Spectroscopic Investigation of Selective Cluster Conversion of Archaeal Zinc-containing Ferredoxin fromSulfolobus sp. Strain 7*

Archaeal zinc-containing ferredoxin fromSulfolobus sp. strain 7 contains one [3Fe-4S] cluster (cluster I), one [4Fe-4S] cluster (cluster II), and one isolated zinc center. Oxidative degradation of this ferredoxin led to the formation of a stable intermediate with 1 zinc and ∼6 iron atoms. The metal centers of this intermediate were analyzed by electron paramagnetic resonance (EPR), low temperature resonance Raman, x-ray absorption, and1H NMR spectroscopies. The spectroscopic data suggest that (i) cluster II was selectively converted to a cubane [3Fe-4S]1+ cluster in the intermediate, without forming a stable radical species, and that (ii) the local metric environments of cluster I and the isolated zinc site did not change significantly in the intermediate. It is concluded that the initial step of oxidative degradation of the archaeal zinc-containing ferredoxin is selective conversion of cluster II, generating a novel intermediate containing two [3Fe-4S] clusters and an isolated zinc center. At this stage, significant structural rearrangement of the protein does not occur. We propose a new scheme for oxidative degradation of dicluster ferredoxins in which each cluster converts in a stepwise manner, prior to apoprotein formation, and discuss its structural and evolutionary implications.


H NMR spectroscopies. The spectroscopic data suggest that (i) cluster II was selectively converted to a cubane [3Fe-4S] 1؉ cluster in the intermediate, without forming a stable radical species, and that (ii) the local metric environments of cluster I and the isolated zinc site did not change significantly in the intermediate. It is concluded that the initial step of oxidative degradation of the archaeal zinc-containing ferredoxin is selective conversion of cluster II, generating a novel intermediate containing two [3Fe-4S] clusters and an isolated zinc
center. At this stage, significant structural rearrangement of the protein does not occur. We propose a new scheme for oxidative degradation of dicluster ferredoxins in which each cluster converts in a stepwise manner, prior to apoprotein formation, and discuss its structural and evolutionary implications.
Ferredoxins, small iron-sulfur (FeS) proteins in archaea, serve as water-soluble electron acceptors of acyl-coenzyme A forming 2-oxoacid:ferredoxin oxidoreductase, a key enzyme involved in the central archaeal metabolic pathways (1)(2)(3)(4)(5). The 2.0-Å resolution x-ray crystal structure (PDB 1 entry 1XER ; Ref. 6) of ferredoxin from Sulfolobus sp. strain 7 (JCM 10545; optimal growth conditions, pH 2.5-3 and 80°C) showed the presence of an unexpected isolated zinc center, tetrahedrally coordinated by three nitrogen atoms from histidine residues in the N-terminal extension region (N␦1 of His 16 , N⑀2 of His 19 , and N␦1 of His 34 ) and one O␦1 atom from Asp 76 in the FeS cluster-binding core fold (Fig. 1). A similar zinc site was also found in ferredoxin from Thermoplasma acidophilum strain HO-62 that also contained one [3Fe-4S] 1ϩ,0 cluster, and one [4Fe-4S] 2ϩ,1ϩ cluster (7). This site was analyzed by zinc K-edge x-ray absorption spectroscopy (XAS) (8) and was found to be essentially identical to the zinc site in Sulfolobus sp. ferredoxin. Thus, these unusual ferredoxins contain both the conventional FeS clusters and a structurally conserved, isolated zinc center. This new class of bacterial type ferredoxin, isolated from phylogenetically diverse members of several aerobic and thermoacidophilic archaea, are thus called "zinc-containing ferredoxins" (2, 6 -10).
Because FeS clusters have a remarkable facility for interconversion under protein-bound conditions (reviewed in Ref. 14), the significant discrepancy of the types of FeS clusters of Sulfolobus sp. zinc-containing ferredoxin in the crystalline and as-isolated states may represent selective degradation of the cluster II to a [3Fe-4S] form in vitro. Although the [4Fe-4S] 7 [3Fe-4S] cluster interconversion is well known in some bacterial type ferredoxins (14 -25) and aconitase (26 -29), there is no conclusive report demonstrating the selective cluster conversion at the cluster II site in bacterial type dicluster ferredoxins. The most extensive spectroscopic analyses of oxidative degra-dation process in dicluster ferredoxins have been conducted with Azotobacter vinelandii ferredoxin I with one [3Fe-4S] cluster and one [4Fe-4S] cluster (30 -35). In this instance, it has been proposed that ferricyanide oxidation results in complete destruction of the [4Fe-4S] cluster, to produce an intermediate containing only one [3Fe-4S] cluster (cluster I), en route to complete disruptions of both clusters and ultimately to unfolded protein (30,31) .
We have recently found conditions where the native 7Fe form of Sulfolobus sp. ferredoxin (5) is slowly and irreversibly converted to a stable intermediate species under aerobic conditions (10). Herein, we report comparisons of EPR, resonance Raman, XAS, and 1 H nuclear magnetic resonance (NMR) analyses of the metal centers in the native 7Fe and the intermediate forms of Sulfolobus sp. zinc-containing ferredoxin. Our spectroscopic data demonstrate the oxidative degradation of cluster II to form a [3Fe-4S] cluster, yielding a 6Fe-containing intermediate of zinc-containing ferredoxin. This will be also discussed with respect to the structure and evolution of a ferredoxin core fold module.
EXPERIMENTAL PROCEDURES DEAE-Sepharose Fast Flow and Sephadex G-50 gels were purchased from Amersham Pharmacia Biotech, and NMR-grade D 2 O was from Wako Pure Chemicals (Tokyo, Japan). Water was purified by the Milli-Q purification system (Millipore). Other chemicals used in this study were of analytical grade.
Sulfolobus sp. strain 7 cells (JCM 10545), originally isolated from Beppu Hot Springs, Japan, were cultivated aerobically and chemoheterotrophically at pH 2.5-3 and 75-80°C, and the 7Fe form of the archaeal ferredoxin was routinely purified as described previously (5,10). The intermediate form of Sulfolobus sp. ferredoxin was obtained by artificial conversion at pH 5.0 as described previously (10), after removal of the unconverted 7Fe form by a DEAE-Sepharose Fast Flow column chromatography (Amersham Pharmacia Biotech) followed by a Sephadex G-50 column chromatography (Amersham Pharmacia Biotech). 2-Oxoacid:ferredoxin oxidoreductase of Sulfolobus sp. strain 7 was purified as described previously (5,36). 2-Oxoacid:ferredoxin oxidoreductase activity was monitored with a horse heart cytochrome c reduction assay at 50°C using purified ferredoxin an intermediate electron acceptor (2,5). Enzymatic reduction of Sulfolobus sp. ferredoxin with the cognate 2-oxoacid:ferredoxin oxidoreductase was conducted under anaerobic conditions at 55°C as described previously (5).
Absorption spectra were recorded using a Hitachi U3210 spectrophotometer or a Beckman DU-7400 spectrophotometer. Matrix-assisted laser desorption ionization-time of flight mass spectrometry of purified apoferredoxin (made in distilled water) was performed by a Finnigan MAT VISION 2000 instrument at an accelerating potential of 5.0 kV, using a 2,5-dihydroxybenzoic acid matrix. Electron paramagnetic resonance (EPR) measurements were performed using a JEOL JES-FE3XG spectrometer equipped with an Air Products model LTR-3-110 Heli-Tran cryostat system and a Scientific Instruments series 5500 temperature indicator/controller. Spin concentrations were estimated by double integration, with 0.1 and 1 mM Cu-EDTA as standards. The spectral data were processed using KaleidaGraph version 3.05 (Abelbeck Software).
Low temperature resonance Raman spectra were recorded at 77 K using 488.0-nm and 457.9-nm Ar ϩ laser excitation (500 mW) as described previously (7,37). The sample was immersed into a liquid nitrogen reservoir and the scattered light was collected at 45 degrees to the incident beam. The spectral slit width was 4 cm Ϫ1 , and a multiscan averaging technique was employed.
Purified zinc-containing ferredoxins in 20 mM potassium phosphate buffer, pH 6.8, were concentrated by pressure filtration with an Amicon YM-3 membrane. Further concentration was achieved by placing the samples under a stream of dry nitrogen gas. The resultant samples (ϳ2-3 mM), containing 30% (v/v) glycerol, were frozen in a 24 ϫ 3 ϫ 2-mm polycarbonate cuvette with a Mylar-tape front window for XAS studies. XAS data were collected at Stanford Synchrotron Radiation Laboratory with the SPEAR storage ring operating in a dedicated mode at 3.0 GeV (Table I), as reported previously (8). EXAFS analysis was performed with the EXAFSPAK software according to standard procedures (38). Curve-fitting analysis was performed as described previously (8,39). Multiple scattering models, calculated using FEFF version 7.02 (40), were based on bis(acetato)-bis(imidazole)-zinc(II) (41) or tetra(imidazole) zinc(II) perchlorate (42). 1 H NMR spectra were recorded on using JEOL GSX-400 spectrometer operating at 399.78 MHz Larmor frequency. The NMR sample was prepared by exchanging the buffer of purified ferredoxin into 10 mM sodium deuterium phosphate buffer, pH 7.5 (pH was uncorrected), using an Amicon ultrafiltration device with a YM-3 membrane (final concentration, ϳ3 mM ferredoxin). Sweep widths of 30 KHz were used for the purified protein. 1 H spectra were recorded with either a slow repetition time (2 s) using presaturation for water suppression or with a fast repetition time (100 ms) using a super water elimination Fourier transform (super-WEFT) pulse sequence (43) with a relaxation delay (60 ms) between the and /2 pulse sequence for water suppression. The spectra were calibrated by referencing to the residual HDO signal, assigned to a 4.74-ppm shift at 303 K. The standard JEOL software package was used for data processing.
Protein concentration of purified ferredoxins was measured as described previously (10). Metal content analyses were carried out by inductively coupled plasma atomic emission spectrometry with a Seiko SPS 1500 VR instrument at Tokyo Institute of Technology and a Jobin-Yvon JY 38S instrument at Rigaku Ltd.

RESULTS
The typical yield of the purified intermediate obtained from the 7Fe form was ϳ30% under the conditions described under "Experimental Procedures." The matrix-assisted laser desorption ionization-time of flight mass spectrometry suggested that masses [M ϩ H] 1ϩ of these apoproteins are identical within the experimental error (data not shown). Chemical analysis by inductively coupled plasma atomic emission spectrometry suggested the iron:zinc ratio of 6.9:1.0 mol/mol in the native 7Fe form and 5 Upon reduction of the native 7Fe form, the sharp g ϭ 2.02 EPR signal attributed to that of a [3Fe-4S] 1ϩ cluster was fully reduced, thus giving rise to a very broad low field resonance at g ϳ 12, which is characteristic of the reduced S ϭ 2 [3Fe-4S] 0 cluster (44) (Fig. 3, trace A). A rhombic EPR signal at g ϭ 2.06, 1.94, and 1.90, attributed to the reduced S ϭ 1/2 [4Fe-4S] 1ϩ cluster, was detected. This signal had additional wings on the high and low field sides of the main EPR signal (g ϭ 2.11 and 1.85), resulting from magnetic interactions with the reduced S ϭ 2 [3Fe-4S] 0 cluster. These EPR signals could be detected up to 30 K (data not shown), and no evidence for the presence of a high multiplicity S ϭ 3/2 [4Fe-4S] 1ϩ cluster was obtained (Fig.  3

, trace A).
Reduction of the intermediate under the same conditions resulted in the disappearance of most of the sharp g ϭ 2.02 EPR signal attributed to the S ϭ 1/2 [3Fe-4S] 1ϩ clusters, and the appearance of the broad low field resonance at g ϭ 12 characteristic of the S ϭ 2 [3Fe-4S] 0 cluster, as the predominant species (Fig. 3, traces B, C, and D). Several weak and minor resonances, mainly consisting of the remaining S ϭ 1/2 [3Fe-4S] 1ϩ cluster at g ϭ 2.02 and a very weak radical feature at g ϭ 2 of unknown origin, were also reproducibly detected in the g ϳ 2 region (Ͻ0.1 spin/mol). All these minor species existed in a substoichiometric amount (Ͻ0.1 spin/mol), indicating minor heterogeneity in the dithionite-reduced intermediate form.
A weak band at 335 cm Ϫ1 appears in the native 7Fe form and is assigned to the Fe-S bridging mode of a regular biological [4Fe-4S] 2ϩ cluster with complete cysteinyl ligation (46) (Fig.  4A). In Pyrococcus furiosus 4Fe ferredoxin and active aconitase, which contain a [4Fe-4S] cluster coordinated by one non-cysteinyl and three cysteinyl ligands (16,48,49), the equivalent band was shifted to a higher frequency (16). It has been reported that the resonance Raman spectra for biological [4Fe-4S] 2ϩ clusters are normally much less intense than those for biological [3Fe-4S] 1ϩ clusters (reviewed in Refs. 45 and 47). Relative intensity of the Fe-S bridging mode of a [4Fe-4S] 2ϩ cluster at 335-cm Ϫ1 band in the native 7Fe form of Sulfolobus sp. ferredoxin (Fig. 4A) is very similar to that reported for Thermus thermophilus 7Fe ferredoxin (45,47). A very weak Fe-S terminal mode at ϳ249 cm Ϫ1 is normally seen for [4Fe-4S] 2ϩ clusters; however, the signal in this region, for the 7Fe form of Sulfolobus sp. ferredoxin, is too weak to analyze. In conjunction with other spectroscopic data reported in this paper (Figs. 2 and 3), the resonance Raman data of the native 7Fe form indicate the presence of a [4Fe-4S] 2ϩ cluster, in addition to a [3Fe-4S] 1ϩ cluster.
The low temperature resonance Raman spectrum at 488.0-nm Ar ϩ ion laser excitation of the intermediate form showed that the [3Fe-4S] 1ϩ cluster also exhibited three Fe-S bridging modes at 260, 285, and 347 cm Ϫ1 , and at least two Fe-S terminal modes at 369 and 385 cm Ϫ1 (Fig. 4D). Detection of these bands in both the native 7Fe (Fig. 4B) and intermediate 6Fe forms (Fig. 4D) suggests that the [3Fe-4S] 1ϩ core structure is structurally not very different in these two forms. On the other hand, the weak band primarily associated with the Fe-S bridging mode at 335 cm Ϫ1 of the [4Fe-4S] 2ϩ cluster of the native 7Fe form, resonantly enhanced upon 457.9-nm Ar ϩ ion laser excitation (Fig. 4A), was not detected in the intermediate form (Fig. 4C). These results are consistent with the absence of a [4Fe-4S] 2ϩ cluster in the intermediate form. It should be noted that the band at 359 cm Ϫ1 , resonantly enhanced upon 457.9-nm Ar ϩ ion laser excitation (Fig. 4, A and C), was detected in both forms of Sulfolobus sp. ferredoxin, whereas the equivalent band was observed in the 7Fe, but not the 6Fe, form of Mycobacterium smegmatis ferredoxin (37,50). This indicates that a weak band at ϳ359 cm Ϫ1 cannot be used diagnostically for the presence of a conventional biological [4Fe-4S] 2ϩ cluster.
X-ray Absorption Spectroscopy-The zinc K-edge x-ray absorption spectra of the purified 7Fe and 6Fe forms of Sulfolobus sp. ferredoxin are very similar (Fig. 5). The absorption edge positions (9663.2 for 7Fe, 9663.3 for 6Fe) for both samples fall at the expected energy for Zn(II) with all light elements (nitrogen or oxygen) in the coordination sphere (51,52). 2 The intensity of the edge is most reminiscent of four-coordinate compounds, and the peak area of the second XANES peak is not as intense as expected for tetraimidazole coordination, nor is it as weak as seen in a ZnO 4 compound (51).
The presence of a carboxylate ligand results in destructive interference with the EXAFS multiple-scattering contributions from outer shell atoms of histidine imidazoles. This interference can be visualized in the Fourier transform of the data as a decrease in the ϳ3-Å peak relative to the ϳ4-Å peak. Previously, we reported that the 7Fe form of Sulfolobus sp. ferredoxin was best fit assuming a Zn(imid) 3,4 (COO Ϫ ) 1 coordination environment with a Zn-O-C bond angle of ϳ126°(Fit 4, Table  II) (8). 3 Similarly, the 6Fe form can also be fit assuming the same coordination geometry (cf. Fits 4 and 8, Table II; Fig. 6). The zinc XAS results clearly show that the zinc site found in the intermediate is an isolated center having very similar coordination environment to the native 7Fe form. The XAS-determined bond distances and bond angles of the intermediate are also in agreement with the crystallographically determined Zn-N and Zn-O bond distances (1.96 and 1.90 Å, respectively) and Zn-O-C angle (ϳ126°) (6).
The iron K-edge x-ray absorption spectra for the two forms of Sulfolobus sp. ferredoxin were almost identical (Fig. 5) 1 H NMR Spectroscopy-1 H NMR spectroscopy has been used to detect and assign the ␤-protons of cysteine ligand residues coordinated to the FeS centers in some 7Fe-containing ferredoxins (54 -57). Bentrop et al. (55) have performed detailed paramagnetic NMR analysis of the native 7Fe form of another zinc-containing ferredoxin from A. ambivalens, which is 95% identical to Sulfolobus sp. ferredoxin at the primary structural level (only 5 out of 103 amino acids are different) (7,12,58). Interestingly, the 1 H NMR spectrum of A. ambivalens ferredoxin shows eight major and eight minor hyperfine-shifted resonances, arising from its ligand protons. This unique pattern apparently results from the superposition of typical spectra from a 3Fe-and a 4Fe-containing cluster (55). This pattern has not been reported for other regular bacterial 7Fe-containing ferredoxins, all of which show only five hyperfine-shifted resonances in the downfield region (54, 56, 57, 59 -62). It has been reported that eight major and eight minor hyperfineshifted resonances likely reflect a heterogeneity of one of the two clusters of A. ambivalens ferredoxin, which does not simply result from sample impurity (55).
The overall spectral features, chemical shift values, and temperature dependences of the hyperfine-shifted resonances in the downfield 1 H NMR spectrum of native Sulfolobus sp. ferredoxin (Fig. 7, A, C, and D)  detected in the downfield region. Each of these hyperfineshifted resonances displays an anti-Curie temperature dependence, with the exception of major signals, C and H, and minor signals, D and E, which have Curie-type temperature dependences (Fig. 7, C and D). Based on the assignment of the hyperfine-shifted resonances of A. ambivalens ferredoxin (55), major signals C, H, G, and K are attributed to ligands of the [3Fe-4S] 1ϩ cluster I, and major signals J, N, O, and P to ligands of the [4Fe-4S] 2ϩ cluster II (Fig. 7, A and C). The sequence specific-assignment of the cysteine resonances in A. ambivalens ferredoxin (55) also indicates that the major signal J of the native 7Fe form is attributable to the H␤ proton of Cys 86 , which is located in the [4Fe-4S] cluster II site. It should be noted that Cys 86 is located in the vicinity of the missing corner (Fe) of the cube at the [3Fe-4S] cluster II site in the x-ray crystal structure of the 6Fe form of Sulfolobus sp. ferredoxin (Fig. 1).
The overall spectral features and chemical shift values of the downfield resonances did not change when stored at 5°C for a short period (within 1 month). However, storage of the native 7Fe form, under oxygenic conditions at 5°C, in NMR tubes for a long term period resulted in gradual and irreversible changes in the spectrum (Fig. 7B). The relative intensities of major signals J, N, O, and P gradually decreased in 7 months, whereas the chemical shift values and relative intensities of other major signals C, H, G, and K, remained unchanged under these conditions (Fig. 7B). This process was sluggish and did

TABLE II
EXAFS curve fitting results R as is the metal-scatterer distance. as 2 is a mean square deviation in R as . The shift in E O for the theoretical scattering functions was optimized, but did not vary more than 1.5 eV. Numbers in square brackets were constrained to be either a multiple of the above value ( as 2 ) or to maintain a constant difference from the above value (R as ). fЈ is a normalized error (chi-squared):    (60). The temperature dependence of the new peak a at 24.0 -24.6 ppm in the downfield spectrum of Sulfolobus sp. ferredoxin suggested splitting into two peaks, a1 and a2, at 30°C (Fig. 7E), as previously observed for the downfieldshifted resonances at around 24.5 ppm of the 3Fe form of P. furiosus monocluster ferredoxin (63)(64)(65). In addition, the new signal a1 at 24.0 -24.6 ppm showed Curie-type temperature dependence (Fig. 7E), as observed with the major signal pair C and H and minor signal pair D and E, which are attributed to one ligand (Cys IV ) of a [3Fe-4S] cluster domain (55) (Fig. 7, C  and D). The details of the sequence-specific assignments of the new hyperfine-shifted resonances are the subject of future analysis, and their correlation with the minor peaks observed in the native 7Fe form remain unclear at this stage. However, some of these resonances (such as signal a) probably belong to a [3Fe-4S] cluster domain and are in line with the increase of the EPR signal intensity at g ϭ 2.02 (data not shown; see Fig. 2).
Taken together, these data suggest that the [4Fe-4S] cluster II in the native 7Fe form is less stable to oxidative degradation than the [3Fe-4S] cluster I, and that the [4Fe-4S] 2ϩ cluster II is gradually and selectively converted to a corresponding cubane [3Fe-4S] 1ϩ cluster II, whereas the [3Fe-4S] cluster I remains essentially unchanged under the applied conditions. Further 1 H NMR analysis is underway to reveal the sequence-specific assignments of the new hyperfine-shifted resonances and the electronic structure of the two [3Fe-4S] clusters in the 6Fecontaining intermediate form.
Biochemical Properties-Archaeal zinc-containing ferredoxins have been shown to serve as water-soluble electron acceptors of acyl-coenzyme A forming 2-oxoacid:ferredoxin oxidoreductase, a key enzyme involved in the central archaeal metabolic pathways (2,3,5,7,12). Previous enzymatic reduction of the native 7Fe form by Sulfolobus sp. 2-oxoacid:ferredoxin oxidoreductase reduced only the [3Fe-4S] cluster I, whereas the [4Fe-4S] cluster II with a lower midpoint redox potential remained in the oxidized state during the steady state (5). The specific activity of the cognate 2-oxoacid:ferredoxin oxidoreductase in the presence of 4 mM 2-oxoglutarate, 50 M coenzyme A, and 23 g of the 6Fe-containing intermediate form at 50°C was 55 mol/min/mg, only slightly lower than that measured with the native 7Fe form under the same conditions (60 mol/min/mg) (5,36). Thus, the reactivity with the cognate oxidoreductase was not lost after the [4Fe-4S] 3 [3Fe-4S] cluster conversion. Enzymatic reduction of the 6Fe form with a catalytic amount of the 2-oxoacid:ferredoxin oxidoreductase (8.6 g/ml) in the presence of 4 mM 2-oxoglutarate and 0.5 mM coenzyme A during the steady-state turnover of the oxidoreductase under anaerobic conditions at 55°C caused bleaching of the g ϭ 2.02 EPR signal attributed to the [3Fe-4S] 1ϩ clusters (see Fig. 3) by ϳ95% in 30 min, with concomitant formation of the broad low field resonance at g ϳ 12, which is indicative of the S ϭ 2 [3Fe-4S] 0 clusters (data not shown). This indicates that the two [3Fe-4S] clusters, i.e. both clusters I and II, of the 6Fe form can be reduced enzymatically by the cognate oxidoreductase under the applied conditions, due to the change of electron distributions within the reduced ferredoxin molecule by relative redox potential upshift of the cluster II upon the selective cluster conversion. DISCUSSION The present spectroscopic investigation indicates that the initial step of oxidative degradation of the native 7Fe form of Sulfolobus sp. zinc-containing ferredoxin, at fairly acidic or neutral pH, is a selective conversion of the [4Fe-4S] cluster II to a [3Fe-4S] cluster. This degradation pathway results in a stable 6Fe-containing intermediate with one isolated zinc center and two [3Fe-4S] clusters, en route to complete protein unfolding. Although the selective interconversion of a [4Fe-4S] 7 [3Fe-4S] cluster at the cluster I site has been well established in some bacterial ferredoxins (14 -25), this novel intermediate is unique in that its formation is triggered by selective cluster conversion at the cluster II site, rather than complete destruction of the cluster. The local metric environments of the [3Fe-4S] cluster I and isolated zinc site in the intermediate form do not change significantly as compared with those in the native 7Fe form, 4 In Sulfolobus sp. zinc-containing ferredoxin, the formation of the 6Fe-containing intermediate is very slow at around neutral pH, which is the main reason why the equivalent intermediate escaped detection in the previous 1 H NMR analysis of A. ambivalens ferredoxin (55). This most likely correlates with the local structural rigidity and/or the complete cysteinyl ligations of the cluster II site in Sulfolobus sp. ferredoxin (see "Discussion"). In this connection, it should be noted that the Asp 14 3 Cys mutant of D. africanus ferredoxin III showed the presence of two [4Fe-4S] 2ϩ,1ϩ clusters, and, unlike in native ferredoxin III, the [4Fe-4S] 7 [3Fe-4S] cluster interconversion reaction was sluggish and did not go to completion (19).  Fig. 4C). The spectral bandwidth was 4 cm Ϫ1 , and a multiscan averaging technique was employed.
indicating that structural rearrangement of the whole molecule does not occur at the initial step of oxidative degradation. The ability of the intermediate to accept electrons transferred from the cognate 2-oxoacid:ferredoxin oxidoreductase is also maintained at this stage.
Correlation with the X-ray Crystal Structure of the 6Fe Form of the Sulfolobus Ferredoxin-The spectroscopic properties of the stable intermediate form of Sulfolobus sp. ferredoxin are consistent with the structural features observed for the 2.0-Å resolution crystal structure of the 6Fe form (6,9). In this structure, the [3Fe-4S] cluster I is coordinated by three cysteinyl ligands contributed from Cys 45 , Cys 51 , and Cys 93 , and the [3Fe-4S] cluster II by another three cysteinyl ligands contributed from Cys 55 , Cys 83 , and Cys 89 (9) (Fig. 1). The two [3Fe-4S] clusters are separated by a crystallographic, center-to-center distance of 12.0 Å, which is similar to that observed in some regular 7Fe-and 8Fe-containing dicluster ferredoxins. The second cysteine residue (Cys 86 ) in the -Cys 83 -Xaa-Xaa-Cys 86 -Xaa-Xaa-Cys 89 -Xaa-Xaa-Xaa-Cys 93 -Pro-motif is located in the vicinity of the missing corner (Fe) of the cluster II cube, and its side chain is exposed to the solvent, away from cluster II ( The electron density for Cys 86 is lower than that of other cysteinyl ligand residues in the crystal structure. Additionally, the average temperature factors for the region surrounding cluster II (ϳ24 Å 2 , on average), especially Cys 86 and Met 87 (Ͼ30 Å 2 ), are markedly higher than those for the region surrounding cluster I (ϳ15 Å 2 ) (9) (Fig. 1). These observations are consistent with the present spectroscopic analyses indicating the structural similarity of the native 7Fe form and the stable intermediate, and suggest that the selective cluster conversion at the cluster II site is a local event. In the crystal structure, the polypeptide conformation in the vicinity of the cubane [3Fe-4S] cluster II and Cys 86 was reported to be intermediate between the [3Fe-4S] and [4Fe-4S] cluster conformations (9). It is likely that the peptide backbone conformation in vicinity of Cys 86 and Met 87 may be somewhat perturbed upon formation of the 6Fe-containing intermediate by oxidative degradation. 5 Based on these considerations, we suggest that oxidative degradation of Sulfolobus sp. zinc-containing ferredoxin is initiated at the [4Fe-4S] cluster II site. As a result of this initial degradation step, the iron atom bound to Cys 86 is released, forming a structurally interconvertible cubane [3Fe-4S] cluster II. Concomitant with the release of iron is a perturbation of the peptide backbone in the vicinity of Cys 86 -Met 87 and rotation of the Cys 86 side chain to solvent, away from the cluster (see Fig.  1). At this stage, no significant structural rearrangement of the entire molecule occurs. Further sluggish oxidative degradation of the two [3Fe-4S] clusters eventually leads to protein unfolding of the ferredoxin core fold and accumulation of apoprotein (10), and may involve other short-lived intermediates not yet characterized.

Oxidative Degradation of Two FeS Clusters in Bacterial Type
Dicluster Ferredoxins-The reaction of the wild-type A. vinelandii ferredoxin I with an excess of ferricyanide is a more complicated oxidative degradation, leading to formation of an unusual intermediate with a three-electron oxidation at sulfur sites (30,31,33,34). Site-directed mutagenesis has suggested that formation of the latter species requires the presence of non-ligating Cys 24 in the vicinity of a cluster sulfide (33). The difference between the oxidative degradation of ferredoxin from A. vinelandii and Sulfolobus sp. strain 7 is likely to reflect the differences in their cluster surroundings, i.e. the absence of the equivalent cysteine residue in the archaeal ferredoxin core fold domain (Fig. 1). An analogous 6Fe-containing ferredoxin species has been isolated from M. smegmatis (37,50) and similar hyperfineshifted resonances appear in the downfield 1 H NMR spectra of ferricyanide-treated ferredoxins from M. smegmatis, Pseudomonas ovalis, and T. thermophilus (59 -61). These observations lend credence to our suggestion that a common step in the oxidative degradation pathway, of regular bacterial type dicluster ferredoxins, is the formation of an intermediate containing two [3Fe-4S] clusters, as illustrated in Fig. 9. In some less stable dicluster ferredoxins, such an intermediate form could be detected only transiently. In other words, detection of this intermediate might depend on the overall structural rigidity of the ferredoxin core fold and/or the cluster surroundings.
Structural and Evolutionary Implications-The polypeptide backbone structure of a bacterial type ferredoxin exhibits a pseudo two-fold symmetry regardless of the number of bound FeS clusters. It has been proposed that the distorted two-fold symmetrical structure has been derived from a putative common ancestor as a result of early gene duplication event, and that the cluster I is strictly conserved in all bacterial type monocluster and dicluster ferredoxins reported so far (5,22,67). There is no monocluster ferredoxin with a cluster bound only to the cluster II site, implying that the two cluster binding sites in a ferredoxin core fold are evolutionary not equivalent (5,67). The selective interconversion of a [4Fe-4S] 7 [3Fe-4S] cluster at the cluster I site has been well established in some bacterial ferredoxins (14 -25), but this was not demonstrated at the cluster II site. Our results indicate that oxidative degradation of the cluster II site in dicluster ferredoxins probably follow the same chemistry with a general structural rule: The missing corner (Fe) of a biological [4Fe-4S] cube is associated with either replacement (e.g. Cys II , Asp), or tilting away to solvent, of the second cysteine residue (Cys II ) in the -Cys I -Xaa-Xaa-Cys II -Xaa-Xaa-Cys III -Xaa-Xaa-Xaa-Cys IV -(Pro)-motif (see Figs. 1 and 9). Although it is currently unclear why this particular site is the most sensitive target for oxidative damage, we expect that increased stability of oxygen-labile bacterial type ferredoxins against oxidant (such as molecular oxygen or ferricyanide) could be conferred by introducing appropriate mutations to the polypeptide region in the vicinity of Cys II .
A ferredoxin core fold of a bacterial type ferredoxin represents a [3Fe-4S]/[4Fe-4S] cluster-binding module in biological systems, and can be found in various simple and complex electron transfer proteins as a cluster-binding subunit and/or domain (14,22). To the best of our knowledge, there are only a few instances of electron transfer proteins which have a [3Fe-4S] cluster at the cluster II site of the module. These include an unusual 7Fe-containing ferredoxin from a hyperthermophilic archaeon, Pyrobaculum islandicum (68) and the C-terminal domain of the cluster-binding subunit (subunit B) of some respiratory fumarate reductase and succinate dehydrogenase complexes (69 -71). In all cases, the second cysteine residue (Cys II ) in the -Cys I -Xaa-Xaa-Cys II -Xaa-Xaa-Cys III -Xaa-Xaa-Xaa-Cys IV -(Pro)-motif is replaced by a non-cysteine residue 5 Preliminary computer modeling analysis by using a SYBYL molecular modeling software (TRIPOS Associates, Inc.) also suggested that simple rotation of the Cys 86 side chain did not allow appropriate coordination of this residue to the missing corner (Fe)  such as Asp, Val, Ile, and Ala. In Escherichia coli respiratory fumarate complex, this might have afforded reduction of the [3Fe-4S] cluster II directly by proximal menaquinol (69). In the native 7Fe form of archaeal zinc-containing ferredoxins, the [4Fe-4S] cluster II is not reduced by the cognate 2-oxoacid: ferredoxin oxidoreductase (5,7,12), although the [3Fe-4S] cluster II of the 6Fe intermediate form can be reduced enzymatically by the oxidoreductase. Hence, the apparent effect of these amino acid replacement at the cluster II site is to modulate electron distributions within a reduced ferredoxin module by upshift of the relative redox potential of the cluster, because a [3Fe-4S] 1ϩ/0 cluster generally has a higher midpoint redox potential than a [4Fe-4S] 2ϩ/1ϩ cluster in biological systems (22). Nevertheless, the cluster conversion at the cluster II site by a single amino acid replacement seems to be not very favored in the course of molecular evolution of a ferredoxin core fold module, which may be related to the overall stability of the polypeptide backbone.
It is known that a ferredoxin core fold module of a bacterial type ferredoxin contains only a few secondary structural elements (5, 22) (see Fig. 1). The FeS cluster binding apparently contributes to the overall protein stability of the module, but the details of structural and thermodynamical roles of a metal cofactor in protein folding and unfolding reactions are little known (72,73). Recent multidimensional heteronuclear NMR analysis of a partially unfolded high potential iron-sulfur protein under severe denaturing conditions indicated that its [4Fe-4S] cluster plays a decisive role in determining the resulting ordered structure, and that the backbone mobility increases with the structural indetermination (74). All the present results are also consistent with the primary role of an FeS cluster in retaining the native-like ordered structure of the ferredoxin core fold module (Figs. 1 and 9). We therefore suggest that the cluster I site of a ferredoxin core fold module is strictly conserved in biological systems not only because of the functional importance in electron transfer (5), but also the structural importance as a possible nucleation site of the folding of the module that shifts equilibrium toward a native-like ordered structure presumably by forming the Fe-S(Cys) bonds in vivo.
The cluster II site is less conserved in the ferredoxin core module (5,22,67), but our results indicate that the native-like ordered structure of the cluster II site is maintained regardless of the type of an FeS cluster ([4Fe-4S] versus [3Fe-4S]) whenever a protein-bound cluster is present (Fig. 9). FIG. 9. A, schematic illustration of the oxidative degradation of the ferredoxin core fold of a regular bacterial type 8Fe ferredoxin and its possible intermediate forms. Cysteines and prolines in two -Cys I/IЈ -Xaa-Xaa-Cys II/IIЈ -Xaa-Xaa-Cys III/ IIIЈ-Xaa-Xaa-Xaa-Cys IV/IVЈ -Pro-motifs are labeled in different colors, and ligating Cys II/IIЈ are highlighted. Ligating cysteines are shown in Roman numerals, and non-ligating cysteines are in brackets. The number of iron atoms in the cluster I site is labeled in blue, and that in the cluster II site in pink. B, the selective degradation of cluster II of the zinc-containing 7Fe ferredoxin from Sulfolobus sp. strain 7, with a concomitant formation of a stable 6Fe-containing intermediate. In the case of Sulfolobus sp. ferredoxin, the initial step probably involves a release of iron atom from Cys IIЈ (Cys 86 ), which then moves away from the cluster, as indicated from the three-dimensional structure of the 6Fe form (see Fig. 1). The selective conversion of a [4Fe-4S] cluster (cluster II) into a [3Fe-4S] cluster occurs without any significant effect on the cluster I and the remote zinc site (B). The native-like ordered structure of the cluster II site is maintained in a ferredoxin core fold when the protein-bound FeS cluster is present.