EPR and Mössbauer spectroscopic studies on enoate reductase.

Enoate reductase (EC) is a protein isolated from Clostridium tyrobutyricum that contains iron, labile sulfide, FAD, and FMN. The enzyme reduces the α,β carbon-carbon double bond of nonactivated 2-enoates and in a reversible way that of 2-enals at the expense of NADH or reduced methyl viologen. UV-visible and EPR potentiometric titrations detect a semiquinone species in redox intermediate states characterized by an isotropic EPR signal at g = 2.0 without contribution at 580 nm. EPR redox titration shows two widely spread mid-point redox potentials (−190 and −350 mV at pH 7.0), and a nearly stoichiometric amount of this species is detected. The data suggest the semiquinone radical has an anionic nature. In the reduced form, the [Fe-S] moiety is characterized by a single rhombic EPR spectrum, observed in a wide range of temperatures (4.2-60 K) with g values at 2.013, 1.943, and 1.860 (−180 mV at pH 7.0). The gmax value is low when compared with what has been reported for other iron-sulfur clusters. Mössbauer studies reveal the presence of a [4Fe-4S]+2/+1 center. One of the subcomponents of the spectrum shows an unusually large value of quadrupole splitting (ferrous character) in both the oxidized and reduced states. Substrate binding to the reduced enzyme induces subtle changes in the spectroscopic Mössbauer parameters. The Mössbauer data together with known kinetic information suggest the involvement of this iron-sulfur center in the enzyme mechanism.

Enoate reductase (EC 1.3.1.31) (1-6) isolated from Clostridium tyrobutyricum catalyzes the NADH or methyl-viologen-dependent reduction of the ␣,␤ carbon-carbon double bond of nonactivated 2-enoates (7) and in a reversible way that of 2-enals (8). The enzyme appears to be a multimer of identical subunits. The total molecular mass is 940 kDa (subunit circa 73 kDa). Sedimentation equilibrium experiments, molecular mass data, and electron microscopy indicate that the native enzyme is composed of a tetramer of trimers. Per enzyme subunit 1 FAD, 0.6 FMN, 4 iron, and 4 labile sulfur atoms were found (9), and thus enoate reductase belongs to the rare class of flavoenzymes containing both FAD and FMN. The involvement of iron-sulfur centers in electron transfer is well established (10), and they have also been shown to be associated with other important physiological nonredox functions, such as those of aconitase and other dehydratases (11,12), glutamine 5-phosphoribosyl-1-pyrophosphate aminotransferase (13), endonuclease III (14), and iron-responsive element-binding protein (15)(16)(17).
In this work, we report the involvement of an iron-sulfur center in a new type of biochemical reaction. EPR and Mössbauer data are analyzed in the oxidized and NADH-and dithionitereduced states, as well as in substrate bound forms in order to identify the type of iron-sulfur core involved. A tentative mechanism is presented that involves a hydride transfer from a flavin group and a carbon-carbon double bond polarized by the presence of the iron-sulfur cluster, thus including enoate reductase in the group of iron-sulfur enzymes in which the metal center interacts with substrate molecules during catalysis.

Growth of C. tyrobutyricum (DSM 1460)-Cells of C. tyrobutyricum
DSM1460 were grown on 57 Fe-or 56 Fe-containing medium on a 50-liter scale according to Bader and co-workers (1) with slight modifications (1)(2)(3). The medium contained 150 mg/liter (NH 4 ) 3 SO 4 instead of (NH 4 ) 2 PO 4 and additionally 32 mg Na 2 SO 4 ⅐10 H 2 O. The source of iron was about 65 mg of 57 Fe in the form of an 57 FeSO 4 solution, i.e., the concentration amounted to about 2 ϫ 10 Ϫ5 M iron. The cells were harvested by centrifugation after reaching the stationary phase 20 h after inoculation.
Purification and Sample Preparation-Enoate reductase was purified under strict anaerobic conditions essentially according to Kuno et al. (9). Wet packed cells (78 g) were suspended in 234 ml of 50 mM potassium phosphate buffer, pH 7.0, containing 10 mM sodium tiglate, 1 mM EDTA, 320 mg of lysozyme, and 32 mg of DNase. The pH of the suspension was adjusted to 7.2 by the addition of 2.5 M Na 2 CO 3 . After incubating the suspension for 40 min at 35°C, it was centrifuged for 20 min at 27000 ϫ g at 0°C. The careful control of the pH is crucial for obtaining the enoate reductase in the supernatant.
All chromatographic steps were performed with oxygen-free buffers containing 0.1 mM EDTA, 250 mM saccharose, 0.02% sodium azide, and 0.02% mercaptoethanol. The supernatant was applied to the following columns in two portions.
It was first applied to a DEAE Sepharose CL-6B column (18.5 ϫ 4.5 cm) equilibrated with 20 mM potassium phosphate buffer (pH 7.0) and then applied directly to a hydroxyapatite column (18 ϫ 2.4 cm) as described previously (3).
Pure enoate reductase both in 57 Fe and in 56 Fe were concentrated in phosphate buffer, pH 7.0, 0.25 M saccharose, and 1 mM crotonate (needed to maintain the enzyme in the oxidized state). In the final steps the enzyme buffer was exchanged with a buffer containing only phosphate and saccharose in the concentrations mentioned above. Enoate reductase in 57 Fe at the final concentration of about 1 mM was used to prepare four samples: native, dithionite-reduced, NADH-reduced, and dithionite-reduced in the presence of 50 mM of cinnamate. The enzyme reductions were performed under strict anaerobic conditions. Spectroscopic/Potentiometric Measurements-Redox potentials of enoate reductase were determined by UV-visible absorbance and EPR potentiometric titrations. The protein solution in phosphate buffer, pH 7.0, was kept under anaerobic conditions by flushing the solution during the titration with purified Argon (with an Indicating Oxygen Trap from Chemical Research Supplies). An optical redox cell and an EPR * 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. potentiometric cell, slightly modified from the design of Dutton (18) was used for the UV-visible and EPR titrations. The potentials were measured with a Crison 2002 potentiometer with a platinum and a Ag/AgCl electrode and are quoted relative to the normal standard hydrogen electrode and calibrated with quinhydrone at pH 7.0. The following redox mediators were present at a final concentration of 3.5 M: 1,4naphtoquinone, methylene blue, triquat, phenosaphranine, benzylviologen, methylviologen, dichlorophenol-indophenol, benzoquinone, anthraquinone-2-sulfonic acid, phenazinamethosulfate, dimethyltriquat, indigo tetrasulfonate, 2-hydroxy-1,4-naphtoquinone, 5-hydroxy-1,4naphtoquinone, duroquinone, phenazil, and safranine. Solution potentials were varied by adding appropriate volumes of deareated 30 mM sodium dithionite or 30 mM NADH as reductant. UV-visible spectra were recorded during titration on a UV-265FS Shimadzu spectrometer. EPR redox titrations were performed in the same conditions as described for the UV-visible titrations, and the potentials were varied as described in Ref. 9. The protein solution was poised at different redox potentials, and 180-l aliquots were transferred under argon to EPR tubes and frozen in liquid nitrogen for later measurement.
EPR spectra were made on a Bruker ER 200 spectrometer equipped with an Oxford continuous flow cryostat. The Mössbauer spectrometer was similar to the one described in Ref. 19. The zero velocity of the Mössbauer spectra is referred to the centroid of metallic iron spectra at room temperature.

UV-visible and EPR Redox Titrations
The UV-visible spectra of enoate reductase are largely dominated by the flavin cofactor in the 300 -550 nm region. UVvisible measurements coupled with potentiometric titrations as shown in Fig. 1 do not distinguish between the flavin and [Fe-S] absorbance contribution due to the similarity of the redox potentials of these prosthetic groups. Also no 580 -600 nm spectral contribution was detected in the UV-visible spectrum during redox titrations. The native form of enoate reductase is EPR silent, and the iron-sulfur and the flavin moieties are both diamagnetic (not shown).
At 10 K, the EPR spectrum of the NADH-reduced state shows a rhombic signal with g values at 2.013, 1.943, and 1.860, which are assigned to an iron-sulfur center with a superimposed isotropic signal centered at g ϭ 2.00, attributable to a flavin semiquinone (see Fig. 2C). At high temperature (120 K), the iron-sulfur center is too broad to be detected, and the EPR features are dominated by the flavin radical signal (Fig.  2B). The dithionite-reduced enzyme shows the same rhombic EPR signal as the NADH-reduced sample, but the isotropic signal is absent ( Fig. 2A). The intensity of the rhombic EPR signals does not increase upon further reduction with dithionite: the spin quantification (relative to a CuEDTA standard) gives 0.94 spin per monomer. The quantitation of the flavin signal is 0.9 spin per monomer, using as a standard Desulfovibrio desulfuricans ATCC 27774 flavodoxin in the semiquinone state at 120 K (20). The temperature dependence of the rhombic signal shows that it can be detected up to 60 K and is better observed at 10 K, using a microwave power of 23.7 W. The line width of this rhombic species is increased in the 57 Fe-enriched sample due to unresolved hyperfine interactions. The optimal temperature to observe the isotropic signal attributable to the flavin is 120 K. A comparison of the EPR signals of the flavin radical in H 2 O and D 2 O shows a line width reduction from 17 to 13 G (21). This observation is not in agreement with what was previously reported for anionic semiquinones where a decrease in the line width from 15 to 14 G was found, whereas in the neutral semiquinones the line width changes from 19 to 15 G due to the permutation of the flavin N-5 exchangeable proton. The span in redox potentials for this redox equilibrium (see below) is large enough (⌬ ϭ 170 mV) to generate a nearly stoichiometric amount of the intermediate redox form. Then, in conclusion, despite the anomalous line width behavior, an anionic semiquinone form is present in this enzyme. Fig. 3 shows the relative intensity of the EPR spectrum of the [FeS] center recorded at 10 K versus the poised redox potential. The experimental data were fitted with a monoelectronic Nernst curve with a midpoint redox potential of Ϫ180 mV when dithionite is used as the reducing agent. When the protein is titrated with NADH, the midpoint potential was found to be Ϫ167 mV (error bar Ϯ 15 mV). The relative intensity of the semiquinone EPR spectrum recorded at 120 K follows a bell shaped curve, which was fitted using a simple sequential redox model (see Fig. 3) and assuming a maximum intensity of 0.9 spin, and the mid-point redox potentials determined were Ϫ190/Ϫ177 mV for the quinone/semiquinone couple (using either dithionite or NADH as reducing agent) and Ϫ350 mV for the semiquinone/hydroquinone couple (using dithionite). The redox titration performed in the presence of NADH, does not reach a redox potential value below Ϫ300 mV.

Mössbauer Spectroscopy Oxidized State
High Temperature Spectra-The Mössbauer spectra of the native enzyme enriched in 57 Fe were recorded at 80, 100, and 130 K.  The high temperature spectrum was fitted with subcomponents that have an average isomer shift (␦) of 0.43 mm/s and an average quadrupole splitting (⌬E Q ) varying from 1.30 to 1.33 mm/s depending on the recording temperature. These Mössbauer parameters are characteristic of high spin iron in an oxidation state intermediate between the ferric and ferrous states, suggesting electron delocalization and tetrahedral coordinated by weak field ligands, such as sulfur atoms.
The analysis of these parameters and the temperature dependence of the quadrupole splitting (not shown) indicate that this cluster has some ferrous character in the native state, excluding the possibility of the presence of an oxidized [2Fe-2S] cluster, which contains only ferric atoms.
Low Temperature Spectra-The Mössbauer spectra of the protein as isolated were recorded at 4.2 K in the presence of an applied magnetic field of 0.0, 4.0 or 8.0 tesla (Fig. 5). The spectra were simulated with four subcomponents, using an S ϭ 0 spin Hamiltonian. The simulated spectra are shown as the solid lines in Fig. 5, with the parameters summarized in Table  I. The good agreement between the theoretical and experimental data indicates that all the iron atoms in the sample are in a diamagnetic environment in agreement with the EPR results. The diamagnetism of the center results from the antiferromagnetic coupling of the d electrons of the two ferric and two ferrous high spin iron atoms. The parameters shown in Table I are characteristic of a [4Fe-4S] ϩ2 cluster, but the mean value of the quadrupole splitting, and in particular the mean value of the isomer shift, is high when compared with reported values for this class of cluster.

Reduced State
High Temperature Spectra-The Mössbauer spectrum of the dithionite-reduced sample observed at high temperature (100 K) shows two resolved quadrupole doublets corresponding to the mixed valence and ferrous pairs (Fig. 6). In the ferrous pair, a small nonequivalence (broader line) is noticed in the peak detected at positive velocity. However, a deconvolution of these two subsites was not attempted due to the poor spectral resolution. For this reason, the data were fitted with two quadrupole doublets in a 1:1 ratio. The mixed valence pair has ⌬E Q ϭ 1.22 mm/s and ␦ ϭ 0.50 mm/s. The ferrous pair in the dithionite-reduced sample has ⌬E Q ϭ 2.32 mm/s and ␦ ϭ 0.61(4) mm/s. This value of quadrupole splitting is the highest reported for a [4Fe-4S] ϩ1 center in biological systems (see Table  II).
Low Temperature Spectra-The Mössbauer spectra recorded at 4.2 K with applied magnetic fields of 0.095, 4.0, and 8.0 tesla are characteristic of a paramagnetic species with an S ϭ 1/2 spin system (Fig. 7). The experimental data were fitted with an S ϭ 1/2 spin Hamiltonian: using the WMOSS analysis program from WEB Research Co. (Edina, MN). Two major sites were identified by the inner and outer movement of the spectral lines when the externally applied magnetic field is increased, corresponding to negative or positive hyperfine coupling constants (22). The site with positive coupling constants (ferrous component) was subdivided for analysis proposes into two subsites (a and b). The sites with negative coupling constants were labeled 2c. This model was adopted to improve the simulation, taking into account small nonequivalence among sites (already detected in the high temperature spectra). The subsite differentiation is more evident when small spectral changes are analyzed on cluster data obtained in the presence or the absence of substrate (see below). The parameters used in the data analysis of the reduced samples are presented in Table III. The ⌬E Q value for the ferrous pair in this cluster is higher than commonly found in ferredoxin type clusters. Site 2c has ␦ ϭ 0.53 mm/s and ⌬E Q ϭ 1.34 mm/s and negative hyperfine coupling constants (Ϫ23.2, Ϫ28.4, and Ϫ24.0 tesla) and was therefore identified as the mixed valence (ferric/ferrous) pair. The ferrous site a has ⌬E Q ϭ ϩ2.4 mm/s and ␦ ϭ 0.65 mm/s, whereas ferrous site b has ⌬E Q ϭ -2.39 mm/s and ␦ ϭ 0.61 mm/s. Analysis of the Hyperfine Parameters-The simulation of the low temperature spectra recorded in the presence of strong externally applied magnetic fields allow determination of the values of the 57 Fe hyperfine coupling constants (Fig. 7). The average value of the splitting constants of the mixed valence pair is found to be similar in most of the analyzed [4Fe-4S] clusters and ranges from Ϫ17 to Ϫ26 tesla. Somewhat higher variability is found in the average splittings of the ferrous pair, which range from ϩ3 to ϩ15 tesla. The intrinsic value of the coupling constant in a iron ion coordinated by an oxygen atom is higher than that with a sulfur ligand. However, as it has been shown (23)(24)(25), the magnitude of A is also dependent on the spin projection, and this parameter cannot be taken as absolutely conclusive of oxygen coordination. Fig. 8 shows the average values of the hyperfine coupling constant of the ferrous and ferric components in iron-sulfur clusters. Enoate reductase shows the second highest hyperfine coupling constants after aconitase.
Interaction of Reduced Enoate Reductase with Substrate-Modifications were observed in the Mössbauer spectra of the reduced protein in the presence of substrate (Fig. 9). The peak at Ϫ3.8 mm/s in the 8 tesla spectrum is sharper in the presence of substrates while some other minor modifications occur in other parts of the spectra (compare Figs. 7 and 9). A simulation of this spectrum was done and required some adjustments in  6. Mö ssbauer spectrum of dithionite-reduced enoate reductase recorded at 100 K. The solid line plotted over the experimental spectrum is a least square fit of two quadrupole doublets to the spectrum (solid lines plotted above).  the overall set of parameters as indicated in Table III. This  table also presents the parameters used for the fitting of the Mössbauer spectra of the NADH-reduced sample. The changes are small and reflect subtle differences in the chemical environment of the cluster.

Comparison of 4.2 K Mössbauer Parameters of the ϩ1 and ϩ2
Oxidation States-Iron-sulfur centers show a wide structural variability in terms of metal stoichiometry and coordination. They may contain from one to six iron atoms, and the metal core can have only sulfur coordination, but other ligands can also be present substituting for cysteinyl residues. [2Fe-2S] clusters can have nitrogen containing ligands (histidine), and [4Fe-4S] clusters can have ligands containing oxygen atoms (aspartic acid) or even hydroxyl or water molecules (26). Recently, the x-ray structure of Desulfovibrio gigas hydrogenase showed a new type of tetranuclear cluster binding with one histidine and three cysteines in the coordination sphere (27).
The presence of one oxygen atom in a biological [4Fe-4S] ϩ2 cluster has been studied by Mössbauer spectroscopy in substrate-free form of aconitase: the site with no sulfur coordination has a ␦ av ϭ 0.46 mm/s and ⌬E Qav ϭ 1.20 mm/s (25).
The amino acid sequence of enoate reductase was recently determined 1 and compared with other related enzymes whose primary structures where available (29). A conserved pattern CX 2 CX 2-3 CX 11-13 C was detected. For enoate reductase, the cysteine anticipated to bind the cluster (i.e., C364, C367, C371, and C384) are supported by x-ray structural data obtained on the related trimethylamine dehydrogenase (30). On the basis of this observation, the enoate reductase center should only involve cysteinyl coordination.
The work performed in reduced aconitase bound to citrate and other substrates has shown a specific iron (Fe a ), which has penta/octahedral coordination, which results in a value of ϩ2.5 mm/s for ⌬E Q and 0.99 mm/s of isomer shift, whereas its tetrahedral counterpart iron (Fe b1 ) has ⌬E Q ϭ -2.5 mm/s and ␦ ϭ 0.64 mm/s (25).  9. Mö ssbauer spectra of dithionite-reduced enoate reductase in the presence of cinnamate recorded at 4.2 K. The spectra were recorded with a magnetic field of 0.095 (A) and 8.0 tesla (B) applied parallel to the ␥ radiation. The solid lines above the spectra represent the components of the theoretical simulation according to the text, and the solid line over the spectra represents the overall simulation.
Enoate reductase exhibits the highest reported average ⌬E Q (1.91 mm/s) and a ␦ average value (0.58 mm/s) that is much more similar to the one found in APS center I (37). These results suggest also that enoate reductase should have a different chemical environment, giving a stronger ferrous character to this cluster component. The Mössbauer studies on aconitase enable us to discard the possibility of having ferrous ions coordinated by five or six ligands; however, some distinction can be made relative to the ferredoxin type of clusters. The strong ferrous character (compared with ferredoxin [4Fe-4S] 1ϩ clusters) in the dithionite-reduced state corroborates the above mentioned character also found in the protein as isolated.
Mechanistic clusters. This could be an indication of non-cysteinyl coordination at one iron site. Other plausible biological ligands can only be oxygen or nitrogen. The possibility of the direct coordination of nitrogen to the cluster was discarded by ESEEM and EN-DOR studies, 2 leaving the possibility of oxygen coordination. These techniques also exclude the possibility of water or hydroxyl coordination, because the magnitudes of the observed 1 H hyperfine coupling constants are much smaller than those observed in P. furiosus ferredoxin and do not show any significant isotopic (H 2 O/D 2 O) effect.
The sequence data together with the x-ray analysis of a related protein strongly suggest that a sufficient number of cysteines are available to coordinate the metal center. We suggest that the unusual properties of the core are due to structural constrains that impose a particular electronic delocalization (ferrous character of subsites a and b) rather than non-sulfur coordination at these subsites.
Kinetic studies proposed that enoate reductase has Bi Bi Ping Pong type mechanism (7). In the first cycle the reducing agent (either NADH or reduced methyl viologen) will reduce the enzyme, making it competent to reduce the substrate (enoate or enal) double bond. Hydride transfer from NADH was proven by isotopic labeling to be stereospecific relative to the R hydrogen atom (7). The hydride transfer from NADH presumably involves the flavin cofactor.
The reduction of enoates with a halogen substituent at the ␤ position lead to a interesting result, that the first reducing equivalents were used not to reduce the double bond but to eliminate HX from the enoate. Sedlmaier et al. (41) interpreted these results by proposing that a positive charge close to the carboxylate group will polarize the double bond in a way that favors the elimination reaction.
Based on the information from the Mössbauer data and the previous kinetic studies, we propose that the iron-sulfur center plays a role in the polarization of the substrate double bond, probably through second sphere coordination of the substrate carboxylate group to the cluster. This idea is supported by theoretical studies of the enzymatic reactions, where the need for a strong polarizing agent is indicated (28,40,42). The strongly polarized double bond is then in a favorable condition for the hydride nucleophilic attack, completing the double bond reduction.