Functional and evolutionary implications of a [3Fe-4S] cluster of the dicluster-type ferredoxin from the thermoacidophilic archaeon, Sulfolobus sp. strain 7.

The dicluster-type ferredoxin is a key electron carrier in the cytoplasm of the aerobic and thermoacidophilic archaeon, Sulfolobus sp. strain 7, and contains 1 aspartate and 7 cysteine residues as possible ligands to two FeS clusters. The optical, electron paramagnetic resonance (EPR), and cyclic voltammetric studies suggest the presence of one each of [3Fe-4S]1+,0 (-280 mV) and [4Fe-4S]2+,1+ (-530 mV) clusters in the purified Sulfolobus ferredoxin, and the lower potential [4Fe-4S] center was scarcely reducible by excess dithionite even at pH 9. While the Sulfolobus ferredoxin has been known to function as an electron acceptor of 2-oxoacid:ferredoxin oxidoreductase (Kerscher, L., Nowitzki, S., and Oesterhelt, D. (1982) Eur. J. Biochem. 128, 223-230), it is not known whether one or both of two clusters is reduced during the steady-state turnover of the enzyme. Here we show by combinations of the optical and EPR spectroscopies that only the higher potential [3Fe-4S] cluster is reduced at the physiological pH during the steady-state turnover of the purified 2-oxoacid:ferredoxin oxidoreductase at 50 degrees C. The functional significance and evolutionary implications of the [3Fe-4S] center in dicluster-type ferredoxins are discussed.


S-3)
is shown to be involved in the catalytic function (9)(10)(11). Thus, not all of the [3Fe-4S] clusters widely spread in nature represent a converted form of FeS cluster due to the oxidative damage.
Sulfolobus sp. strain 7 (formerly Sulfolobus acidocaldarius strain 7)' is a typical thermoacidophilic archaeon (archaebacterium) isolated from Beppu hot springs, Kyushu, Japan. It has at least two major redox systems; one is the membrane-bound aerobic respiratory chain: and the other is the cytoplasmic Fd-dependent redox ~y s t e m .~ Both systems are coupled at the different sites of the archaeal "oxidative" tricarboxylic acid cycle, and faciliate downstream electron transfer from the physiologically important intermediate metabolites. Our final goal is to elucidate the details of these redox systems of this particular archaeon in molecular details and to discuss it from the bioenergetic and evolutionary points of view. As an initial step, we have recently purified all constitutes of the succinateoxidizing aerobic respiratory system of Sulfolobus, which could reconstitute a cyanide-sensitive electron transport chain i n We now turn our attention to the Sulfolobus Fd-dependent redox system, because none of the aerobic archaeal Fddependent redox systems has been investigated in details while the physiological roles of Fds are more closely linked to the central metabolisms of archaea than in the cases of aerobic respiratory (eu-)bacteria or eukarya (12,13). For this purpose, we have previously isolated and crystallized a dicluster-type Fd, which is present in a large quantity in the archaeal cytoplasm (141, and more recently cloned and sequenced the gene encoding t h e p r~t e i n .~ The deduced amino acid sequence of the protein contains 7 cysteine residues, which are sufficient in number as potential ligands to hold two FeS clusters, as in the case of the structural homologue from S. acidocaldarius strain DSM 639 (15). Thus, two parts of the sequence, namely C Y S~~-The organism had been isolated from Beppu hot springs, Kyushu, Japan, originally named 5'. acidocaldarius strain 7, but was recently redesignated tentatively as "Sulfolobus sp. strain 7" due to a small difference in 16 S rRNA base sequences of strain 7 and 5'. acidocaldarius type strain DSM 639. The preliminary 16 S rRNA sequence analysis suggests that the isolate is a novel species belonging to the genus Sulfolobus.
T. Iwasaki, T. W a k a~, K. Matsuura. and T. Oshima. manuscript in preparation (preliminary'results have been presented at the Eleventh
T. Iwasaki, T. Wakagi Pro and Cys83-Ile-Phe-Cyss6-Met-Ala-Cys8g, provide a typical binding motif for one [4Fe-4S] cluster, and the other two parts, namely Cy~~~-Leu-Ala-Asp~~-Gly-Ser-Cys~' and Cysg3-Pro, provide a second binding motif, re~pectively.~ Interestingly, the latter motif contains aspartate residue (number 48) in a position normally for cysteine (4,15,16).5 These, together with the quantitative iron and amino acid composition analyses indicating the presence of 7-8 mol of Fe/mol of protein, suggested that the Sulfolobus Fd is probably of the dicluster-type (14,15,17). However, the types of the FeS centers in the purified protein remain to be characterized.
In Sulfolobus, Fd is known to serve as an electron acceptor of a 2-oxoacid:Fd oxidoreductase (17), which is distinct from the well known 2-oxoacid dehydrogenase complexes from mitochondria and most of aerobic respiratory prokaryotes in many respects (12,18). This enzyme has been purified recently from Sulfolobus sp. strain 74 and appears to be similar to those from Halobacterium salinarium (f. halobium) (19)(20)(21)(22). Since the Halobacterial Fd is a monocluster plant-type functioning as a single electron carrier (23-261, while the Sulfolobus protein has been proposed as a dicluster bacterial-type (15,171, it is of interest to study whether both FeS clusters of the latter protein participate in electron transfer from 2-oxoacid:Fd oxidoreductase. In this study, we have carried out the preliminary spectroscopic and cyclic voltammetric characterization of Fd from Sulfolobus sp. strain 7, and show that it is in fact a 7Fe-containing dicluster-type. In addition, evidence is presented that only the [3Fe-4S]'+~0 center (-280 mV uersus normal hydrogen electrode) of the Sulfolobus Fd is reducible during the steadystate turnover of the purified 2-oxoacid:Fd oxidoreductase. The functional significance of the [3Fe-4S] cluster in the archaeal dicluster-type Fd is demonstrated for the first time, and its evolutionary implication is discussed.

EXPERIMENTAL PROCEDURES
Materials-DEAE-Sephacel, Superose-6, and Sephadex G-50 were purchased from Pharmacia LKB Biotechnology Inc, hydroxylapatite HTP from Bio-Rad, and HW-55C and phenyl-Toyopearl 650M from Tosoh Corp., respectively. Tobramycin was purchased from Sigma, and MES, PIPES, and HEPES were from Nakarai Tesque. Water was purified by the Milli-Q purification system (Millipore). Other chemicals mentioned in this study were of analytical grade.
Protein Preparation-Sulfolobus sp. strain 7' was cultivated aerobically and chemoheterotrophically at pH 2.5-3 and 75-80 "C as described previously (27,281, and was harvested in the late exponential phase of growth and stored at -80 "C until use. The cells were suspended i n 40 mM Tris-C1 buffer, pH 7.5, containing 0.5 mM phenylmethanesulfonyl fluoride and 1 mM EDTA and disrupted with a French press (Otake Works, Tokyo, Japan) at 1500 kg/cm2 twice, and the membrane fraction was removed by ultracentrifugation with a Beckman 45Ti rotor at 130,000 x g for 100 min at 15 "C; the supernatant thus obtained contained negligible amounts of cytochrome and was used as the cytosolic fraction.
A bacterial-type Fd present in the cytoplasm was purified by a DEAE-Sephacel (Pharmacia Biotech Inc.), a hydrophobic HW-55C (Tosoh Corp.), and a Sephadex G-50 gel filtration column chromatography to an electrophoretically homogeneous state, by following the patterns on polyacrylamide gel electrophoresis in the presence of SDS and the absorption at 280, 408, and 450 nm of each fraction a t different purification steps as described elsewhere (141, and was stored at -80 "C. These modifications of the original procedure by Kerscher et al. (17), used for the purification of S. acidocaldarius Fd, provide a reproducibly crystallizable Fd; the purified Fd gave a single band on 20% analytical gel electrophoresis and had a purity index (A4,,8/A,8,,) of 0.70 (14). Approximately 3 0 4 0 mg of purified material could be routinely obtained from about 150 g (wet weight) of the cells.
Purification of a 2-oxoacid:Fd oxidoreductase was also carried out also from the cytoplasmic fraction of Sulfolobus sp. strain 7, as will be described e l~e w h e r e .~ I n summary, the enzyme was purified by conventional column chromatographies (using DEAE-Sephacel (Pharmacia centrifugation, and the purified enzyme was stored a t -80 "C until use. Approximately 6-8 mg of pure enzyme could be obtained from about 100 g (wet weight) of the cells.
Measurement of Enzymatic Actzuity-The 2-oxoacid:Fd oxidoreductase activities were monitored with a horse heart cytochrome c reduction assay using the Sulfolobus purified dicluster Fd as a n intermediate electron acceptor (171, except that the reaction was initiated by addition of the purified enzyme. The assay was performed a t 50 "C in 10 mM potassium phosphate buffer, pH 6.8. Analytical Methods-Absorption spectra were recorded with a Hitachi U-3210 spectrophotometer equipped with a thermoelectric cell holder. EPR measurements were carried out using a JEOL JEX-RE1X spectrometer equipped with a n Air Products model LTR-3 Heli-Tran cryostat system, in which temperature was monitored with a Scientific Instruments series 5500 temperature indicator/controller. Spin concentrations were estimated by double integration, with Cu-EDTA as a standard (29).
Electrochemical measurements were made using a Huso Electrochemical System potentiostat model 315A. DC cyclic voltammetry was carried out with the extensively degassed protein solution, typically 1 ml of 70-100 p~ Fd from Sulfolobus sp. strain 7, contained in a capped all-glass cell at room temperature and under continuous flow of 0,-free N, gas, with a three-electrode configuration as described by Armstrong et al. (30). The pyrolytic graphite "edge" electrode was polished and cleaned before measurements. An aminoglycoside, tobramycin (Sigma), was added as aliquots from 100 mM stock solutions (adjusted at pH 7.0) to promote electrochemical response of the protein, which was buffered by a mixture of MES, PIPES, and HEPES (each 5 mM) in 100 mM NaCl. All redox potentials quoted in this paper are uersus normal hydrogen electrode.
Protein was measured by the BCA assay (Pierce) with bovine serum albumin as a standard and by dividing the results by 1.48 for calibration (see "Results and Discussion"). Metal content analysis was carried out by inductively coupled plasma atomic emission spectrometry with a Seiko SPS 1500 VR instrument.
Polyacrylamide gel electrophoresis in the presence of SDS was carried out according to Laemmli (31) on 13 or 15% gels after treating proteins with 2% SDS in the presence or absence of 2% 2-mercaptoethanol a t 90 "C for 5 min, and proteins were visualized by Coomassie Brilliant Blue staining.

Optical and EPR Characterization of the Sulfolobus
Ferredoxin-A dicluster-type ferredoxin is present in a large quantity in chemoheterotrophically grown Sulfolobus sp. strain 7. It has been reported as an [8Fe-8Sl type in the case of S. acidocaldarius strain DSM 639 (171, although no detailed study was carried out for the assignment of the cluster type. Fd purified from Sulfolobus sp. strain 7 has an apparent molecular weight of 12,000 on a calibrated size exclusion fast protein liquid chromatography in the presence of 1.0 M NaCl, in agreement with the previous reports (14,17). However, the average iron content of the purified materials (4.73 mol of Fe/ mol of Fd) on the basis of the BCA protein assay and inductively coupled plasma atomic emission spectrometry were unexpectedly low, and the subsequent quantitative analysis indicated the overestimation of the amount of protein by BCA assay method with BSA (bovine serum albumin) as a standard, as reported by others (17); this has been also experienced in other FeS proteins, including the bacterial-type Fds (16,(32)(33)(34)(35). We therefore used the protein concentration by the BCA assay method using BSA as a standard and by dividing the result by 1.48 for calibration throughout this study. The presence of two FeS centers in the Sulfolobus Fd has been indicated from its deduced primary ~e q u e n c e ,~ and the preliminary cyclic voltammetric analysis in the presence of 2 mM tobramycin as a promoter as described below. In the following studies, we used the pH value of 6.8 throughout the experiments, because the value is very close to the physiological pH of Sulfolobus (44). prepared; truce A). Anaerobic addition of excess sodium dithionite in 40 m M potassium phosphate buffer (pH 6.8) resulted in -18.5-20% reduction of the 408 nm absorbance (Fig. 1, truce  B ) . When the reduction of Fd by addition of excess sodium dithionite was carried out anaerobically in 100 m M Tris-C1 buffer (pH 9) over 40 min, further bleaching of the 408 nm band to -30% of the original absorbance was observed (Fig. 1, truce C). While further reduction of FeS cluster could not be achieved under the experimental conditions, the extent of the reduction of the 408 nm absorbance (30%) clearly indicates that the archaeal Fd is still in the partially reduced state, since most bacterial-type Fds show the bleaching of 400-nm shoulder to -40% or more in the fully reduced state. Thus, these properties are very similar to the dicluster-type Fds with at least one low potential FeS cluster (cf. Ref. 8). Fig. 2 shows the effect of temperature on the X-band EPR spectra of the air-oxidized form of the as-isolated Sulfolobus Fd. It elicits EPR signals, which give rise to a sharp peak at g = 2.03, a broad trough around g = 1.94 with an extended tail toward the high field, and a distinct shoulder at g = 1.98 (at 11 K or below; truces A and B ) . This spectrum could not be observed clearly at temperatures above 15 K (truce C), indicating the extremely rapid spin relaxation of this center. These EPR properties are very similar to those of the [3Fe-4SI1+ cluster of the dicluster Fds from ! l ! thermophilus and A. vinelundii (7, 8, 36), and quantitation of the signal using Cu-EDTA as a standard under the non-saturating conditions resulted 0.9-1.1 spin/ mol Fd. No other EPR signal could be detected in the lower field region under the conditions (data not shown). These data clearly suggest the presence of at least one [3Fe-4SI1+ cluster in the Sulfolobus Fd.  the [3Fe-4Sl cluster was reduced under the conditions, where -20% reduction of the 408 nm absorbance occurred (Fig. 1,  truce B ) . Further anaerobic reduction ofthe purified Sulfolobus Fd at pH 9, where -30% reduction of the 408 nm absorbance occurred (Fig. 1, truce C), resulted in a slight alternation of the overall EPR line shapes in addition to the appearance of several new signals ( g = 2.09, 2.03, 2.01, 1.99, 1.92, and 1.86; Fig.   3, truce C). The complex spectrum, though not very clearly, resembles those of highly reduced 7Fe Fds (7,8,30,34,36) exhibiting a spin-spin interaction between two FeS clusters. Quantitation of the remaining S = 1/2 resonance resulted only 0.04 spidmol Fd, indicating that the extremely low potential [4Fe-4S] cluster is probably diamagnetic (in the S = 0 ground state) even under the conditions (see below).
Cyclic Voltummetry- Fig. 4 shows the preliminary cyclic voltammogram of the dicluster-type Fd from Sulfolobus sp. strain 7 in the presence of 2 mM tobramycin as a promoter. Well defined quasi-reversible waves (not observed in buffer alone) are obtained only in the presence of an aminoglycoside, tobramycin, as an electrode promoter (typically 1-3 mM), as in the cases of the 7Fe Fds from Azotobacter chroococcum 130) and Desulfovibrio africanus (Fd 111) (37-391, though even in the presence of aminoglycoside, the Sulfolobus Fd shows a poorer electrode response than the cases of these bacterial Fds, and thus slightly higher concentrations of protein were required.
The parameters obtained from Fig. 4 are: E,,, = -280 2 10 mV, and the cathodic-to-anodic peak potential separation AEp (scan rate 40 mV.s", pH 7.0) = 60-70 mV for couple " A ; E,, = -530 2 10 mV, and AEp (scan rate 40 mV-s", pH 7.0) = 60-70 mV for couple "B"; and E,, = -690 2 10 mV, and AEp (scan rate 40 mV.s-', pH 7.0) = 40 mV, for couple "C," respectively. Couples A and B are of similar intensity to each other, and have the reproducible and almost constant AEp values (60-70 mV) over the scan-rate range 3-200 m V d . In addition, a linear relationship between the anodic-peak current and the square root of the potential scan rate at least up to 100 m V 4 suggests that these electrode processes are quasi-reversible and effectively diffusion-controlled (data not shown). Couple C has a slightly smaller intensity than those of couples A and B and has almost constant A E p values (30-40 mV) up to 100 mV.s-', where the electrode process is quasi-reversible and effectively diffusion-controlled, in contrast to the case of D. africanus Fd I11 (37). On the other hand, all of these couples (especially couple C) show the increments of AEp and of the intensity of both cathodic and anodic waves at scan rates above 100 mV.s" (especially a t 200-400 mV.s"), indicating the occurrences of increased relative current contributions from adsorbed protein under these conditions. Thus, the preliminary cyclic voltammetric studies suggest the presence of two reducible centers with E,, values at pH 6.8-7.0 of -280 2 10 mV and -530 & 10 mV, at room temperature (Table I). A further reduction process is observed at E,,, values at pH 6.8-7.0 of -690 2 10 mV, which can be assigned to be the further two-electron reduction process of the one-electron reduced [3Fe-4SIo cluster like the cases of other 7Fe Fds (30, 37, 39, 40) (in the cases ofA. vinelandii and D. africanus Fds (39,40), the products have been assumed to be a strong base, which binds protons possibly at the bridging sulfide ligands). These data, in conjunction with the optical and EPR studies, suggest that the Sulfolobus dicluster Fd is in fact a 7Fe type similar to those from A. vinelandii (40421, T thermophilus (8,16,36), and Streptomyces griseus (34,431, in that it has one each of the dithionite-reducible [3Fe-4S11+,0 (-280 mV) and the dithionite-unreducible extremely low potential [4Fe-4S12+~'+ clusters (-530 mV). The midpoint redox potentials of the FeS centers of the Sulfolobus Fd resemble most closely to those of the corresponding centers of T thermophilus Fd (36) (see Table   I). The midpoint redox potentials of the [4Fe-4S12+J+ clusters of these 7Fe Fds are considerably lower, therefore requiring either the photochemical (6, 34) or direct electrochemical reduction method (30,37)  Functional Importance of the 13Fe-4SI Center of the Sulfolobus Ferredoxin-Kerscher et al. (17) previously reported a preliminary survey indicating the presence of a 2-oxoacid:Fd oxidoreductase activity in thermoacidophilic archaea including S. acidocaldarius and Thermoplasma acidophilum. This activity was recently purified from Sulfolobus sp. strain 7 t o an apparent homogeneity using the cognate Fd as an electron a~c e p t o r .~ As in the case ofH. salinarium pyruvate:Fd oxidoreductase (20,21), the purified Sulfolobus enzyme consisted of two non-identical subunits and exhibited a broad substrate specificity toward 2-oxoacid (with the specific activities of -69 unitdmg in the case of 2-oxoglutarate (K, = 870 PM), and of -39 units/mg in the case of pyruvate (K, = 250 PM), respectively); the details of the molecular properties of the Sulfolobus enzyme will be reported e l~e w h e r e .~ Using the purified 2-oxoacid:Fd oxidoreductase and the dicluster Fd of Sulfolobus, an in vitro 2-oxoacid-dependent Fdreducing system was constructed to test whether one or both of the FeS clusters of the Fd were reduced during the steady-state turnover of the enzyme. This system consists of 56 VM Fd and 8.6 pg of the purified enzyme in 20 mM potassium phosphate buffer, pH 6.8 (the value close to the reported intracellular pH of Sulfolobus; Ref. 441, in the presence of 2 mM 2-oxoglutarate and 100 PM coenzyme A. It should be mentioned that Fd is not reoxidized under the anaerobic conditions due to the absence of any electron acceptor in the in vitro system, whereas oxygen reoxidizes Fd when tested aerobically. On the other hand, the archaeal 2-oxoacid:Fd oxidoreductase catalyzes the steadystate turnover in either conditions, due to the presence of excess amounts of both Fd (56 PM) and 2-oxoglutarate (2 mM) over enzyme in the system. One typical result is shown in Fig. 5, which indicates the maximal reduction of the 408 nm absorbance of the dicluster Fd to be -19-20% under the experimental conditions even when tested anaerobically at 50 "C. The enzymatically reduced spectra of Fd were neither altered at least for 6 h a t 50-60 "C, nor affected by further addition of the enzyme or substrates (data not shown). Furthermore, when the reaction was performed anaerobically in an EPR tube at 56 "C for 5 h, the resulting spectrum at 10 K was completely EPR silent in the g = 2 region, indicating that the [3Fe-4S] cluster was almost fully reduced under the conditions (data not shown). These data, together with the spectroscopic properties of the Sulfolobus Fd (see Fig. 1 and 31, clearly suggest that only the dithionite-reducible [3Fe-4S1 cluster is reduced during the steady-state turnover of the enzyme, while the bulk of the * Monocluster-type L4Fe-4SI Fd; -, not reported.

I .50
Enzymatic Reduction of Sulfolobus Fd at 50 "C extremely low potential [4Fe-4S] cluster remains in the oxidized state. This is thermodynamically in agreement with the preliminary EPR analysis of the Sulfolobus 2-oxoacid:Fd oxidoreductase that suggested the presence of a low potential [4Fe-4S12+.'+ cluster only partially reducible by dithionite at pH 6.8 in the purified e n~y m e .~ Therefore, it seems most likely that the [3Fe-4S]1+~0 center of the Sulfolobus dicluster Fd functions as a single electron carrier in vivo, as in the case of H. salinarium monocluster planttype Fd (23, 24, 26). To our knowledge, this is the first report that clearly shows the functional importance of a [3Fe-4Sl'+~0 center of a dicluster-type Fd, as established in the succinate dehydrogenase and fumarate reductase complexes (9)(10)(11). In addition, the role of the low potential [4Fe-4S]2+z1+ cluster of the Fd is most likely a structural rather than a functional one, as in the case of, e.g., Bacillus subtilis glutamine phosphoribosylpyrophosphate amidotransferase (45) 48, and references therein) proposed a possible evolutionary scheme of the bacterial-type Fds on the basis of their comparative studies on the three-dimensional (three-dimensional) structures of various bacterial-type Fds, as illustrated in Fig. 6 (top). In their model, early dicluster Fds have been derived from a putative common ancestor of bacterial-type Fds with a single FeS cluster as a result of a gene duplication event (4); the polypeptide backbone structures of such dicluster-type Fds might be similar to those of the 8Fe-type dicluster Fds (designated as the "clostridial-type"), in that they exhibit a marked two-fold symmetry as examplified by Peptococcus aerogenes Fd (49). The 7Fe-type dicluster Fds ("Azotobacter-type") have been also derived from early dicluster-type Fds, but have a distorted two-fold symmetry due to the insertions of several loops. On the other hand, the three-dimensional structures of the "monocluster-type" Fds known so far exhibit a remarkable pseudo-two-fold symmetry in spite of the presence of only a single [3Fe-4S] or [4Fe-4S] cluster at the position corresponding to the "cluster I" site of the dicluster-type Fds; Fukuyama et al. (48) therefore proposed that this might be due to the loss of the [4Fe-4S] cluster at the "cluster 11" site and the subsequent insertion of an a-helix structure and a disulfide bond into the corresponding site for stabilization of the protein. Thus, cluster I, which corresponds to a [3Fe-4S] center in Desulfovibrio gigas and Azotobacter-type Fds, is strictly conserved in all bacterial-type Fds, while the cluster I1 is missing (replaced by an a-helix) in the monocluster-type Fds (Fig. 6).
The primary structure of the dicluster Fd of Sulfolobus sp. strain 7 indicated the overall homology (87% identity) to that of S. acidocaldarius Fd The alignment of the primary structures of several Azotobacter-type Fds with that of the Sulfolobus Fd indicated the strict conservation of 7 cysteine residues5 (see Fig. 6, bottom), all of which have been assigned as ligands to the FeS clusters at least in the case ofA. vinelandii Fd I (41,42). These data indicate that the [3Fe-4Sl'+~o cluster of the Sulfolobus Fd corresponds to the cluster I, and the low potential [4Fe-4S]2+.1+ cluster to the cluster 11, respectively (Fig. 6). This is very intriguing in conjunction with the reduction of the Sulfolobus Fd enzymatically with 2-oxoacid, coenzyme A, and the archaeal 2-oxoacid:Fd oxidoreductase (Fig. 5). In the case of the Sulfolobus Fd, only the cluster I plays a redox role, which is conserved in all bacterial-type Fds. On the other hand, the cluster I1 (the low potential [4Fe-4SI2+~'+ center) apparently plays only a structural role, and is depleted in the cases of the monocluster-type Fds (Fig. 6, bottom). In this connection, we have also noted that an extent of the reduction of the clostridial 8Fe Fds was only partial when tested enzymatically with hydrogen and hydrogenase (see Refs. 46 and 47, and references therein), leaving a possibility that even the cluster I1 of several clostridial-type 8Fe Fds may not be involved in the electron transfer in vivo. Of course, these are not in disagreement with the ability of the dicluster-type Fds to act as a two-electron carrier in vitro (see Ref. 47 ., A z . vi., Ds. gi., and Ba. th.; from top to bottom) were prepared by the MOLSCMPT program using the coordinates deposited in the Protein Data Bank and are shown so that the cluster I is on the left of each model, and the cluster I1 of the diclustertype Fds on the left, respectively; shaded a-helices indicate those absent in the dicluster-type Fds. For details, see Refs. 4 and 48 and "Evolutionary Implication." Below: Amino acid sequence comparison of several bacterialtype Fds. Only conserved residues and those replaced by others are shown, otherwise indicated by X and a gap (-). Possible ligand residues to the cluster I (boxed) are conserved in all bacterial-type Fds, except for the N-terminal second cysteine residues which are replaced by aspartate Sulfolobus 7Fe Fd is also capable of undergoing two-electron reduction in a reversible manner i n vitro as indicated by the cyclic voltammetric analysis (Fig. 4).

Alignment of possible ligand residues
These data indicate the following hypotheses on the evolution of the bacterial-type Fds from their dicluster prototype. (i) The cluster I has been conserved because of the functional importance. (ii) The cluster I1 may have lost its original redox role in the early stage of the evolution; the presence of a single [4Fe-4S] cluster in the Fd of the hyperthermophilic archaeon Pyrococcus furiosus (33, 50, 51) (see also Fig. 6 , bottom) indicates that this might have taken place even before the divergence of the Archaea and Bacteria domains (52,53). (iii) This possibly allowed the replacement of the cluster I1 by structural polypeptides in some Fds in the later stage (cf. Fig. 6 and Ref.

4).
If these are indeed the cases, the cluster I1 of the diclustertype Fds may possibly represent an "evolutionary relic," since one can easily speculate that the extremely low potential nature of cluster I1 (cf Table I) is functionally no more necessary after adaptation to the oxidative atmosphere on the ancient earth during the evolution.
Although our hypotheses require more discriminations for further proofs, it should be added that the presence of such evolutionary relics has also been reported for certain chromophores of the bacterial photosynthetic reaction centers (54) and the heme group of succinate dehydrogenase complexes (11),3 and presumably such prosthetic groups may be found in some other redox proteins.