Communication

Activation of SoxR-dependent Transcription in Vitro by Noncatalytic or NifS-mediated Assembly of [2Fe-2S] Clusters into Apo-SoxR*
(Received for publication, December 18, 1995, and in revised form, February 5, 1996) Elena Hidalgo and Bruce Demple From the Department of Molecular and Cellular Toxicology, Harvard School of Public Health, Boston, Massachusetts 02115

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 271, No. 13, Issue of March 29, pp. 7269 7272, 1996 1996 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

SoxR is a transcriptional activator that senses superoxide and nitric oxide stress in Escherichia coli. The active protein isolated from E. coli contains a pair of [2Fe-2S] clusters per SoxR dimer. We previously demonstrated that the iron-free protein (apo-SoxR), isolated during purification in thiol-containing buffers, binds soxS promoter DNA with an affinity equal to that of the metalloprotein (Fe-SoxR), but lacks significant ability to activate transcription in vitro. Here we demonstrate the reversibility of this process: the full transcriptional activity of SoxR can be restored by in vitro assembly of iron-sulfur clusters into the apoprotein. Two methods were used to synthesize the metallocenters of SoxR: (i) nonenzymatic, in which apo-SoxR, incubated in the presence of iron, inorganic sulfide, and a reducing agent, regained full transcriptional activity in 5 6 h; (ii) enzymatic, in which NifS protein of Azotobacter vinelandii regenerated active Fe-SoxR in as little as 2 min. Analysis by electron paramagnetic resonance spectroscopy indicated that binuclear [2Fe-2S] clusters were restored by both the enzymatic and nonenzymatic reconstitutions. A mutant SoxR protein missing one of its four cysteine residues failed to undergo either transcriptional activation or the formation of [2Fe-2S] centers, even in the presence of NifS. Thus, only the presence of an iron-sulfur center is required to restore transcriptional activity to apo-SoxR. Moreover, the catalytic generation of [2Fe-2S] centers extends the known specificity of this enzyme beyond that already shown for [4Fe4S] centers. Catalytic generation of [2Fe-2S]-containing SoxR could allow for rapid activation of this transcription factor in vivo.

The functions of iron-sulfur (FeS)1 clusters in electron trans* This work was supported in part by National Institutes of Health Grant CA37831 (to B. D.), the Amyotropic Lateral Sclerosis Association, and American Cancer Society Grant NP-899. 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. Supported by a Fulbright Fellowship. To whom correspondence should be addressed. Tel.: 617-432-3462; Fax: 617-432-0377; E-mail: demple@mbcrr.harvard.edu. 1 The abbreviations used are: FeS, iron-sulfur; apo-SoxR, the protein lacking iron; DTT, dithiothreitol; Fe-SoxR, iron-containing SoxR protein; IRP, iron-responsive protein; SoxR-C117A, noninducible and ironlacking mutant protein with cysteine 117 substituted by alanine (T. M. Bradley, E. Hidalgo, H. Ding, and B. Demple, in preparation).

fer reactions are well established, but new roles have recently been found for these metallocenters (Beinert, 1990; Johnson, 1994). [4Fe-4S] clusters are involved directly in hydrolytic catalysis by dehydratases such as aconitase (Klausner et al., 1993) and dihydroxy-acid dehydratase. (Flint et al., 1993a). A recent report proposed that the [4Fe-4S] center of endonuclease III of Escherichia coli is involved in DNA recognition (Thayer et al., 1995). Regulatory roles for FeS centers have been described for the mammalian iron response protein (IRP; Klausner et al. (1993)) and are implicated for two bacterial regulators, Fnr protein (Khoroshilova et al., 1995) and SoxR protein (Hidalgo and Demple (1994); see below). Some tetranuclear [4Fe-4S] clusters appear to be very sensitive to damage via oxidation. For example, both E. coli dihydroxy acid-dehydratase (Flint et al., 1993b) and Bacillus subtilis phosphoribosyl diphosphate 5-amidotransferase (Bernlohr and Switzer, 1981; Grandoni et al., 1989) lose their respective clusters in the presence of hyperbaric oxygen in vivo and in vitro. The amidotransferase apoprotein is immediately degraded upon cluster disassembly (Bernlohr and Switzer, 1981; Grandoni et al., 1989), while the dehydratase apoprotein is stable and, when oxygen levels return to normal, reactivates by the reinsertion of iron and inorganic sulfide (Flint et al., 1993b). Cluster assembly/disassembly seems to be employed deliberately as a regulatory mechanism in mammalian IRP. In the latter case, iron limitation (Klausner et al., 1993) or oxidative damage (Drapier et al., 1993) lead to formation of the apoprotein in vivo; apo-IRP specifically binds certain mRNAs to effect post-transcriptional control, which is lost when the metalloprotein is regenerated. Little is known about the biochemical mechanisms of FeS cluster disassembly or assembly, either into newly synthesized proteins or in the cases of reversible processes cited above. The Azotobacter vinelandii nifS gene was identified as essential for the in vivo formation of active nitrogenase, which contains oxygen-sensitive [4Fe-4S] centers (Kennedy and Dean, 1992). NifS protein catalyzes the formation of these FeS centers in nitrogenase (Zheng et al., 1993), probably by mobilization of inorganic sulfide from L-cysteine (Zheng et al., 1994). In vitro NifS-mediated reconstitution of [4Fe-4S] clusters into the E. coli Fnr regulatory protein has also been described recently (Khoroshilova et al., 1995). Dean and collaborators have proposed a general role for NifS-type proteins in FeS cluster assembly (Zheng and Dean, 1994), although the reconstitution of other types of clusters has not been reported. SoxR protein is a transcriptional activator that triggers a defense response against excess superoxide or nitric oxide in E. coli (Hidalgo and Demple, 1995). SoxR is post-translationally activated to stimulate transcription of a second regulatory gene, soxS (Nunoshiba et al., 1992; Wu and Weiss, 1992). The newly synthesized SoxS protein then triggers expression of 10 genes involved in defenses against oxidative damage (Amabile-Cuevas and Demple, 1991; Li and Demple, 1994; Wu and Weiss, 1992) and antibiotic resistance (Chou et al., 1993; Ma et al., 1996). SoxR is a dimeric FeS protein that contains two [2Fe-2S] centers (Hidalgo and Demple, 1994; Wu et al., 1995; Hidalgo et al., 1995). The SoxR FeS centers are not required for DNA binding, but are essential for promoting open-complex formation by RNA polymerase and triggering expression of the soxS

Downloaded from www.jbc.org at ARIZONA STATE UNIVERSITY, on December 7, 2011

7269

7270

In Vitro Assembly of [2Fe-2S] Clusters into Apo-SoxR

gene (Hidalgo and Demple, 1994; Hidalgo et al., 1995). The biochemical mechanism by which SoxR senses oxidative stress and activates transcription of soxS is unknown, but almost certainly relates to the proteins [2Fe-2S] clusters (Hidalgo and Demple, 1994). Two possible mechanisms of SoxR activation have been considered: (i) redox reactions involving pre-existing, reduced [2Fe-2S] clusters and (ii) assembly of the [2Fe-2S] clusters into pre-existing apoprotein. Here we demonstrate that purified apo-SoxR, which is unable to activate soxS transcription, can be fully activated for this function by in vitro reconstitution of its [2Fe-2S] centers, and that this process is catalyzed by A. vinelandii NifS protein.
MATERIALS AND METHODS

Purification of SoxRIron-containing SoxR (Fe-SoxR) and apo-SoxR were obtained from E. coli containing the SoxR expression plasmid pKOXR using purification buffers containing 1 mM 2-mercaptoethanol (apo-SoxR) or lacking added thiols (Fe-SoxR), as described previously (Hidalgo and Demple, 1994). Fractions eluted from heparin-agarose (Life Technologies Inc.) columns (SoxR purity of 80%) were used for these studies. Noncatalytic Reconstitution of SoxRTwo-ml aliquots of 50 mM HEPES-NaOH, pH 7.6, 0.1 M NaCl, 0.14 M 2-mercaptoethanol were deoxygenated by bubbling with 100% argon gas. FeCl2 and Na2S, each at a final concentration of 0.15 mM, were then added from previously deoxygenated 150 mM stocks. After 1530 min incubation at room temperature, 200- l aliquots of deoxygenated 10 M apo-SoxR were added to the reconstitution mixtures. The reactions were then incubated at room temperature under argon, and samples were removed at the indicated times to measure SoxR transcriptional activity in vitro. Where indicated in the text, reconstituted samples were concentrated by ultrafiltration in Centricon microconcentrators (Amicon), treated with 1 mM dithionite, and analyzed by electron paramagnetic resonance (EPR) spectroscopy. NifS-mediated Reconstitution of SoxRThe following solutions were deoxygenated during a 30-min bubbling with argon gas: 200 mM ferrous ammonium sulfate, 100 mM L-cysteine in 500 mM DTT, 4.1 M apo-SoxR or SoxR-C117A,2 and 10 M NifS. Purified recombinant NifS protein (Zheng et al., 1994), kindly provided by Limin Zheng and Dennis Dean (Virginia Polytechnic University, Blacksburg) and stored at 80 C, was freshly diluted in 25 mM Tris-HCl buffer, pH 7.4. NifS that had been refrozen at 80 C was ineffective in reconstitution reactions.3 The above solutions were moved inside an anaerobic chamber (Vacuum Atmospheres Co. model HE-493) kept at 1 ppm O2. The final reconstitution mixtures (0.5 ml) contained 4 M apo-SoxR or SoxR-C117A, 1 mM L-cysteine, 5 mM DTT, 2 mM ferrous ammonium sulfate, and 0.1 M NifS. The solutions were incubated at 30 C and removed from the anaerobic chamber at the indicated times. Aliquots of 300 l were then immediately incubated with 1 mM dithionite (added from a 100 mM stock freshly prepared in 1 M HEPES, pH 7.6), placed in an EPR tube, and frozen in a liquid nitrogen bath until spectroscopic analysis. Samples of the reconstitution reactions were also immediately assayed in in vitro transcription reactions. In Vitro TranscriptionThe activity of SoxR protein was determined by in vitro transcription of the soxS gene in plasmid pBD100 by E. coli RNA polymerase (Hidalgo et al., 1995). The soxS transcript (SoxR-dependent) and a control bla transcript (SoxR-independent) were quantified by primer-extension analysis as described previously (Hidalgo et al., 1995), except that the primer soxS-1 (5 -GCGATAAGATCCTGAATAAT-3 ) was used instead of primer 1 (Hidalgo et al., 1995), which is less stable.3 Primer soxS-1 hybridizes 75 base pairs downstream of the transcription start site of the soxS gene; its extension product migrated with the expected mobility in urea-polyacrylamide gels. EPR Spectroscopy300- l samples of reconstituted SoxR were incubated with freshly dissolved 1 mM dithionite and immediately placed inside 4-mm EPR sample tubes. The tubes were quickly frozen in a liquid nitrogen bath and kept at 80 C until analysis. X-band EPR spectra were recorded at 20 K on a Bruker model ESP300 spectrometer maintained at constant temperature, either with an Oxford Instruments model ESR910 continuous flow cryostat or with a Bruker model ER4111VT variable temperature controller. The amount of reconsti2 T. M. Bradley, E. Hidalgo, H. Ding, and B. Demple, manuscript in preparation. 3 E. Hidalgo and B. Demple, unpublished data.

FIG. 1. Spontaneous formation of active SoxR with iron and inorganic sulfide. Apo-SoxR (1 M) was incubated anaerobically at room temperature in the presence of ferrous salts, sodium sulfide, and the reducing agent 2-mercaptoethanol (see Materials and Methods). At the times indicated, 50- to 100- l aliquots were removed using a syringe and diluted in 0.1 M NaCl, 50 mM HEPES-NaOH, pH 7.6. The transcriptional activity of RNA polymerase (RNAP) alone (lane 1) or in the presence of 10 ng of purified Fe-SoxR (lane 2), untreated apo-SoxR (lane 3) or treated apo-SoxR (lanes 4 7) was determined as described under Materials and Methods. Similar results were obtained in four independent experiments. The primer extension products (soxS, bla) are indicated by arrows. tuted SoxR was determined by comparison in the same experiment to standardized Fe-SoxR samples, previously reduced with dithionite as described above. Protein DeterminationsProtein concentrations were determined by Coomassie staining of SDS-polyacrylamide gels, using as a standard SoxR previously quantified by amino acid analysis (Hidalgo and Demple, 1994).
RESULTS

Downloaded from www.jbc.org at ARIZONA STATE UNIVERSITY, on December 7, 2011

Anaerobic Reactivation of Apo-SoxRTranscriptionally inactive apo-SoxR is formed upon exposure of Fe-SoxR to 2-mercaptoethanol-containing buffers during purification (Hidalgo and Demple, 1994). We wished to determine whether SoxR inactivation could be reversed in vitro by reconstitution of the [2Fe-2S] clusters into the apoprotein. Reconstitution reactions were carried out by anaerobic incubation of diluted apo-SoxR (final concentration 1 M) with iron salts and Na2S in the presence of the reducing reagent 2-mercaptoethanol. Samples were removed at various times to assess their in vitro transcriptional activity specific for the soxS promoter. As shown in Fig. 1, a slow reconstitution process was observed that was essentially complete by 6 h. There was no activating effect of SoxR on bla transcription under any circumstances (Fig. 1). We confirmed the presence of [2Fe-2S] clusters in this spontaneously reconstituted SoxR by concentrating samples of the reconstitution reactions, reduction with dithionite, and EPR spectroscopy.3 Although clear signatures of [2Fe-2S] clusters were observed after a 6-h reconstitution, the yield measured by EPR was lowered than expected. It seems likely that the instability of the newly synthesized FeS clusters in the presence of thiols and oxygen prevented observation of the full complement of [2Fe-2S] centers in reactivated SoxR. NifS Catalyzes Apo-SoxR ReconstitutionThe slow reconstitution of SoxR described above prompted us to determine whether the assembly of the SoxR FeS clusters could be accelerated. We employed for this purpose the A. vinelandii NifS protein, which has been reported to catalyze the assembly of two different [4Fe-4S] centers (Khoroshilova et al., 1995; Zheng et al., 1994). Incubation of apo-SoxR under anaerobic condi-

In Vitro Assembly of [2Fe-2S] Clusters into Apo-SoxR

7271

FIG. 2. NifS-mediated reactivation of SoxR transcriptional activity. A, 4.1 M sample of apo-SoxR was incubated inside an anaerobic chamber with 0.1 M NifS protein in the presence of ferrous salts, the sulfide donor L-cysteine, and DTT. The reconstitution reaction was incubated at 30 C for 10 min. The sample was immediately removed from the chamber, and SoxR was analyzed for in vitro transcription activity (lane 4). Control reactions show the primer extension products for RNA polymerase (RNAP) alone (lane 1), purified Fe-SoxR (lane 2), or untreated apo-SoxR (lane 3). Although the amount of soxS transcription relative to bla appears higher in lane 4 than in lane 2, this effect was not seen in two other experiments in which the activities (measured as the soxS/bla transcription ratio) of NifS-reconstituted SoxR and Fe-SoxR were equal. B, as for A, but using SoxR-C117A (cysteine 117 substituted by an alanine2) instead of apo-SoxR in the reconstitution reaction; the mixture was incubated for 60 min at 30 C before transcriptional activity assay. The primer extension products (soxS, bla) are indicated by arrows.

Downloaded from www.jbc.org at ARIZONA STATE UNIVERSITY, on December 7, 2011

tions with inorganic iron and a reducing reagent (DTT), as for the nonenzymatic process, and the further addition of L-cysteine and catalytic amounts of NifS (40-fold lower concentration than SoxR) gave a dramatic increase in the reactivation rate (Fig. 2A). Under these conditions, full transcriptional activity was reproducibly restored to apo-SoxR in as little as 10 min. In fact, full activity was restored to apo-SoxR by NifS within 2 min even at 25 C; uncatalyzed reconstitution under the same conditions reached only 40% after 55 min.3 The recovery of SoxR activity required the full complement of SoxR cysteine residues. When partially purified SoxR-C117A ( 80%), a noninducible mutant protein with cysteine 117 substituted by an alanine and which lacks detectable iron,2 was incubated with NifS up to 60 min, no transcriptional activity was detected in vitro (Fig. 2B). The amount of SoxR in the NifS-mediated reconstitution reaction was 4 M, high enough for direct EPR analysis without additional concentration. We compared the EPR spectrum of a reduced Fe-SoxR sample (8 M; Fig. 3A) with those of untreated apo-SoxR or NifS-reconstituted apo-SoxR after dithionite treatment (Fig. 3, B and C, respectively). Comparison of the signals in A and C indicated that essentially complete reconstitution of the [2Fe-2S] centers had been achieved (4 M estimated from EPR signal versus 4 M SoxR protein in Fig. 3C).
DISCUSSION

FIG. 3. EPR analysis of NifS-reconstituted Fe-SoxR. Apo-SoxR was incubated anaerobically with NifS as for Fig. 2A, then reduced and subjected to EPR spectroscopy. A 400- l sample was analyzed at 20 K, with a microwave frequency of 9.42 GHz, modulation frequency of 100 kHz, microwave power of 1 milliwatt, modulation amplitude of 1.29 mT, time constant of 40.96 ms, and magnetic field from 310 to 370 mT. A, purified Fe-SoxR (printing scale 15); B, untreated apo-SoxR (printing scale 14); C, reconstituted Fe-SoxR (printing scale 14).

The results presented here demonstrate two novel points. First, reassembly of the SoxR iron-sulfur center is sufficient to restore full transcriptional activity to the protein. No other physical difference between apo-SoxR and the transcriptionally active protein (e.g. oxidation or reduction of an amino acid side chain) needs to be postulated to account for the difference in activity. Second, the reassembly of [2Fe-2S] clusters into SoxR is efficiently catalyzed by NifS, extending the known specificity of this enzyme. Since the NifS activity generates S2 (Zheng et al., 1993), this result implies that apo-SoxR does not contain sufficient residual inorganic sulfide for the reassembly process to occur. SoxR is a transcriptional activator that responds to superox-

ide and nitric oxide stress. We have been exploring the role of the SoxR-FeS centers in the oxidative stress-sensing and transcription-activating functions of this unusual protein. The mechanism by which SoxR is activated by intracellular superoxide or nitric oxide has not been established, in part because the resting (inactive) state of SoxR in vivo remains unknown. The experiments presented here demonstrate that such activation can occur in vitro by the reconstitution of SoxRs [2Fe-2S] centers, and that the process mediated by NifS protein is sufficiently rapid to mediate in vivo activation. SoxR activation occurs within 10 min in response to the superoxide-generating agent paraquat, as determined using a single-copy soxS -lacZ operon fusion (Nunoshiba and Demple, 1993). The sluggishness of spontaneous SoxR reactivation by FeS cluster assembly indicates that catalysis would be involved if in vivo activation occurs by this process. The only E. coli activity isolated so far that is able to favor FeS cluster assembly seems to be related to NifS.4 It will be important to determine whether this activity also restores [2Fe-2S] centers and transcriptional activity to apo-SoxR. The mechanisms by which FeS centers are incorporated into proteins have not been established and could differ for different cluster types. It has been proposed that clusters are preassembled from iron and sulfide in a process mediated by NifStype enzymes, followed by integration of the clusters into apoproteins (Zheng and Dean, 1994). If so, suitable forms are generated by NifS for both [4Fe-4S] (Zheng and Dean, 1994; Khoroshilova et al., 1995) and [2Fe-2S] proteins, as shown by the catalysis of Fe-SoxR reassembly. Although genetic evidence implicates A. vinelandii NifS in the formation of active nitrogenase (Kennedy and Dean, 1992), the generality of this role has not been established for other FeS proteins. It is also unknown whether NifS-mediated reconstitution of FeS centers applies to both newly synthesized and pre-existing apoproteins. Reconstitution of FeS centers into proteins that have lost them
4

D. Flint, personal communication.

7272

In Vitro Assembly of [2Fe-2S] Clusters into Apo-SoxR

REFERENCES
Amabile-Cuevas, C., and Demple, B. (1991) Nucleic Acids Res. 19, 4479 4484 Beinert, H. (1990) FASEB J. 4, 24832491 Bernlohr, D. A., and Switzer, R. L. (1981) Biochemistry 20, 56755681 Chou, J. H., Greenberg, J. T., and Demple, B. (1993) J. Bacteriol. 175, 1026 1031 Drapier, J. C., Hirling, H., Wietzerbin, J., Kaldy, P., and Kuhn, L. C. (1993) EMBO J. 12, 36433649 Flint, D. H., Emptage, M. H., Finnegan, M. G., Fu, W., and Johnson, M. K. (1993a) J. Biol. Chem. 268, 1473214742 Flint, D. H., Smyk-Randall, E., Tuminello, J. F., Draczynska-Lusiak, B., and Brown, O. R. (1993b) J. Biol. Chem. 268, 2554725552 Flint, D. H., Tuminello, J. F., and Emptage, M. H. (1993c) J. Biol. Chem. 268, 22369 22376 Gardner, P. R., and Fridovich, I. (1991) J. Biol. Chem. 266, 1478 1483 Grandoni, J. A., Switzer, R. L., Makaroff, C. A., and Zalkin, H. (1989) J. Biol. Chem. 264, 6058 6064 Hidalgo, E., and Demple, B. (1994) EMBO J. 13, 138 146 Hidalgo, E., and Demple, B. (1995) in Regulation of Gene Expression in Escherichia coli (Lin, E. C. C., and Lynch, A. S., eds) R. G. Landes Co., Austin, TX, in press Hidalgo, E., Bollinger, J. M., Jr., Bradley, T. M., Walsh, C. T., and Demple, B. (1995) J. Biol. Chem. 270, 20908 20914 Johnson, M. K. (1994) in Encyclopedia of Inorganic Chemistry (King, R. B., ed) Vol. 4, pp. 1896 1915, John Wiley & Sons Ltd., Chichester, UK Kennedy, C., and Dean, D. (1992) Mol. & Gen. Genet. 231, 494 498 Khoroshilova, N., Beinert, H., and Kiley, P. J. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 2499 2503 Klausner, R. D., Rouault, T. A., and Harford, J. B. (1993) Cell 72, 19 28 Li, Z., and Demple, B. (1994) J. Biol. Chem. 269, 1837118377 Liochev, S. I., and Fridovich, I. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 58925896 Ma, D., Alberti, M., Lynch, C., Nikaido, H., and Hearst, J. E. (1996) Mol. Microbiol. 19, 101112 Nunoshiba, T., and Demple, B. (1993) Cancer Res. 53, 3250 3252 Nunoshiba, T., Hidalgo, E., Amabile-Cuevas, C., and Demple, B. (1992) J. Bacteriol. 174, 6054 6060 Nunoshiba, T., deRojas-Walker, T., Wishnok, J. S., Tannenbaum, S. R., and Demple, B. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 99939997 Sanyal, I., Cohen, G., and Flint, D. H. (1994) Biochemistry 33, 36253631 Thayer, M. M., Ahern, H., Xing, D., Cunningham, R. P., and Tainer, J. A. (1995) EMBO J. 14, 4108 4120 Wu, J., and Weiss, B. (1992) J. Bacteriol. 174, 39153920 Wu, J., Dunham, W. R., and Weiss, B. (1995) J. Biol. Chem. 270, 1032310327 Zheng, L., and Dean, D. R. (1994) J. Biol. Chem. 269, 1872318726 Zheng, L., White, R. H., Cash, V. L., Jack, R. F., and Dean, D. R. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 2754 2758 Zheng, L., White, R. H., Cash, V. L., and Dean, D. R. (1994) Biochemistry 33, 4714 4720

would lead to inactivation of regulatory activity in some cases (e.g. IRP) and activation in others (Fnr, SoxR). Several interesting possibilities exist that relate the formation of FeS centers to the activation of SoxR in vivo. If the resting state is apo-SoxR, oxidative stress might generate available Fe from oxidant-sensitive [4Fe-4S] proteins (Gardner and Fridovich, 1991; Liochev and Fridovich, 1992; Flint et al., 1993c) to serve in the formation of Fe-SoxR. However, we have consistently detected EPR signatures in vivo in untreated E. coli corresponding to 2550% occupancy of the SoxR [2Fe-2S] centers when SoxR is overproduced to 1% of the cell protein. Thus, the nonactivated state of SoxR could be a hemi-metalloprotein with a single [2Fe-2S] center per (SoxR)2 dimer. Still another possibility is that Fe-SoxR in the reduced form is constantly being generated, but is unstable and rapidly converted to the apoprotein. In this mechanism, oxidative stress accompanied by superoxide or nitric oxide would oxidize FeSoxR to stabilize the metalloprotein. In fact, oxidized SoxR is sufficiently stable to withstand considerable handling during purification (Hidalgo and Demple, 1994; Wu et al., 1995). Other [2Fe-2S] proteins, such as E. coli biotin synthase (Sanyal et al., 1994), are also more stable in the oxidized than the reduced form. In such a fashion, the activity of SoxR in vivo would be connected to both the redox activity of its [2Fe-2S] centers and to their assembly.
AcknowledgmentsWe greatly appreciate the generosity of Limin Zheng and Dennis Dean in providing purified NifS protein, as well as their advice on performing the NifS-mediated reconstitutions described in this paper. We thank Dennis Flint for helpful discussions regarding the nonenzymatic SoxR reconstitution described above and for communicating unpublished results. We are also indebted to Dr. Huangen Ding for recording the EPR spectra shown in this work. We are grateful to Professor Joanne Stubbe and members of her laboratory and to the Chemistry Dept. of the Massachusetts Institute of Technology, for allowing us the use of their EPR spectrometer.

Downloaded from www.jbc.org at ARIZONA STATE UNIVERSITY, on December 7, 2011