Redox Components of Cytochrome bc-type Enzymes in Acidophilic Prokaryotes

The Rieske proteins of two phylogenetically distant acidophilic organisms, i.e. the proteobacteriumThiobacillus ferrooxidans and the crenarchaeonSulfolobus acidocaldarius, were studied by EPR. Redox titrations at a range of pH values showed that the Rieske centers of both organisms are characterized by redox midpoint potential-versus-pH curves featuring a common pK value of 6.2. This pK value is significantly more acidic (by almost 2 pH units) than that of Rieske proteins in neutrophilic species. The orientations of the Rieske center’s g tensors with respect to the plane of the membrane were studied between pH 4 and 8 using partially ordered samples. At pH 4, theSulfolobus Rieske cluster was found in the “typical” orientation of chemically reduced Rieske centers, whereas this orientation changed significantly on going toward high pH values. TheThiobacillus protein, by contrast, appeared to be in the “standard” orientation at both low and high pH values. The results are discussed with respect to the molecular parameters conveying acid resistance and in light of the recently demonstrated long-range conformational movement of the Rieske protein during enzyme turnover in cytochrome bc 1 complexes.

The cytochrome bc complex is the only energy-conserving enzyme that is common to photosynthetic and respiratory electron transport systems. The respective complex has been studied for several decades in mitochondria and proteobacteria (bc 1 complex) (1,2) as well as in chloroplasts and cyanobacteria (b 6 f complex) (3). The functional core of the enzyme was found to be made up of three subunits containing two b-type hemes (b H and b L ; cytochrome b or b 6 ), a c-type heme (cytochrome c 1 or f), and a [2Fe-2S] cluster called the Rieske center.
Recently, crystal structures of mitochondrial cytochrome bc 1 complexes from various sources were reported (4 -6). These structures found the soluble domain of the Rieske protein in substantially differing positions, suggesting conformational flexibility of this subunit within the complex. The conformational flexibility seen in the structures together with independent evidence (7,8) 1 indicated a long-range conformational movement of the extramembrane domain of the Rieske protein as an essential step in enzyme turnover. This movement appears to swing the [2Fe-2S] cluster from a position close to heme b L toward another one close to heme c 1 , i.e. promoting electron transfer from the quinone in the Q p site 2 to cytochrome c 1 . Such a domain movement represents a hitherto unknown mechanism for bridging long electron transfer distances between redox centers.
Cytochrome bc-type complexes appear to be spread over the entire phylogenetic tree of bacteria since, in addition to proteoand cyanobacteria, the enzyme was found to be present in green sulfur bacteria (10 -12), in green filamentous bacteria (13), in Deinococci (14) and in Firmicutes (15)(16)(17)(18)(19). The Rieske proteins in the cytochrome bc complexes from all the abovementioned species are rather similar with respect to EPR spectroscopic and redox properties despite the fact that significant variability is observed concerning the remaining subunits of the complex. In all cases, the Rieske cluster is characterized by a typical g av ϭ 1.91 EPR spectrum; sensitivity of the spectral features (especially in the region of g x ) to the presence of inhibitors and the redox state of the quinone (3, 16 -18, 20); and a dependence of redox midpoint potential on pH, indicating pK values of ϳ8 and 9 -10 on redox-linked deprotonatable groups in the oxidized form of the cluster (12,17,(21)(22)(23).
In addition to the mentioned eubacterial species, Rieske proteins have been found in the thermo-and acidophilic archaeon Sulfolobus acidocaldarius. In Sulfolobus, cytochrome b and a Rieske protein are present in a supercomplex with components of a cytochrome oxidase-related enzyme (24). The Rieske protein contained in this fused cytochrome bc-cytochrome oxidase supercomplex has subsequently been characterized in detail (purification, sequence, and E m versus pH) (25). Intriguingly, the Sulfolobus Rieske cluster falls out of the pattern of characteristics observed on the cluster in bacteria. The EPR spectrum was reported to be insensitive to the presence of inhibitors; its redox potential does not correspond to those of the other mentioned systems; and the pK values of the E m -versus-pH curve are shifted by almost 2 pH units toward the acidic region. This raises the question of whether the unusual properties of the Sulfolobus Rieske protein are due to the large phylogenetic distance of its parent organism (perhaps indicating a significantly different mode of functioning of the whole enzyme) or whether all or part of these particularities represent adaptation to the acidophilic growth conditions. Fur-thermore, since a cytochrome c subunit appears to be absent in Sulfolobus (25), the structural organization of the Rieske protein in the complex may well differ from that of typical cytochrome bc 1 or b 6 f complexes.
To address the ensemble of these questions, we have carried out a comparative study of the Rieske protein's electrochemical and structural characteristics in two phylogenetically distant acidophilic species, i.e. the crenarchaeon S. acidocaldarius and the proteobacterium Thiobacillus ferrooxidans. Both organisms strive at pH values of ϳ2 and use molecular oxygen as a terminal electron acceptor in (otherwise rather different) respiratory chains (for reviews, see Refs. 25 and 26). The presence of a cytochrome bc complex in T. ferrooxidans has recently been demonstrated (27), and the enzyme has been characterized in detail in the accompanying article (50). T. ferrooxidans belongs to the ␤-subgroup of proteobacteria (28), and its cytochrome bc complex can therefore be expected to be phylogenetically closely related to the bc 1 complex of purple bacteria, of Paracoccus, or of mitochondria.

EXPERIMENTAL PROCEDURES
Bacterial growth and isolation of membrane fragments from T. ferrooxidans were carried out according to Elbehti and Lemesle-Meunier (27). S. acidocaldarius (DSM 639) was grown and harvested, and the membranes were prepared as described (29).
To obtain partially ordered membrane multilayers at various pH values, membrane fragments from T. ferrooxidans and S. acidocaldarius were first washed in 50 mM sodium acetate (pH 4), 2 mM EDTA, and 5 mM ascorbate or in 50 mM Tricine (pH 8), 2 mM EDTA, and 5 mM ascorbate. The pellet was subsequently resuspended in unbuffered water at pH 4, 7, or 8 and resedimented by ultracentrifugation for 1 h at 300, 000 ϫ g. Oriented membrane multilayers were obtained as described by Rutherford and Sétif (30). The membranes were resuspended in unbuffered water at pH 4, 7, or 8; applied to sheets of Mylar; and dried in a humidity-controlled atmosphere, under argon, for ϳ72 h at 4°C.
Redox titrations of the Rieske cluster from T. ferrooxidans were performed at 15°C as described by Dutton (31) at several pH values (between pH 4 and 8.1) using 50 mM sodium acetate, 30 mM MES, 30 mM MOPS, 10 mM potassium phosphate, and 30 mM Tricine for titrating at pH 4, 6, 6.7, 7.4, and 8.1, respectively. pH values were controlled at the beginning and end of each redox titration. The following redox mediators were used: benzoquinone, potassium ferricyanide, ferrocenemonocarboxylic acid, and 1,1Ј-ferrocenedicarboxylic acid (all at 100 M). Reductive titrations were carried out using sodium dithionite, and oxidative titrations were done using potassium hexachloroiridate. No hysteresis was observed.
EPR spectra were recorded at liquid helium temperatures with a Bruker ESP 300e X-band spectrometer fitted with an Oxford Instruments cryostat and temperature control system. All chemicals used were reagent-grade. Fig. 1 shows a comparison of EPR spectra obtained at neutral pH on membrane fragments from T. ferrooxidans and S. acidocaldarius in the ascorbate-reduced state. Both spectra were characteristic for the [2Fe-2S] Rieske cluster with its typical g y line at g ϳ 1.9 and a broad g x trough in the region of g ϭ 1.80 to 1.70. As shown in the accompanying article (50), the shape and position of the g x trough of the Thiobacillus Rieske center were sensitive to the redox state of the quinone and to the presence of inhibitors, resembling the effects observed on many other Rieske clusters from cytochrome bc complexes (3,12). By contrast, the Sulfolobus Rieske protein appeared to be unaffected by the presence of the tested inhibitors (data not shown), as already reported previously (32). As mentioned in the Introduction, a salient particularity of the Sulfolobus Rieske cluster is the close to 2 pH units-downshifted pK value of its E m -versus-pH dependence (32). Since respective data were so far missing for the Thiobacillus Rieske center, we have studied the pH dependence of the Thiobacillus protein.

Electrochemical Properties of the T. ferrooxidans Rieske
Center-A series of redox titrations in the region between pH 4 and 8.1 monitoring the intensity of the g y EPR signal of the Rieske center as a function of ambient redox potential was carried out on isolated cytoplasmic membranes from T. ferrooxidans. All titration data obtained could be fitted with simple n ϭ 1 Nernst curves. Fig. 2 shows the redox midpoint potential as a function of pH. The determined data points yielded a pK value of 6.2 and an E m value of ϩ490 mV in the pH-independent region. Above the pK, the redox potential was seen to decrease with a slope of about Ϫ60 mV/pH unit. This behavior corresponded well (apart from a difference of 90 mV in absolute values of the redox potential) with that of the Sulfolobus Rieske cluster (32).
Orientation Properties of the T. ferrooxidans and S. acidocaldarius Rieske g Tensors-To obtain structural information, the Rieske clusters from both species were studied in partially oriented membrane multilayers. In general, studies on oriented membrane multilayers are performed at roughly neutral pH values. Due to the acidophilic character of both species, however, orientations were performed at several pH values in the range from pH 4 to 8. Good two-dimensional orientations (as judged by the anisotropy of EPR signals from all observable paramagnetic species in the membrane) were obtained at all pH value with membranes from Sulfolobus and at high pH values with membranes from Thiobacillus. Unexpectedly, at pH ϳ4, only very mediocre anisotropies were observed upon orienting Thiobacillus membranes. Since this effect was observed systematically in all of several attempts to produce oriented samples from Thiobacillus at low pH, we concluded that this failure arises from intrinsic properties of the Thiobacillus membranes at low pH values (see also "Discussion"). Fig. 3 shows EPR spectra of partially ordered membrane multilayers from T. ferrooxidans and S. acidocaldarius at pH 8 (panel a) and pH 4 (panel b) recorded parallel (solid lines) and perpendicular (dashed lines) to the membrane plane with the magnetic field. Anisotropic g y and g x signals of both species could easily be discerned. The g z signal was obscured by a strong signal at g ϭ 2 arising from radical-type paramagnetic species in the membranes.
Substantial differences were observed both (i) between the low and high pH experiments in Sulfolobus (cf. Fig. 3, a versus b) and (ii) between Sulfolobus and Thiobacillus at high pH values (Fig. 3a). (i) At pH 4, the Sulfolobus g y signal was slightly more intense when the magnetic field was parallel to the membrane, whereas at pH 8, a significantly stronger signal was observed with the field perpendicular to the membrane. (ii) Thiobacillus at pH 8, by contrast, showed a stronger g y signal parallel to the membrane. These differences indicate differing conformations of the Rieske proteins as a function of both species and pH value. The full dependence of the Sulfolobus Rieske center's g y signal amplitudes versus angle is shown in the polar plots of Fig. 4a. At pH 4, the polar plot of the g y direction was characterized by a pronounced maximum parallel to the membrane and a broad, badly defined maximum close to 90°. The g x trough (Fig. 4b) yielded only a single, well defined maximum perpendicular to the membrane. g y and g x directions parallel and perpendicular, respectively, to the membrane correspond to what has been seen in cytochrome bc 1 complexes from mitochondria (33) and purple bacteria (34) and in cyto-chrome bc-type complexes from Chlorobium limicola (12) and Chloroflexus aurantiacus. 3 At pH 8, drastically different polar plots for the g y direction of the Sulfolobus Rieske center were obtained (Fig. 4a). The perpendicular component had become largely dominant at pH 8, whereas the parallel component (although probably still present) could not be discerned anymore. The g x direction still showed a single maximum perpendicular to the membrane. It is of note, however, that the relative intensity of the g x trough as compared with the g y line significantly decreased on going from pH 4 to 8 (cf. Fig. 3, b versus a). Since a corresponding decrease was observed for the parallel component of the g y line, the perpendicular g x trough and the parallel g y line must be attributed to the same paramagnetic species. The intensity of the g x trough corresponding to centers pointing their g y direction at steep angles with respect to the membrane was obviously below detection (see "Discussion").
A different behavior was observed for the Thiobacillus Rieske center at pH 8. The polar plots of g y and g x showed clear maxima parallel and perpendicular to the plane of the membrane, respectively, i.e. corresponding to the typical orientation of neutrophilic bc 1 complexes measured at neutral pH values.
As stated above, oriented samples from Thiobacillus at pH 4 yielded only slightly anisotropic signals for all paramagnetic centers present in the membrane. In the best orientations, a weak anisotropy of the Rieske cluster's g y and g x signals could be observed. The resulting directions were similar to those observed at pH 8, i.e. g y parallel and g x perpendicular to the membrane. This indicates that the orientation pattern of the 3  dominant species in Thiobacillus does not change between pH 4 and 8. The low quality of orientation at pH 4, however, does not allow us to decide whether conformational heterogeneity is present at this pH (as observed for the Sulfolobus Rieske protein).

Electrochemical Properties
As shown in Fig. 2, the redox midpoint potential of the T. ferrooxidans Rieske center remains independent of pH up to ϳ6 at E m ϭ ϩ490 mV and decreases with an apparent slope of Ϫ60 mV/pH unit above this pK value. Apart from a shift of 90 mV (E m,low pH ϭ ϩ400 mV for Sulfolobus) (32), the pH dependence of the Thiobacillus Rieske center thus strongly resembles that of the protein from Sulfolobus (Fig. 5) (32). In the titration of the isolated Rieske protein from Sulfolobus, a second pK value at pH 8.5 (32) has been determined. The examination of the Thiobacillus protein for the presence of a second pK value at higher pH was not feasible by experiments performed on membrane fragments.
All Rieske centers studied previously in bacteria, mitochondria, and chloroplasts 4 fall into one of two distinct classes ( Fig.   5) with respect to electrochemical behavior: (a) the ubiquinone/ plastoquinone group with an E m value of about ϩ300 mV and (b) the menaquinone group with a redox potential of about ϩ120 mV in the pH-independent regions. In contrast to this difference in absolute potential, the global shape of the E mversus-pH curve and especially the pK value of close to 8 appear to be common to all ubiquinone-, plastoquinone-, and menaquinone-oxidizing Rieske centers studied so far. In a few selected systems, the presence of a second pK value at significantly higher pH has been demonstrated (22,17,35) or suggested (12) on the basis of the detailed shape of the E m -versus-pH curve in the pH-dependent region, analogous to the results obtained with the Sulfolobus Rieske protein. These two pK values have tentatively been attributed (17,35) to the N ⑀ protons on the two histidine ligands to the cluster. Although this interpretation appears sensible, no experimental confirmation has been presented so far. Since the two pK values are roughly similar in ubiquinone/plastoquinone and menaquinone systems, they apparently are not involved in controlling the absolute redox midpoint potentials in the different species (in contrast to what had been proposed previously). The respective redox potentials in the ubiquinone/plastoquinone and menaquinone groups were recently shown to be predominantly adjusted by the presence or absence of hydrogen-bonding interactions between a specific amino acid side chain and an acid-labile sulfur as well as one of the cysteine sulfur ligands of the [2Fe-2S] cluster (36,37). Hydrogen-bonding interactions have previously been proposed as factors strongly influencing E m values of Rieske-type proteins on the basis of electron nuclear double resonance and electron spin echo envelope modulation data (38).
The Rieske centers of T. ferrooxidans and of the archaeon S. acidocaldarius do not belong to either of the two described groups (Fig. 5). Their E m -versus-pH curves showed pK values close to 6.2, i.e. almost 2 pH units lower than that of the above-mentioned species. The strongly similar pH dependences of Rieske proteins in the phylogenetically extremely distant species S. acidocaldarius and T. ferrooxidans dismiss the hypothesis that the low pK in the archaeon could reflect a phenomenon related to the evolution of the Rieske proteins. They rather argue strongly in favor of the downshifted pK being an adaptive response to the acid (pH ϳ2) solvent environment of the periplasmic Rieske proteins in the two acidophiles. This raises the question of whether (a) the observed pK shift is required to assure the functional mechanism of the enzyme or whether (b) it simply confers stability to the Rieske protein at pH 2.
Is the First pK Value Involved in the Functional Mechanism?-At first sight, a functional implication of the pK value of 8 in the neutrophilic species appears unlikely since this value is at least 1 pH unit above the highest pH reached during turnover of the cytochrome bc complexes. This gap goes up to 4 pH units for the studied acidophilic species (pK of ϳ6 versus medium pH of ϳ2). It is of note, however, that the reported pK values were determined under equilibrium conditions. We have observed that the presence of the inhibitor 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone in the Q p site of the plastidic b 6 f complex downshifts the pK value of the Rieske protein's E mversus-pK curve by 2 pH units (39). Such a pK shift makes perfect sense in light of the x-ray structures of the mitochon- 4 In a recent study of the soluble Rieske fragment from spinach chloroplasts, Zhang et al. (47) observed differing pK values and differing absolute E m values when using optical (pK ϳ6.5) or EPR (no pH dependence in the neutral region) spectroscopy. The EPR data are consistent with the earlier study on thylakoids (23). Zhang et al. rationalized this discrepancy by invoking temperature-dependent pK and E m values. This explanation appears unlikely to us for the following reasons. (a) None of the other Rieske proteins examined by optical methods at room temperature showed this kind of temperature effect (22,35,48,49). (b) An E m value obtained by CD at room temperature for the b 6 f complex Rieske protein (42) corresponds well with the EPR titrations and is in conflict with the data reported by Zhang et al. (c) In contrast to what is stated by Zhang et al., a pK of 6.5 is not closer to the value of free histidine than that observed in the other systems. Since the histidine residue serves as ligand to the [2Fe-2S] cluster, the pK value in question corresponds to the second deprotonation reaction. The pK value of this deprotonation in free histidine is far beyond pH 10, and a downshift of this pK value in the Rieske proteins by ϳ1.5 pH units, to our mind, would require substantial structural differences in the vicinity of the histidine ligands. This is, however, not observed (9). We could not help noticing that the optical titration reported by Zhang et al. was carried out in the absence of redox mediators, and we are therefore presently initiating a reexamination of this pK value in the spinach complex using CD spectroscopy at room temperature. (32). UQ, ubiquinone; PQ, plastoquinone; MK, menaquinone. drial bc 1 complexes, which show that the N ⑀ proton on His 181 points into the Q p pocket and is therefore likely to be affected by occupancy of this site. Respective alterations of pK values, e.g. by the presence of reduced quinone in the Q p site, can therefore not be excluded. The dependence of conformational parameters of the Rieske cluster on pH described in this work further indicates that protonation/deprotonation reactions probably play a role in the conformational rearrangement of the Rieske protein during turnover. However, as discussed below, the deprotonatable groups involved in domain movement of the Rieske protein are not necessarily identical to the groups giving rise to the pK values of the E m -versus-pH curve.

FIG. 5. Schematic representation of the E m -versus-pH curves of cytochrome bc-type enzymes studied in species from the domain of bacteria containing ubiquinones, plastoquinones, or menaquinones as quinone pool as well as those measured on T. ferrooxidans (this work) and on the archaeon S. acidocaldarius
Structural Integrity of the Rieske Proteins at Low pH Values-The [2Fe-2S] cluster of the Rieske proteins from neutrophilic purple bacteria and from chloroplasts (see above) is irreversibly lost at pH values below 4. 5 This instability at low pH values is a common property of iron-sulfur clusters due to the presence of the "acid-labile" bridging sulfur atoms. In addition to coping with the lability of these sulfurs, Rieske proteins from acidophilic species have to overcome the problem that the histidines are labile ligands at low pH due to protonation of the ␦-nitrogen of the imidazole moiety. The breakage of the metal histidine bond at acidic pH values has been extensively studied in blue copper proteins (40). To assure the integrity of the [2Fe-2S] cluster in the acidic region, the respective Rieske proteins need to shield their histidine ligands from solvent access. An excellent example how this may be achieved is provided by the blue copper protein rusticyanin from Thiobacillus (41), which caps the solvent-exposed histidine ligand to the copper ion by two Trp residues, thereby lowering the pK values of both N ␦ and N ⑀ by several pH units. 6 Placing the two histidine ligands to the [2Fe-2S] Rieske cluster in a more hydrophobic environment would correspondingly render the N ␦ -Fe bond more stable at low pH values, but also downshift the pK value of the proton on N ⑀ . The observed pK shifts can therefore be considered as being entailed by the stabilization of the imidazole-iron bond.
A major difference between the Sulfolobus Rieske protein and those from neutrophilic eubacterial sources is the insertion of 10 amino acids (compared with the bovine protein) in the case of the SoxF protein and 29 amino acids for the SoxL protein between the two conserved cluster-binding motifs (43). It is tempting to speculate that this stretch folds such that the environment of the histidine ligands is rendered more hydrophobic. The lack of sequence data for the Thiobacillus Rieske protein unfortunately precludes analogous speculations regarding this organism at the present time. Due to the large phylogenetic distance between Sulfolobus and Thiobacillus, we would tend to assume that the strategy employed to achieve acid stability differs between the two species. Detailed structural information on the Thiobacillus and Sulfolobus Rieske proteins can be expected to significantly improve our understanding of the adaptation to acidic conditions. Orientation S. acidocaldarius-As shown in Fig. 4, the orientation of the Sulfolobus Rieske cluster's g tensor with respect to the membrane varies with pH. At low pH values, the majority of Rieske centers point their g y axis roughly parallel to the membrane plane, i.e. similar to the orientation typically observed for neutro-and alkalophilic organisms. A minority of centers, however, were observed with the g y direction at high angles with respect to the membrane. A respective minority fraction was not seen in the cytochrome bc 1 complexes from mitochondria (33), purple bacteria (34), and Gram-positive bacteria (17), but was seen in the bc-type complexes from the green sulfur bacterium C. limicola (12) and C. aurantiacus. 3 It is of note that in all studies on the orientation of the Rieske cluster published so far, the oriented samples have been prepared at roughly neutral pH, i.e. well below the pK values of the E m -versus-pH curve. At high pH values (pH 8), the orientation dependence of the Sulfolobus Rieske g y line changed substantially, and a sizable fraction of the centers appeared to point their g y direction at rather steep angles with respect to the membrane. The most straightforward interpretation of the observed orientation dependences at low and high pH values (Fig. 4) is to assume that at least two differently oriented populations of Rieske centers coexist in these samples. At pH 4, a large majority of centers pointed g y parallel to the membrane, whereas at pH 8, the population directing g y at high angles to the membrane had substantially increased at the expense of the "standard" orientation. The fact that an intermediate orientation dependence was observed at pH 7 (data not shown) corroborates this hypothesis. Concomitantly with the decrease in the parallel g y component, a decrease in the intensity of the (perpendicular) g x signal was observed. A corresponding new g x trough at low angles with respect to the membrane, however, could not be identified. This is probably due to the fact that (a) parallel components are characterized by a much lower intrinsic signal amplitude than perpendicular components (44) and that (b) the g x trough of this conformation of Rieske centers can be expected to be very broad, as was also observed for the purple bacterial complex (9).
In Sulfolobus, two Rieske proteins with somewhat differing primary structures are present (25). Part of the observed heterogeneity may therefore arise from these two distinct proteins. At least one of those, however, must show pH-dependent changes in the orientation of its g tensor. This raises the question of whether the respective pH effects are a unique feature of Sulfolobus Rieske proteins. We have therefore studied the pH dependence of the orientation properties of the Rieske center in a purple bacterium and have observed the same phenomenon, although upshifted by 2 pH values. 7 Thus, the Sulfolobus Rieske protein(s) behave like their purple bacterial counterparts in altering the equilibrium between differing orientations as a function of pH.
The mitochondrial and purple bacterial Rieske proteins were recently discovered to perform a long-range movement in order to transfer an electron from quinol to cytochrome c 1 . The Sulfolobus bc-type complex lacks a c-type heme subunit and rather seems to be fused to its physiological electron acceptor, i.e. cytochrome oxidase. The lack of the c-type heme subunit thus, in principle, raises the question of whether the Sulfolobus Rieske protein displays differences as compared with its eubacterial counterparts. The data shown above indicate similar properties of the Rieske protein from both the archaeon Sulfolobus and mitochondria. This suggests a comparable functional mechanism involving movement of the Rieske center in Sulfolobus, most probably shuttling the electron directly to the binuclear copper center of the oxidase subunits.
It is of note that the apparent pH correlation of the orientation and redox effects may well be coincidental. Orientation by partial dehydration necessarily occurs in a buffer-free medium (since the presence of buffers at typical concentrations hampers orientation), and slight deviations of pH values from the starting conditions during drying cannot be excluded. It is therefore not possible to determine an exact pK value for the orientation effects.
From an inspection of the recently published structures (5,6), the histidine ligands to the [2Fe-2S] cluster are certainly tempting candidates for controlling the association of the Rieske protein with the Q p site. Ding et al. (45) have proposed several possible arrangements and different interactions of the ubiquinone and ubihydroquinone in the Q p site and the resulting interactions with the adjacent [2Fe-2S] Rieske cluster. In particular, the possibility of hydrogen bonding from the ⑀-nitrogen of one or both histidines of the [2Fe-2S] cluster to the carbonyl oxygen of ubiquinone and the phenoxyl of the ubihydroquinone or other occupants of the Q p site has been pointed out. As mentioned above (see "Electrochemical Properties"), occupation of the Q p site by the inhibitor 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone shifts the pK of the Rieske cluster from 8 to ϳ6 in spinach chloroplasts (39), indicating strong interactions of the quinone analog with the residues responsible for the pH dependence of the [2Fe-2S] cluster.
The pK values of these residues, however, concern the oxidized state of the cluster. From the shape of the E m -versus-pH curve, one must conclude that this pK value is much higher in the reduced state of the [2Fe-2S] cluster and that therefore the respective residues are always protonated when the Rieske center is reduced. The EPR experiments discussed above were performed with the Rieske center in the paramagnetic, i.e. reduced, state. We therefore favor a model assuming that the residues responsible for the pH-dependent orientation properties differ from those affecting the pH dependence of the redox midpoint potential.
In an oriented study performed on the cytochrome b 6 f complex from spinach chloroplasts, heterogeneous populations of g tensor orientations of the Rieske cluster have already been observed (20). In this work, several possible rationalizations for this phenomenon have been proposed, including structural heterogeneity. Guided by the observation of differing relaxation properties of the clusters giving rise to the differing orientation dependences, however, a purely physical interpretation was favored. In light of the recently observed conformational mobility of the Rieske protein (4 -8) 1 and the data reported above, we now tend to reinterpret the respective data as in fact arising from conformational heterogeneity, possibly affected by the pH value of the sample. A reexamination of the orientation characteristics of the Rieske protein in the cytochrome b 6 f complex has therefore been initiated.
T. ferrooxidans-Despite several efforts, we did not succeed in producing well oriented multilayers from Thiobacillus cytoplasmic membranes at low pH values (pH ϳ4). At neutral pH values, by contrast, Thiobacillus membranes oriented readily, yielding strongly anisotropic angular dependences of signal amplitudes. We therefore concluded that (a) the failure to obtain oriented samples at low pH arose from sample-inherent properties and that (b) these properties change in a pH-dependent manner, permitting satisfactory orientation at neutral pH values. The existence of permanently charged membrane surfaces at low pH values invoked to explain maintenance of neutral interior pH values even in the resting state of cells (46) is possibly involved in the experimental problems in dehydrating Thiobacillus membranes at acidic pH.
Unexpectedly, at neutral pH (i.e. well above the physiological ambient pH values of Thiobacillus), the orientation of the Rieske cluster's g y value was found to be parallel to the membrane, i.e. the reverse of what was observed for the Sulfolobus cluster and for the Rieske centers in the neutrophilic purple bacterium at comparatively high pH values. Whereas the Sulfolobus Rieske protein with respect to orientation characteris-tics therefore behaves like a standard Rieske protein in other examined organisms (apart from the acidophilicity-induced pH shifts), the Thiobacillus protein stands out as an exception. It is tempting to attribute this unique feature of the Thiobacillus Rieske cluster to the singular growth conditions of this organism. Ferrous iron as sole source of electrons does not provide sufficiently reducing equivalents to enter the chain upstream of the cytochrome bc 1 complex. As discussed in the accompanying article (50), this complex is therefore liable to work in reverse gear, possibly entailing substantial deviations in functional parameters from typical cytochrome bc 1 complexes.