The Scs disulfide reductase system cooperates with the metallochaperone CueP in Salmonella copper resistance

The human pathogen Salmonella enterica serovar Typhimurium (S. Typhimurium) contains a complex disulfide bond (Dsb) catalytic machinery. This machinery encompasses multiple Dsb thiol-disulfide oxidoreductases that mediate oxidative protein folding and a less-characterized suppressor of copper sensitivity (scs) gene cluster, associated with increased tolerance to copper. To better understand the function of the Salmonella Scs system, here we characterized two of its key components, the membrane protein ScsB and the periplasmic protein ScsC. Our results revealed that these two proteins form a redox pair in which the electron transfer from the periplasmic domain of ScsB (n-ScsB) to ScsC is thermodynamically driven. We also demonstrate that the Scs reducing pathway remains separate from the Dsb oxidizing pathways and thereby avoids futile redox cycles. Additionally, we provide new insight into the molecular mechanism underlying Scs-mediated copper tolerance in Salmonella. We show that both ScsB and ScsC can bind toxic copper(I) with femtomolar affinities and transfer it to the periplasmic copper metallochaperone CueP. Our results indicate that the Salmonella Scs machinery has evolved a dual mode of action, capable of transferring reducing power to the oxidizing periplasm and protecting against copper stress by cooperating with the cue regulon, a major copper resistance mechanism in Salmonella. Overall, these findings expand our understanding of the functional diversity of Dsb-like systems, ranging from those mediating oxidative folding of proteins required for infection to those contributing to defense mechanisms against oxidative stress and copper toxicity, critical traits for niche adaptation and survival.

The bacterial cell envelope harbors disulfide bond (Dsb) 2 proteins, a group of redox-active enzymes characterized by a thioredoxin (TRX)-like domain and a conserved CXXC catalytic motif (1). Dsb enzymes introduce disulfide bonds to diverse substrates, including secreted toxins, bacterial adhesins, and nutrient acquisition and secretion systems, which are required for bacterial fitness and virulence (1,2). Typically, Dsb proteins form two separate redox pathways, the oxidative DsbA/B pathway, which oxidizes thiols in substrate proteins, and the reductase/isomerase DsbC/D pathway, which corrects nonnative disulfide bonds (3).
Salmonella enterica serovar Typhimurium (S. Typhimurium) is a main cause of food-borne illnesses such as acute gastroenteritis and can cause life-threatening bacteremia in children and immunocompromised individuals (4 -6). This Gram-negative facultative intracellular pathogen displays an unusual complexity in the Dsb machineries it encodes. In addition to the disulfide bond-forming DsbA/B system (7) and a DsbC/D isomerization pathway (8), S. Typhimurium contains a separate DsbL/I redox pair (9, 10), a plasmid-encoded DsbA homologue SrgA (9,11), and a set of related and less characterized proteins encoded by the suppressor of copper sensitivity (scs) locus ( Fig. 1A) (12).
The scs locus encompasses four proteins (ScsA-D). ScsA and ScsB are predicted integral membrane proteins; ScsA has two membrane-spanning ␣-helices (12), and ScsB contains two periplasmic domains (n-and c-ScsB) that frame a central transmembrane domain (t-ScsB) (13). ScsC is a periplasmic soluble protein, and ScsD has a predicted N-terminal membrane anchor joined to a periplasmic domain (12). Scs proteins contain the hallmarks that characterize Dsb proteins; they all incorporate a putative catalytic CXXC motif, which in the case of ScsC, c-ScsB, and ScsD, is imbedded in a TRX fold (12,14). The conservation of the Dsb protein features in the Scs family suggests that they are participants in dithiol/disulfide interchange reactions.
The Scs proteins were first identified for their ability to confer resistance to copper stress in Salmonella (12). Copper plays a central antibacterial role in the innate immune system of many organisms through its oxidizing effects on damaging cell membranes and generating reactive oxygen species along with its ability to displace enzyme iron-sulfur complexes (15). During systemic infection, Salmonella is exposed to elevated copper concentrations within the macrophage phagosome, and copper tolerance is a critical adaptive trait for its survival (16 -18). Copper resistance mechanisms in Salmonella include the cue/gol regulon, which contains a cytoplasmic sensor, CueR, that upon stimulation by toxic Cu(I) induces the expression of plasma-membrane P-type copper-transporting P1B-type ATPases such as CopA and GolT, which remove excess Cu(I) from the cytoplasm; the periplasmic cuprous oxidase CueO; and the metallochaperone CueP (17,(19)(20)(21). The latter is a major periplasmic Cu(I)-binding protein required for Cu tolerance and Cu delivery to the stress response Cu/Zn-superoxide dismutase (Cu/Zn-SOD) (22,23). CueP is also regulated by the Cu-depandant CpxR/CpxA two-component regulatory system, which additionally regulates the transcription of the scs operon (23,24). The coregulation of CueP and Scs proteins suggests a concerted action in copper homeostasis; however, the mechanism underlying Scs-mediated copper resistance had remained unknown.
In this study, we have investigated the Salmonella Scs electron transfer pathway. Through redox and kinetics studies, we demonstrate that ScsB and ScsC form a redox pair and that this system is maintained separate from the DsbA oxidizing machinery but cross-talks with the DsbD reductase system. We also investigated the molecular basis of ScsB/ScsC-mediated copper tolerance. Combination of quantitative copper binding and copper transfer experiments revealed that both ScsB and ScsC tightly bind Cu(I) and are able to deliver it to periplasmic CueP. These results provide an unprecedented biochemical insight into the mechanism underlying the ScsB/ScsC-mediated copper resistance trait that involves an interplay between the Scs system and the cue regulon, a major copper resistance machinery in Salmonella.

Distribution of Scs systems across bacteria
Although the scs cluster was first identified in S. Typhimurium (12), genes encoding for Scs proteins are widespread across Proteobacteria (13). Using STRING analysis (25), we reevaluated the distribution of the scs operon across the everincreasing number of publicly available bacterial genomes. Analysis of the 1,678 bacterial genomes showed that the components of the scs operon are more frequently found in ␣and ␥-Proteobacteria and are less common in ␤and ␦-Proteobacteria (Table S1). The full scs operon appears to be limited to some Enterobacteriaceae, including Salmonella enterica, Serratia proteamaculans, Citrobacter koseri, and Klebsiella pneumoniae along with Aeromonas hydrophila (Aeromonadaceae family) and Photobacterium profundum (Vibrionaceae family). Other species in the Proteobacterium phylum contain all scs genes but scsA, whereas most ␣-Proteobacteria primarily encode ScsB and ScsC proteins, which also occur in Fibrobacteres, Bacteroidetes, and Planctomycetes (Table S1). The widespread distribution of ScsB and ScsC is important given the role these two proteins play in bacterial fitness (24). In this study, we have investigated the Salmonella ScsB and ScsC redox system and the molecular basis for their role in reducing copper toxicity.

Salmonella ScsC structurally resembles DsbA but lacks dithiol oxidase activity
The recent characterization of ScsC proteins from different organisms has uncovered a remarkable diversity in structure and function. Structural and biochemical studies of S. Typhimurium ScsC revealed that this protein resembles the DsbAlike thiol oxidase, both structurally and in redox properties (14) (Fig. 1B). Given the similarities between Salmonella ScsC and DsbA, we investigated whether ScsC displayed thiol oxidase activity in a peptide oxidation assay where we employed a fluorescently labeled peptide that is known to be oxidized by DsbAlike proteins (26). The oxidation activity was monitored by the increment of europium fluorescence resulting from peptide cyclization via disulfide bond formation between the terminal cysteines (Fig. 1C). Although DsbA was able to efficiently oxidize the standard peptide substrate as indicated by the increase in fluorescence intensity over time, equivalent concentrations of ScsC showed a fluorescence profile comparable with that of background oxidation (Fig. 1C). These results indicate that, despite the similarities between ScsC and DsbA in their structures and reduction potentials, they have different oxidoreductase functions probably with different substrates.

ScsB and ScsC form a redox pair in S. Typhimurium
Previous work has shown that ScsC exists in the reduced state in the Salmonella periplasm (14). We assessed the ability of ScsB, the predicted thiol reductase encoded immediately upstream of ScsC, to catalyze its reduction. A gel-shift assay was used to monitor the redox state of ScsC upon incubation with the N-terminal domain of ScsB (n-ScsB). Stoichiometric amounts of reduced n-ScsB (n-ScsB red ) and oxidized ScsC (ScsC ox ) were incubated, and samples were taken from the reaction mixture at 15-, 120-, and 300-s time points. Determination of the redox state of ScsC was done by alkylating free thiols with 4-acetoamido-4Ј-maleimidylstilbene-2,2Ј-disulfonic acid (AMS), which increased the mass by 500 Da for each free cysteine (1 kDa total) that could be detected by SDS-PAGE.
These experiments showed that ScsC ox ( Fig. 2A, lane 2) was rapidly reduced by n-ScsB red (Fig. 2A, lane 6) as shown by the SDS-PAGE analysis of AMS-derivatized samples ( Fig. 2A, lanes  3 and 4) as well as Western immunoblotting using anti-ScsC antibody (Fig. S1A). These data show that Salmonella ScsC and ScsB form an efficient redox relay. We also investigated the reverse electron flow by mixing n-ScsB ox and ScsC red (Fig. 2B, lanes 6 and 2, respectively) and showed that reduced ScsC was unable to transfer electrons to oxidized n-ScsB (Figs. 2B, lanes 3-5, and S1B).

Kinetic characterization of the ScsB-ScsC interaction
To further characterize the interaction between n-ScsB and ScsC, we used surface plasmon resonance (SPR) to determine binding affinity (K D ) of these two proteins in different redox states. We first analyzed the n-ScsB red -ScsC ox electron transfer complex by immobilizing n-ScsB red onto a CM5 sensor chip and injecting a concentration series of ScsC ox over the chip Salmonella ScsB and ScsC protect against toxic copper surface, which gave a K D of 310 Ϯ 46 M ( Fig. 2C and Table 1). This micromolar binding affinity is comparable with the previously reported K D between Escherichia coli n-DsbD ox and c-Ds-bD red (K D of 86 M (27)) and characteristic of the transient protein-protein interactions between Dsb-like proteins. Interestingly, a similar K D value of 287 Ϯ 34 M (Table 1 and Fig.  S2B) was obtained for the reverse interaction between ScsC red and n-ScsB ox , which, as shown before, does not favor an electron transfer reaction. SPR experiments were also carried out for the two proteins in the same redox state, which yielded K D values of 233 Ϯ 56 and 361 Ϯ 12 M for n-ScsB red -ScsC red and n-ScsB ox -ScsC ox , respectively (Table 1 and Fig. S2, A-C). To validate these binding data, we immobilized ScsC on the chip and injected n-ScsB, which yielded similar results (data not shown). These kinetic studies show that the interaction between n-ScsB and ScsC is oxidation state-independent, as evidenced by the similar K D values obtained for all investigated complexes ( Table 1).

The interaction between ScsB and ScsC is thermodynamically driven
The reduction potential of n-ScsB in equilibrium with different concentrations of oxidized and reduced dithiothreitol (DTT) was determined by monitoring the relative amounts of oxidized and reduced protein using AMS alkylation of free thiols. Protein-AMS adducts with increased molecular mass were detected by SDS-PAGE analysis (Fig. 2D). An equilibrium constant of (1.24 Ϯ 0.15) ϫ 10 Ϫ2 was calculated by plotting the relative amount of reduced n-ScsB versus [DTT red ]/[DTT ox ]. This K eq value corresponds to a standard reduction potential of Ϫ256 Ϯ 10 mV, which is substantially more reducing than that of ScsC (Ϫ132 mV) (14). Collectively, these results show that, although kinetics factors may contribute to the interaction between n-ScsB red and ScsC ox (Table 1), thermodynamic factors favor the electron flow between these proteins and prevent the unproductive reverse electron transfer from n-ScsB ox to ScsC red .

The ScsB-ScsC reducing system interacts with DsbA less efficiently
Given the similarity in three-dimensional structures and reduction potentials between ScsC and DsbA, we wondered whether n-ScsB could also reduce the thiol oxidase DsbA. Using the AMS gel-shift assay, we could only detect a partial reduction of DsbA (Fig. 3A, lane [3][4][5] when DsbA ox (Fig. 3A, lane 2) was incubated with n-ScsB red (Fig. 3A, lane 6). Indeed, although n-ScsB red reduced 75% of ScsC ox after 15 s, only 28% of DsbA was reduced by n-ScsB red in the same time frame (gel

Salmonella ScsB and ScsC protect against toxic copper
band intensity analysis using ImageJ 1.45). This indicates a comparatively slower electron transfer process from n-ScsB red to DsbA ox . Furthermore, although the amount of reduced DsbA increased as time progressed, reduction of the protein did not go to completion after 5-min incubation (57% DsbA reduction after 5-min incubation versus 100% ScsC reduction). We also attempted to determine the dissociation constant for the interaction between n-ScsB red and DsbA ox by SPR (Fig. S2D). DsbA has a molecular weight comparable with that of ScsC; however, the binding response for DsbA at 500 M was only 11 RU compared with an ϳ400 RU binding response for ScsC at 585 M. These results suggest that the interaction between DsbA and ScsB is too weak to be detected by SPR and to be biologically relevant.
As oxidized DsbA and reduced ScsC coexist in the periplasm, the cross-talk between these proteins was also tested. When DsbA ox (Fig. 3B, lane 2) was mixed with ScsC red (Fig. 3B, lane 6), no reduction of DsbA was detected even after 5-min incubation. Together, these results support the hypothesis that in the periplasm of Salmonella the reducing Scs system is kept separate from the DsbA oxidative pathway, probably by means of enzyme specificity.

Cross-talk between Salmonella ScsC/B and DsbC/D reducing systems
We investigated the interaction between the Salmonella ScsB/ScsC and DsbD/DsbC reducing pathways by assessing the ability of the two membrane-bound reductases, n-DsbD and n-ScsB, to reduce ScsC and DsbC, respectively. AMS-based gelshift assays showed that reduced n-DsbD efficiently transfers electrons to oxidized ScsC (Fig. 3C), whereas the reverse reac-  [3][4][5] indicates no reduction of this protein upon mixing with an equimolar amount of reduced ScsC. C, SPR analysis of ScsC binding to immobilized n-ScsB. The representative sensorgram of n-ScsB red binding to ScsC ox was double referenced and is shown together with equilibrium binding analysis. A series of concentrations (0 -650 M of oxidized ScsC) were injected over the reduced n-ScsB immobilized surface. The apparent equilibrium dissociation constant (K D ) was determined using a steady-state affinity model. The data are expressed as mean Ϯ S.E. All experiments were conducted on at least three independent occasions with fresh immobilization. D, reduction potential determination of n-ScsB and redox equilibrium of n-ScsB. n-ScsB was equilibrated with redox buffers containing various concentrations of reduced and oxidized DTT. After 24 h, all samples were treated with 10% TCA followed by AMS alkylation and separation using SDS-PAGE (bottom panel). The fraction of reduced n-ScsB was plotted against the [DTT red ]/[DTT ox ] ratio (top panel). The plots were fitted to Equation 1 to determine K eq as described under "Materials and methods." Error bars correspond to the standard deviation calculated from three independent experiments.

Salmonella ScsB and ScsC protect against toxic copper
tion did not result in any apparent change in the redox state of the proteins (Fig. S1C). The n-DsbD red -ScsC ox interaction was also analyzed by SPR, which yielded a K D of 422 Ϯ 26 M, which is similar to the K D of the n-ScsB red -ScsC ox interaction (Table  1). With regard to the n-ScsB red and DsbC ox electron transfer, this occurred less readily under the conditions tested ( Fig. 3D), although the two proteins seem to interact by SPR, yielding a K D of 334 Ϯ 26 M (Table 1).

n-ScsB, ScsC, and CueP bind Cu(I) with affinities in the subpicomolar to femtomolar range
The scs operon, particularly ScsB and ScsC, provides copper tolerance to Salmonella (14,24). To understand the mechanism underpinning the scs-mediated increased copper tolerance, the copper binding properties of n-ScsB and ScsC were quantified and compared with that of CueP, a Salmonella specific copper chaperone required for the activation of Cu/Zn-SOD in the periplasm (17). CueP together with the scs cluster are coregulated by the CpxR/CpxA envelope stress response (22,24).
We used a competition assay involving the chromophoric Cu(I) chelator bicinchoninic anion (Bca) to accurately determine the Cu(I) binding affinities to each protein (28). Titration of increasing concentrations of n-ScsB red into a series of solutions containing the same amount of [Cu I (Bca) 2 ] 3Ϫ with 0.25fold excess Bca led to an initial linear decrease in absorbance at 562 nm. This decrease is known to occur when copper is removed from the [Cu I (Bca) 2 ] 3Ϫ (Fig. 4A, i) (Table 2).
Equivalent experiments with CueP red generated two very different binding curves i and ii for the conditions [Bca] tot ϭ 80 and 300 M, respectively (Fig. 4C) ii was nonlinear, indicating competitive Cu(I) binding under higher concentrations of Bca. This allows a satisfactory curve fitting of the experimental data to a two-site binding model and derivation of a K D of 10 Ϫ15.0 M for CueP assuming identical affinity for the two sites ( Table 2).
The above experiments demonstrate that CueP possesses a Cu(I) binding affinity comparable with that of n-ScsB but its Cu(I) binding capacity doubles that of n-ScsB. In contrast, the Cu(I) affinity of ScsC is ϳ100 times weaker than that of CueP or n-ScsB.

ScsC binds Cu(II) with subnanomolar affinity
The environment of periplasm is generally more oxidizing than that of cytosol, and the copper ions in periplasm may exist in redox equilibrium between Cu(I) and Cu(II) (21). Therefore, the possibility of periplasmic proteins ScsC and CueP interacting with Cu(II) was investigated. The first experiment was designed to test whether Cu(II) may be reduced by the free Cys thiols in ScsC and CueP upon interaction. ScsC red and CueP red were each incubated with 1 eq of Cu(II) (added as CuSO 4 solution). After 0.5-h incubation under anaerobic conditions, unreacted Cys thiols were trapped with excess iodoacetamide (IAA; 100 eq) in the presence of bathocuproinedisulfonate (BCS) and EDTA (each 5 eq). BCS and EDTA exhibit high affinity for Cu(I) and Cu(II) (log ␤ 2 ϭ 19.8 for [Cu I (BCS) 2 ] 3Ϫ and log K a ϭ 15.9 for Cu(II)-EDTA at pH 7.4) (28) and can free Cys thiols in each protein from copper binding, allowing an effective trapping by IAA.
The reaction mixture was analyzed for Cys oxidation state by electrospray ionization MS. Control samples of ScsC red and CueP red without added Cu(II) were also analyzed. Approximately 15% of the reduced ScsC was oxidized by Cu(II) to the disulfide form, less than the predicted 50% oxidation for a full Cu(II) reduction. In contrast, the reduced CueP was oxidized stoichiometrically by 1 eq of Cu(II) to produce about 50% oxidized form, which contained one disulfide bond (Fig. 5, A and B, and Table S2). These experiments demonstrate that reduced ScsC is somewhat resistant to oxidation by Cu(II), but reduced CueP reacts with Cu(II) and therefore is not compatible for Cu(II) binding studies.
The interaction of ScsC with Cu(II) was evaluated quantitatively with the two dansyl peptide (DP) probes, DP1 and DP2, recently reported (30). These probes fluoresce intensively at max ϳ550 nm upon excitation at ϳ330 nm, but Cu(II) binding induces fluorescence quenching. They each bind 1 eq of Cu(II) with different affinities (K D ϭ 10 Ϫ8.1 M for DP1 and 10 Ϫ10.1 M for DP2 at pH 7.4) and are effective probes for detection of Cu(II) binding to proteins with comparable affinity. ScsC ox was tested initially to avoid potential complication of Cu(II)-induced protein thiol redox chemistry. Titration of Cu 2ϩ (taken as CuSO 4 ) into a solution containing an equal molar concentration of DP1 and ScsC ox induced initially minor fluorescence quenching with two apparent turning points at [Cu(II)]/[DP1] of ϳ1.0 and ϳ2.0, respectively, whereas a control titration of DP1 probe generated only one titration turning point at [Cu(II)]/[DP1] of ϳ1.0 (Fig. 5C). This indicates that ScsC ox can bind 1 eq of Cu(II) with affinity considerably higher than that of DP1, which was estimated to be K D ϭ 10 Ϫ9.5 M at pH 7.4 ( Table   Table 2

With DP1 With DP2
n-ScsB(SH) 2 Ϫ9.6 (1) CueP(SH) 3 2 The parenthetical values are the errors in the last digits estimated from two sets of experiments with a total of about 10 individual experimental data points. b Not determined due to the redox sensitivity between Cu 2ϩ and the reduced protein form.

Salmonella ScsB and ScsC protect against toxic copper
2). This value was matched by an equivalent independent estimation with the DP2 probe that derived a K D of 10 Ϫ9.6 M for ScsC ox (Fig. 5D and Table 2). Intriguingly, equivalent experiments with ScsC red estimated an indistinguishable K D of 10 Ϫ9.6 M with either DP1 or DP2 probe (Fig. 5, C and D). This revealed that the Cys-XX-Cys motif is not involved in Cu(II) binding, although they are the key ligands for Cu(I). Although an equal molar mixture of Cu(II) and ScsC red detected partial oxidation of the two Cys thiol groups (Fig. 5A), such redox chemistry is less likely under the experimental conditions where Cu(II) is constrained stably by the DP probes. However, the nature of the Cu(II)-binding site in ScsC has yet to be determined. An equivalent experiment with the DP1 probe and n-ScsB ox estimated a Cu(II) binding affinity of K D of ϳ10 Ϫ8.4 M, which is weaker than that of ScsC ox but slightly stronger than that of DP1 (Fig. S3). Furthermore, with the higher-affinity probe DP2, we were not able to detect the weak Cu(II) binding to the n-Scs-B ox protein. However, the low levels of Cu(II) reduction by n-ScsB red prevented a meaningful estimation of its Cu(II) binding affinity.

ScsB and ScsC transfer Cu(I) to CueP
To further explore the biological role of Scs proteins in copper resistance (12, 24), we investigated the ability of Scs proteins to transfer bound copper to CueP using a well-established pro-tocol for detection of copper transfer (29,31,32). The periplasmic Cu chaperone CueP binds copper in its reduced state (22). The Cu(I)-bound forms of n-ScsB and an MBP-fused ScsC, which was prepared to increase the molecular weight of ScsC for separation from CueP, were incubated for 0.5 h with reduced apo-CueP in a 1:1 molar ratio. The lack of Cu(I) in reduced apo-CueP was confirmed prior to mixing with the Cu(I)-loaded Scs proteins (Fig. S4). The protein mixture was then loaded onto a size-exclusion column, and elution fractions were analyzed by SDS-PAGE as well as spectrophotometrically after the addition of the Cu(I) probe BCS. These experiments showed that Cu(I) is primarily present in the CueP-containing elution fractions but not in those corresponding to n-ScsB and ScsC (Fig. 6, A and B), indicating that both n-ScsB and ScsC transferred Cu(I) to CueP in vitro. The similar Cu(I) binding affinities of ScsB and CueP are consistent with an incomplete transfer of Cu(I) between the two proteins (Fig. 6B). Conversely, the Cu(I) binding affinity of ScsC is ϳ100 times weaker than that of CueP (Table 2), allowing a complete transfer of Cu(I) from ScsC to CueP (Fig. 6A).
Given that the reduced form of CueP is the functional form for copper binding (33), we also investigated whether CueP is maintained reduced by the Scs reducing system. AMS free thiol-trapping assays did not show any productive redox interactions between ScsB red and ScsC red with CueP ox (Fig. S5, A  and B). Similarly, under our experimental conditions, we did not observe reduction of CueP ox by DsbC red (Fig. S5C). These results seem to contradict previous work that described DsbC as a CueP reductase (33).

Discussion
S. Typhimurium contains an extended collection of periplasmic oxidoreductases, including the Dsb proteins, which contribute to Salmonella fitness and virulence (2,34,35), and the less-studied scs gene cluster, which plays a role in tolerance to high levels of copper (12,24). The work presented here provides the first mechanistic evidence of the dual function of the Scs machinery, which acts as a Dsb-like disulfide reductase system with an additional function in copper homeostasis.
The limited number of detailed studies examining Scs proteins are beginning to uncover an intriguing diversity among individual Scs proteins across different organisms, particularly ScsC. For example, Caulobacter crescentus and Proteus mirabilis ScsCs are dimeric and trimeric isomerases, respectively, that form redox relays with the membrane protein ScsB (13,36). Conversely, S. Typhimurium ScsC is a monomeric protein that structurally resembles DsbA (14). In this study, we aimed to characterize the S. Typhimurium ScsC/ScsB pair and define the molecular basis for the ScsC/ScsB-mediated copper resistance trait.
We first tested the oxidase activity of S. Typhimurium ScsC and showed that, despite the remarkable structural similarity with the DsbA thiol oxidase (Fig. 1B) and their very close reduction potentials (Ϫ132 versus Ϫ126 mV), ScsC is not able to catalyze disulfide formation in the standard peptide substrate used to assay DsbA disulfide transfer rates (9,14). This suggests that ScsC may have a different substrate specificity or act as a reductase of substrates with higher reduction potentials. We

Salmonella ScsB and ScsC protect against toxic copper
have shown that ScsC is maintained in the reduced state in the periplasm of Salmonella, suggesting that this is the functional form of the protein (14). The next question to address was to identify its cognate reductase. The scsC gene is often located directly downstream of scsB, a predicted DsbD-like membrane disulfide reductase (12,13). Using AMS gel-shift assays, we showed that oxidized S. Typhimurium ScsC is rapidly reduced upon addition of reduced n-ScsB, which demonstrates that the two proteins form a redox relay. Previous work showed functional redox interactions between ScsB and oligomeric ScsC in P. mirabilis and C. crescentus (13,36), and our new data demonstrate that monomeric ScsC is also efficiently reduced by n-ScsB. SPR analysis of the interactions between n-ScsB and ScsC in different redox states revealed relatively low binding affinities (ϳ300 M; Table 1), which are consistent with K D values previously reported for transient electron transfer interactions between Dsb proteins (27,37). Remarkably, these affinities were independent of the redox state of the interacting proteins and did not provide a mechanism for the unidirectional flow of electrons between n-ScsB and ScsC. Determination of the reduction potential of the n-ScsB active-site disulfide by equilibration with DTT showed a more reducing reduction potential of n-ScsB (Ϫ256 mV) compared with ScsC (Ϫ132 mV (14)), specifying that the unidirectional electron transfer between these two proteins is thermodynamically driven.
In the Salmonella periplasm, the ScsC/ScsB pair coexists with Dsb oxidative and reductive pathways, and therefore there is the possibility of an interplay between these thiol-disulfide exchange systems. As shown by SPR and AMS gel-shift assays, n-ScsB does not efficiently interact with and reduce DsbA under the conditions tested. This is remarkable given the structural similarity between DsbA and ScsB's cognate substrate ScsC. Similarly, we did not detect any functional interaction between ScsC and DsbA, indicating that these pathways are maintained separate with little or no cross-talk. This is consistent with previous work that showed large kinetic barriers between E. coli oxidative DsbA/B and reductive DsbC/D pathways (38). Additionally, we measured the interplay between the ScsC/ScsB and DsbC/DsbD reducing systems and showed that, although reduced n-DsbD efficiently transfers electrons to oxidized ScsC, n-ScsB reduction of DsbC is inefficient. The apparent redundancy in the catalytic activity of ScsB and the common disulfide reductase DsbD questions the need for a specific reductase for ScsC. We postulate that the presence of a dedicated reductase in the scs gene cluster may be necessary to facilitate the transfer of a fully functional gene cluster across bacteria and to ensure cotranscription and functional coupling of all scs genes. Furthermore, under copper stress, cotranscription of ScsC with its reductase ScsB supports the importance of maintaining reduced ScsC to remove toxic Cu(I) from the cell. Fractions under the peak labeled as "excess copper" were also analyzed by BCS assay and SDS-PAGE, which in all cases showed a high abundance of copper but no protein content (data not shown).

Salmonella ScsB and ScsC protect against toxic copper
Salmonella Scs proteins have been shown to play a role in protection against copper toxicity (14,24,39). Although all Scs proteins contribute to Salmonella copper tolerance, ScsB is the most important Scs factor followed by ScsC (24). In this work, we aimed to decipher the molecular basis of ScsB/ScsC-mediated copper resistance. Our data showed that the two main Scs players in Salmonella copper homeostasis bind Cu(I) with high affinities (K D values of 10 Ϫ14.8 and 10 Ϫ13.1 M for ScsB and ScsC, respectively). The stoichiometry was 1:1, and only the proteins in the reduced form were able to bind Cu(I), suggesting that the catalytic cysteines play a dominant role in Cu(I) binding, which is consistent with the role played by cysteines in other comparable systems (40 -44). Under aerobic conditions, periplasmic copper may cycle between Cu(I) and Cu(II) oxidation states (45). Similar to other metallochaperones (46), ScsC is also able to bind Cu(II), but the determined affinity is much weaker than that for Cu(I).
The transcription of scs and cueP genes is coregulated by the CpxR/CpxA signal transduction system (24). Although the CueP metallochaperone has been shown to contribute to copper tolerance in the absence of oxygen, CueP has also been shown to express and to contribute to copper resistance under aerobic conditions (16,17,47). We therefore investigated a possible interplay between these two machineries. Our studies showed that both ScsB and ScsC can efficiently bind Cu(I) and transfer it to the periplasmic chaperone CueP.
As such, our data allowed us to propose a mechanism where the Scs and CueP systems work in concert in Salmonella copper resistance. Under copper stress conditions, the envelope stress response system CpxR/CpxA drives the production of Scs proteins and the metallochaperone CueP. These periplasmic proteins can bind Cu(I) with high affinity, combating copper toxicity in the periplasm and limiting its movement to the cytosol. The different Cu(I) binding affinities displayed by ScsC, ScsB, and CueP may play a crucial role in copper homeostasis by allowing a dynamic transfer of toxic copper ions from Scs proteins to CueP, which in turn can deliver it to the stress response enzyme Cu/Zn-SOD for its functional activation (Fig. 7). Copper-loaded CueP has been found to be essential for the activation of Cu/Zn-SOD in Salmonella (22). The inner-membrane copper transporters CopA and GolT are known to transfer copper from the cytoplasm to CueP (22). Similarly, we show that ScsB/ScsC fulfill a similar role but specialized for the periplasm, being able to bind copper and pass it to CueP independently of ATP hydrolysis. Additionally, ScsC also binds Cu(II) with a nanomolar affinity, and this may help to scavenge periplasmic Cu(II).
In light of previous work, our findings reveal the ability of the n-ScsB tandem immunoglobulin structural fold to efficiently interact with and reduce a diverse set of ScsC targets, including monomers as shown here (Salmonella ScsC), dimers (Caulobacter ScsC) (13), and trimers (Proteus ScsC) (36,48), while discriminating against closely related Dsb proteins such as monomeric DsbA and dimeric DsbC. Future studies will be required to determine the molecular basis for the intriguing specificity of ScsB for diverse ScsC proteins. The Scs proteins may also transfer reducing power to other periplasmic proteins yet to be identified that could play a role in defense mechanisms against oxidative stress.
In summary, our findings show that ScsB and ScsC form a reducing system in the Salmonella cell envelope. We have provided the first biochemical insight into the molecular mechanism underlying Scs-mediated copper tolerance in Salmonella. In this machinery, Scs reducing proteins provide protection against copper toxicity by sequestering Cu(I) with subpicomolar affinities and transferring it to a major periplasmic copper chaperone, CueP, that binds Cu(I) at femtomolar affinity. Given the distribution of the scs operon across bacteria, similar Scsmediated copper resistance traits may also be present in other organisms, including C. koseri, K. pneumoniae, and P. mirabilis, that encode the important Scs proteins required for copper homeostasis.

Production of purified proteins
The coding sequence for the N-terminal domain of Salmonella ScsB (n-ScsB) lacking the signal peptide (residues 26 -231) was cloned into a modified version of the expression vector pMCSG7, which encodes the targeted gene as a fusion Under copper stress conditions, the envelope stress response system CpxR/CpxA drives the production of Scs proteins and the metallochaperone CueP. ScsB and ScsC form a functional redox relay that efficiently transfers electrons from the cytoplasm via ScsB to ScsC through a cascade of thiol-disulfide exchange reactions. Reduced ScsC and ScsB can then bind Cu(I) with subpicomolar affinity and transfer it to the metallochaperone CueP, which in turn delivers it to the stress response Cu/Zn-SOD. Collectively, these systems contribute to combating copper toxicity in the Salmonella periplasm and limit copper movement to the cytosol as well as contributing to defense mechanisms against reactive oxygen species. The figure was created with BioRender.

Salmonella ScsB and ScsC protect against toxic copper
protein containing an N-terminal His 6 tag, TRX, and a TEV protease cleavage site. The coding sequences of the mature form (lacking the signal peptide) of Salmonella n-DsbD (residues 23-128) and DsbC (residues 22-237) were cloned into a pET28a-His-TEV vector, which encodes the targeted gene as a fusion protein containing an N-terminal His 6 tag followed by a TEV protease cleavage site. The recombinant proteins were produced using a method similar to that descried previously for other S. Typhimurium DsbA-like proteins (14,49). Briefly, all proteins were overexpressed using an autoinduction method (50) and then purified by Ni 2ϩ -affinity chromatography. TEV protease was used to cleave the His 6 -TRX tag, and the cleaved proteins were further purified by reverse Ni 2ϩ -affinity chromatography followed by size-exclusion chromatography. Similarly, S. Typhimurium CueP (residues 22-180; accession locus tag STM3650) was purified using a similar method to that described for n-DsbD and DsbC (49,51).
For copper transfer experiments, an MBP fusion construct of ScsC was prepared to increase the size of ScsC relative to CueP. The coding DNA sequence for ScsC lacking the signal peptide (residues 19 -189) was cloned between EcoRI and HindIII sites in pMAL-c2x vector using the Gibson assembly reagent kit (New England Biolabs) (see Table S3 for cloning primers) (52). The recombinant fusion protein was overexpressed using autoinduction and purified using amylose resin (New England Biolabs) with 10 mM maltose.

Peptide oxidation assay
The ability of ScsC to catalyze disulfide bond formation in a synthetic peptide substrate was assessed as described previously (26). Briefly, the peptide substrate CQQGFDGTQNSCK with a 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid chelate attached to the N terminus and a methylcoumarin amide group coupled to the ⑀-amino group of the lysine was purchased from AnaSpec (Fremont, CA). The lyophilized peptide was reconstituted at 2 mM in 100 mM imidazole, pH 6.0. Europium trifluoromethanesulfonate (Sigma-Aldrich) was added to the peptide solution at a molar ratio of 1:2. Assays were performed in a 96-well Optiplate on an Envision multilabel plate reader (PerkinElmer Life Sciences) using excitation at 340 nm and emission at 615 nm. The total reaction volume in each well was 50 l, containing 80 nM ScsC or DsbA and 2 mM oxidized GSH (GSSG) in a buffer of 50 mM MES, pH 5.5, 50 mM NaCl, 2 mM EDTA. The reactions were initiated by the addition of 8 M peptide substrate. The disulfide bond formation was measured using time-resolved fluorescence with a 100-ms delay before reading and a 400-ms reading time. The test was carried out in triplicate per plate and performed on two occasions. The data were analyzed using GraphPad Prism version 5.0 (GraphPad Software, Inc.).

Electron transfer experiments
All proteins (n-ScsB, ScsC, DsbA, n-DsbD, and DsbC) were prepared at 50 M and oxidized and reduced by incubating at 4°C for 1 h using 10 mM GSSG (Sigma-Aldrich) and DTT red (Astral Scientific, Australia), respectively. Excess GSH and DTT in the reaction mixture were removed by size-exclusion chromatography (Superdex 200 10/300 GL, GE Healthcare) equilibrated in 10 mM NaPO 4 , pH 7.0, 50 mM NaCl, 1 mM EDTA. The final redox state was confirmed by AMS gel-shift analysis.
Electron transfer reactions were performed as described previously (53) with some modifications. Briefly, 100 l of 20 M reduced catalysts (N-ScsB and N-DsbD) were mixed with 100 l of 20 M oxidized substrate proteins (ScsC, DsbA, and DsbC) (1:1 ratio). Immediately after mixing, 50-l samples were taken at 15-, 120-, and 300-s time points and quenched with 10% (w/v) trichloroacetic acid (TCA). The samples were centrifuged (16,100 ϫ g for 10 min at 4°C), and the pellets were washed with acetone (100% (v/v)) and resuspended in 50 mM Tris, pH 7.0, 1% SDS containing 2 mM AMS (Life Technologies) to alkylate free thiols (-SH). The alkylation reaction was stopped by adding nonreducing loading dye to each sample, and the separation of oxidized and reduced proteins was carried out by SDS-PAGE. Experiments were carried out on three independent occasions.

SPR binding analysis
A Biacore T200 biosensor instrument (GE Healthcare) was used to measure the affinity of the protein-protein interactions. All experiments were performed at 25°C in 20 mM HEPES, pH 7.4, 150 mM NaCl, 0.001% Tween 20. Ligands were immobilized using a standard amine coupling method. Briefly, following activation of the CM5 sensor chip (GE Healthcare) with 0.1 M 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride and 0.25 M N-hydroxysulfosuccinimide (NHS), 20 g/ml ligand proteins in 10 mM acetate buffer, pH 4.0 -4.5, was injected at 10 l/min until an immobilization level of 500 -1,000 RU was reached. Residual active NHS esters were blocked by injecting 1 M ethanolamine HCl, pH 8.5, for 7 min. All analytes were exchanged into the running buffer and injected over the surface at a constant flow of 50 l/min. To restore the redox state of the immobilized protein after each run, 10 mM reduced DTT was injected to the relevant flow cell at 10 l/min for 2 min twice. A steady-state affinity model was used to determine the affinity of the protein-protein interactions using BIAcore Evaluation 4.1 software. Experiments were performed in triplicate using different immobilization levels (500 -1,000 RU).

Measurement of reduction potential
The reduction potential of n-ScsB was measured using AMS gel-shift analysis as described previously (37). Recombinant n-ScsB (22 M) was incubated in degassed nitrogen-purged buffer containing 100 mM sodium phosphate, pH 7.0, 1 mM EDTA supplemented with 100 mM oxidized DTT (DTT ox ) (Sigma-Aldrich) and varying concentrations (8 M-120 mM) of reduced DTT (DTT red ) (Astral Scientific). After overnight equilibration at room temperature, the reactions were quenched with 10% (w/v) TCA, sedimented (21,600 ϫ g for 10 min at 4°C), and processed as described previously (37). Separation of oxidized and reduced proteins was carried out by SDS-PAGE, and the quantities of reduced versus oxidized n-ScsB were determined using density analysis software (ImageJ 1.45) (54). The amount of reduced protein (R) in each reaction was plotted against DTT red /DTT ox to obtain the equilibrium constant (K eq ) of the redox reaction using Equation 1.

Salmonella ScsB and ScsC protect against toxic copper
The corresponding reduction potential values were calculated by substituting the K eq value in the Nernst equation (Equation 2), where E DTTred/DTTox 0Ј ϭ Ϫ312 mV, F denotes the Faraday constant (96,485.3 coulombs mol Ϫ1 ), n ϭ 2 (number of electron transferred), and RT is the product of the gas constant (8.314 J K Ϫ1 mol Ϫ1 ) and the absolute temperature (298.15 K) (55).
[Cu I L 2 ] 3Ϫ ϩ P º Cu I P ϩ 2 L 2Ϫ (L ϭ Bca) Under the condition of an effective competition for Cu(I) between the probe ligand Bca and the target protein P of Equation 3, the Cu(I) speciation is quantitatively described by Equation 4. Consequently, the dissociation constant K D for the complex Cu I -P can be estimated via curve fitting of the experimental equilibrium concentration of the probe complex [Cu I (Bca) 2 ] 3Ϫ to Equation 4 together with the known ␤ 2 for [Cu I (Bca) 2 ] 3Ϫ and the known total concentrations of related components (i.e. Cu(I), P, and Bca). The equilibrium concentration of [Cu I (Bca) 2 ] 3Ϫ may be determined directly from its characteristic absorbance at 562 nm. Decreasing the probe ligand Bca concentration in Equation 3 may lead to a noncompetitive Cu(I) transfer from [Cu I (Bca) 2 ] 3Ϫ to protein P, and this allows estimation of the Cu(I) binding stoichiometry of protein P. The experiments were performed under anaerobic condition in deoxygenated buffers according to the reported protocols (31,56).

Quantification of Cu(II) binding
Cu(II) binding to protein was analyzed with two fluorescence DP probes, DP1 and DP2, according to a recent report (30). The quantification was based on the competition reaction (5) and the two associated equations, Equations 6 and 7.
Cu II DP ϩ P º Cu II P ϩ DP  The Cu(II) affinity of the target protein P (expressed as K D (Cu II P)) may be estimated from Equation 6 with the known K D (Cu II DP) of the DP probe. Because the fluorescence quenching responds to the Cu(II) binding to the DP probe only, the equilibrium concentration [Cu II DP] in Equations 5 and 6 can be estimated via Equation 7 where F 0 , F, and F 1 are the fluorescence intensity of the DP probe upon binding 0, Ͻ1.0, and 1.0 eq of Cu(II), respectively. Other terms in Equation 6 were determined via mass balance relationships. The experimental protocols followed those reported previously (30).

Cu(I) transfer experiments
All proteins used in the copper transfer experiments were fully reduced prior to each experiment, and redox state was confirmed by the Ellman assay (57). For the copper transfer experiments, an MBP-fused form of ScsC (ScsC-MBP) was prepared to increase the molecular weight of ScsC relative to CueP (MBP has been previously shown not to bind copper (58,59)). Cu(I)-loaded forms of n-ScsB and ScsC-MBP were prepared by incubation of each metal-free apoprotein with CuSO 4 and NH 2 OH at a 1:5:10 molar ratio in 20 mM MOPS, 50 mM NaCl, pH 7.0, for 30 min. To remove free (excess) copper, the reaction mixture (ϳ1 ml) was applied onto a size-exclusion chromatography column (Superdex 75 10/300 GL, GE Healthcare) equilibrated in 20 mM MOPS, 50 mM NaCl, pH 7.0. Peak fractions were analyzed by SDS-PAGE, and the Cu(I) chromophoric ligand BCS was added to detect copper content at 483 nm (31).
To assess the copper transfer ability of these proteins, Cu(I)bound forms of n-ScsB and ScsC-MBP were incubated for 0.5 h with reduced apo-CueP in a 1:1 molar ratio. The protein mixture was loaded onto a size-exclusion column (Superdex 75 10/300 GL), and elution fractions were analyzed by SDS-PAGE for protein and by BCS for Cu(I).