The l-Cysteine/l-Cystine Shuttle System Provides Reducing Equivalents to the Periplasm in Escherichia coli*

Intracellular thiols like l-cysteine and glutathione play a critical role in the regulation of cellular processes. Escherichia coli has multiple l-cysteine transporters, which export l-cysteine from the cytoplasm into the periplasm. However, the role of l-cysteine in the periplasm remains unknown. Here we show that an l-cysteine transporter, YdeD, is required for the tolerance of E. coli cells to hydrogen peroxide. We also present evidence that l-cystine, a product from the oxidation of l-cysteine by hydrogen peroxide, is imported back into the cytoplasm in a manner dependent on FliY, the periplasmic l-cystine-binding protein. Remarkably, this protein, which is involved in the recycling of the oxidized l-cysteine, is also found to be important for the hydrogen peroxide resistance of this organism. Furthermore, our analysis of the transcription of relevant genes revealed that the transcription of genes encoding FliY and YdeD is highly induced by hydrogen peroxide rather than by l-cysteine. These findings led us to propose that the inducible l-cysteine/l-cystine shuttle system plays an important role in oxidative stress tolerance through providing a reducing equivalent to the periplasm in E. coli.

A key building block of proteins, L-cysteine is an amino acid with a thiol side chain. Because of its high reactivity, L-cysteine is an important structural and functional component of many proteins. Although L-cysteine is pivotal for various protein functions, the molecule itself is toxic to cells even at low concentrations in both prokaryotes and eukaryotes (1)(2)(3). It has been reported that threonine deaminase, an enzyme in L-isoleucine biosynthesis, is inhibited by L-cysteine (4), which could be part of the reason for the cytotoxicity of L-cysteine in this organism. To maintain the L-cysteine concentrations below the threshold of cytotoxicity, the intracellular L-cysteine level is strictly controlled.
Serine acetyltransferase, a key enzyme in the L-cysteine synthesis pathway of Escherichia coli, is under the control of feedback inhibition by L-cysteine. In addition, E. coli has five or more enzymes having L-cysteine desulfhydrase activity (TnaA, CysK, CysM, MalY, and MetC). These systems may prevent the accumulation of excess L-cysteine in cells.
E. coli has L-cysteine transporters in the inner membrane (YdeD, YfiK, and Bcr) (5)(6)(7), and in the outer membrane (TolC) (8). It is known that TolC associates with the inner membrane and accessories, e.g. AcrAB or AcrEF, forming tripartite efflux pumps which export toxic compounds directly from the cytoplasm to the outside of the cells. However, in the L-cysteine export system, TolC does not associate with the L-cysteine transporters in the inner membrane (YdeD, YfiK, and Bcr) (8). These findings suggest that L-cysteine, transported from the cytoplasm, is first pooled in the periplasm, and then exported through TolC in the outer membrane. Despite this knowledge, the role of the periplasmic L-cysteine remained elusive.
The electron transport chain in the inner membrane of E. coli is thought to generate reactive oxygen species (ROS), 3 such as superoxide and hydrogen peroxide (H 2 O 2 ), due primarily to the leakage of electrons (9). H 2 O 2 in the cytoplasm is eliminated by two catalases (KatE and KatG) and a peroxidase (AhpCF). An Hpx Ϫ mutant lacking all of these three major enzymes accumulates H 2 O 2 in cells (10). However, these enzymes do not exist in the periplasm, but superoxide dismutase (SodC), which generates the H 2 O 2 , localizes in this compartment. This fact raises a question concerning how H 2 O 2 generated in the periplasm is eliminated.
In addition, E. coli is exposed to H 2 O 2 , which is produced by phagocytes, in the environment. If the cells could detoxify H 2 O 2 in the periplasm before its penetration into the cytoplasm, it would diminish its toxicity. Thus, the possession of such H 2 O 2 removal ability in the periplasm may be beneficial for the cells. It is known that the sulfhydryl group of L-cysteine can react with H 2 O 2 to yield H 2 O and L-cystine (11) as in Equation 1.
Thus, we speculated that an L-cysteine transporter such as YdeD exports L-cysteine as a scavenger of H 2 O 2 into the periplasm. These considerations have led us to study the role of L-cysteine transporters in E. coli. In this report, we show evidence that the L-cysteine transporter YdeD indeed functions against H 2 O 2 stress in E. coli. We also provide evidence that the periplasmic L-cystine-binding protein FliY is involved not only in the uptake of L-cystine, a product of oxidation of L-cysteine, but also in the H 2 O 2 tolerance of this organism. Further, our data show that H 2 O 2 stress highly induces the expression of the genes encoding YdeD and FliY. From these findings, we propose that the inducible L-cysteine/L-cystine shuttle system plays an important role for the resistance of cells to H 2 O 2 in the periplasm.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Oligonucleotides-E. coli strains and plasmids used in this work are listed in Table 1, and oligonucleotides used are listed in supplemental Table S1. Gene cloning and DNA manipulation and the transformation of E. coli strains were performed according to standard methods (12). E. coli wild-type strain, BW25113, their derivatives (deletion mutants), and plasmids, pCA24N and pDsbA were supplied by the National BioResource Project (NBRP). A Hpx Ϫ mutant strain lacking two catalases (KatE and KatG) and a peroxidase (AhpCF)(10) was kindly provided by James A. Imlay. High L-cysteine-producing plasmid pDES (supplied by Ajinomoto) is a derivative of pACYC184 containing the altered cysE gene encoding the L-cysteine feedback inhibition-insensitive mutant SAT (T167A), the wild-type ydeD gene encoding inner membrane L-cysteine transporter (5), and the altered serA gene encoding the L-serine feedback inhibition-insensitive mutant of D-3-phosphoglycerate dehydrogenase (T410stop). Each gene fragment is under the control of the constitutive promoter of the E. coli ompA gene encoding outer membrane protein A precursor (13). The medium copy number vector pSTV29 was purchased from Takara Bio Co. (Kyoto, Japan). The construction of pYdeD has already been described (8).
Media and Growth Conditions-Luria-Bertani (LB) complete medium or SM1 medium that is supplemented with LB broth, L-methionine, and thiosulfate (8) was used for the general cultivations of E. coli. When appropriate, antibiotics were added at 50 g/ml (for kanamycin), 30 g/ml (for chloramphenicol), and 10 g/ml (for tetracycline). Growth of cultures was monitored by measuring of the optical density at 660 nm (OD 660 ). H 2 O 2 was added to the medium at the indicated concentration. For solid medium, 1.5% agar was added.
L-Cystine Uptake Assays-Cells grown to mid-exponential phase were harvested by centrifugation, washed twice with cold KPM solution (10 mM MgSO 4 , 0.1 M K 2 HPO 4 ; pH was adjusted to 6.5 with H 3 PO 4 ), and suspended in cold KPM solution to a density of 10 8 cells per ml. Portions of the cell suspension (6 ml each) were energized, by the addition of 0.57 ml of 40% D-glucose, followed by incubation for 10 min at 37°C. The L-cysteine uptake assay was initiated by the addition of 2.5 l of L-[ 14 C]cystine (291.3 mCi/mmol). Following the incubation of the cells at room temperature for the indicated time, the cells were collected by filtration through a GF/C filter (Whatman), and the cells collected on the filter were washed three times with KPM solution. Then, the radioactivity derived from 14 C incorporated into the cells on the filter was determined by liquid scintillation counter LS6500 (Beckman).
Preparation of Intracellular L-Cysteine-After E. coli cells were grown to stationary phase at 30°C in LB medium or SM1 medium that was supplemented with L broth, L-methionine, and thiosulfate, 1 ml of the cell culture was harvested, washed with distilled water, and suspended in 0.2 ml of distilled water. The intracellular amino acids were then extracted from the cells by boiling the cell suspension for 10 min using a heat block. After centrifugation (1 min at 15,000 ϫ g) of the heated sample, the supernatant was used as an intracellular amino acid extract (7).
Quantification of L-Cysteine-The amount of L-cysteine in culture supernatants was determined according to the method of Gaitonde (14). 100 l of the sample was incubated with 200 l of Gaitonde reagent (250 mg ninhydrin dissolved in a mixture of 4 ml of concentrated HCl and 16 ml of glacial acetic acid) at 100°C for 15 min. Under the strongly acidic conditions, ninhydrin reacts specifically with L-cysteine even in the presence of other thiols, forming a pink-colored product (E max 560 nm). The reaction product was immediately cooled on ice and diluted to 1.8 ml with 99.5% (v/v) ethanol. The concentration of L-cysteine in the original sample was then determined by measuring the absorbance at 560 nm of the diluted sample.
Quantification of Total Free L-Cysteine (L-Cysteine Plus L-Cystine)-To determine the amount of the total L-cysteine, L-cystine in the sample was reduced by incubation with 5 mM lacpromoter, pACYC184-based, Cm R (8) pYdeD pSTV29, ydeDgene on 1.5 kb DNA fragment (8) dithiothreitol in 200 mM Tris-HCl buffer (pH 8.6) for 10 min.
Then the amount of L-cysteine in the reduced sample was determined using the method by Gaitonde.
Determination of L-Cystine-The amount of L-cystine was calculated by subtracting the amount of L-cysteine from that of total free L-cysteine.
Real-time Quantitative Polymerase Chain Reaction Analysis-Primers used in this study were designed using the Primer Express (Applied Biosystems, Foster City, CA; supplemental Table S1). E. coli cells were lysed with 1 mg/ml lysozyme in 10 mM Tris-HCl (pH 8.0) containing 1 mM EDTA buffer. Total RNA was prepared using the RNeasy Mini Kit (Qiagen, Valencia, CA) and the RNase-Free DNase Set (Qiagen). Complementary DNA was synthesized from 1 g of total RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). For real-time quantitative polymerase chain reaction (qPCR), cDNA was amplified with oligonucleotide primers specific to each target gene using the 7300 Real-Time PCR System (Applied Biosystems). Reactions contained Power SYBR PCR Master Mix (Applied Biosystems), forward and reverse primers (0.1 M each), and a cDNA template (20 ng). For the dissociation curve analysis, the following conditions were used: initial steps at 50°C for 2 min, 95°C for 10 min; 40 cycles of PCR at 95°C for 15 s, 60°C for 1 min; and final steps at 95°C for 15 s, 60°C for 30 s, and 95°C for 15 s (15). The melting curve for each PCR product was determined according to the supplier's guidelines, ensuring specific amplification of the target gene. Quantitative values were obtained as the threshold PCR cycle number (Ct) when the increase in the fluorescent signal of the PCR product showed exponential amplification. The mRNA level of each gene was normalized to that of rrsH in the same sample. The cycle threshold (Ct) value for each reaction was determined using the 7300 Real-time PCR System software package (Applied Biosystems). The Ct values were used to calculate the mean-fold change of the reactions via the 2 Ϫ⌬⌬Ct method for each sample in triplicate, in which 1 indicates no change in abundance (16).
Hydrogen Peroxide Measurements-The concentration of H 2 O 2 in the sample was measured using coloring reagent mixture containing peroxidase, 4-aminoantipyrine, and phenol as a specific pink product forms from oxidative condensation of phenol, 4-aminoantipyrine, and H 2 O 2 . 5 l of sample was incubated with 1 ml of the coloring reagent mixture (Wako) at room temperature for 10 min. Then, the amount of the pink product derived from H 2 O 2 in the original sample was determined by measuring absorbance at 505 nm (A 505 ).
Redox State Analysis on DsbA-To determine the in vivo redox states of DsbA, the free cysteine residues of the protein were acid trapped and alkylated with the high molecular mass reagent AMS (Invitrogen) as described (17). Alkylated samples were separated by SDS-PAGE, and detected by Western blot analysis with anti-DsbA that has been described (18).
Microscopic Analysis-Cultures, grown in LB complete medium at 37°C overnight, were diluted 1:100-fold into the same medium with or without 100 M H2O2. After growth was continued for 8 h, cells were harvested for microscopic analyses. The pictures of the cells were taken with the Axiovert 200 M microscope (ZEISS, Osaka, Japan). Images were collected and processed using the AxioVision 4.5 software (ZEISS, Osaka, Japan).

RESULTS
The L-Cysteine Transporter YdeD Contributes to Hydrogen Peroxide Tolerance in E. coli-To address our hypothesis concerning the role of L-cysteine transporters in the detoxification of H 2 O 2 , we first investigated whether YdeD is involved in H 2 O 2 resistance in E. coli. For this purpose, ⌬ydeD mutant and wild-type cells were transformed with either pYdeD, a middle copy plasmid carrying the ydeD gene with the original promoter, or the empty vector pSTV29. Serial dilutions of the overnight culture in LB medium were spotted onto the surface of a H 2 O 2 -containing LB agar plate. The cells were then grown at 30°C for 24 h. The ydeD deletion mutant cells displayed increased H 2 O 2 sensitivity (Fig. 1A). In contrast, mutants missing either one of the other L-cysteine transporter genes, yfiK or bcr, did not show an increased H 2 O 2 sensitivity (supplemental Fig. S1). Remarkably, both wild-type cells and ⌬ydeD cells overexpressing YdeD showed higher levels of H 2 O 2 tolerance than that of wild-type cells (Fig. 1A). These results demonstrate that YdeD contributes to H 2 O 2 resistance in E. coli.
The levels of H 2 O 2 tolerance were clearly elevated by H 2 O 2 pretreatment (Fig. 1B), suggesting that E. coli has an inducible H 2 O 2 tolerance system. Interestingly, qPCR showed an 11.2fold increase in the ydeD gene expression after treatment with 0.88 mM H 2 O 2 for 30 min (Fig. 1C).
Glutathione is a redox-active tripeptide (L-␥-glutamyl-Lcysteinylglycine; GSH) that is present in the cytoplasm of many organisms. It is reported that GSH also exists in the periplasm of E. coli (19). To examine whether GSH plays a role in the H 2 O 2 resistance of E. coli cells, we tested the effects of the deletion of the gshA gene on the H 2 O 2 resistance of E. coli. This gene encodes ␥-glutamylcysteine synthetase, a key enzyme in GSH synthesis. We found that H 2 O 2 tolerance was increased by YdeD overexpression even in the ⌬gshA mutant (Fig. 1D). Importantly, no significant difference in H 2 O 2 tolerance was observed between the wild-type cells and the ⌬gshA mutant. These findings suggest that, in contrast to L-cysteine, GSH may not contribute to the H 2 O 2 resistance of E. coli cells.
L-Cysteine Is Oxidized to L-Cystine in the Periplasm-As the periplasm is under oxidative conditions, it is inferred that L-cysteine is oxidized into L-cystine in this oxidative cellular compartment. Thus, we investigated whether L-cystine is, in fact, produced in the periplasm. Because E. coli cells maintain an extremely low cellular L-cysteine level, we could not determine the cellular contents of L-cysteine and L-cystine. Therefore, we used cells transformed with pDES, which allows the cells to produce a large amount of L-cysteine and effectively export it into the periplasm. The pDES plasmid encodes serine acetyltransferase (T167A) and D-3-phosphoglycerate dehydrogenase (T410stop), which are released from feedback inhibition by L-cysteine and L-serine, respectively, and YdeD (8).
As shown in Fig. 2A, L-cystine accounted for about 30% of the total free cysteine (L-cysteine plus L-cystine) in E. coli cells carrying the pDES plasmid. The further overexpression of DsbA, which oxidizes peptidyl-L-cysteine residues into a disulfide bridge in the periplasm (20), facilitated L-cystine formation: the ratio of L-cystine to the total cysteine reached about 50% (Fig.  2A). These findings suggest that a significant fraction of endogenously produced L-cysteine is in fact exported into the periplasm, where it is oxidized into L-cystine even in the absence of a specific exogenous oxidative stressor. These observations are consistent with the proposed role of L-cysteine as a reducing equivalent in the periplasm.
The Intracellular L-Cystine/L-Cysteine Ratio Increases Upon Treatment of the Cells with Hydrogen Peroxide-If L-cysteine is used as a reducing equivalent to detoxify H 2 O 2 in vivo, a portion of L-cysteine will be oxidized after treatment of the cells with H 2 O 2 . To test this, we also determined the intracellular contents of L-cysteine and L-cystine after treatment of cells with H 2 O 2 . To detect the changes in the intracellular L-cystine/Lcysteine ratio before and after H 2 O 2 treatment, we used E. coli cells carrying the pDES plasmid. As shown in Fig. 2B, the ratio of L-cystine to the total L-cysteine of the cells after treatment with 0.88 mM H 2 O 2 (52%) was higher than that of the cells before treatment with H 2 O 2 (41%). This ratio reached 73% after treatment with 8.8 mM H 2 O 2 (data not shown). These results are consistent with the finding that endogenous L-cysteine can detoxify exogenous H 2 O 2 in the periplasm.
Overexpression of the L-Cysteine Transporter YdeD Can Affect the Redox Environment of the Periplasm-We showed evidence that the overexpression of the L-cysteine transporter, YdeD, confers H 2 O 2 resistance to E. coli (see above). We next wanted to examine the influence of the L-cysteine transporter overexpression on the redox environment of the periplasm. To this end, we investigated the redox state of DsbA after the overexpression of the L-cysteine transporter from the pYdeD plasmid because DsbA is a periplasmic protein with a pair of redoxactive L-cysteines that can alternate between the oxidized and the reduced states. Importantly, these two L-cysteines are normally maintained in the fully oxidized state in vivo by the membrane protein DsbB, which passes electrons from DsbA to quinones in the respiratory chain (17). To study the oxidative state of DsbA in vivo, we used an alkylating reagent, AMS, which modifies free L-cysteines. The modification of FIGURE 2. L-Cysteine is oxidized to L-cystine in the periplasm. A, ratio of the oxidized L-cysteine (L-cystine) to the total intracellular L-cysteine. The wildtype strain (BW25113) and its ⌬fliY-derivative (JW1905) were transformed with plasmid pDES for the hyperproduction of L-cysteine. Where indicated, pDsbA plasmid was co-transformed to express DsbA. The pCA24N plasmid is an empty vector. Cell extract was prepared by incubating the cell suspension in hot water for 10 min. To determine the total L-cysteine, L-cystine in the sample was once reduced with DTT to L-cysteine. The L-cysteine content was then determined using the method by Gaitonde (14). B, wild-type strain (BW25113) was transformed with pDES plasmid. The transformant was grown to stationary phase in LB supplemented with 3% glucose and 10 mM sodium thiosulfate. The culture was then incubated with 0.88 mM H 2 O 2 for 10 min and the cells were harvested. The ratio of the L-cystine to the total intracellular L-cysteine was determined as described under "Experimental Procedures." free L-cysteines with AMS allows the separation of the oxidized and the reduced form of DsbA on a gel (17). Upon overexpression of the L-cysteine transporter, a fraction of DsbA was indeed reduced in the cells (supplemental Fig. S2). This observation is consistent with the hypothesis that L-cysteine exported into the periplasm by YdeD can function as a reducing equivalent in the periplasm L-Cystine Is Imported Dependently of the Periplasmic L-Cystine-binding Protein, FliY-In Lactobacillus fermentum BR11, the L-cystine binding protein BspA is responsible for L-cystine uptake. FliY of E. coli was identified as a BspA ortholog and has been shown to bind L-cystine (21). However, the mechanism of the L-cystine import remained unclear in this organism. To examine whether the uptake of L-cystine is dependent on FliY, the rate of uptake of L-[ 14 C]cystine in a ⌬fliY strain was compared with that of E. coli wild-type strain. It was found that the intracellular uptake of L-[ 14 C]cystine was significantly impaired in the ⌬fliY mutant, only 37% of that in wild-type cells (Fig. 2C), indicating that FliY is required for the efficient uptake of L-cystine by E. coli.
Consistent with the role of FliY in the uptake of L-cystine, the disruption of the fliY gene increased the ratio of intracellular L-cystine to L-cysteine plus L-cystine by about 50% (Fig.  2A). Remarkably, the ratio reached 96% in the ⌬fliY cells that overexpress DsbA from the pDsbA plasmid. It should be noted that, in these experiments, we used an E. coli strain that exports a large amount of L-cysteine so that we could detect L-cysteine and L-cystine in the periplasm (see above). These results led us to draw a model in which endogenous L-cysteine is exported to the periplasm by YdeD and oxidized to L-cystine, which is then imported back into the cytoplasm in a FliY-dependent manner.
FliY Is Also Involved in the H 2 O 2 Resistance of E. coli-We envisaged that, in addition to YdeD, FliY may also contribute to H 2 O 2 tolerance as it can be expected that L-cystine uptake will promote the recycling of the L-cystine into L-cysteine (22). To test this possibility, serial dilutions of ⌬fliY or ⌬ydeD mutant cells were spotted onto the surface of an H2O2-containing LB agar plate. The cells were then grown at 30°C for 24 h. The ⌬fliY and ⌬ydeD mutants showed a higher sensitivity to H 2 O 2 than wild-type cells (Fig. 3A). Moreover, in liquid medium, the growth of the ⌬fliY cells was completely inhibited in the presence of 0.5 mM H 2 O 2 , though the mutant would normally grow in the same medium without H 2 O 2 (Fig. 3B). The growth defect of ⌬fliY was partially restored by the transformation of the mutant cells with the pFliY plasmid. These results demonstrate that FliY contributes to growth in liquid medium with H 2 O 2 .
It should be noted that, as we showed before, overexpression of YdeD conferred an increased H 2 O 2 resistance to the wild-type cells. However, such an increased resistance was not observed when FliY was overproduced in the wild-type cells (not shown and Fig. 3D). The difference in the effect of protein overexpression between YdeD and FliY may reflect the fact that, while YdeD directly promotes the export of L-cysteine into the periplasm, FliY is involved in the uptake of L-cystine that has been exported in the form of L-cysteine and is then oxidized in the periplasm. Nevertheless, the increased sensitivity of the ⌬fliY mutant to H 2 O 2 indicates the importance of the L-cystine uptake system in the H 2 O 2 defense mechanism of E. coli. Taken together, we conclude that YdeD and FliY are the key components in the L-cysteine/L-cystine shuttle system, which confers H 2 O 2 resistance to E. coli cells.
The L-Cysteine/L-Cystine Shuttle System Is Induced by H 2 O 2 -Next, we elucidated whether this system is inducible under oxidative stress. For this purpose, we quantified the expression level of related genes using real-time qPCR. As positive controls for the H 2 O 2 treatment of the cells, we also examined the expression of the katG and ahpC genes, which are known to be up-regulated in response to H 2 O 2 in an oxyR-dependent manner. The levels of katG and ahpC expression increased significantly under the stress conditions (Fig. 4), consistent with the observation by Strorz et al. (23). Among the genes encoding L-cysteine/L-cystine transporters, ydeD, yfiK, and fliY were highly induced, at 11.2-, 4.0-, and 10.2-fold, respectively, after treatment of the cells with 0.88 mM H 2 O 2 for 20 min (Fig. 4). Both YdeD and FliY are key components involved in H 2 O 2 resistance of E. coli. The dramatic increases in the expression of the genes encoding these proteins are consistent with our observation in Fig. 1B that the levels of tolerance of E. coli to H 2 O 2 were elevated by pretreatment of the cells with this reagent.
Curiously, in addition to H 2 O 2 , exogenous L-cysteine also induced the expression of ydeD, yfiK, and fliY genes (Fig. 5). However, the induction of these genes by exogenous L-cysteine was not as significant as that by H 2 O 2 (Fig. 4). Further, in contrast to the katG and ahpC genes whose H 2 O 2 -dependent induction was abolished in an ⌬oxyR mutant, expression of the ydeD and fliY genes was still highly induced by H 2 O 2 even in the same ⌬oxyR mutant. 4 Thus, the induction of the ydeD and fliY genes by H 2 O 2 may not depend on oxyR, which is a regulator of oxidative stress response.
Our analysis of the expression of the other relevant genes during H 2 O 2 stress also revealed their unique regulatory features. First, the expression of the cysE gene encoding serine acetyltransferase, a key enzyme in L-cysteine biosynthesis and the tnaA gene encoding a major L-cysteine desulfhydrase, was not induced, suggesting that H 2 O 2 treatment does not promote L-cysteine production. Second, the system to uptake periplasmic GSH is composed of L-glutamyl transferase (Ggt), which degrades GSH into L-glutamate and L-cysteinylglycine in the periplasm; the machinery importing L-cysteinylglycine encoded by yliA, B and C; and cytosolic aminopeptidase, pepA, B, D, and N (24). Among them, H 2 O 2 treatment induced the expression of ggt (12.7-fold), yliA (4.0-fold), yliB (5.4-fold), yliC (4.8-fold), pepB (4.2-fold), and pepN (7.4-fold). However, no expression of gshA, a gene encoding a key enzyme of GSH synthesis, or cydD (25), encoding a GSH exporter, was induced. In addition, cellular L-cysteine was undetectable even under oxi-dative stress conditions. These findings indicate that, under oxidative stress conditions, E. coli up-regulates the expression of the genes involved in the export of L-cysteine or the uptake of L-cystine but not the expression of the genes involved in the synthesis of L-cysteine. Thus, to resist oxidative stress, E. coli appears to promote the L-cysteine/Lcystine shuttle system rather than enhancing the synthesis of L-cysteine.
The  Fig. 6, the growth of the Hpx Ϫ mutant was partially but clearly restored by the transformation of the mutant cells with the pYdeD plasmid. In addition, overexpression of YdeD in the Hpx Ϫ mutant decreased the number of the cells forming filaments. These finding further supports the protective role of YdeD against H 2 O 2 .

DISCUSSION
Mutants missing the L-cysteine transporter, YdeD, or the L-cystine-binding protein, FliY, exhibited an increased sensitivity to H 2 O 2 . Further, our data indicated that E. coli FliY is involved in the uptake of L-cystine from the periplasm to the cytoplasm. Interestingly, the expressions of the genes encoding these proteins were dramatically increased upon treatment of the cells with H 2 O 2 . These findings led us to propose that E. coli removes periplasmic H2O2 using L-cysteine supplied to the periplasm from the cytoplasm by a L-cysteine/L-cystine shuttle system (Fig. 7).
The question then is why does E. coli require such a L-cysteine/L-cystine shuttle system? The inner membrane is the place where E. coli produces ATP via the respiratory chain. The process often generates superoxide (O 2 . ) due primarily to the leakage of electrons (Fig. 7) (9). The respiratory electron 4 I. Ohtsu, unpublished observation.    (27). We suggest that the absence of an efficient enzyme catalyst to remove H 2 O 2 from the periplasm may be one of the reasons why L-cysteine is exported to the periplasm.
Interestingly, H 2 O 2 but not exogenous L-cysteine (Fig. 5) significantly induced the expression of the ydeD gene (Figs. 1C, 4, and 5). Thus, YdeD seems to be intrinsically used to protect cells from H 2 O 2 . However, other L-cysteine transporters, such as Yfik and Bcr, might not contribute much to this defense mechanism because the mutants that are missing one of the latter genes displayed no increase in H 2 O 2 sensitivity (supplemental Fig. S1) GSH is a predominant low molecular weight thiol in E. coli, as in the case of many other organisms, and it has been proposed to protect cells from oxidative damages such as those caused by H 2 O 2 and ionizing radiations (28). However, it has been reported that a ⌬gshA mutation does not cause increased sensitivities to H 2 O 2 or cumene hydroperoxide (29). In addition, we, ourselves, observed here that there was no detectable change in the levels of resistance to H 2 O 2 between the wild-type cells and the ⌬gshA mutant whether or not YdeD is overproduced (Fig. 1D). Thus, we propose that E. coli uses L-cysteine to protect cells from periplasmic H 2 O 2 even though GSH exists in the periplasm as well as in the cytoplasm (17). Interestingly, under H 2 O 2 stress, genes involved in the utilization of periplasmic GSH as a source of L-cysteine were induced (Fig. 4). This finding may imply that the periplasmic GSH is also used as a source of L-cysteine that is to be provided to the periplasm to reduce oxidative stress.
Furthermore, we showed evidence that FliY is involved in L-cystine uptake in E. coli (Fig. 2C). FliY is a homologue of BspA, an L. fermentum BR11 L-cysteine binding protein. BspA is a component of the L-cysteine transport system in this organism.
Here we showed that FliY is an important component of L-cystine import and that this protein is necessary for resistance of E. coli to oxidative stress (Fig. 3, A and B). The latter finding is consistent with the previous observation in L. fermentum (30). It should be noted that, as FliY is a periplasmic solute-binding protein, FliY probably cooperates with a certain transporter for the uptake of L-cystine (Fig. 7).