Direct Oxidation of the [2Fe-2S] Cluster in SoxR Protein by Superoxide

Background: The [2Fe-2S] transcription factor SoxR is activated by reversible one-electron oxidation of its [2Fe-2S] cluster. Results: Superoxide oxidized the reduced form of SoxR directly. Conclusion: SoxR-mediated transcription in Escherichia coli is activated by reaction of superoxide with the reduced form of the [2Fe-2S] cluster. Significance: SoxR proteins play distinct a regulatory role in the activation of superoxide. The [2Fe-2S] transcription factor SoxR is activated by reversible one-electron oxidation of its [2Fe-2S] cluster, leading to enhanced production of various antioxidant proteins through induction of the soxRS regulon in Escherichia coli. Recently, there has been considerable debate about whether superoxide (O2̇̄) activates SoxR directly. To elucidate the underlying activation mechanism, we investigated SoxR interaction with O2̇̄ using pulse radiolysis. Radiolytically generated hydrated electrons reduced the oxidized form of the [2Fe-2S] cluster of SoxR within 2 μs. A subsequent increase in absorption in the visible region corresponding to reoxidation of the [2Fe-2S] cluster was observed on a time scale of milliseconds. Addition of human copper/zinc superoxide dismutase inhibited this delayed oxidation in a concentration-dependent fashion (I50 = 1.0 μm), indicating that O2̇̄ oxidized the reduced form of SoxR directly. The second-order rate constant of this process was estimated to be 5 × 108 m−1 s−1. A similar result was observed after pulse radiolysis of Pseudomonas aeruginosa SoxR. However, superoxide dismutase inhibited the oxidation of reduced SoxR much more effectively in P. aeruginosa, even at a lower concentration (I50 = 80 nm), indicating that the soxRS response is much more sensitive to O2̇̄ in E. coli than in P. aeruginosa. These results suggest that SoxR proteins play a distinct regulatory role in the activation of O2̇̄.


The [2Fe-2S] transcription factor SoxR is activated by reversible one-electron oxidation of its [2Fe-2S] cluster, leading to enhanced production of various antioxidant proteins through induction of the soxRS regulon in
Reactive oxygen species, byproducts of cellular aerobic metabolism, produce cellular and genetic damage in aerobic organisms. Cells have thus evolved defense systems against reactive oxygen species and respond to sublethal levels of oxidative stress through coordinated activation of a battery of antioxidant genes (1)(2)(3)(4)(5)(6)(7). The molecular signals that activate these multifunctional defense systems have been the object of considerable interest. The protective responses initiated in response to reactive oxygen species exposure have been most thoroughly investigated in Escherichia coli, where induction of antioxidant genes is mediated by activation of the soxRS regulon. The reactive oxygen species sensor for this regulon is SoxR, which contains a [2Fe-2S] cluster that is essential for its transcription-enhancing activity. In E. coli cells, SoxR is maintained in its reduced inactive form (8,9), but when cells are exposed to superoxide (O 2 . )-generating compounds, the [2Fe-2S] cluster of SoxR undergoes univalent oxidation to yield the oxidized active form (10). The only known target of activated SoxR is the soxS gene in E. coli (11). In turn, increased SoxS levels induce expression of various antioxidant proteins and repair proteins (12). Therefore, SoxR senses oxidative stress using the redox states of the [2Fe-2S] cluster and regulates transcription of the soxS gene by undergoing structural changes between the oxidized and reduced forms (8,9). X-ray crystallographic structures have been obtained for the oxidized (active) state of E. coli SoxR and its complex with a 20-bp oligodeoxynucleotide (ODN) 2 of the target promoter ( Fig. 1) (13). The structure reveals N-terminal winged helix DNA-binding, dimerization helix, and [2Fe-2S] cluster-binding domains. The [2Fe-2S] cluster of SoxR is coordinated to four cysteine residues (Cys-119, Cys-122, Cys-124, and Cys-130). One of the sulfur atoms (S 2 ) and two iron atoms are fully exposed to solvent (Fig. 1B). The structural alterations of SoxR induced by redox changes of [2Fe-2S] clusters are communicated to the DNA-binding domain, leading to target promoter DNA distortion (13,14). The overall architecture of the SoxR-DNA complex is similar to that of MerR family proteins (15)(16)(17), which function as transcriptional activators in response to a variety of stresses, including exposure to heavy metals (16,18,19), reactive oxygen species (8,10), and antimicrobials (15,17,20). The DNA structure in the MerR family complex is in a bent conformation with local untwisting, allowing RNA polymerase to initiate transcription (15, 18 -21).
A crucial unanswered question concerns the nature of the cellular signal sensed by SoxR. O 2 . was the first candidate suggested as the SoxR-activating signal, reflecting the early observation that SoxR responded to agents that generate O 2 . (10).
Experiments in intact cells showed that expression of the soxRS regulon member fumarase C was increased by an elevation of O 2 . caused by the deletion of superoxide dismutase (SOD) (22).
The induction of fumarase C in the parental strain was eliminated by mutational deletion of the soxRS response (23). Similarly, the expression of soxRS-dependent glucose-6-phosphate dehydrogenase was higher in a control strain compared with E. coli cells overexpressing Mn-SOD (24). However, there is no evidence that the direct oxidant of the [2Fe-2S] cluster of SoxR is O 2 . itself (25)(26)(27). On the other hand, Gu and Imlay (26) recently demonstrated that the soxRS response of E. coli is directly activated by redox-cycling drugs rather than by O 2 . .
Reduced SoxR (SoxR red ) was reported to be oxidized directly by redox-cycling drugs such as paraquat in the absence of SOD. However, there were arguments against this proposal (27). As an alternative, it was suggested that the soxRS regulon does not respond to a change in O 2 . concentration, but rather responds to the redox state (NADPH/NADP ϩ ratio) (28). Accordingly, NADPH could maintain SoxR red by reactions linked by NADPH (29,30), whereas depletion of available reducing equivalents would produce a shift in the redox balance toward oxidized SoxR (SoxR ox ). SoxR homologs have been identified in many bacterial species (31). All of these homologs contains the SoxR-specific cysteine motif CI(G/Q)CGC(L/M)(S/L)XXXC, required for binding of the [2Fe-2S] cluster (31). Pseudomonas aeruginosa SoxR, a homolog with 62% sequence identity and 77% sequence similarity to E. coli SoxR (32), exhibits similar physicochemical properties as the E. coli protein, such as redox potential of the [2Fe-2S] cluster and visible absorption spectra (33). However, the P. aeruginosa SoxR function differs dramatically from that of E. coli. The SoxR regulon in P. aeruginosa, in which there is no obvious SoxS homolog, is composed of genes encoding an efflux pump, a transporter, and a putative monooxygenase (32-34) and does not control any of the genes typically involved in O 2 . resistance and detoxification (35). It was shown that P. aeruginosa SoxR is activated by the endogenous redox-cycling antibiotic compound pyocyanin (31,34), which is the physiological signal for the up-regulation of quorum sensingcontrolled genes. This reaction can occur anaerobically in intact cells, suggesting O 2 .
This study was designed to investigate the hypothesis that O 2 . has a direct role as a signal for the soxRS regulon. Using the pulse radiolysis method, we have confirmed that O 2 . reacts rapidly and directly with SoxR red , but not with SoxR ox . In addition, we present evidence that the sensitivity of SoxR to O 2 . is much higher in E. coli than in P. aeruginosa.

EXPERIMENTAL PROCEDURES
Materials-Human Cu/Zn-SOD, overproduced by E. coli, was kindly provided by Nippon Kayaku. All other reagents were commercially obtained and analytical grade.
Expression and Purification-The expression plasmids for E. coli (29) and P. aeruginosa (33) SoxR were transformed into E. coli C41(DE3) and coexpressed with the isc operon (36) as described previously (37). SoxR was purified essentially as described previously (37). SoxR protein samples were purified as the oxidized form and confirmed as Ͼ95% homogeneous by SDS-polyacrylamide gel electrophoresis. The concentration of SoxR was determined using an extinction coefficient of 12.7 mM Ϫ1 cm Ϫ1 at 417 nm (38).
SoxR-DNA complexes were prepared according to the crystallization method described previously (37). The palindromic oligonucleotide GCCTCAAGTTAACTTGAGGC (Sigma-Aldrich) was formed into a double-stranded DNA structure by dissolving in an aqueous solution containing 20 mM potassium phosphate (pH 7.0), 50 mM KCl, and 10 mM potassium/sodium tartrate; heating to 94°C; and gradually cooling to room temperature. SoxR-DNA complexes were prepared by mixing 40 -70 M SoxR (20 mM potassium phosphate (pH 7.0), 250 mM KCl, and 10 mM potassium/sodium tartrate) and ODN solutions at molar ratios of 2:1.05-1.1 and incubating for at least 4 h at 4°C (37). Ferredoxin from spinach was purified by a combination of DEAE-cellulose and Sephadex G-75 column chromatographies essentially as described (39).
Pulse Radiolysis-Samples of SoxR for pulse radiolysis were prepared by bubbling a solution containing 20 mM potassium phosphate (pH 7.0), 10 mM potassium/sodium tartrate, 0.5 M KCl, and 0.1 M sodium formate (for scavenging OH) with O 2 gas for 2 min and then adding a concentrated solution of SoxR (1-2 mM). Pulse radiolysis experiments were performed with a linear accelerator at the Institute of Scientific and Industrial Research of Osaka University (40 -42). The pulse width and energy were 8 ns and 27 MeV, respectively. A 1-kilowatt xenon lamp was used as a light source. After passing through an optical path, the transmitted light intensities were analyzed and monitored by a fast spectrophotometric system composed of a Nikon monochromator, an R-928 photomultiplier, and a Unisoku data analysis system. For the time-resolved transient absorption spectral measurement, the monitor light was focused into a quartz optical fiber, which transported the electron pulse-induced transmittance changes to a gated multichannel spectrometer (Unisoku TSP-601-02). The initial concentration of superoxide radical generated by pulse radiolysis was 20 -30 M, which was estimated using the relationship ⑀ 260 ϭ 1925 M Ϫ1 cm Ϫ1 (43).
Spectrophotometric Measurements-Optical absorption spectra were measured with a Hitachi U-3000 spectrometer.

RESULTS
To follow the reaction of O 2 . with SoxR, we studied absorption changes after pulse radiolysis of oxygen-saturated phosphate buffer in the presence of 50 M SoxR ox . Under these conditions, although most hydrated electrons generated reacted with O 2 to form O 2 . , a portion of the hydrated electrons were found to react very rapidly with SoxR ox , resulting in reduction of the [2Fe-2S] cluster of the protein. This reduction was reflected in a decrease in absorbance at 420 nm. The initial transient spike in absorbance indicated the formation and decay of hydrated electrons (Fig. 2). Subsequently, these initial changes in absorption partially reversed (ϳ38%) on a time scale of milliseconds (Fig. 2). Fig. 3 shows difference spectra obtained at 20 s and 20 ms after pulse radiolysis. Both spectra are sim-ilar to the SoxR red -minus-SoxR ox spectrum. From these findings, it can be concluded that SoxR ox was initially reduced, and subsequently, the [2Fe-2S] cluster became partially reoxidized. As shown in Fig. 4, this millisecond time scale recovery of oxidation was inhibited by addition of Cu/Zn-SOD (11 M), which completely eliminated the delayed change in absorption without affecting the initial rapid decrease. From this, we conclude that the recovery process reflects the oxidation of SoxR red by O 2 . .
Following the spectral changes, the absorbance very slowly returned to the initial SoxR ox state and was almost recovered after ϳ20 s (data not shown). Under the deaerated condition, the spectrum was persistent for at least 30 s (data not shown). Therefore, the spectroscopic changes at 20 s account for the direct oxidation of the [2Fe-2S] cluster of SoxR red with O 2 . This suggests that the spectral changes after pulse radiolysis corresponds to reversible reduction of the protein, not by irreversible destruction of the iron-sulfur cluster.
Electrochemical studies have shown that the reduction potential of SoxR bound to promoter DNA is positively and dramatically shifted from Ϫ285 mV (8,44) to 200 mV (45). Thus, we were particularly interested in investigating the spectroscopic changes and rate constant after pulse radiolysis in the presence of the promoter ODN. As was the case without the promoter ODN, SoxR ox was initially reduced and subsequently oxidized. Although the spectra and kinetics obtained after pulse radiolysis were not affected by the presence of ODN (Fig. 5B), the absorbance recovery 10 ms after pulse radiolysis was ϳ50%,   which is larger than that in the absence of ODN (ϳ38%) (Fig.  5A). It is important to note that we did not observe a reduction in To extend our analysis, we examined the effect of Cu/Zn-SOD on the oxidation of SoxR red . The dependence of the absorption changes at 420 nm on Cu/Zn-SOD is presented in Fig. 6C, which shows that SOD dose-dependently inhibited the recovery of oxidation. Half-maximal inhibition was observed at 1 M SOD. Inhibition of the oxidation of SoxR red by SOD can be expressed as inhibition (%) ϭ k 1  A similar result was also obtained with P. aeruginosa SoxR, as shown in Fig. 6B. However, subsequent millisecond scale oxidations were different between E. coli and P. aeruginosa. The recovery of oxidation for P. aeruginosa (ϳ50%) was larger than that for E. coli. Notably, the oxidation of P. aeruginosa was nearly completed inhibited by 0.5 M SOD, whereas this same concentration resulted in only ϳ20% inhibition in E. coli. Halfmaximal inhibition of oxidation of P. aeruginosa was observed at 80 nM SOD, from which we can estimate a rate constant of 4 ϫ 10 7 M Ϫ1 s Ϫ1 for the reaction of O 2 . with P. aeruginosa SoxR red .

DISCUSSION
In this investigation, we demonstrated the direct reaction of O 2 . with SoxR red to form SoxR ox . The reaction schemes after pulse radiolysis can be summarized as follows (Reactions 1 and 2), where e aq Ϫ represents hydrated electrons.

REACTION 2
The reaction sequence is reversible; thus, iron-sulfur clusters are not irreversibly destroyed. The second-order rate constant of E. coli SoxR (5 ϫ 10 8 M Ϫ1 s Ϫ1 ) is in the same range as that of Mn-SOD (15-1.6 ϫ 10 8 M Ϫ1 s Ϫ1 ) (49) and Fe-SOD (3.3 ϫ 10 8 M Ϫ1 s Ϫ1 ) (50). In fact, Reaction 2 was not inhibited completely, even in the presence of 5 M Cu/Zn-SOD (Fig. 6C). This suggests that O 2 . formed intracellularly reacts with SoxR red . However, it is not clear that O 2 . is available to interact with SoxR because high levels of SOD are present, and the SoxR protein is produced constitutively at a low level of ϳ100 molecules/cell (3). It has been calculated that the level of SOD found in aerobic cells (ϳ10 Ϫ5 M) would efficiently reduce O 2 . at steady-state con-    (32,34), which is the physiological signal for up-regulating quorum sensing-controlled gene expression and serves as an electron acceptor to balance the intracellular redox state (54). Our present data also support the view that the major function of the SoxR regulon in P. aeruginosa, unlike that in E. coli, is not O 2 . responsiveness.
Importantly, this study is the first to provide a direct observation of the reaction between an iron-sulfur protein and O 2 . , revealing an interesting effect of O 2 . on iron-sulfur clusters.
O 2 . has been reported to inactivate several iron-sulfur proteins with second-order rate constants in the range of 10 6 -10 7 M Ϫ1 s Ϫ1 (55)(56)(57)(58). In each of these cases, the process was irreversible and was associated with concomitant cluster degradation. We have also examined reactions of O 2 . with spinach ferredoxin and adrenodoxin using the pulse radiolysis method. However, unlike the case with SoxR (Fig. 2), after reduction of the [2Fe-2S] cluster of these proteins, there was no recovery of absorbance due to oxidation on a time scale of milliseconds (data not shown). A slower recovery of direct O 2 oxidation of the reduced form of ferredoxin was observed, as in the case of SoxR (data not shown). Therefore, it can be said that Reaction 2 observed here is a specific characteristic of the SoxR protein. is favored in the presence of a proton source, as is observed in the reaction of superoxide reductase (60 -62). Both proton and solvent may play important roles in the SoxR mechanism. An analysis of the crystal structure revealed that the [2Fe-2S] clusters of oxidized SoxR are solvent-exposed (13), creating an asymmetrically charged environment and a possible conformational change in SoxR. The unusual positioning of the metal ion on the surface of SoxR provides an active site with a readily available source of protons.
In conclusion, SoxR-mediated transcription is activated by oxidation of the [2Fe-2S] ϩ cluster in SoxR. We found that the SoxR [2Fe-2S] ϩ cluster of E. coli was more readily oxidized by O 2 . compared with that of P.