the Opp system is also present in K. vulgare. GSH, a tripeptide in the form of L-GluCysGly, could be imported into the cell through the Opp system. In this study, we have identified several bacterial extracellular solute-binding proteins, family 5 middle family protein, and also the protein OppF, which links transport to the hydrolysis of ATP as an ATP binding protein [28]. The profiles for these proteins were shown in Figure 8. Among which SBP-1 had a 2.4-fold up-regulation compared with the control (115/114) at 1 h after GSH addition, which revealed that the oligopeptide transport in K. vulgare started to be enhanced shortly after the additon of GSH. In our former proteomic study of the co-culture system, similar up-regulation of bacterial extracellular solute-binding proteins, family 5 middle family protein also occurred when the cell lysis of B. megaterium took place during the mix-culture fermentation, which suggests the cell lysis of B. megaterium also stimulated the uptake of oligopeptide by K. vulgare. Considering the conversion between reduced GSH and GSSG, glutathione reductase (GR) is important so as to keep a relative higher GSH/GSSG ratio. Not only the transport of GSH, but also the GR abundance should play important role in the function of GSH inside the cells. However, perhaps due to the separation limitation, we didn't find this protein by LC-MS detection. Then in order to find this protein, we tried 2-D gel separation for its good separation properties. After theoretical molecular weight/PI calculation based location, and further peptide mass fingerprint identification, we found that in the control, the spot for GR was hardly visible, while in the GSH treated K. vulgare, GR had a certain expression, as shown in Figure 9. In this way, we suppose GSH could positively function inside the cells. Protein TolR is a part of the Tol/Pal system which forms a five-member, membrane-spanning, multiprotein complex. The complex has been found conserved in gram-negative bacteria and involves the integrity of the outer membrane [30,31,32]. In our study, the Previous studies revealed that the conversion of L-sorbose to 2-KGA comprised two successive reactions: one is the oxidation of L-sorbose into L-sorbosone by sorbose dehydrogenase [33], and the other is the subsequent oxidation of L-sorbosone into 2-KGA by sorbosone dehydrogenase [13]. Both of these two dehydrogenases are membrane bounded and pyrroloquinoline quinone (PQQ)-dependent [34]. In our investigation, these two enzymes in the form of sorbose/sorbosone dehydrogenase, had an upregulation of 3.6 folds in K. vulgare at 15 h after GSH addition. This phenomenon together with cell elongation, resembled the case of enhanced sorbitol dehydrogenase activity and cell elongation in G. oxydans as a result of the ICM formation [15]. Considering the above similarities, it is quite probable that ICM forming occurred in K. vulgare under GSH treatment.

GSH and Reactive Oxygen Species Related Proteins
GSH is the major nonprotein thiol compound in living cells, including human, yeast, and microorganism cells [35,36,37], which plays an important role in protecting cells against oxidative damage, toxic compounds, radiation and heavy metals [38]. Oxidation stress is caused by intracellular ROS, including superoxide anion, hydrogen peroxide, and hydroxyl radical as a result of oxygen metabolism in aerobic organisms [39]. ROS can cause damages to cells by attacking DNA, resulting in chain breaks, and modification of the carbohydrate parts and nitro bases [40]. In addition, ROS can also modify amino acids in proteins [39]. Several systems have been described to combat ROS in microorganisms, including catalase, superoxide dismutase, GSH metabolism related enzyme, thioredoxin and glutaredoxin [41]. In this study, we found that seven proteins involved in ROS detoxification were up-regulated after adding GSH in K. vulgare ( Figure 10). Relative abundance variations of these seven proteins in the GSH treated K. vulgare compared with the control at 15 h of fermentation were as follows: superoxide dismutase, which reduces superoxide anion into molecular oxygen, had a 1.7-fold upregulation; catalase, which reduces hydrogen peroxide into water and oxygen [41], had a 1.5-fold up-regulation; thioredoxin, thirodoxin-disulfide reductase and glutathione S-transferase respectively had a 3.3-fold, a 1.5-fold and a 2.9-fold up-regulation; 6phosphogluconate dehydrogenase and NAD(P)H: quinine oxidoreductase had a 2.6-fold and a 1.9-fold up-regulation, respectively. It is well known that NADPH is the principal reducing power in cells, playing essential roles in GSH and thioredoxin metabolism [42]. The source of NADPH is glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase in the PPP [43]. Our results provided a speculation that the enhancement in PPP resulting from improved TPP transport, could generate more NADPH, which could then be used in the reduction of oxidized thioredoxin and GSSG, and in this way helped to combat against intracellular ROS.
As an aerobic bacterium, K. vulgare was inevitably subjected to ROS stress. In our former research, proteins for combating ROS stress in K. vulgare had similar up-regulation after the cell lysis of B. megaterium [18]. This coincidence highlighted the essential requirement for a reducing environment for the cells to grow and metabolize. In the case of this study, the positive effect of GSH also reflected in enhanced TPP transport and other processes associated with cell membranes of K. vulgare.
In this study, we aim to explore the mechanism of GSH's positive effect on K. vulgare by determine proteomic changes of K. vulgare when subjected to the reduced GSH. We generated comparative proteomic profiles for K. vulgare with and without GSH addition during the time course of the fermentation. Although no distinctive changes at early fermentation phase (1 h), at 15 h of the fermentation, the cells of the test group had a 4-6-fold increase in cell length, and the proteome profiles of K. vulgare showed significant differences. A proposed mechanism of GSH to K. vulgare was illustrated in Figure 11. In response to the GSH stimulation, the expressions of several proteins involved in cell membrane function, including oligopeptide transport protein and thiamin/TPP transport protein, membrane-bound dehydrogenase and outer membrane integrity maintainer were significantly increased. The coenzyme TPP obtained by thiamin transport would enhance the enzymatic activities of PPP and TCA cycle, and thus generate more reducing power in the form of NADPH and energy in the form of ATP, which could then be used in combating against intracellular ROS. This work provides valuable information regarding the molecular mechanism of GSH's enhancing effects on K. vulgare growth and 2-KGA production, and reveals the demand of K. vulgare for substrate transport for thiamin/TPP and oligopeptides, and the demand for antioxidant protection. Similar antioxidant protection requirement in K. vulgare also occurred in the mix-culture of K. vulgare and B. megaterium, and this coherence emphasized the importance of a reducing environment to K. vulgare, thus this requirement could be potential target in K. vulgare for future gene modification, which may eventually lead to the establishment of high-efficiency monostrain fermentation for vitamin C production in industry.

Cell Culture and Growth Conditions
The strain of K. vulgare was kindly provided by Prof. Yuezhong Li of Shandong University in China. It was first inoculated in a seed medium containing: 20 g/L L-sorbose, 1 g/L KH 2 PO 4 , 3 g/L yeast powder, 0.2 g/L MgSO 4 , 3 g/L beef extract, 1 g/L urea, 10 g/L peptone, and 3 g/L corn steep liquor. The seed was then transferred to the fermentation medium with a 10% inoculum. The fermentation medium contains: 80 g/L L-sorbose, 20 g/L corn steep liquor, 1 g/L KH 2 PO 4 , 0.2 g/L MgSO 4 , and 12 g/L urea. The pH values of seed and fermentation media were adjusted to 6.8 and 7.0, respectively using 5 M NaOH before cultivation. K. vulgare was cultivated in a 250 mL flask with a 50 mL fermentation volume, shaken at a speed of 250 rpm at 30uC for about 70 h. GSH was added into the test group at the beginning of the fermentation with a final concentration of 1 g/L, and other cultivation conditions were identical as the control group.

Analyses of 2-KGA and L-sorbose
The analyses of extracellular 2-KGA and L-sorbose were performed by the use of High Performance Liquid Chromatography (Waters Corp., Massachusetts, USA) coupled with a refractive index detector. 5 mM H 2 SO 4 was used as the eluent on an Aminex HPX-87H column (BioRad, CA) at the temperature of 65uC with a flow rate of 0.6 mL/min.

Determination of Extracellular GSH
The concentration of GSH in the fermenting liquor was determined by LC-MS/MS using a Venusil XBP C 18 column coupled with a LCQ DecaXP Maxa mass spectrometer (Thermo Finnigan, Palo Alto, CA). A 30-min elution gradient for the reverse-phase chromatography separation consisting of a 3-min of 100% buffer A (5% acetonitrile, 0.1% formic acid in water); a 4min gradient to 95% buffer B (0.1% formic acid in acetonitrile) followed by a 5-min maintenance; a 2-min gradient to 100% buffer A; and a 16-min re-equilibration at 100% buffer A was adopted. Selected reaction monitoring (SRM) mode was chosen to perform the mass detection. It works by isolating the target parent ion with a mass to charge ratio of 308 in the form of [GSH+H] + , initiating collision induced dissociation (CID), detecting and quantifying its unique daughter ion with a 179 m/z, derived from the breakage of amide bond between glutamyl and cysteinyl groups, to achieve the precise quantification of parent ion. This mass detection mode has the virtue of confirmative quantification of chemicals at lower concentrations. A tolerance of 0.5 m/z was allowed. GSH content in the fermentation liquor was determined according to a GSH standard calibration curve.

Scanning Electron Microscopy and Transmission Electron Microscopy of K. vulgare
The sample preparation method for scanning electron microscopy was according to Ridgway [44] with minor modifications. 5 mL of bacteria suspensions was subjected to a 50006 g centrifugation at 4uC for 5 min and washed twice with phosphate buffer solution (PBS, pH 7.0) to obtain the cells for subsequent glutaraldehyde fixation and ethanol gradient dehydration. K. vulgare cells were fixed in 3% (w/v) phosphate-buffered glutaraldehyde for 2 h. Then a 5-min centrifugation was applied at a speed of 50006 g, followed by twice washes with distilled H 2 O. Increasing concentrations of ethanol in water (30%, 50%, 70%, 90%, 100%) were used in dehydration of the cells with each concentration lasting for 10 min. Dehydrated cells were dried on a SpeedVac, mounted on metal stubs with silver paint, and coated  with gold. Finally, specimens were observed in a scanning electron microscope (Nova NanoSEM 430, FEI.).
For transmission electron microscopy observation, the cells were first fixed in 3% glutaraldehyde for 4 h, and then in 1% osmium tetroxide for postfixation. After dehydration in graded ethanol as described before, the cells were embedded in SPI-PON 812 resin, sectioned with a LKBV, mounted on uncoated grids, stained with uranyl acetate followed by lead citrate, and examined with Philips EM400ST microscope.

Sampling, Extracting of Proteins and iTRAQ Labelling
K. vulgare cells with and without GSH were harvested at 1 h and 15 h of fermentation, the time points that respectively represented the beginning and middle phases of extracellular GSH depletion according to our preliminary experiment result, by a centrifugation at 50006g for 5 min at 4uC. The supernatants were collected for 2-KGA, L-sorbose and GSH analyses. Meanwhile, the pellets were quenched by liquid nitrogen after washing twice with PBS (pH 7.2) and once with distilled water.  Extraction of proteins was conducted according to Cheng [45] with minor modifications. First, 0.1 g cells were grounded in liquid nitrogen with mortar and pestle, suspended in 0.5 mL of lysis buffer (8 M urea, 4% m/v CHAPS, 40 mM Tris-HCl, and 1 mM PMSF) on ice, and then vortexed. Next, the suspensions went through a pulsed ultrasonication for 8 times, each time lasting for 6 s followed by a 10 s pause, and then were stored at 4uC for 2 h. After centrifugation at 120006 g for 40 min at 4uC, the supernatants were transferred to new tubes, and protein contents were determined using the Bradford method [46]. Then a total of 100 mg proteins were precipitated using 4 times volume of ice-cold acetone at 240uC overnight. Finally, following twice washes of the precipitant with 80% cold acetone in water, the pellets were dried on a SpeedVac, waiting for iTRAQ labelling.
The protein pellets were dissolved in 20 mL of dissolution buffer (50 mM TEAB) with 1 mL of denaturant (0.1% SDS), reduced by 2 mL of reducing agent (50 mM TCEP) and incubated at 60uC for 1 h according to the iTRAQ labelling protocol supplied by the manufacturer (Applied Biosystems) and then cysteine blocked using 1 mL of MMTS at room temperature for 10 min.

On-line 2-D Nano-LC and Mass Spectrometry Analysis
An Agilent 1200 series nanoflow LC system (Agilent Technologies, USA) interfaced with an electrospray ionization micro-Q-TOF II mass spectrometer (Bruker Daltonics, Germany) was applied for the peptide separation and detection. First, 5 mL of peptide mixture was loaded onto a ZORBAX BIO-SCX II column (3.5 mm, 3560.3 mm) by SCX buffer A at a flow of 10 mL/min pumped by a capillary pump, and was then eluted onto a ZORBAX 300SB-C 18 trap column (5 mm, 560.3 mm) by sequentially injecting 5 mL increasing concentrations of ammonium chloride (NH 4 Cl) solution (2 mM, 10 mM, 20 mM, 40 mM, 60 mM, 80 mM, 100 mM, 300 mM, 500 mM, 1 M) to achieve the first dimension separation and the enrichment. In the second dimension separation, a ten-port valve for the switch of the first and second dimension separation switched the trap column into the solvent path of the nanopump, and the peptides retained on Figure 11. A proposed mechanism of GSH influencing K. vulgare: yellow arrow means positive effect; pink arrow means generation; white arrow means transport; black arrow means conversion; brown arrow means ATP supply. The transport of GSH into cell by the Opp system, improved the TPP transport, and enhanced the PPP and TCA cycle, thus generated more ATP and NADPH, which could then be used in solute transport and intracellular ROS depletion. doi:10.1371/journal.pone.0032156.g011 the trap column were then eluted by a gradient eluent consisted of buffer B (5% acetonitrile in water, 0.1% formic acid) and buffer C (5% water in acetonitrile, 0.1% formic acid) at a flow of 300 nL/min on a ZORBAX 300SB-C 18 analytical column (3.5 mm, 150 mm675 mm). The 120-min reverse-phase gradient elution contained: a 5-min of buffer A; a 10-min linear ramp to 10% buffer B; a 70-min ramp to 35% buffer B; a 5-min ramp to 80% buffer B; a 8-min ramp to 90% buffer B; a 3-min ramp back to 100% buffer A; and a 19-min re-equilibration of 100% buffer A.
After LC separation, mass data acquisitions were conducted using micrOTOFcontrol 3.0 (Bruker Daltonics, Germany). The electrospray of ions was generated by applying a 1500 V high voltage on the ionization tip (1061 mm, SilicaTip), with a 150uC interface temperature and a 2 L/min flow of dry gas. A mass range between 1200 and 2500 m/z was acquired with 5 most abundant precursors selected for further MS/MS analysis, and the fragments covered a mass range between 70 and 2500 m/z. Each peptide was selected twice and then dynamically excluded for 0.2 min to have more peptides be detected.

Database Searching and Analysis
The mass spectrums after smoothed and background subtracted were analyzed by the tool of DataAnalysis (Bruker Daltonics, Germany). After finding auto MS (n) peak lists and deconvouluting the mass spectra, compound lists were exported in mgf files. Different mgf files from different runs of NH 4 Cl solution elution were combined into one file and was then used in protein identification using the Mascot database search algorithm with the threshold set at p,0.05. The database was an in-house protein database for K. vulgare with 3179 sequences. The searching parameters were set as follows: one maximum missed cleavage of trypsin; fixed modifications including MMTS modification of cysteine (Methylthio (C)), iTRAQ 4-plex (N-term) and iTRAQ 4plex (K); variable modifications consisting of oxidation of methionine and iTRAQ 4-plex (Y); 0.1 Da peptide mass tolerance; 0.05 Da fragment mass tolerance. We required a match of at least two peptides per protein as a confident detection. The quantification of labelled proteins was conducted also by the Mascot server by performing the iTRAQ 4-plex quantification. Three biological replicates were labelled and analyzed in parallel. InterPro (http://www.ebi.ac.uk/interpro/) was used for protein sequence analyses and classification, and the Kyoto Encyclopedia of Genes and Genomes (http://www.genome.jp/kegg/pathway. html) was used to reconstruct major metabolic pathways of K. vulgare. Database of Clusters of Orthologous Groups of proteins (http://www.ncbi.nlm.nih.gov/COG) [47,48] was utilized to perform phylogenetic classification of proteins. After mean centering, the data of protein abundances was subjected to PCA analysis by Matlab 7.0.

Extraction, 2-DE, MALDI-TOF Identification of Glutathione Reductase
Cells of K. vulgare with and without GSH at 15 h were harvested, respectively. Extraction of proteins and 2-DE were performed according to the method described before [18]. Each experiment was carried out at least twice for replicates. After calculating the theoretical molecular weight and PI of glutathione reductase by ExPASy Compute pI/Mw tool (http://web.expasy.org/ compute_pi/), we found the possible location area of GR on gel according to the calculated Mw of and PI of 5.24. Then peptide mass fingerprint (PMF) identification was performed by Autoflex TOF-TOF II (Bruker Daltonics, Germany) for several spots around this area. Mascot PMF searching was performed with a required peptide mass tolerance of 6100 ppm. Finally, we identified glutathione reductase with 6 matches of peptides, 22% protein coverage and a mascot score of 65, and ultimately we found its exact location on gel.