Mechanism of [4Fe-4S](Cys)4 Cluster Nitrosylation Is Conserved among NO-responsive Regulators*

Background: Fumarate nitrate reduction (FNR) regulator, the master switch between aerobic and anaerobic respiration, responds in vivo to nitric oxide (NO). Results: [4Fe-4S] FNR reacts rapidly with eight NO molecules in a complex, multistep reaction. Conclusion: FNR nitrosylation is remarkably similar to that of WhiB-like proteins. Significance: A common mechanism of nitrosylation exists for phylogenetically unrelated iron-sulfur regulatory proteins. The Fumarate nitrate reduction (FNR) regulator from Escherichia coli controls expression of >300 genes in response to O2 through reaction with its [4Fe-4S] cluster cofactor. FNR is the master switch for the transition between anaerobic and aerobic respiration. In response to physiological concentrations of nitric oxide (NO), FNR also regulates genes, including the nitrate reductase (nar) operon, a major source of endogenous cellular NO, and hmp, which encodes an NO-detoxifying enzyme. Here we show that the [4Fe-4S] cluster of FNR reacts rapidly in a multiphasic reaction with eight NO molecules. Oxidation of cluster sulfide ions (S2−) to sulfane (S0) occurs, some of which remains associated with the protein as Cys persulfide. The nitrosylation products are similar to a pair of dinuclear dinitrosyl iron complexes, [Fe(I)2(NO)4(Cys)2]0, known as Roussin's red ester. A similar reactivity with NO was reported for the Wbl family of [4Fe-4S]-containing proteins found only in actinomycetes, such as Streptomyces and Mycobacteria. These results show that NO reacts via a common mechanism with [4Fe-4S] clusters in phylogenetically unrelated regulatory proteins that, although coordinated by four Cys residues, have different cluster environments. The reactivity of E. coli FNR toward NO, in addition to its sensitivity toward O2, is part of a hierarchal network that monitors, and responds to, NO, both endogenously generated and exogenously derived.

Nitric oxide (NO) is a gaseous lipophilic radical molecule that functions in eukaryotes both as a signal molecule (at nanomolar concentrations) and as a cytotoxic agent (at micromolar concentrations) (1). The latter arises from the ability of NO to react readily with a variety of cellular targets leading to thiol S-nitrosation (2), amino acid N-nitrosation (3), and nitrosative DNA damage (4). Iron-sulfur proteins, which are vital for many important cellular functions, can also be highly reactive toward NO (5).
Bacteria that carry out anaerobic respiration with nitrate/ nitrite as terminal electron acceptors can produce NO endogenously (6,7). Furthermore, many Gram-positive bacteria contain NO synthases, and the NO they generate is proposed to play a protective role against antibiotics (8). Pathogenic bacteria will encounter exogenous NO generated by the host as a defense mechanism. To survive and thrive, bacteria, both pathogenic and nonpathogenic, must be able to respond to fluctuations in NO concentrations, spanning a wide range, and to remove it when it reaches cytotoxic levels (9).
Several of the regulatory proteins known to respond to NO contain iron-sulfur clusters that function as the sensory module (10). Until recently, little was known about the reactions of iron-sulfur clusters with NO in regulatory proteins. Wbl family proteins, found exclusively in the actinomycetes, contain a [4Fe-4S] cluster that undergoes a remarkably rapid and complex multistep reaction involving a total of eight NO molecules per cluster (11). The products of the reaction have the stoichiometry of [Fe(I) 2 (NO) 4 (SR) 2 ], similar to RRE 2 complexes. Similar species have recently been recognized as a major product of cluster nitrosylation in a Rieske-type [2Fe-2S] protein (12). The products of the NO reaction are EPR-silent (most likely with electron spin S ϭ 0), unlike the tetrahedral dinitrosyl iron complexes (DNICs), [Fe(I)(NO) 2 (SR) 2 ] Ϫ , previously identified as products of iron-sulfur cluster nitrosylation, that do give EPR signals (S ϭ 1 ⁄ 2).
The FNR regulator, which contains a [4Fe-4S] cluster, senses O 2 and controls the switch between anaerobic and aerobic respiration in many bacteria (13,14). FNR has sequence homology with cyclic-AMP receptor protein and, like the cyclic-AMP receptor protein, consists of two distinct domains that provide DNA binding and sensory functions (15). FNR becomes activated under anaerobic conditions by the insertion of an O 2 -labile [4Fe-4S] 2ϩ cluster, coordinated by four Cys residues (Cys-20, -23, -29, and -122), into the N-terminal sensory domain (14). Cluster incorporation promotes dimerization, enabling the C-terminal DNA binding domain to recognize specific binding sites located in FNR-regulated promoters (16).
In addition to its response to O 2 , FNR also responds to NO. Anaerobic exposure of E. coli cells to physiologically relevant concentrations (ϳ5 M) of NO led to up-regulation of multiple FNR-repressed genes and down-regulation of FNR-activated genes, suggesting that NO inactivates FNR (20,21). Among those genes up-regulated by FNR in the presence of NO is hmp, encoding a flavohemoglobin (21) that constitutes a principal NO detoxification pathway in E. coli and a wide range of other bacteria. It also plays an important role in the establishment of infection by pathogens (22). Reaction of [4Fe-4S] FNR with NO in vitro was shown to generate ironnitrosyl species associated with FNR, which bound DNA with reduced affinity (21).
Here we report detailed kinetic and thermodynamic investigations that reveal a rapid, multistep reaction of NO with the [4Fe-4S] cluster of FNR that, in terms of both rate and mechanism, is remarkably similar to that previously observed for Streptomyces and Mycobacterium Wbl proteins (11). These results suggest a common mechanism for the nitrosylation of the tetra-Cys ligated iron-sulfur clusters of different regulator proteins.

EXPERIMENTAL PROCEDURES
RNA Isolation and RT-PCR-Escherichia coli JRG6348, which lacks a chromosomal copy of the fnr gene and carries a single-copy FNR-dependent lacZ reporter gene, was transformed with pBAD-fnr and cultured at 37°C under anaerobic conditions in M9 minimal medium supplemented with 5% L-broth, 0.4% glycerol, 20 mM trimethylamine-N-oxide, and 20 mM fumarate (7) until the optical density at 600 nm was 0.175-0.200. For the time-course experiments, anaerobic supplements were then added to the cultures as follows: water; 10 mM sodium nitrite; 10 M each NOC-5 and NOC-7 (Enzo Life Sci-ences). In addition, one culture was exposed to air with shaking. Samples after supplementation were taken at t ϭ 0, 5, 10, and 15 min. For dose dependence experiments, anaerobic NOC-5 and NOC-7 mixtures were added to the cultures such that the indicated amounts of NO were released at the 15-min sampling point. All samples were mixed with 0.4 volumes of ice-cold 95% ethanol, 5% phenol, pH 4.5, to rapidly stabilize the mRNA. Total RNA was prepared using the RNeasy RNA purification kit (Qiagen) according to the manufacturer's instructions and quantified on a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific). Relative lacZ RNA quantities were determined on a Mx3005P Thermocycler using SYBR Green detection of amplification in a one-step protocol. RNA (100 ng) was converted to cDNA and subsequently PCR-amplified using Brilliant III Ultra-Fast SYBR Green quantitative RT-PCR master mix (Agilent Technologies) following the manufacturer's instructions (reverse transcription, 50°C for 10 min; annealing temperature, 60°C; elongation time, 20 s) using 5 pmol each of primers specific to lacZ (CGT-GACGTCTCGTTGCTG, GTACAGCGCGGCTGAAAT) and gyrA (ACCTTGCGAGAGAAATTACACC, AATGAC-CGACATCGCATAATC). The housekeeping gene gyrA was used to control for differences in starting material, whereas a genomic DNA dilution series was used to correct for any differences in primer amplification efficiencies. All experiments were done in triplicate.
Spectroscopy-UV-visible absorbance measurements were made with a JASCO V500 spectrometer, and CD spectra were measured with a JASCO J810 spectropolarimeter. X-band EPR spectra were recorded with a Bruker EMX spectrometer equipped with an ESR-900 helium flow cryostat (Oxford Instruments). Spin intensities of paramagnetic samples were estimated by double integration of EPR spectra using 1 mM Cu(II), 10 mM EDTA as the standard. Rapid Reaction Kinetics-UV-visible stopped-flow experiments were performed with a Pro-Data upgraded Applied Photophysics Bio-Sequential DX.17 MV spectrophotometer, with a 1-cm path length cell. Absorption changes were detected at a single wavelength (360 or 420 nm), as described previously (11). All stopped-flow experiments were carried out in assay buffer using gas tight syringes (Hamilton). Prior to use, the stopped-flow system was flushed with ϳ30 ml of anaerobic assay buffer. All solutions used for stopped-flow experiments were stored and manipulated inside an anaerobic cabinet (Belle Technology). Fitting of the overall multiphase kinetic data at 360 and 420 nm (separately and together) was performed using DynaFit (BioKin) (27), which employs numerical integration of simultaneous first order differential equations, and verified by fitting individual phases to single or double exponential functions using Origin (version 8, OriginLab). Observed rate constants (k obs ) obtained from the fits were plotted against the corresponding initial concentration of nitric oxide to obtain the apparent second order rate constant.
Analytical Methods-Stock solutions of the NO donor PROLI-NONOate (t1 ⁄ 2 ϭ 1.5 s; Cayman Chemicals) were prepared in 25 mM NaOH, quantified optically (⑀ 252 nm 8400 M Ϫ1 cm Ϫ1 ), and calibrated as described previously (21). For kinetic experiments, an aliquot of PROLI-NONOate was combined with assay buffer (25 mM Hepes, 2.5 mM CaCl 2 , 100 mM NaCl, 100 mM NaNO 3 , pH 7.5) and allowed to decompose in a gastight syringe (Hamilton) to achieve the desired NO concentration before addition to FNR samples. Elemental sulfur (S 0 ) was determined by Sörbo's method, as described previously (11), in which the addition of cyanide ion is used to generate SCN Ϫ that yields an absorption band at 460 nm on the addition of Fe 3ϩ ions. For LC-MS analyses, samples were diluted into 10% acetonitrile, 0.1% formic acid at a concentration of 0.5-1 pmol l Ϫ1 , and 10 l were loaded onto a ProSwift RP-1S monolithic column (1.0 ϫ 50 mm) (Dionex, Leeds, UK) on a U3000 HPLC System (Dionex). Proteins were eluted with three gradient steps of acetonitrile (in 0.05% formic acid) from 12 to 35% in 5 min, then from 35 to 60% in 10 min, and from 60 to 70% in 2 min. The eluent was continuously infused into a SYNAPT G2 High Definition MS TM mass spectrometer (Waters, Manchester, UK) using electrospray ionization (ESI), and full MS spectra were recorded with Masslynx 4.1 (Waters) under standard conditions over the whole time of the LC run. The mass spectrometer was calibrated using sodium iodide. Spectra under the LC protein peak (at half-height) were combined, background-subtracted (5/20%), and smoothed (Savitzky-Golay, 3/2), and a mass range of approximately m/z ϭ 650 -1100 was selected for processing with MaxEnt1 in Masslynx 4.1 (Waters) until convergence. Gel filtration was carried out under anaerobic conditions using assay buffer and a Sephacryl S-100HR 16/50 column (GE Healthcare), at a flow rate of 1 ml min Ϫ1 .

Escherichia coli FNR Responds to Exogenous and Endogenous
NO in Vivo-Exposure of E. coli cultures to micromolar concentrations of exogenous NO results in changes in FNR-regu-lated gene expression and nitrosylation of the FNR iron-sulfur cluster (20,21,28). However, it was recently reported that exogenous NO, at a concentration similar to the highest measured in vivo (ϳ10 M), did not invoke an FNR-dependent transcriptional response (7). Because the dedicated NO-sensor NsrR failed to respond, it seems that in these experiments, NO did not reach the cytoplasm. Therefore, we compared in vivo transcription from a semisynthetic FNR-dependent reporter gene after exposure to exogenous and endogenous NO with the response to O 2 . FNR-dependent transcription decreased ϳ4-fold after a 15-min exposure to a controlled release of exogenous NO provided by a mixture of NOC-5 and NOC-7 (10 M each) (Fig. 1A). A more rapid FNR response (ϳ3-fold decrease after 5 min and ϳ5-fold after 15 min) was observed when the cultures were supplemented by nitrite (10 mM; Fig.  1A), a source of endogenous NO (6, 7); anaerobic suspensions of E. coli contained 28 M NO 15 min after the addition of 12.5 mM nitrite (6). The FNR response to aeration was FIGURE 1. In vivo sensitivity of FNR to NO. A, dynamics of FNR-dependent in vivo transcription when E. coli cultures were exposed to NO, nitrite, or O 2 . Anaerobic cultures of E. coli with a chromosomal FNR-dependent lacZ reporter were exposed to anaerobic water (filled diamonds); a mixture of the NO-releasing compounds (NOC-5 and NOC-7, 10 M each; release of NO is indicated by the dashed line) as a source of exogenous NO (filled squares); nitrite, 10 mM, as a source of endogenous NO (open squares); or aeration (filled triangles). The cultures were sampled at the indicated times, and the relative transcription of lacZ, which is entirely dependent on FNR in this experiment, was quantified by quantitative RT-PCR. The abundance of the lacZ mRNA at t ϭ 0 min was set at 1.0 and normalized to the housekeeping gyrA mRNA. Data from a typical experiment are shown. B, effects of increasing NO concentrations on FNR-dependent in vivo transcription. The experimental conditions were as described above except that cultures were amended with mixtures of NOC-5 and NOC-7 (0, 0.33, 0.66, 1, and 2 M of each NOC) for 15 min at which point the cultures were sampled for the lacZ quantitative RT-PCR measurements. The concentration of NO released from the NOC mixtures after 15 min was calculated from the measured half-lives of NOC-5 and NOC-7. Data are mean values (n ϭ 3) with standard deviations indicated.
initially similar to that observed upon the addition of nitrite (ϳ4-fold after 5 min) but was ultimately more pronounced (ϳ20-fold after 15 min; Fig. 1A). We conclude that in vivo, FNR responds to exogenous and endogenous NO and that the amplitude of the NO response is ϳ4-fold lower than that observed for O 2 , suggesting that FNR is partially protected by actions initiated by the dedicated NO sensors NorR and NsrR.
The dose dependence of the FNR response to NO was determined by measuring the response of the FNR-dependent lacZ reporter after anaerobic cultures were exposed for 15 min to different concentrations of NO, generated by the decay of mixtures of NOC-5 and NOC-7 (Fig. 1B). These experiments showed that FNR responded progressively to increases in NO concentration, with a maximum response (ϳ5.5-fold decrease in FNR activity) observed in the presence of 1.2 M NO.
The [4Fe-4S] Cluster of FNR Reacts with Eight NO Molecules-Having established that FNR responds to NO in vivo, the stoichiometry of the reaction of the FNR cluster with NO was investigated by measuring absorbance changes following sequential additions of NO to anaerobic FNR ( Fig. 2A). A progressive decrease in A 406 nm , an increase in A 362 nm , and a lesser increase at 500 -700 nm were observed. These changes are consistent with previous observations (21) and indicate the formation of iron-nitrosyl species (11 (Fig. 2B) was quite different, with changes complete at 4 -5 NO. The different stoichiometries and lack of isosbestic points reveal a complex reaction. The final UV-visible spectrum ( Fig.  2A), with a principal absorption band at 362 nm and a shoulder at ϳ430 nm, was consistent with that of an RRE complex. Based on a model RRE extinction coefficient (⑀ 362 nm ϭ 8530 M Ϫ1 cm Ϫ1 (29), the spectrum indicated the presence of two RREs per FNR cluster, consistent with the observed reaction stoichiometry and previous observations with Wbl proteins (11). The reaction was also studied using near UV-visible CD spectroscopy. Signals arising from the [4Fe-4S] cluster decreased almost to zero as the reaction with NO proceeded (Fig. 2C) (Fig. 2B).
An EPR titration of wild-type [4Fe-4S] FNR with NO was conducted under conditions identical to those of the UV-visible absorption titration above, with spectra recorded at 15 (not shown) and 74 K at increasing levels of NO (Fig. 3A). Complex signals centered on g(ϳ) 2.03 were observed, similar to those previously reported for nitrosylated FNR in vivo (30). Spectra could be deconvoluted into signals arising from three distinct S ϭ 1 ⁄ 2 species: two thiol-ligated DNIC species with g Ќ ϭ 2.045, g ϭ ϭ 2.023 and g Ќ ϭ 2.023, g ϭ ϭ 2.016, respectively, and a third, giving rise to the sharp feature observed at g(ϳ) 2.04 (g Ќ ϭ 2.044, g ϭ ϭ 2.032) (Fig. 3, B and C). This latter species is similar to that previously assigned as an FNR-derived persulfide ligated DNIC (31). Quantification of the final signal envelope, including all of the EPR-detectable species, obtained with excess NO gave a concentration corresponding to ϳ16% of the original cluster (or ϳ4% of the original iron). We previously reported a signal due to a small but variable amount of a [3Fe-4S] 1ϩ species, in addition to the DNIC signal, upon nitrosylation of wild-type FNR (21); this was not observed in the samples studied here.
The RRE-like species remained associated with the protein following passage down a gel filtration column (Fig. 4A). Analysis by gel filtration of molecular mass changes upon nitrosylation of [4Fe-4S] FNR revealed a decrease in mass from 54 to 34 kDa (Fig. 4B). Given that wild-type FNR has an actual mass of 28 kDa but has previously been observed by gel filtration to run at ϳ30 kDa following reaction with O 2 (32), we conclude that nitrosylation, like reaction with O 2 , results in monomerization. In combination, the spectroscopic and gel filtration data indicate that the addition of NO results in FNR monomers containing two protein-bound RREs, similar to recent observations of the reactivity of NO with Wbl proteins (11).
The [4Fe-4S] Cluster of FNR Reacts Rapidly with NO-The reaction of [4Fe-4S] FNR with excess NO was followed using stopped flow by monitoring A 360 nm and A 420 nm , the maxima of the final nitrosylated product and the initial iron-sulfur cluster, respectively (Fig. 5). Three distinct kinetic phases were observed at both wavelengths, indicating a complex multistep process, with a minimum of three steps in the overall reaction, A 3 B 3 C 3 D. The data at the two wavelengths were fitted separately and simultaneously to exponential functions (Fig.  5D), as described previously (11). In both cases, the intermediate phase had quite different kinetic characteristics at the two wavelengths, indicating that they report on different processes in the intermediate part of the reaction. Therefore, the overall reaction was remodeled as a four step reaction, i.e. A 3 B 3 C 3 D 3 E, where the initial and final steps A 3 B and D 3 E, respectively, are detected at both wavelengths, whereas step B 3 C is detected at 360 nm and step C 3 D is detected at 420 nm. Data corresponding to step D 3 E at 420 nm were of very low amplitude as compared with 360 nm, consistent with the thermodynamic titrations that showed relatively little change at 420 nm during the second half of the titration. Overall, these observations are remarkably similar to those made previously with the Wbl proteins Streptomyces coelicolor WhiD and Mycobacterium tuberculosis WhiB1 (11).
Plots of the observed pseudo-first order rate constants (k obs ) against NO concentration for FNR were linear for each step, indicating a first order dependence on NO for each step (Fig. 6, A-D, and Table 2), consistent with the stepwise addition of NO. The rate constant for the slowest step of the NO reaction is at least ϳ3 orders of magnitude greater than that for the slowest step with O 2 , indicating that FNR is actually much more sensitive to NO than it is to O 2 . The rate constants for each step are of the same order of magnitude as those previously observed for Wbl proteins (11) ( Table 2).
FNR Cluster Sulfide Is Oxidized to S 0 and Retained as Persulfide during Nitrosylation-The addition of NO to a [4Fe-4S] 2ϩ cluster to generate DNIC/RRE species entails reduction of the mixed valence, iron core cluster [2Fe(III)-2Fe(II)] 10ϩ to the [4Fe(I)] 4ϩ state. The six electrons required could be derived from oxidation of cluster sulfide ion, S 2Ϫ , to sulfane, S 0 , as shown previously in the case of WhiD (11). After reaction with excess NO, FNR was indeed found to contain ϳ0.5 S 2Ϫ and ϳ1.9 S 0 per [4Fe-4S] ( Table 1). Although only ϳ2-3 of the original sulfides could be detected following nitrosylation, the results demonstrate that S 0 is generated through oxidation of S 2Ϫ (11). Importantly, ϳ2 S 0 per cluster remained associated with the protein following gel filtration, whereas the majority of S 2Ϫ was lost (Fig. 4A and Table 1). The signal shape is indicative of the presence of more than one S ϭ 1 ⁄2 species, as observed previously (11). Spectra were recorded at 74 K. Microwave power and frequency were 2 milliwatts and 9.68 GHz, respectively, and field modulation amplitude was 1 millitesla. The sample buffer was 25 mM HEPES, 2.5 mM CaCl 2 , 100 mM NaCl, 100 mM NaNO 3 , pH 7.5. B, simulated EPR spectra of persulfide-ligated DNIC (g Ќ ϭ 2.044, g ϭ ϭ 2.032) (31,46) and two thiol-coordinated DNICs at g(ϳ) 2.04 (g Ќ ϭ 2.045, g ϭ ϭ 2.023) (29,44) and g(ϳ) 2.02 (g Ќ ϭ 2.023, g ϭ ϭ 2.016) (11,47) that were used to deconvolute the recorded (74 K) EPR spectra in A. C, EPR spectra for nitrosylated FNR (black lines), together with simulations of the experimental data (R 2 (ϳ) 0.90) (gray lines) obtained by the combination of varying amounts of the spectra shown in B.
Although ESI-MS of acidified iron-sulfur proteins leads to loss of the cluster, cysteine persulfides/polysulfides remain attached to the protein (33,34), thus providing a means to detect S 0 associated with FNR following nitrosylation. The mass spectrum of [4Fe-4S] FNR (Fig. 7) was dominated by the FNR monomer peak at 29,163 Da (theoretical mass of 29,165 Da). Exposure of [4Fe-4S] FNR to excess NO prior to analysis by MS resulted in significant additional peaks at ϩ32, ϩ64, ϩ96, and ϩ128 Da (Fig. 7). These peaks correspond to apo-FNR containing between 1 and 4 S 0 adducts per protein.
Recently, it was shown that the reaction of O 2 with [4Fe-4S] FNR results in oxidation of cluster sulfide to S 0 , with FNR containing one or two S 0 species per protein as the major species (34). The ESI-TOF MS spectrum of [4Fe-4S] FNR following the addition of oxygen (Fig. 7, inset) shows the same persulfide adducts, although more abundant following reaction with O 2 as compared with NO. Although this suggests that more persulfide species are generated during reaction with O 2 than with NO, our S 0 analytical data are not consistent with this conclusion. This may arise from instability of the persulfide species under the conditions of the MS experiment following nitrosylation (35). In this case, one possibility is that persulfide species may become re-reduced to cysteine plus S 2Ϫ and iron reoxidized from Fe(I) to Fe(II)/(III) upon disruption of iron-nitrosyl-FNR complex(es) during acidification in preparation for MS. Resonance Raman studies of O 2 -treated [4Fe-4S] FNR showed that the S 0 is carried as persulfide that coordinates the [2Fe-2S] cluster (34). Therefore we propose that the S 0 adduct species formed in FNR by the action of NO are also persulfides that ligate the Fe(I) cluster core.

DISCUSSION
The reactivity of O 2 /reactive oxygen species and NO toward protein-bound iron-sulfur clusters is a major cause of their toxicity, but also underlies the biological mechanisms through which such species are sensed, often as a first step of stress response (10,13). Understanding the chemistry of the reaction between NO and iron-sulfur clusters is, therefore, an area of growing importance. A recent mechanistic study of the nitrosylation of the [4Fe-4S] clusters of Wbl proteins, some of which have been proposed to function as NO sensors (36,37), revealed a rapid and complex multistep reaction involving a total of eight NO molecules per cluster (11).
E. coli FNR, which functions as a global O 2 sensor that controls the switch between aerobic and anaerobic metabolism, is mechanistically the best characterized of the iron-sulfur cluster-containing regulators (13). Here and elsewhere it has also been shown to regulate gene expression in response to physiological concentrations of NO (20,21). Previous in vitro studies of this reaction led to the conclusion that ϳ3 NO molecules per cluster are involved in the reaction, resulting in the formation of a mixture of DNIC and RRE-like species (21). The data presented here demonstrate that the addition of ϳ4 -5 NO molecules results in loss of the [4Fe-4S] cluster, but 8 NO molecules are required for complete reaction.
Stopped-flow kinetics showed that nitrosylation of the cluster proceeds via a multiphasic reaction. The first step of the reaction very probably corresponds to the binding of one NO molecule to the FNR [4Fe-4S] cluster (A 3 B; first order with respect to NO) (Fig. 8A). This most likely enhances accessibility to further NO binding in steps B 3 C (detected at 360 nm) and C 3 D (detected at 420 nm), which are both first order in NO, indicating either that a single NO is involved in each or that NO binding occurs independently to different irons of the cluster, giving an overall first order dependence. During the intermediate step C 3 D, a major decrease in absorbance occurs at 420 nm, suggesting that iron-sulfide and/or cysteine interactions (which give rise to the cluster absorbance) are significantly disrupted and may also be simultaneous with reduction of iron (to the ϩ1 state) and oxidation of S 2Ϫ (to S 0 ). The optical titration data (absorbance and CD) revealed a stable intermediate at a stoichiometry of approximately four NO molecules per [4Fe-4S] cluster, and EPR showed that this is diamagnetic. Thus, intermediate D may be an EPR-silent tetra-nitrosylated multiiron species. The last step of the reaction (D 3 E) resulted in a large increase in absorbance at 360 nm, consistent with significant further nitrosylation, and was again first order in NO, resulting in an EPR-silent protein-bound product composed of four irons and eight NO molecules. This could be a pair of EPR-silent RRE-like species, [Fe I 2 (NO) 4 (Cys) 2 ], with only minor amounts of EPR-active DNIC, as recently proposed for Wbl proteins (11).
Nitrosylation of the [4Fe-4S] cluster results in the oxidation of bridging sulfide ions to S 0 and most likely addition to cysteine forming a persulfide. The O 2 reaction of [4Fe-4S] FNR also leads to sulfide oxidation, resulting in a [2Fe-2S] cluster coordinated by one or two cysteine persulfides (17,34). It is likely therefore that the iron-nitrosyl species formed upon cluster nitrosylation are coordinated by one or more cysteine persulfides (Fig. 8A). A possible overall reaction is given in Reaction 1. We note that, as described in the case of Wbl proteins (11), the two RRE-like species may dimerize to form a novel cluster, e.g.
[Fe I 4 (NO) 8   Nitrosylation of [4Fe-4S] FNR led to monomerization of the protein, consistent with previous studies that showed a significantly reduced DNA binding affinity following NO reaction (21) and consistent with the in vivo data reported here indicating that NO alleviates FNR-mediated transcriptional repression.
The kinetic and thermodynamic characteristics of FNR nitrosylation are similar to those recently reported by us for the Wbl proteins S. coelicolor WhiD and M. tuberculosis WhiB1 (11). Although both FNR and WhiD/WhiB1 contain [4Fe-4S] 2ϩ clusters ligated by four cysteines (37,38), they represent phylogenetically unrelated protein families that have different residues surrounding the cluster (Fig. 8B). Clearly, nitrosylation of these Cys-coordinated clusters is not strongly influenced by the protein environment, and we suggest that this pathway is likely to occur widely in proteins during the reaction of [4Fe-4S] clusters with NO.
FNR regulates the aerobic-anaerobic switch via its sensitivity toward O 2 ; what, then, is the biological significance of its reac-tivity toward NO? Many transcriptional regulators respond to NO in E. coli. Principal among these are NorR, NsrR, and FNR (9, 21, 39). NorR is a non-heme iron-containing regulator that, when bound to NO, activates the transcription of norVW, FIGURE 6. Dependence of the rate of each step of nitrosylation on NO. A-D, plots of the observed (pseudo-first order) rate constants (k obs ) from fits of the kinetic data at 360 (white circles) and 420 nm (black circles), over a range of NO concentrations. Note that panels A-D correspond to steps 1-4, respectively (see under "Results"). Least squares linear fits are shown giving apparent second order rate constants ( Table 2). The buffer was 25 mM HEPES, 2.5 mM CaCl 2 , 100 mM NaCl, 100 mM NaNO 3 , pH 7.5.   The peak at 29,163 Da corresponds to the monomer molecular ion peak of FNR, and the peaks at ϩ32, ϩ64, ϩ96, and ϩ128 Da correspond to the addition of one, two, three, and four covalently bound sulfur atoms, respectively, as indicated. The inset is the ESI-TOF mass spectrum of wildtype [4Fe-4S] (34 M in 25 mM HEPES, 2.5 mM CaCl 2 , 100 mM NaCl, 100 mM NaNO 3 , pH 7.5) after exposure to O 2 (102 M) for 60 s at 20°C prior to dilution in the MS solvent. Peaks at ϩ32, ϩ64, ϩ96, and ϩ128 Da again correspond to the addition of one, two, three, and four covalently bound sulfur atoms.
which encodes the flavorubredoxin NorV and its associated oxidoreductase NorW that, together, detoxify NO to form nitrous oxide (N 2 O) under anaerobic conditions (40). NsrR regulates at least 60 genes involved in detoxification and/or repair, including nrfA and hmp, for which NO is a substrate. It is suggested that if these dedicated NO detoxification systems are unable to lower the NO concentration sufficiently to counteract the ensuing nitrosative stress, FNR will become nitrosylated (30), leading to lowered expression of the nar, nir, nrf, and nap operons (that require [4Fe-4S] FNR for activation) and derepression of hmp expression. We note that NarG is the most important source of endogenously derived NO during nitrate/ nitrite respiration (7,(41)(42)(43) (20). Thus, the nitrosylation of FNR should only occur when the cellular NO detoxification systems are overwhelmed. This implies that FNR functions as a final safeguard against NO damage in a hierarchal network that has evolved to monitor, and respond to, NO both endogenously generated during anaerobic respiration with nitrate or nitrite and exogenously generated by the host immune system.