Binuclear [2Fe-2S] clusters in the Escherichia coli SoxR protein and role of the metal centers in transcription.

SoxR protein of Escherichia coli is activated by superoxide-generating agents or nitric oxide as a powerful transcription activator of the soxS gene, whose product activates ∼10 other promoters. SoxR contains non-heme iron essential for abortive initiation of transcription in vitro. Here we show that this metal dependence extends to full-length transcription in vitro. In the presence of E. coli σ70 RNA polymerase, iron-containing SoxR mediates open complex formation at the soxS promoter, as determined using footprinting with Cu-5-phenyl-1,10-phenanthroline. We investigated the nature of the SoxR iron center by chemical analyses and electron paramagnetic resonance spectroscopy. Dithionite-reduced Fe-SoxR exhibited an almost axial paramagnetic signature with g values of 2.01 and 1.93 observable up to 100 K. These features, together with quantitation of spin, iron, and S2−, and hydrodynamic evidence that SoxR is a homodimer in solution, indicate that (SoxR)2 contains two [2Fe-2S] clusters. Treatment of Fe-SoxR with high concentrations of dithiothreitol caused subtle changes in the visible absorption spectrum and blocked transcriptional activity without generating reduced [2Fe-2S] centers, but was also associated with the loss of iron from the protein. However, lowering the thiol concentration by dilution allowed spontaneous regeneration of active Fe-SoxR.

dative stress and is controlled by the oxyR gene (Demple, 1991;Storz et al., 1991). In contrast, the soxRS system governs an inducible response to superoxide-generating agents (Nunoshiba et al., 1992;Wu and Weiss, 1992) or nitric oxide (Nunoshiba et al., 1993(Nunoshiba et al., , 1995. The soxRS response occurs in two stages: an intracellular signal of oxidative stress converts existing SoxR protein into a potent transcriptional activator of the soxS gene; the resulting increase in SoxS levels then triggers expression of the various regulon genes (Amá bile-Cuevas and Demple, 1991).
SoxR, isolated from bacteria that overproduce the protein, contains an FeS cluster(s) essential for in vitro transcriptional activation of the soxS promoter . The metal is not required for the binding of SoxR to the soxS promoter nor for the subsequent binding of RNA polymerase . These observations suggest that the critical effect of activated SoxR on transcription occurs at a later stage, perhaps by specific conformational effects of Fe-SoxR 1 on DNA, as proposed for the homologous MerR protein (Ansari et al., 1992). It seems likely that the FeS center(s) of SoxR is involved in the signal transduction mechanism that links O 2 . or NO ⅐ stress to gene activation in the soxRS system.
Although protein FeS centers have been most commonly associated with electron transfer and some enzymatic dehydratase reactions (Johnson, 1994), iron has recently been proposed as an important regulatory component of other genetic responses (Beinert, 1990). In the cytoplasmic form of mammalian aconitase, for example, the iron center regulates the activity of the protein as an RNA-binding factor: the apoprotein binds mRNAs encoding ferritin (blocking its translation) and the transferrin receptor (stabilizing the message), while the [4Fe-4S] and [3Fe-4S] forms do not (Klausner et al., 1993). This protein is thus linked to the iron status of the cell and coordinates key constituents of iron assimilation, utilization, and storage. The Fnr protein of E. coli coordinates transcription of a large number of genes that allow cells to take advantage of electron acceptors other than oxygen (Green and Guest, 1993). Although iron seems to be required for in vitro DNA binding by Fnr, the structure of its metal center has not been elucidated.
The structure of the iron center in SoxR is of importance both for the transcriptional activity described above and for the role of this protein as a sensor of oxidative stress. We describe here experiments that define [2Fe-2S] clusters in SoxR, and we show that only the metalloprotein activates in vitro transcription and is apparently associated with formation of the open complex at the soxS promoter.

MATERIALS AND METHODS
Purification of SoxR-Purification of SoxR protein from E. coli containing the SoxR expression plasmid pKOXR was performed as described previously . Fractions eluted from heparin-agarose columns (Life Technologies, Inc.) (purity of ϳ65-80%) were used for these studies; the more purified material obtained by DNA-affinity chromatography  could not be obtained in a high enough concentration for the physical studies conducted.
In Vitro Transcription-All the solutions and reagents used during the assay were prepared RNase-free (Sambrook et al., 1989). Protein-DNA binding reactions were performed in 19 l containing 75 mM KCl, 2 mM dithiothreitol (DTT), 10% glycerol, 15 mM MgCl 2 , 10 mM Tris-HCl, pH 7.5, ϳ200 -400 ng of pBD100 DNA, and the indicated amounts of SoxR. Plasmid pBD100 is a pBR322 derivative containing an E. coli genomic insert of ϳ4 kilobases that includes the whole soxRS locus (Amá bile-Cuevas and Demple, 1991). After a 5-min incubation at room temperature, 1 l of 1 mg/ml E. coli 70 -containing RNA polymerase (ϳ300 units/ml; kindly provided by Drs. Linda D'Ari and Michael Chamberlin, University of California, Berkeley) was added to each sample to achieve a final concentration of 0.1 M, and the reactions were incubated for 15 min at 37°C. One l of a mixture of the four NTPs (25 mM each) was then added, and the polymerase extension reaction proceeded for another 5 min at 37°C. The in vitro transcription reaction was then stopped by mixing into each sample 330 l of a mixture of a solution of 73% ethanol, 7 g/ml tRNA, 0.11 M sodium acetate. After a 45-min precipitation at Ϫ20°C, the samples were centrifuged for 20 min at 10,000 ϫ g at room temperature. The precipitates were resuspended in 10 l of H 2 O treated with diethyl pyrocarbonate.
The transcription products were quantified by primer extension analysis. Primer 1 (5Ј-CTGAATAATTTTCTGATGGG-3Ј; Nunoshiba et al. (1992)) hybridized 64 bp downstream of the transcriptional start site of the soxS gene. As an internal control for transcription activity, ␤-lactamase (bla) gene transcript, also directed by pBD100 (Amá bile-Cuevas and Demple, 1991), was quantified in parallel by primer extension analysis using primer pBR-1 (5Ј-GGGTGAGCAAAACAGGAA-3Ј), which hybridizes to a site 105 bp downstream from the 5Ј-end of this message (Russell and Bennett, 1981). The primers were labeled at the 5Ј-end with [␥-32 P]ATP (3,000 Ci/mmol; DuPont NEN) and T4 polynucleotide kinase (New England Biolabs). Fifty fmol of the indicated 32 P-labeled primer was annealed to 5-l samples from the in vitro transcription reactions, and primer extension reactions were performed with avian myeloblastosis virus-reverse transcriptase (Promega) as recommended by the manufacturer. Samples corresponding to 40% of each reaction were analyzed by electrophoresis on an 8% polyacrylamide, 6 M urea gel (Sambrook et al., 1989). 32 P-Labeled X174 DNA digested with HinfI (Promega) was used for size calibration.
5-Phenyl-1,10-phenanthroline-Copper(I) (CuPPA) Footprinting-A 180-bp DNA fragment containing the soxS promoter was isolated and labeled in one strand as described previously . For binding, labeled DNA (ϳ0.5-1 nM final concentration; ϳ10,000 cpm) and SoxR protein (ϳ25 nM in the reaction mixture) were incubated in an 18-l reaction containing 10 mM Tris⅐HCl, pH 8.0, 50 mM KCl, 2 mM mgCl 2 , 5% glycerol, 0.05% Nonidet P-40, 0.1 mM EDTA, pH 8.0, 50 g/ml bovine serum albumin, 10 g/ml salmon sperm DNA, and 1 mM 3-mercaptopropionic acid. After 10 min at room temperature, 1 l of E. coli RNA polymerase (130 units/ml; Promega) was added (final concentration 50 nM), and the incubation continued for 15 min at 37°C. The cleavage reagent was freshly prepared by mixing 10 l of 32 mM 5-phenyl-1,10-phenanthroline (Sigma) in ethanol with 10 l of 7.2 mM cupric sulfate (Sigma) and immediately diluting it with 80 l of H 2 O. One l of the cleavage reagent was added to each 19-l binding reaction, and strand scission was initiated by addition of 1.2 l of 100 mM 3-mercaptopropionic acid (Sigma). After 2 min at 37°C, the reactions were quenched by adding 1.3 l of 50 mM 2,9-dimethyl-1,10-phenanthroline (Sigma) (dissolved in ethanol) and further incubation for 2 min at room temperature. The DNA was then precipitated by the addition of a 100-l aliquot of the following mixture: 1 ml of ethanol, 50 l of 3 M sodium acetate, pH 5.2, and 10 l of 1 mg/ml yeast tRNA. After centrifugation, the precipitated DNA was redissolved in 5 l of formamide loading buffer and analyzed by electrophoresis in 5% polyacrylamide, 6 M urea gel (Sambrook et al., 1989). The same 180-bp probe was cleaved in a Maxam-Gilbert guanine-specific reaction (Sambrook et al., 1989) and used as a DNA sequence ladder.
Sedimentation Analysis-SoxR protein was analyzed by zonal centrifugation as described by Freifelder (1973) in 5-20% sucrose gradients in 0.5 M NaCl, 50 mM HEPES, pH 7.6. The relatively high salt concen-tration was necessary to maintain SoxR solubility. Bovine carbonic anhydrase (29.0 kDa; Sigma) and soybean trypsin inhibitor (20.1 kDa; Sigma) were used as markers. After sedimentation for 24 h at 50,000 rpm in a Beckman SW50.1 rotor, 0.3-ml fractions were collected from the bottoms of the tubes using a peristaltic pump. Samples were analyzed by SDS-PAGE, silver staining, and quantitative densitometry (BioImage system, Millipore).
Protein, Iron, and Labile Sulfide Determinations-Protein concentrations were determined by Coomassie staining of SDS-polyacrylamide gels using as a standard SoxR previously quantified by amino acid analysis .
Iron concentration was determined by two different methods: by inductively coupled plasma emission spectrometry  and colorimetrically by using the iron chelator ferrozine (Stookey, 1970). For the colorimetric determinations, 460-l samples of SoxR were mixed with 100 l of ultrapure concentrated HCl (Baker analyzed), and the mixture was incubated at 80°C for 20 min with occasional vigorous shaking. After centrifugation at 15,000 ϫ g for 5 min to remove denatured protein, 510 l of the supernatant was mixed with 20 l of 10 mM ferrozine and 20 l of 75 mM ascorbic acid. The mixture was neutralized by the addition of 120 l of saturated ammonium acetate to allow ferrozine chelation. After a 20-min incubation at room temperature, the absorbance at 562 nm was determined and iron concentration was calculated using ⑀ ferrozine-Fe ϭ 27,900 M Ϫ1 cm Ϫ1 (Stookey, 1970).
Labile inorganic sulfide was determined by using a modification of the basic methylene blue procedure (Fogo and Popowsky, 1949) as described by Beinert (1983).
Electron Paramagnetic Resonance (EPR) Spectroscopy-Samples of purified Fe-SoxR (500 l at ϳ10 M) inside an anaerobic chamber (Vacuum Atmospheres Co. model HE-493) at an O 2 concentration of ϳ0.5-1 ppm were incubated with 10 l of 3 mg/ml dithionite (freshly dissolved; this amount contained ϳ25 reducing equivalents ). Oxidized and reduced SoxR were sealed in cuvettes and removed from the chamber, and the spectra were recorded with a Hewlett-Packard model 8452A spectrophotometer to determine the extent of reduction. The cuvettes were then returned to the anaerobic chamber, opened, and 300 l of each sample was placed inside a 4-mm EPR sample tube. Tubes were sealed with rubber septa, removed from the anaerobic chamber, and immediately frozen in a bath of liquid nitrogen until analysis. EPR spectra at X-band were recorded at 10 -100 K on a Bruker model ESP 300 spectrometer maintained at constant temperature either with an Oxford Instruments ESR 910 continuous flow cryostat or with a Bruker ER4111VT variable temperature controller. Spin quantitations were carried out under nonsaturating conditions and were calculated by comparison of the integrated intensity of samples with that of a 1 mM Cu 2ϩ prepared as described by Malmström et al. (1970). Intensities of samples and the standard were corrected for differences in g value as described previously (Aasa and Vä nngård, 1975).
DTT Treatment of Purified Fe-SoxR Fractions-SoxR protein was incubated aerobically with increasing concentrations of DTT (10 to 100 mM) for 10 min at room temperature. The spectra of the oxidized and DTT-treated preparations were recorded in a Perkin-Elmer Lambda 3A UV/VIS spectrophotometer. In vitro transcription reactions for the DTT-treated samples were performed under the conditions described above, except that the same concentration of DTT used for the SoxR treatment was maintained in the transcription reaction. Reversibility of the DTT treatment was analyzed by omitting DTT from the in vitro transcription reaction. Stability of the FeS center in DTT-treated SoxR was analyzed by gel filtration chromatography of 100-l samples using NICK columns (Pharmacia), following the manufacturer's instructions. Protein concentration in the eluted fractions (ϳ150 l) was measured as described above. Colorimetric determination of iron was performed by adding 0.1 M DTT to the eluted fractions (except for those which already contained it), and a further addition of 10 l of 10 mM ferrozine. The absorption at 562 nm was determined after 20 min of incubation, and iron concentrations were estimated as described above.

Fe-SoxR-dependent Transcription on a Supercoiled
Template-We previously showed that purified Fe-SoxR, but not apo-SoxR, was able to initiate abortive transcription from a linear template containing the soxS promoter . However, we had not examined whether this difference is retained in the more relevant situation of a supercoiled template and transcription of the intact soxS gene in the [2Fe-2S] Clusters in E. coli SoxR Protein presence of competing promoters. A supercoiled plasmid containing the whole soxRS locus (pBD100; Amá bile-Cuevas and Demple, 1991) was used as the template for these in vitro reactions and allowed for the simultaneous detection of bla (␤-lactamase gene) transcription, which was expected to be SoxR-independent. In the presence of E. coli 70 -containing RNA polymerase and Fe-SoxR, a 64-nucleotide primer extension product was detected that corresponded to the size expected for soxS mRNA, while a 105-nucleotide product corresponded to the size expected for the bla transcript. In these experiments (Fig. 1), the amount of soxS mRNA increased strongly with the amount of Fe-SoxR added to the in vitro transcription reaction and was near-maximal with 10 ng of this protein. In contrast, even with 100 ng of apo-SoxR, only a small increase in the production of soxS transcript was observed compared to control reactions with RNA polymerase alone. Thus, the transcriptional activity of SoxR in this assay also depends strongly on its iron content.
5-Phenyl-1,10-phenanthroline-Copper(I) Footprinting-Our previous studies  showed that Fe-SoxR has only a small effect on the binding of RNA polymerase to the promoter, which suggested that Fe-SoxR acts at a subsequent step, such as conformational changes leading to the initiation of transcription. We addressed this issue by conducting footprinting studies with 5-phenyl-1,10-phenanthrolinecopper(I) (CuPPA), a reagent that is also sensitive to proteininduced changes in DNA structure (Sigman et al., 1991). Footprints of both Fe-SoxR and apo-SoxR across the Ϫ10 to Ϫ35 region of both strands of the soxS promoter were apparent in the absence and in the presence of RNA polymerase ( Fig.  2A). These footprints were not as sharply defined as those The Ϫ10 and Ϫ35 sites of the soxS promoter are indicated, as well as the transcriptional start site (ϩ1). Each strand was also chemically digested with dimethyl sulfate and piperidine in a "Gspecific" reaction (Sambrook et al., 1989) to generate sequence markers (G in the figure). For the left panel, the marker positions were established from a longer autoradiographic exposure. B, comparison of the DNase I  and CuPPA footprints and hypersensitive sites. The Ϫ10 and Ϫ35 sites of the soxS promoter are indicated with brackets. The 9-bp inverted repeat is shown in bold, the center of the dyad symmetry being labeled with a dot. The outlined areas show the protection exerted by SoxR against cleavage by DNase I (closed box, upper) or CuPPA (horizontal lines, lower). The hypersensitive sites are indicated by arrows.
[2Fe-2S] Clusters in E. coli SoxR Protein observed with DNase I , but were detected consistently.
CuPPA-hypersensitive sites were observed in the vicinity of the soxS transcription start site only with Fe-SoxR and RNA polymerase together, in both the nontranscribed ( Fig. 2A, left) and the transcribed strand ( Fig. 2A, right). CuPPA-hypersensitive sites at Ϫ3 to Ϫ7 in the transcribed strand and at ϩ4 and ϩ5 of the nontranscribed strand have been associated with open complex formation at other promoters (Sigman et al., 1991;Thederahn et al., 1990). CuPPA-hypersensitive sites were also reported at ϩ4 to ϩ6 in the transcribed strand of the lacUV5 open complex (Thederahn et al., 1990), but were not apparent in our experiments. These sites may have been difficult to detect because they are much weaker than those seen at Ϫ3 to Ϫ7 in the same strand (Sigman et al., 1991;Thederahn et al., 1990). Taken together, these data indicate that Fe-SoxR specifically leads to open complex formation by RNA polymerase.
In addition, three sites in the nontranscribed strand, in the center of the SoxR binding site, became unprotected against CuPPA only with Fe-SoxR and RNA polymerase together ( Fig.  2A, left side). This deprotection was observed consistently when the background cleavage by CuPPA was high, but was difficult to detect under milder cleavage conditions. We therefore determined whether the footprinting reagents had specific effects on the in vitro transcription of soxS. CuPPA at 0.16 mM inhibited soxS-specific transcription (relative to bla transcription) by 30%, but 6.7 mM mercaptopropionic acid eliminated detectable transcription of both bla and soxS. Because of the general inhibition of transcription by mercaptopropionic acid, we checked for specific effects on Fe-SoxR. Fe-SoxR retained 90% of the visible absorption characteristic of the intact metalloprotein . Therefore, the integrity of Fe-SoxR was not strongly compromised by the footprinting reagents, although partial effects were noted.
The CuPPA footprinting results are compared in Fig. 2B with those found previously for DNase I . The cleavage sites in the nontranscribed strand that became unprotected with Fe-SoxR and RNA polymerase are in the center of dyad symmetry of the SoxR binding site (Fig. 2B). Interestingly, CuPPA-hypersensitive sites were described for Hg-MerR, also positioned in the center of the protein binding site (Frantz and O'Halloran, 1990).
Iron and S 2Ϫ Content of Dimeric SoxR-Before determining the iron stoichiometry of SoxR, we first established the oligomeric state of the protein. Two features seemed to indicate that SoxR could exist in solution as a dimer (or higher oligomer). First, the SoxR homologue, MerR, is dimeric both in the presence and in the absence of the Hg 2ϩ (Helmann et al., 1990). Second, SoxR binds a DNA site in the soxS promoter that has perfect dyad symmetry . This hypothesis was tested by sedimentation analysis of both Fe-and apo-SoxR. In these experiments, a clear peak of SoxR protein was observed sedimenting just slower than carbonic anhydrase (29 kDa) and significantly faster than soybean trypsin inhibitor (20 kDa) (Fig. 3). These data indicate that both Fe-SoxR and apo-SoxR are predominantly homodimers at 0.5 M NaCl, even in dilute solution (Ͻ1 M).
The amount of iron in freshly purified SoxR was quantified by two independent methods: colorimetrically with the iron chelator ferrozine and by plasma emission spectrometry. As summarized in Table I, the ratio of Fe to SoxR monomer averaged 2.7 Ϯ 0.3.
For the determination of inorganic sulfide, the low solubility of SoxR necessitated using the assay described by Beinert (1983). These measurements revealed ratios of S 2Ϫ :SoxR monomer near 2 in 11 independent, freshly prepared samples (Table I and data not shown). Together with the iron determinations described above, these data indicate that (SoxR) 2 might contain either a single [4Fe-4S] cluster or a pair of [2Fe-2S] centers.
EPR Spectroscopy of Fe-SoxR-EPR spectroscopy was used

[2Fe-2S] Clusters in E. coli SoxR Protein
to obtain more information about the structure and properties of the SoxR FeS center(s). Both oxidized Fe-SoxR and the protein reduced with an excess of dithionite (Fig. 4A) were analyzed by EPR. No signal was observed for oxidized SoxR at temperatures between 10 and 100 K (Fig. 4B). However, the dithionite-treated protein showed a nearly axial signature with apparent g values of 2.01 and 1.93 (Fig. 4B). This signal was observed at temperatures as high as 100 K. The integrated intensity indicated the presence of 1 spin/2. should exhibit a rhombic signal, which should be observed only at low temperature (Ͻ20 K) and which should integrate to Ͻ1 spin/4 Fe. In contrast, the data are entirely consistent with the presence in (SoxR) 2 of two [2Fe-2S] clusters. The fact that the spin/SoxR monomer ratio exceeds 1 suggests that the SoxR concentration is routinely underestimated (by ϳ30%) by our method.

Destabilization of the SoxR [2Fe-2S]
Center by Thiols-Some transcriptional regulators are thought to be controlled by reversible redox reactions of the protein (Storz et al., 1990;Abate et al., 1990). We therefore tested whether reduction of the [2Fe-2S] centers of SoxR might modulate its transcriptional activity. Preliminary experiments showed that the dithionitetreated protein could not be readily used for this purpose, owing to its easy reoxidation . We therefore examined the effects of treating Fe-SoxR with DTT. Concentrations of this thiol Յ50 mM did not alter the visible spectrum of SoxR. However, incubation of Fe-SoxR with DTT at Ն75 mM decreased the absorption peaks at 414,462 and 548 nm and generated a new maximum at 424 nm (Fig. 5A). The activity of these samples was analyzed in in vitro transcription reactions also containing DTT. In the presence of 100 mM DTT, the transcriptional activity of Fe-SoxR was abolished (Fig. 5B, lane 3 versus lane 2). However, if the DTT concentration was diminished either by gel filtration chromatography in elution buffer lacking the thiol (data not shown) or by omitting the thiol from the transcription reaction (Fig. 5B), Fe-SoxR transcriptional activity was completely recovered. The high DTT concentrations did not affect the transcription of the bla gene (Fig. 5B).
Although the visible absorption spectrum of Fe-SoxR was changed by the DTT treatment (Fig. 5A), DTT-treated SoxR did not show an EPR spectrum (data not shown). Therefore, DTTtreated Fe-SoxR does not correspond to the form of the protein with reduced [2Fe-2S] centers. We further analyzed the effect DTT had on Fe-SoxR by gel filtration chromatography in the absence or the presence of 100 mM DTT in the column buffer. The amounts of SoxR protein and Fe were determined in the eluted fractions. SoxR and Fe co-eluted at a 1:2 ratio in the samples without DTT (Fig. 5C). Chromatography in the presence of DTT yielded a peak of SoxR associated with diminished amounts of iron (a 1:0.9 ratio in the experiment shown) accompanied by a considerable amount of iron of slower mobility (Fig.  5C). These data indicate that DTT reversibly blocks the transcriptional activity of Fe-SoxR by destabilizing the [2Fe-2S] centers, which can then be physically removed in the continuing presence of high levels of DTT.

DISCUSSION
The studies presented here indicate that SoxR protein is a homodimer that, in its activated form, contains a pair of [2Fe-2S] centers. These metal clusters are not required to maintain the overall structure of SoxR, since the apoprotein is also a homodimer that binds the soxS promoter with high affinity . It remains to be established whether the two SoxR [2Fe-2S] clusters are arranged as one per subunit or as a pair of clusters coordinated between the subunits. With respect to the latter possibility, it is interesting to note that a single Hg is coordinated to the homologous MerR protein by distinct cysteine ligands from each subunit of a dimer (Helmann et al., 1990). Two of these cysteine residues of MerR are positioned identically in alignments with the SoxR protein (Amá bile-Cuevas and Demple, 1991). At least one intersubunit iron-sulfur cluster has been described, the [4Fe-4S] center of Azotobacter vinelandii nitrogenase (Georgiadis et al., 1992).
The [2Fe-2S] centers of SoxR are necessary for the protein's FIG. 4. Spectroscopic analysis of Fe-SoxR. A, UV/visible absorption. SoxR protein, ϳ10 M in 50 mM HEPES-NaOH, pH 7.6, 0.5 M NaCl, was reduced with dithionite under anaerobic conditions (see text for details). The absorption spectrum over the range 300 -600 nm was recorded immediately after transfer of the sample to a sealed cuvette (reduced SoxR). An untreated Fe-SoxR sample is also shown (oxidized SoxR). The strong absorption below 380 nm is due to the relatively large amount of dithionite used for reduction in this experiment (25 eq relative to SoxR, compared to 5 eq in previous work ). B, EPR spectroscopy. Oxidized or dithionite-reduced SoxR (300-l samples) were processed for EPR spectroscopy as described under "Materials and Methods." Upper trace, EPR spectrum of ϳ10 M reduced Fe-SoxR at 40 K; the g values are indicated. Lower trace, oxidized (not dithionite-treated) Fe-SoxR. The upper spectrum was recorded at 40 K with the following spectrometer conditions: microwave frequency, 9.42 GHz; microwave power, 200 microwatts; modulation amplitude, 1 millitesla (mT); receiver gain, 10 5 . The lower spectrum was recorded at 10 K, with a modulation amplitude of 0.4 millitesla, and all other conditions identical with those described for the upper spectrum. No signal was observed for oxidized SoxR at temperatures between 10 and 100 K, and magnetic fields between 290 and 370 millitesla. function as a transcriptional activator. The features of CuPPA footprinting of Fe-SoxR and RNA polymerase at the soxS promoter show that only active SoxR leads to open complex formation by the polymerase. For the homologous MerR protein, transcriptional activation has been proposed to result from localized underwinding that compensates for the suboptimal spacing (19 bp) between the Ϫ10 and Ϫ35 elements of the merT promoter (Ansari et al., 1992). The SoxR-regulated soxS promoter also seems to be overwound, with 19-bp spacing .
The FeS centers of many proteins are damaged when cells are exposed to intracellular superoxide-generating agents or to nitric oxide. Superoxide-sensitive FeS proteins typically contain [4Fe-4S] centers (Gardner and Fridovich, 1991;Liochev and Fridovich, 1992). Tetranuclear FeS centers, as found in aconitase, are inactivated in mammalian cells exposed to nitric oxide (Drapier et al., 1993;Weiss et al., 1993), perhaps by peroxynitrite (Hausladen and Fridovich, 1994) formed from the combination of NO ⅐ and O 2 . (Koppenol et al., 1992). In contrast, Fe-SoxR must remain active when E. coli is exposed to high intracellular fluxes of O 2 . (Nunoshiba et al., 1992;Wu and Weiss, 1992) or NO ⅐ (Nunoshiba et al., 1993(Nunoshiba et al., , 1995. Perhaps binuclear [2Fe-2S] clusters are well suited to this requirement. The [2Fe-2S] clusters of SoxR seem to be quite stable in the oxidized form that is rapidly generated upon exposure of the protein to air   (Flint et al., 1993).
Although the metal centers of SoxR are clearly essential for the transcriptional activity of the protein ; this work), the mechanism that activates SoxR in vivo is unknown. As demonstrated by the EPR experiments present here, the oxidized form of the protein is certainly active. The question then is whether the nonactivated state for SoxR is the reduced form or the apoprotein, or perhaps some other species. Some recent experiments 2 suggest that dithionite-reduced Fe-SoxR is still active as a transcription factor (in contrast to DTT-treated Fe-SoxR), but the ease with which this protein is reoxidized by O 2 in vitro indicates that caution should be applied in this interpretation. It is also possible that apo-SoxR is the physiologically relevant inactive state. If this is so, the [2Fe-2S] centers would be reconstituted adventitiously during extraction of the protein, even from cells not treated to activate SoxR. This possibility is being explored.
So far, apo-SoxR has been isolated only following purification of the protein in buffers containing ␤-mercaptoethanol . Since the mere addition of ␤-mercaptoethanol to Fe-SoxR did not affect its spectroscopic or transcriptional properties, 2 it seemed likely that this thiol destabilizes the [2Fe-2S] centers and allows their removal during chromatography . A similar destabilizing effect seems to be mediated by DTT, high concentrations of which also abolish the transcriptional activity of Fe-SoxR. DTT-treated Fe-SoxR and apo-SoxR are the only two transcriptionally inactive forms of the protein thus far identified. This opens the possibility that the DTT-treated protein could correspond to the inactive form of SoxR in vivo, perhaps with partially disassembled [2Fe-2S] centers. However, the high thiol 2 E. Hidalgo and B. Demple, unpublished data. concentrations necessary to generate this inactive state do not prevail in cells, which typically contain ϳ5 mM glutathione accounting for most of the low molecular weight thiol (Meister and Anderson, 1983).