Discovery of a Novel Ferredoxin from Azotobacter vinelandii Containing Two [4Fe-4S] Clusters with Widely Differing and Very Negative Reduction Potentials*

Ferredoxins that contain 2[4Fe-4S]2+/+ clusters can be divided into two classes. The “clostridial-type” ferredoxins have two Cys-Xaa-Xaa-Cys-Xaa-Xaa-Cys-Xaa-Xaa-Xaa-Cys-Pro motifs. The “chromatium-type” ferredoxins have one motif of that type and one more unusual Cys-Xaa-Xaa-Cys-Xaa7–9-Cys-Xaa-Xaa-Xaa-Cys-Pro motif.Here we report the purification of a novel ferredoxin (FdIII) from Azotobacter vinelandii which brings to 12 the number of small [Fe-S] proteins that have now been reported from this organism. NH2-terminal sequencing of the first 56 amino acid residues shows that FdIII is a chromatium-type ferredoxin with 77% identity and 88% similarity to Chromatium vinosumferredoxin. Studies of the purified protein by matrix-assisted laser desorption ionization–time of flight mass spectroscopy, iron analysis, absorption, circular dichroism, and electron paramagnetic resonance spectroscopies show that FdIII contains 2[4Fe-4S]2+/+clusters in a 9,220-Da polypeptide. All 2[4Fe-4S]2+/+ferredoxins that have been studied to date, including C. vinosum ferredoxin, are reported to have extremely similar or identical reduction potentials for the two clusters. In contrast, electrochemical characterization of FdIII clearly establishes that the two [4Fe-4S]2+/+ clusters have very different and highly negative reduction potentials of −486 mV and −644 mVversus the standard hydrogen electrode.

Ferredoxins are small, generally acidic, electron transfer proteins that are found in all types of living systems (1). They contain iron-sulfur ([Fe-S]) clusters ligated to the polypeptide backbone primarily via cysteine residues. When the first ferredoxin was discovered in Clostridium pasteurianum, it was thought to be a general cellular electron carrier that could participate in several different metabolic processes (2). Today it is clear that a single organism may have numerous ferredoxins, each distinguished by sequence, [Fe-S] cluster type, and [Fe-S] cluster reduction potential (1,3,4).
The most common cluster type is the [4Fe-4S] 2ϩ/ϩ cluster, and the first ferredoxins isolated were found to contain two such clusters in an ϳ6,000 molecular weight polypeptide (1,5). These "clostridial-type" ferredoxins, represented by the structurally characterized Peptococcus aerogenes ferredoxin, contain two Cys-Xaa-Xaa-Cys-Xaa-Xaa-Cys-Xaa-Xaa-Xaa-Cys-Pro motifs for ligation of the two [4Fe-4S] 2ϩ/ϩ clusters (6,7) which are believed to have evolved by gene duplication (8). The two clusters have reduction potentials that are effectively identical and in the region of Ϫ0.4 V versus SHE. 1 These proteins evolved further to produce other types of ferredoxins. Replacement of the central Cys in one of the two Cys-Xaa-Xaa-Cys-Xaa-Xaa-Cys motifs by another residue or insertion of two residues between the second and third Cys resulted in formation of a [3Fe-4S] ϩ/0 cluster and a new class of ferredoxins, the so-called 7Fe ferredoxins, as represented by the structurally characterized FdI from Azotobacter vinelandii (9). The two different cluster types in these proteins have two very different reduction potentials (10). Another evolutionary modification of clostridial-type ferredoxins involved retention of the two [4Fe-4S] 2ϩ/ϩ clusters but insertion of additional residues into one of the two Cys motifs to give a Cys-Xaa-Xaa-Cys-Xa 7-9 -Cys-Xaa-Xaa-Xaa-Cys-Pro motif (1). This class is represented by the structurally characterized 8Fe ferredoxin from Chromatium vinosum (11). Surprisingly, despite the considerable difference in binding motifs for the two clusters, it has been reported that they have the same reduction potentials (12)(13)(14). We have now discovered an 8Fe ferredoxin (FdIII) from A. vinelandii which contains two different binding motifs analogous to the Chromatium protein. As described in this paper, FdIII is the first example of any 8Fe ferredoxin having two [4Fe-4S] 2ϩ/ϩ clusters with very different reduction potentials.

EXPERIMENTAL PROCEDURES
Cell Growth and Protein Purification-The A. vinelandii strain used in this study is designated DJ138/pBS122. The parent strain DJ138 was described previously (15). The plasmid pBS122, which was constructed as described elsewhere (16,17), is a derivative of pKT230 with insertion of a site-directed mutant fdxA gene encoding a C16S variant of FdI. For protein purification, cells were grown under N 2 -fixing conditions in a 200-liter New Brunswick fermentor (17). For the experiment to determine the effect of ammonia levels on the intracellular FdIII levels, a separate batch of cells was grown in the presence of excess ammonium acetate.
Cells were harvested, cell-free extracts were prepared, and the first DEAE-cellulose column was run as for the purification of nitrogenase (18) and FdI (16,19) except that the heat step was omitted. FdIII eluted at 70% of the linear 0.1-0.5 M NaCl gradient as a well resolved brown peak exactly where the FdI peak was expected (16). This fraction was then diluted with 2 volumes of 0.1 M potassium phosphate buffer (pH 7.4) and loaded onto a 2.5 ϫ 12-cm DEAE-cellulose column that was then washed slowly with 4 liters of 0.12 M KCl in the same buffer. The greenish brown fraction containing FdIII was then eluted with 0.3 M KCl, and ammonium sulfate was added to 75% saturation. After centrifugation at 8,000 ϫ g for 20 min, the pellet was resuspended in 0.025 M Tris-HCl (pH 7.4), and the resulting solution was loaded onto a 2.5 ϫ 100-cm Sephadex G-75 superfine gel filtration column. FdIII eluted as a well resolved greenish brown band.
Protein Characterization-NH 2 -terminal protein sequencing was carried out after the protein was reduced and alkylated by ␤-mercaptoethanol and 4-vinylpyridine at the Biotechnology Resource Facility at the University of California, Irvine. For spectroscopic studies all samples were prepared anaerobically under argon in a Vacuum Atmospheres glove box using fully degassed buffers. The protein was initially purified in the presence of dithionite. To prepare oxidized samples the dithionite was first removed by gel filtration, and then the samples were exposed to oxygen for at least 2 h. Circular dichroism (CD) and electron paramagnetic resonance (EPR) samples of oxidized FdIII were then prepared by concentrating the protein and exchanging it into 0.1 M potassium phosphate (pH 7.4) using a Centricon-3 microconcentrator. The reduction of FdIII was carried out by mixing well degassed FdIII, 5Ј-deazariboflavin and EDTA (final concentrations 100 M, 200 M, and 20 mM, respectively) in 0.1 M potassium phosphate (pH 7.4) and then illuminating for 1 min using white light from a slide projector. UVvisible absorption spectra were obtained with a Hewlett-Packard 8452 diode array UV-visible spectrophotometer, CD spectra were recorded using a JASCO J720 spectropolarimeter, and EPR spectra were obtained using a Bruker 300 Ez spectrophotometer. To determine iron content, samples were digested, and the analysis was carried out as described elsewhere (18) using FeCl 3 ⅐6H 2 0 to generate a standard curve with FdI (9) and FixFd (3) as controls. Matrix-assisted laser desorption ionization-time of flight mass spectrometry was conducted at the Protein/Peptide Micro Analytical Facility, California Institute of Technology.
Electrochemistry-Purified water (ϳ18 megohms⅐cm; Millipore) was used in all electrochemical experiments. The buffers Mes, Hepes, and Taps, and co-adsorbates neomycin sulfate and polymyxin B sulfate, were purchased from Sigma. Other reagents were purchased from Aldrich or British Drug House and were of at least analytical grade. Neomycin and polymyxin solutions were prepared as concentrated stocks (0.2 M and 15 mM (i.e. 20 mg/ml), respectively) and adjusted to pH 7.4.
An AutoLab electrochemical analyzer (Eco Chemie, Utrecht, The Netherlands) was used to measure cyclic voltammograms. Controlled potential reductions were carried out using an Ursar Instruments potentiostat in conjunction with a Kipp and Zonen YT recorder. The all-glass cell and three-electrode system used for protein film voltammetry and the bulk solution voltammetry have been described previously (10,17,20). All potential values are given with reference to the SHE. The saturated calomel reference electrode (SCE) was held at 22°C at which we have adopted E(SCE) ϭ ϩ243 mV versus SHE. Reduction potentials were calculated as the average of the anodic and cathodic peak potentials, E°Ј ϭ 1/2(E pa ϩ E pc ). The sample compartment was maintained at 0°C. The pyrolytic graphite edge electrode (surface area typically 0.18 cm 2 ) was polished prior to each experiment with an aqueous alumina slurry (Buehler Micropolish: 0.3 m for solution electrochemistry or 1.0 m for protein film voltammetry) and then sonicated extensively to remove traces of Al 2 O 3 .
Controlled potential electrolysis was carried out using a cell that featured a graphite pot constructed so that the internal walls project edge surface to the solution. The cell, which has been described previously (21), was set up in an anaerobic glove box (Belle Technology, Poole, U. K.) with an inert atmosphere of N 2 (O 2 Ͻ 2.0 ppm). For cyclic voltammetry, only the base was connected. For controlled potential electrolysis, all parts of the electrode were used, the solution being stirred by magnetic microflea. As a precaution to ensure homogeneity in electrochemical studies, samples of FdIII were prepurified using FPLC under an anaerobic atmosphere. Solutions were dialyzed into the required buffer solutions using an Amicon 8MC unit equipped with a microvolume assembly and a YM3 membrane. For protein film experiments, the ferredoxin solution (80 -100 M) used to coat the electrode contained 0.1 M NaCl, 25 mM Tris-HCl, and polymyxin (200 g/ml). The pH of this solution was adjusted to 7.0 at 0°C. The buffer-electrolyte solution in the electrochemical cell consisted of 0.1 M NaCl, a 60 mM mixed buffer system (15 mM in each of acetate, Mes, Hepes and Taps), 0.1 mM EGTA, and 0.2 mg/ml polymyxin, adjusted to the desired pH using HCl or NaOH at 0°C. The freshly polished electrode surface was painted with about 1 l of chilled protein solution from a fine capillary and then placed promptly into the cell solution. To stabilize protein films, the cell solution also contained 200 g/ml polymyxin or 2 mM neomycin. The pH of this solution was checked after each set of experiments, with the pH electrode calibrated at 0°C. For bulk solution voltammetry and controlled potential electrolysis, ferredoxin solutions contained 0.1 M NaCl with 60 mM mixed buffer. Small aliquots of neomycin stock solution were added (final concentration 2.0 mM) to promote a strong and persistent electrochemical response.

RESULTS AND DISCUSSION
The Discovery of FdIII and Its Relationship to FdI-When crude extracts from nitrogen-fixing cells of wild-type A. vinelandii are separated on DEAE-cellulose with a 0.1-0.5 M NaCl gradient, three major brown peaks are observed (Fig. 1). The first two peaks correspond to the MoFe and Fe proteins of nitrogenase, respectively, and the third corresponds to FdI (16). The size of the third peak is proportional to and therefore an indicator of the FdI level present. In recent years we have constructed and purified many site-directed mutant variants of FdI (e.g. 9, 10, 17) some of which accumulate to much lower levels than the wild-type protein. We have observed a strong correlation between the size of the FdI peak on the first column ( Fig. 1) and the amount of material present in the cell-free extracts which cross-reacts with polyclonal antibodies raised against denatured gel-purified native FdI (Fig. 2). We were therefore very surprised to find that cells expressing one particular variant, C16S FdI, which had only very low levels of FdI that cross-reacted to the antibody (Fig. 2), had a normal, wildtype FdI size peak on the first DEAE-cellulose column (Fig. 1).
Further purification of the brown "FdI" fraction from the C16S preparation (as described under "Experimental Procedures") resulted in a greenish brown protein solution that exhibited a single band on Coomassie-stained SDS-polyacrylamide gels (Fig. 3). This protein did not cross-react with antibodies raised against FdI, and for reasons described below the new protein was designated FdIII. Once purified FdIII was available, polyclonal antibodies were raised, and we reexamined cell-free extracts from wild-type, FdI Ϫ , and C16S strains of A. vinelandii using the FdIII antibody. As illustrated in Fig.  2, the results show that the levels of FdIII are much greater in cells expressing the C16S variant of FdI than they are in either wild-type cells or in cells that make no FdI. The levels of FdIII in the C16S variant are also observed to be much higher than in all other FdI mutants tested to date based on Western analysis (data not shown). Thus FdIII specifically accumulates in response to expression of the C16S FdI variant. The reason for this is not currently understood. However, it should be noted that the [3Fe-4S] ϩ/0 cluster is implicated in a regulatory function carried out by FdI in A. vinelandii and that C16 is a ligand to this cluster (9, 15, 17, 22-24). The purification yields of FdIII from cells expressing C16S FdI are typically 15 mg/1 kg, wet weight, of cells, compared with the approximately 8 mg of FdI usually obtained from 1 kg, wet weight, of wild-type cells (19).
FdIII Is Closely Related to C. vinosum Ferredoxin-To identify FdIII, the first 56 NH 2 -terminal amino acid sequence of the purified protein was obtained after modification of cysteine residues by 4-vinylpyridine (Fig. 4). Before this study 11 small [Fe-S] proteins had been identified in A. vinelandii, many by gene sequencing (3). Surprisingly, the sequence shown in Fig. 4 did not correspond to any of the known ferredoxin-like proteins from this organism. Data base searches, however, revealed that the first 56 amino acids of FdIII exhibit 77% identity and 88% similarity with the low potential 8Fe ferredoxin from C. vinosum. Fig. 4 compares the sequence obtained here with sequences from other ferredoxins in the C. vinosum class. In general, ferredoxins that contain 2[4Fe-4S] 2ϩ/ϩ clusters can be divided into two classes. The clostridial-type ferredoxins have two Cys-Xaa-Xaa-Cys-Xaa-Xaa-Cys-Xaa-Xaa-Xaa-Cys-Pro motifs, whereas the ferredoxins in the "chromatium" class shown in Fig. 4 have one motif of that type and one more unusual Cys-Xaa-Xaa-Cys-Xa 7-9 -Cys-Xaa-Xaa-Xaa-Cys-Pro motif (1). The ligand assignment shown in Fig. 4 is derived from the x-ray structure of C. vinosum ferredoxin (11). All eight ligand cysteine residues lie within the available FdIII sequence and coincide in position with the [Fe-S] cluster ligand cysteines of C. vinosum ferredoxin, thereby eliminating the possibility that FdIII is a clostridial-type ferredoxin. Most of the ferredoxins shown in Fig. 4 contain nine Cys residues, one of which is not a cluster ligand. This ninth Cys is also conserved in FdIII and is in a position identical to that of the ninth Cys of C. vinosum ferredoxin.
The FdIII sequencing data presented here appear to bring to four the number of ferredoxins from one organism, A. vinelandii, which have a chromatium-type sequence (3). The other three were identified originally by gene sequencing, and all appear to be related somehow to nitrogen fixation based on their relationships to other genes. For example, fdxN is cotranscribed with nifB (25), vnfFd is cotranscribed with vnfH (26,27), and fixFd is cotranscribed with fixABCX (28). Only FixFd has been purified to date (3). Because of the relationship of these other proteins to nitrogen fixation we tested whether or not FdIII was nif-regulated by growing the cells in the presence and absence of ammonia. There was no difference in the amounts of FdIII present under the two conditions as measured either by Western analysis or by the purification of the protein. Thus, like FdI (29), FdIII is not nif-regulated.
FdIII Is a Monomer with a Molecular Mass of 9,920 Da-Although SDS-polyacrylamide gel electrophoresis is often used to determine subunit molecular masses, we have found that it is not useful when studying small acidic [Fe-S] proteins. This is illustrated in Fig. 3, which compares the migration of the denatured FdIII with the migration of molecular mass standards and two A. vinelandii ferredoxins of known molecular mass, FdI and FixFd. As shown in Fig. 3, FdI and FixFd migrate as if their molecular masses were 18,800 and 9,000 Da, respectively, whereas their actual polypeptide molecular masses are known to be 12,071 (29) and 7,758 (28). Therefore other methods were employed to determine the molecular mass.
First, matrix-assisted laser desorption ionization-time of flight mass spectrometry shows that the FdIII apoprotein has a mass of 9,220 Ϯ 2 Da. Based on a composition of 2[4Fe-4S] 2ϩ/ϩ clusters (see below) this method gives a molecular mass of 9,924 Da for the holoprotein. To obtain a second estimate and to compare the native protein with the denatured protein, an FPLC-Superdex 75 gel filtration method was used, with FdI and FixFd as standards in addition to the commercially available standards. Again the migration of FdI and FixFd deviated significantly from the standard curve derived from the other four non-iron-sulfur proteins. Using only FdI and FixFd as standards, the native molecular mass of FdIII by this method was calculated to be 10,600 Ϯ 680 Da. Thus, like the other proteins in Fig. 4 and most ferredoxin-like proteins, FdIII is a monomer. Fig. 4 compares the molecular mass of FdIII with values reported for homologous chromatium-type ferredoxins. Clearly this class can be subdivided further into two groups, proteins with molecular masses of 6,000 -7,000 Da and proteins with molecular masses of 9,000 -10,000. The increase in molecular mass results from a COOH-terminal extension. Even when only the NH 2 -terminal sequences are considered, however, the four proteins having greatest sequence similarity to FdIII, including C. vinosum ferredoxin, also fall into the same molecular mass class (Fig. 4).
FdIII Contains Two [4Fe-4S] 2ϩ/ϩ Clusters-The UV-visible absorption spectra of FdIII are shown in Fig. 5. The spectrum of air-oxidized FdIII contains a broad peak at 390 nm and a shoulder at 315 nm. The shape of this spectrum is indistinguishable from that obtained for air-oxidized C. vinosum ferredoxin (30), FixFd (3), and other proteins that contain two [4Fe-4S] 2ϩ/ϩ clusters and is quite different from that obtained for the 7Fe FdI (10). The spectrum shown in Fig. 5 has an  [4Fe-4S] 2ϩ clusters do not exhibit EPR signals at low temperature (31). As shown in Fig. 7a (32). This further eliminates the possibility that FdIII is a 7Fe protein and also shows that unlike some 8Fe ferredoxins that lose iron to form 3Fe clusters upon exposure to air (5), the 4Fe clusters of FdIII are extremely stable. Some protein-bound [4Fe-4S] clusters can also convert to [3Fe-4S] ϩ clusters upon addition of ferricyanide (33). However, incubation with a 10-fold excess of ferricyanide produced no new EPR signals attributable to [3Fe-4S] ϩ or indeed to [4Fe-4S] 3ϩ (34), the latter observation eliminating the albeit remote possibility (based on sequence comparisons and visible spectra (1,30)) that FdIII is actually a HiPIP.
Taken together the above data lead to the conclusion that the new protein isolated here is a 2[4Fe-4S] 2ϩ/ϩ ferredoxin of the chromatium-type. Ferredoxins that are first identified by gene sequencing are often named based on the location of the gene relative to other known genes (e.g. FdxN (25), FixFd (28)). Ferredoxins that are first identified by protein purification, as  (58); and trace c, the reduced sample was exposed to air for 5 min. The spectra were recorded with 1-nm increments for 10 scans.
is the case here, are generally numbered in the order in which they are identified. We have chosen to name this ferredoxin FdIII because FdI and a [2Fe-2S]-containing protein that is designated [Fe-S]II but is sometimes referred to as FdII have both been characterized extensively from A. vinelandii (1,(35)(36)(37). Fig. 4, the following have been isolated and at least partially characterized with respect to reduction potentials: Rhodobacter capsulatus FdI (the fdxN gene product) (38); C. vinosum ferredoxin (11-14, 39, 40); FixFd from A. vinelandii (3) and recombinant Rhizobium meliloti FdxN (41). As monitored by the appearance of a g ϭ 1.94 EPR signal due to [4Fe-4S] ϩ clusters, at least three of these proteins could be at least partially reduced in solution by dithionite or a combination of dithionite, methyl viologen, and zinc (14). As originally reported for C. vinosum ferredoxin (12), the addition of dithionite (at different pH values ranging from 6.0 to 9.0) or electrochemically reduced methyl viologen to FdIII did not lead to the reduction of its [4Fe-4S] 2ϩ clusters as evidenced by the lack of change in either the UV-visible (Fig. 5), CD, or EPR spectra. Attempted reduction with dithionite/ methyl viologen/zinc, or Ti(III) citrate unfortunately led to irreversible denaturation of the protein. Fig. 6 shows that the protein could be reversibly reduced with a 5Јdeazariboflavin/ EDTA/light system. However, the CD data show that unlike the situation with 8Fe ferredoxins that have two clusters with potentials around Ϫ400 versus SHE even this powerful reductant fails to reduce the protein fully (3).

Unlike All Known 8Fe Ferredoxins, FdIII Has Two Very Different Reduction Potentials for the Two [4Fe-4S] 2ϩ/ϩ Clusters-Of the proteins shown in
Using the same 5Ј-deazariboflavin/EDTA/light system, EPR samples were prepared of FdIII (Fig. 7b). In general, [4Fe-4S] ϩ clusters exhibit EPR signals with the g values and the intensity depending on the spin state. Most [4Fe-4S] ϩ clusters exhibit S ϭ 1 ⁄2 spin and anisotropic EPR with g av ϳ1.94. If two S ϭ 1/2 clusters are present, separated by short distances, the signals are broadened and show additional structure (3,14,31). The complex g av ϭ 1.94 (Fig. 7) exhibited by reduced FdIII is therefore consistent with the presence of two [4Fe-4S] ϩ clusters. Again the reduction is not complete, and the spectrum integrates to only about 1.2 spins/molecule. This result combined with the complex nature of the spectrum would be consistent with complete reduction of one cluster and partial reduction of the other cluster to give a mixture of FdIII molecules, either with only one cluster reduced or with both clusters reduced. This in turn leads to the conclusion that the two clusters must have substantially different reduction potentials. Direct electrochemical methods were employed to measure those potentials. Fig. 8a shows the cyclic voltammetry measured for a 60 M solution of FdIII at pH 7.0, 0°C, at a scan rate of 10 mV s Ϫ1 . Two well defined redox couples (we refer to these as A and B) are observed which have reduction potentials of Ϫ486 Ϯ 10 and Ϫ644 Ϯ 10 mV, respectively. The corresponding peak separations (oxidation minus reduction) are 120 and 60 mV, and peak currents are proportional to the square root of the scan rate up to 20 mV s Ϫ1 . These observations are as expected for a freely diffusing redox couple at a planar electrode (42). Fig. 8b shows the voltammetry obtained for a film of FdIII measured at 100 mV s Ϫ1 . Two symmetrical signals, pairs of oxidation and reduction peaks, are observed (AЈ and BЈ, the "prime" denoting film configuration) with reduction potentials of Ϫ466 Ϯ 10 and Ϫ681 Ϯ 10 mV versus SHE, and approximately equal areas, consistent with both couples being present in a 1:1 ratio. Small differences are commonly observed when comparing reduction potentials measured for bulk solution versus the protein film.
The former values should correspond to potentiometric data, whereas the advantages of the film configuration lie in economy, sensitivity, the capability of probing kinetics and easy determination of relative stoichiometries of different electron transfer reactions as shown here. The generally sharp appear- ance of the signals (particularly AЈ) with just modest peak separations at 100 mV s Ϫ1 shows that the clusters in the film-bound protein behave uniformly and are able to shuttle rapidly between redox levels (43). As measured also by film voltammetry, each couple exhibits a pH dependence of approximately Ϫ15 mV/pH unit with no obvious pK a . A pH dependence of similar magnitude has been observed for the [4Fe-4S] 2ϩ/ϩ cluster of A. vinelandii FdI (10). Controlled potential coulometry at pH 7.0 gave 1.1 Ϯ 0.3 electron equivalents for the first electron and 1.4 Ϯ 0.3 electrons for the second electron. These results are consistent with 1:1 stoichiometry if allowance is made for interference by trace O 2 .
In general, the reduction potentials of [4Fe-4S] 2ϩ/ϩ clusters are known to be very sensitive to protein structure with potentials ranging from Ϫ280 to Ϫ700 mV for different proteins (44). It is therefore surprising that for all 8Fe proteins that have been characterized to date both [4Fe-4S] 2ϩ/ϩ clusters are reported to have similar or even identical reduction potentials (44). This is true not only for the clostridial-type ferredoxins that have the same Cys motif (although different sequences) for both clusters, but also for the chromatium-type ferredoxins that have two very different motifs for the two clusters. For example, FixFd, which is in the 6,000 -7,000-Da molecular mass chromatium-type class, very clearly shows only a single signal in cyclic voltammetry, i.e. two couples having indistinguishable reduction potentials. For C. vinosum ferredoxin, which is in the larger molecular mass group, the two clusters are also reported to have the same reduction potential; this is surprising since the local structures around the two clusters are now known to be very different (11).
The data presented here establish for the first time that a ferredoxin can contain two [4Fe-4S] 2ϩ/ϩ clusters having two very different reduction potentials. In contrast to typical 8Fe ferredoxins in which the two clusters have indistinguishable reduction potentials and are thus suited to both one-electron or two-electron transfers, FdIII is expected to be more suited to undertake independent one-electron processes. This is similar to the situation for 7Fe ferredoxins in which replacement of one [4Fe-4S] 2ϩ/ϩ by [3Fe-4S] ϩ/0 results in about a 200-mV difference in potential for the two clusters. Thus, the 8Fe FdIII and the 7Fe FdI are very similar to each other not only in size and charge but also in the reduction potentials for their two [Fe-S] clusters. The results also highlight the increasing number of Fe-S clusters being identified which have extremely low reduction potentials, i.e. below the thermodynamic limit for water. These include A. vinelandii FdI (10), Center X of photosystem I (45), "hyper-reduced" [3Fe-4S] 2Ϫ clusters (21), and most recently the [4Fe-4S] cluster of nitrogenase iron protein, which can be generated in the all-Fe(II) level (46,47). It is highly probable that these newly discovered activities underpin important physiological functions yet to be established.