The redox state of the [2Fe-2S] clusters in SoxR protein regulates its activity as a transcription factor.

SoxR protein is a redox-responsive transcription factor that governs a regulon of oxidative stress and antibiotic resistance genes in Escherichia coli. Purified SoxR contains oxidized [2Fe-2S] clusters and stimulates in vitro transcription of its target gene soxS up to 100-fold. SoxR transcriptional activity, but not DNA binding, is completely dependent on the [2Fe-2S] clusters; apo-SoxR prepared in vitro binds the soxS promoter with unchanged affinity but does not have transcription activity. Thus, modulation of the SoxR [2Fe-2S] clusters was proposed to control the protein's function in transcription. Here, we provide evidence that SoxR with reduced [2Fe-2S] clusters is inactive. Redox titration of purified SoxR revealed a midpoint potential of −285 ± 10 mV (pH 7.6). In vitro transcription assays showed that SoxR was inactivated when the [2Fe-2S] cluster was reduced (−380 mV), and full activity was restored upon reoxidation (+100 mV). The results suggest that one-electron oxidation and reduction of the [2Fe-2S] cluster regulate SoxR transcriptional activity.

The soxRS regulon is a multifunctional cellular defense system against oxidative stress in Escherichia coli (1,2). The regulon is controlled by two transcription factors, SoxR and SoxS, which act sequentially. SoxR is a constitutive protein that is activated post-translationally by superoxide stress (3,4) or nitric oxide (5) to stimulate expression of the soxS gene. The increased amount of SoxS protein then stimulates expression of the regulon genes (6).
SoxR protein is a homodimer containing a pair of [2Fe-2S] clusters (7,8) anchored to four cysteine residues in the Cterminal region of the polypeptide. Mutations changing any of these cysteines to alanine eliminates the [2Fe-2S] clusters from SoxR and prevents in vivo activation of the protein by the superoxide-generating agent paraquat. 1 In vitro experiments with wild-type SoxR protein showed that disruption of the [2Fe-2S] clusters by thiols to generate apo-SoxR does not affect soxS-specific DNA binding but does eliminate transcriptional activity (9,10). These results demonstrated that the [2Fe-2S] clusters are essential for the SoxR regulatory function both in vivo and in vitro.
The mechanism for SoxR activation is not established. One possibility is that assembly and disassembly of the SoxR [2Fe-2S] clusters could contribute to the regulation of SoxR transcriptional activity. It has been shown that glutathione-based free radicals disrupt the [2Fe-2S] clusters, thereby inactivating SoxR (10). Moreover, assembly of SoxR [2Fe-2S] clusters from Fe 2ϩ and S 2Ϫ can occur rapidly in vitro (11), which is a prerequisite for such a possible regulatory role. Although assembly and disassembly of the [2Fe-2S] clusters must occur during SoxR metabolism, whether these processes are rate-limiting in the activation of SoxR in vivo has yet to be demonstrated.
An alternative regulatory mechanism proposed for SoxR is through reduction and oxidation of the [2Fe-2S] clusters (9). Although such an experiment was obvious to do, reoxidation of reduced SoxR [2Fe-2S] clusters during in vitro transcription reactions yielded inconsistent results. 2 In this report, we have used a specially designed anaerobic redox cuvette to prevent reoxidation of reduced SoxR [2Fe-2S] clusters during the in vitro transcription assay. The results show that SoxR protein is reversibly inactivated when the [2Fe-2S] clusters are reduced.

MATERIALS AND METHODS
Purification of SoxR Protein-Purification of SoxR from E. coli containing the expression plasmid pKOXR was done as described previously (9,10). The purity of SoxR protein used in this work was Ͼ 90% as judged by staining of sodium dodecyl sulfate-polyacrylamide gels.
Redox Titration of SoxR [2Fe-2S] Clusters-A specially designed anaerobic cuvette was used for redox titrations as described by Dutton (12). Before titration, solutions containing SoxR protein were equilibrated with ultra-pure argon for about 1 h at room temperature. During titration, the argon flow was maintained with gentle stirring by a small magnet on the bottom of the cuvette. The redox potential of the solution was adjusted by adding freshly prepared sodium dithionite or potassium ferricyanide using a gas-tight 10-l Hamilton microsyringe (Hamilton Co., Reno, NV). The redox potential was monitored directly with a redox microelectrode (Microelectrodes Inc., Bedford, NH).
The redox state of SoxR [2Fe-2S] clusters was monitored by absorbance at 414 nm (one of the maxima for oxidized SoxR; see Refs. 8 and 10) to minimize the absorbance interference from the redox mediator safranine O (present at 3 M). After examining the absorbance spectra of reduced and oxidized SoxR [2Fe-2S] clusters, an approximate isoasbestic point at 391 nm was chosen as the reference wavelength for the redox titration experiment. Absorbance was recorded in a UV-visible spectrophotometer (Perkin-Elmer Lambda 3A).
Transcriptional Activity of SoxR Protein in Vitro-SoxR activity was assayed by measuring in vitro transcription of the soxS gene in plasmid pBD100, which also contains the SoxR-independent control gene bla (8). Transcription reaction mixtures contained 20 nM SoxR protein, 10 nM pBD100 DNA, 1.25 mM each of ATP, GTP, CTP, and UTP, 75 mM KCl, 1 mM dithiothreitol, 10% glycerol, 15 mM MgCl 2 , 100 mM Tris-HCl (pH 7.5), and 10 M of redox mediator safranine O. The reaction mixtures were equilibrated with ultra-pure argon in anaerobic cuvette for about 1 h, and the redox potential of solution was adjusted and monitored as described above. After the redox potential was adjusted by the addition of dithionite for reduction or potassium ferricyanide for oxidation, ali-* This work was supported by Grant CA37831 (to B. D.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ quots of 40 l were removed, 1 l containing 2 units of de-gassed 70 RNA polymerase (Promega, Madison, WI) was added under anaerobic conditions. These reaction mixtures were incubated at 37°C for 3 min, and the reactions were stopped by the addition of 360 l of stop solution containing 73% ethanol, 0.11 M sodium acetate (pH 7.9), and 7 g/ml yeast tRNA. The transcription products, the SoxR-dependent soxS transcript and the SoxR-independent bla transcript, were quantified by primer extension analysis as described previously (8,10).
A typical redox titration of the SoxR [2Fe-2S] cluster at pH 7.6 is shown in Fig. 1. Each redox potential was equilibrated for Ն2 min before absorbance measurement. The solid line drawn through the data points is the theoretical calculation of the Nernst equation with n ϭ 1. The midpoint potential for the SoxR [2Fe-2S] clusters under these conditions was estimated to be Ϫ285 Ϯ 10 mV. It is worth noting that this value is close to the intracellular redox potential estimated for E. coli during exponential growth (13).

SoxR with Reduced [2Fe-2S] Clusters Is Inactive in Vitro-
Redox titration of the in vitro transcription activity of SoxR protein was done under the conditions described above, except that SoxR protein was present at a 500-fold lower concentration. Fig. 2 shows the result of in vitro transcription under various redox potentials. The synthesis of the SoxR-dependent soxS transcript was dramatically decreased (ϳ20-fold) when the redox potential was adjusted from ϩ200 to Ϫ380 mV (where the SoxR [2Fe-2S] clusters are Ͼ95% reduced; see Fig.  1). When the redox potential was increased from Ϫ380 mV back to ϩ100 mV (where the SoxR [2Fe-2S] clusters are reoxidized; Fig. 1), SoxR-dependent soxS transcription was fully restored (Fig. 2). In contrast, the amount of the SoxR-independent bla transcript remained virtually unchanged as the redox potential was varied from ϩ200 to Ϫ380 mV and back to ϩ100 mV. Thus, the 70 RNA polymerase was not significantly affected over the range of redox potentials used in these experiments. These results clearly show that reduced SoxR is specifically inactive for soxS transcription and that SoxR activity is restored by reoxidation of the [2Fe-2S] clusters. DISCUSSION Activated SoxR bound to DNA exerts structural changes in the soxS promoter that compensate for the ϳ70°overwinding of the Ϫ10 and Ϫ35 promoter elements (8,9). A key question is how this structural effect is transmitted via changes in the oxidation state of SoxR [2Fe-2S] clusters. Like many other DNA binding proteins, SoxR is a homodimer and binds a symmetric DNA site. Gel filtration chromatography under denaturing conditions indicates that each SoxR monomer contains a single [2Fe-2S] cluster 2 and that apo-SoxR is also homodimeric (9). Thus, the association between two SoxR monomers is not mediated by the [2Fe-2S] clusters, and oxidation would involve the loss of one electron from each SoxR monomer. One mechanism driving a significant conformational change in the protein would be electrostatic repulsion between the two positive charges, especially if the two [2Fe-2S] clusters are physically close. Because Fe is not required for DNA binding by SoxR (9) and the reduced [2Fe-2S] clusters of SoxR are stable (7,8), SoxR activation seems to be an electron-driven conformational change in the DNA-protein complex.
The E. coli intracellular redox potential is estimated to be Ϫ260 to Ϫ280 mV (15), a value close to the redox midpoint potential of the SoxR [2Fe-2S] clusters determined in vitro (Ϫ285 Ϯ 10 mV; Fig. 1). If these values were directly comparable, about half of the SoxR in cells would be in the oxidized state. However, the in vivo estimates of redox potential are indirect, and intracellular conditions (e.g. pH) could also shift the midpoint potential of SoxR [2Fe-2S] clusters to a higher value than found in vitro. In the range we are considering, a cellular redox potential 60 mV lower than that of the SoxR [2Fe-2S] clusters would maintain Ͼ90% of the SoxR protein in the reduced (inactive) state in vivo.
Oxidative activation of SoxR in vivo could occur in various ways. In cells challenged with redox cycling agents such as paraquat, NADPH and perhaps other reductants could be ex-  (12). Redox titration was performed in an anaerobic redox cuvette as described under "Materials and Methods." The x axis shows the redox potential measured with microelectrode. The y axis shows the absorbance difference between 414 and 391 nm, normalized to 0 or 100% for fully reduced or oxidized SoxR, respectively. The redox potential was measured against a standard hydrogen electrode. The solid line drawn through the data points was generated from the Nernst equation with n ϭ 1. The experiment was repeated three times. The mean value for redox midpoint potential was Ϫ285 Ϯ 10 mV .   FIG. 2. In vitro transcription  hausted, increasing the intracellular redox potential and indirectly the proportion of oxidized [2Fe-2S] clusters in SoxR. Consistent with this hypothesis, Fridovich and Liochev noted that SoxR is more readily activated by paraquat in bacteria deficient in glucose-6-phosphate dehydrogenase, an enzyme expected to contribute to the NADPH pool (14). This mechanism does not posit a direct role for superoxide as a signal, but deficiency in superoxide dismutase activates SoxR in the absence of paraquat (4). Moreover, SoxR is activated by nitric oxide more dramatically under anaerobic than aerobic conditions (5). Thus, superoxide and nitric oxide could also be direct oxidants of the SoxR [2Fe-2S] clusters to activate the system. It also seems likely that the [2Fe-2S] clusters of SoxR are maintained in the reduced state by an enzymatic reductase yet to be identified. Thus, a third possibility is that the putative reductase is inactivated by superoxide or nitric oxide to allow oxidized SoxR to accumulate. As this manuscript was being completed, we learned of independent studies that indicate reversible inactivation of SoxR in vitro by reduction of its [2Fe-2S] clusters (15).