Evidence for the Head Domain Movement of the Rieske Iron-Sulfur Protein in Electron Transfer Reaction of the Cytochromebc 1 Complex*

The three-dimensional structure of the mitochondrial cytochrome bc 1 complex suggests that movement of the extramembrane domain (head) of the Rieske iron-sulfur protein (ISP) may play an important role in electron transfer. Such movement requires flexibility in the neck region of ISP, since the head and transmembrane domains of the protein are rather rigid. To test this hypothesis, Rhodobacter sphaeroidesmutants expressing His-tagged cytochrome bc 1complexes with cysteine substitution at various positions in the ISP neck (residues 39–48) were generated and characterized. The mutants with a single cysteine substitution at Ala42 or Val44 and a double cysteine substitution at Val44 and Ala46 (VQA-CQC) or at Ala42 and Ala46 (ADVQA-CDVQC) have photosynthetic growth rates comparable with that of complement cells. Chromatophore membrane and intracytoplasmic membrane (ICM) prepared from these mutants have cytochrome bc 1complex activity similar to that in the complement membranes, indicating that flexibility of the neck region of ISP was not affected by these cysteine substitutions. Mutants with a double cysteine substitution at Ala42 and Val44 (ADV-CDC) or at Pro40 and Ala42 (PSA-CSC) have a retarded (50%) or no photosynthetic growth rate, respectively. The ADV-CDC or PSA-CSC mutant ICM contains 20 or 0% of the cytochromebc 1 complex activity found in the complement ICM. However, activity can be restored by the treatment with β-mercaptoethanol (β-ME). The restored activity is diminished upon removal of β-ME but is retained if the β-ME-treated membrane is treated with the sulfhydryl reagent N-ethylmaleimide orp-chloromercuribenzoic acid. These results indicate that the loss of bc 1 complex activity in the ADV-CDC or PSA-CSC mutant membranes is due to disulfide bond formation, which increases the rigidity of ISP neck and, in turn, decreases the mobility of the head domain. Using the conditions developed for the isolation of His-tagged complement cytochrome bc 1 complex, a two-subunit complex (cytochromes b andc 1) is obtained from all of the double cysteine-substituted mutants. This suggests that introduction of two cysteines in the neck region of ISP weakens the interactions between cytochromes b, ISP, and subunit IV.

The cytochrome bc 1 complex (ubiquinol-cytochrome c reductase) is an essential segment of the energy-conserving electron transfer chains of mitochondria and many respiratory and photosynthetic bacteria (1). This complex catalyzes electron transfer from ubiquinol to cytochrome c and concomitantly translocates protons across the membrane to generate a membrane potential and pH gradient for ATP synthesis. Although the cytochrome bc 1 complexes from different sources vary in their polypeptide compositions, they all contain four redox prosthetic groups: two b-type cytochromes (b 566 or b L and b 562 or b H ), one c-type cytochrome (cytochrome c 1 ), and one high potential Rieske iron-sulfur cluster [2Fe-2S]. 1 The proton-motive Q cycle model (2) has been favored for electron transfer and proton translocation in the complex. The key feature of this model is the presence of two separate ubiquinone-or ubiquinol-binding sites: a ubiquinol oxidation site near the P side of the inner mitochondrial membrane and a ubiquinone reduction site near the N side of the membrane.
Recently, the cytochrome bc 1 complex from beef heart mitochondria was crystallized and its three-dimensional structure solved at 2.9-Å resolution (3,4). The structural information obtained not only answered a number of questions concerning the arrangement of the redox centers, transmembrane helices, and inhibitor binding sites but also suggested movement of an extramembrane domain within the iron-sulfur protein (ISP) during electron transfer (4). This suggestion arose from observation of an uneven electron density in the I4 1 22 crystal data of native bovine cytochrome bc 1 complex. A particularly low electron density area is observed in the intermembrane space portion of the complex, where the extramembrane domains of ISP and cytochrome c 1 reside (4). This movement hypothesis was further supported by the finding that the position of the ironsulfur cluster in the complex is affected by ubiquinol oxidation site inhibitor binding (5,6,8) and by the crystal form (7,9).
The anomalous light scattering signal of the [2Fe-2S] cluster is enhanced in co-crystals with stigmatellin or UHDBT but is diminished in the co-crystal with (E)-methyl-3-methoxy-2-(4Јtrans-stilbenyl) acrylate (5). Thus, binding of stigmatellin or UHDBT arrests the movement of the extramembrane domain of ISP, fixing the iron-sulfur cluster 27 Å from heme b L and 31 Å from heme c 1 (referred to as the "fixed state" of ISP), the same position it occupies in the I4 1 22 crystal of native bovine cytochrome bc 1 complex. The position of the iron-sulfur cluster changes from the fixed state to somewhere closer to heme c 1 (referred to as the "released state" of ISP) upon (E)-methyl-3-methoxy-2-(4Ј-trans-stilbenyl) acrylate binding. The recent report of Iwata et al. (9), showing the iron-sulfur cluster at two different positions in two crystal forms, further supports the presence of a variable position of in the "released state" of the iron-sulfur cluster.
Movement of the head domain of ISP during electron transfer in cytochrome bc 1 complex can be explained as follows. The [2Fe-2S] cluster is reduced by the first electron of ubiquinol at a position 27 Å from heme b L and 31 Å from heme c 1 (ISP in "fixed state"). Since a reduced [2Fe-2S] cluster cannot donate an electron to cytochrome c 1 before the second electron of ubiquinol is transferred to heme b L , it was postulated that either the change of the ubiquinone binding position during reduction of b L or the electron transfer from b L to b H causes a conformational change in cytochrome b, which forces or allows reduced [2Fe-2S] to move close enough to heme c 1 (ISP in "released state") for electron transfer (5,7). This model would also explain why ubisemiquinone, a more powerful reductant than ubiquinol, reduces b L , but not the [2Fe-2S] cluster, during ubiquinol oxidation.
ISP has three domains: the membrane-spanning N-terminal domain consisting of residues 1-62 (tail), the soluble C-terminal extramembrane domain consisting of residues 73-196 (head), and the flexible linking domain comprising residues 63-72 (neck). ISP is associated with the complex primarily via the membrane-spanning N-terminal domain (4,7,9). The [2Fe-2S] cluster is located at the tip of the head domain (12,13). Since the three-dimensional structures of the head and tail domains are the same in the fixed and released states, movement of the head domain of ISP in the bc 1 complex requires flexibility in the neck region.
If movement of the head domain of ISP is required for bc 1 catalysis and the neck region of ISP confers the required mobility, decreasing the flexibility of the neck region of ISP should affect bc 1 complex activity. This hypothesis can be tested by site-directed mutagenesis followed by biochemical and biophysical characterization of mutant expressing cytochrome bc 1 complexes with altered ISP necks. However, site-directed mutagenesis in bovine heart mitochondria is not practical. R. sphaeroides is an ideal system to study the neck region of ISP by molecular genetics approach. The four-subunit complex is functionally analogous to the mitochondrial enzyme; the largest three subunits are homologous to their mitochondrial counterparts; and this system is readily manipulated genetically. In addition, the recent generation of R. sphaeroides expressing His 6 -tagged cytochrome bc 1 complex greatly speeds up the preparation of the bc 1 complex from wild-type or mutant cells (14).
The R. sphaeroides ISP neck is composed of residues 39 -48 (corresponding to residues 63-72 of bovine ISP) with a sequence of NPSADVQALA. Ala 42 , Asp 43 , Val 44 , Ala 46 , and Ala 48 are the conserved amino acid residues. We have previously generated mutants with increased ISP neck rigidity by double or triple proline substitution of the conserved residues. The results demonstrated that flexibility in the ISP neck is important in bc 1 catalysis (14). The ALA-PLP and ADV-PPP mutant membranes have, respectively, 30 and 0% of the cytochrome bc 1 complex activity found in the complement membrane. The ALA-PLP mutant complex has a larger activation energy than the wild-type complex, suggesting that movement of the head domain decreases the activation energy barrier of the bc 1 complex.
To better define the structure and function relationship of the neck region of ISP in the bc 1 complex, we recently generated mutants expressing His 6 -tagged bc 1 complex with single or double cysteine substitution at various positions in the ISP neck. We predict that formation of a disulfide bond between a pair of genetically engineered cysteines in the neck will decrease its flexibility and thus decrease electron transfer activity. Measuring the bc 1 activity in these cysteine-substituted mutants should give insight into the dynamic state of the ISP neck. Herein we report procedures for generating R. sphaeroides mutants expressing His 6 -tagged cytochrome bc 1 complexes with altered ISP necks by introducing single cysteines or a pair of cysteines at different positions. The photosynthetic growth behavior, the cytochrome bc 1 complex activity, and the EPR characteristics of the Rieske [2Fe-2S] cluster in membranes and the purified state from complement and mutant strains are examined and compared. The effect of sulfhydryl reagents on cytochrome bc 1 complexes from complement and mutant membranes is also examined. The method for construction of ISP mutants is essentially the same as previously reported by Tian et al. (14). The ADVQA(42-46)-CDVQC mutant was constructed by annealing the oligonucleotide primer with single-stranded pSELNB3503 carrying a A42C mutation in ISP. The presence of engineered mutations were confirmed by DNA sequencing before and after photosynthetic or semiaerobic growth of the cells. Expression plasmid pRKDfbcF m BC H Q was purified from an aliquot of a photosynthetic or semiaerobic culture using the Qiagen plasmid Mini Prep kit. Since R. sphaeroides cells contain four types of endogenous plasmids, the isolated plasmids are not pure and concentrated enough for direct sequencing. Thus, a 1.2-kilobase pair DNA segment containing the mutation sequence was amplified from the isolated plasmids by polymerase chain reaction and purified by 1% agarose gel electrophoresis. The 1.2-kilobase pair polymerase chain reaction product was recovered from the gel by a gel extraction kit from Qiagen.
Growth of Bacteria-Escherichia coli was grown at 37°C in an enriched medium (TYP) in order to shorten growth time and increase plasmid yield (15). For photosynthetic growth of the plasmid-bearing R. sphaeroides BC17 cells, an enriched Sistrom's medium containing 5 mM glutamate and 0.2% casamino acids was used. The pH of the medium was adjusted to 7.1 with a mixture of 6 N NaOH and 2 N KOH to increase the sodium and potassium ion content of the medium to a more optimal level (16). Photosynthetic growth condition of R. sphaeroides was essentially as described previously (14); cells harboring mutated fbc genes on the pRKDfbcFBCQ plasmid were grown photosynthetically for one or two serial passages to minimize any pressure for reversion. For semiaerobic growth of R. sphaeroides, an enriched Sistrom's medium supplemented with 20 amino acids and extra rich vitamins was used. These semiaerobic cultures were grown in 0.5 liters of enriched medium in 2-liter Bellco flasks with vigorous shaking (220 rpm) for 26 h. The inoculation volumes used for both photosynthetic and semiaerobic cultures were always at least 5% of the total volume. Antibiotics were added to the following concentrations: ampicillin (125 mg/liter), tetracycline (10 mg/liter for E. coli and 1 mg/liter for R. sphaeroides), kanamycin sulfate (30 mg/liter for E. coli and 20 mg/liter for R. sphaeroides), trimethoprim (87.5 mg/liter for E. coli).
Enzyme Preparations and Activity Assays-Cells were harvested, washed, and passed twice through a French pressure cell at 846 p.s.i. as described previously (17). The chromatophore fraction was pelleted by ultracentrifugation of broken cells in a Beckman Ti 50.2 rotor at 48,000 rpm for 2 h at 4°C. Membrane was washed with 50 mM Tris-Cl, pH 8.0, containing 1 mM MgSO 4 and stored at Ϫ80°C in the presence of 20% glycerol. The His 6 -tagged cytochrome bc 1 complexes were purified from chromatophores by the method of Tian et al. (14). His 6 tag is located at the C terminus of the cytochrome c 1 subunit.
Ubiquinol-cytochrome c reductase activity was measured at 23°C in a 1-ml assay mixture containing 100 mM sodium/potassium phosphate buffer, pH 7.4, 0.3 mM EDTA, 100 M cytochrome c, 25 M 2,3-dimethoxy-5-methyl-6(10-bromodecyl)-1,4-benzoquinol, and an appropriate amount of membrane or purified cytochrome bc 1 complex. Chromatophores or ICM were diluted with 50 mM Tris-Cl, pH 8.0, containing 20% glycerol and 1 mM MgSO 4 to a final concentration of cytochrome b of 5 M. No detergent was added to the diluted mixture in order to preserve the bc 1 activity. 5 l of diluted membrane was added to the assay mixture. Activity was determined by measuring the reduction of cytochrome c (the absorbance increase at 550 nm), using a millimolar extinction coefficient of 18.5 cm Ϫ1 mM Ϫ1 . Nonenzymatic oxidation of 2,3-dimethoxy-5-methyl-6(10-bromodecyl)-1,4-benzoquinol, determined under the same conditions in the absence of enzyme, was subtracted. The purified bc 1 complex activity assay is essentially as described previously (14).
Treatment of Membranes with ␤-ME-␤-ME was added to membrane preparations (40 M cytochrome b) to a final concentration of 100 M. After incubation on ice for 10 min, aliquots were removed from the mixture for assay. ␤-ME was removed from the treated membrane by washing with 50 mM Tris-Cl, pH 8.0, and 1 mM MgSO 4 and centrifuging at 80,000 rpm for 30 min with a Beckman TL ultracentrifuge. This process was repeated three times.
Reaction of Genetically Engineered Cysteine Residues in Membranes with NEM or PCMB-Freshly prepared NEM (100 mM in H 2 O) was added to the ␤-ME-pretreated chromatophore or ICM (concentration is at 40 M cytochrome b) to a final concentration of 500 M. The mixture was flushed briefly with nitrogen, sealed, and incubated at room temperature for 15 min before measuring cytochrome bc 1 complex activity. The stock solution of PCMB was prepared by first dissolving PCMB powder in 0.5 N NaOH followed by neutralization. Modification of cysteine with PCMB was carried out as described for NEM. Excess NEM and PCMB was removed by repeated washing and centrifuging, and enzymatic activity was redetermined.
Molecular Modeling-Molecular modeling was carried out in a Indigo II Silicon Graphics Station. A peptide with NCSCQVQALA sequence, corresponding to the ISP neck sequence of the R. sphaeroides PSA-CSC mutant, was built using the Builder Module from Insight II software from Molecular Simulation, Inc. Whether this peptide has acceptable geometry for disulfide bond formation was examined by minimizing the structure as follows: (a) continued iterations using steepest descents minimization until the maximum derivation was less than 10 kcal/Å; (b) continued iterations using conjugate minimization until the maximum derivation was less than 1.0 kcal/Å; (c) continued iterations using va09a minimization until the maximum derivation was less than 0.01 kcal/Å; (d) creation of a disulfide bond; (e) minimization of the peptide using the Optimize command in Builder; and (f) visual examination of geometry and comparison with a control (no disulfide bond) using the print energy per residue command.
Other Biochemical and Biophysical Techniques-Protein concentration was determined by the method of Lowry et al. (18). Cytochrome b (19) and cytochrome c 1 (20) were determined according to published methods. SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was performed according to Laemmli (21) using a Bio-Rad Mini-protean dual slab vertical cell. Western blotting was performed using rabbit polyclonal antibodies against cytochrome b, cytochrome c 1 , ISP, and subunit IV of the R. sphaeroides bc 1 complex. The polypeptides sepa-rated in the SDS-PAGE gel were transferred to polyvinylidene difluoride membrane for immunoblotting. Goat anti-rabbit IgG conjugated to alkaline phosphatase or protein A conjugated to horseradish peroxidase was used as the second antibody.
EPR spectra were recorded with a Bruker ER 200D apparatus equipped with a liquid N 2 Dewar at 77 K. Instrument settings are detailed in the figure legends.

Characterization of the Mutants Carrying Cysteine Substitutions in the Neck Region of the Iron-Sulfur Protein-Six R.
sphaeroides mutants expressing His 6 -tagged cytochrome bc 1 complexes with single or double cysteine substitutions at various positions in the ISP neck region were generated to test the hypothesis that neck flexibility allows ISP head domain movement required for bc 1 catalysis. The flexibility of the neck should decrease when a disulfide bond is formed between a pair of substituted cysteines.
Of the six mutants, two, A42C and V44C, are single substitutions, in which Ala-42 or Val-44 is replaced with cysteine. The other four are double cysteine substitutions, PSA-CSC, ADV-CDC, VQA-CQC, and ADVQA-CDVQC, in which Pro 40 and Ala 42 , Ala 42 and Val 44 , Val 44 and Ala 46 , or Ala 42 and Ala 46 are replaced with cysteines. A plate mating technique was used (22) to transfer the pRKDfbcF m BC 6H Q plasmid from E. coli S17 to R. sphaeroides BC17. The mating took place in less than 16 h on the LB/SIS plates. R. sphaeroides BC17 cells harboring pRKDfbcF m BC 6H Q plasmid were selected by spreading the conjugated cell mixture on enriched Sistrom's plate containing tetracycline and kanamycin sulfate. It took 4 days for the A42C, V44C, ADVQA-CDVQC, and VQA-CQC mutant colonies to show up on the plate, the same time period as that required for complement colonies. However, it took about 7 days for the ADV-CDC and PSA-CSC mutant colonies to appear. This slower growth rate on the plates is an indication of zero or reduced bc 1 activity in the virtual absence of environmental selection pressure. A similar phenomenon was observed with several cytochrome b mutants that had no bc 1 activity. 2 When mid-log phase, aerobically grown complement and mutant cells were inoculated into enriched Sistrom medium and subjected to anaerobic photosynthetic growth conditions, the A42C, V44C, VQA-CQC, and ADVQA-CDVQC mutants grow at a rate comparable with that of complement cells, the ADV-CDC mutant has a retarded (50%) growth rate, and PSA-CSC does not grow photosynthetically (Table I)  complex activity found in complement chromatophores. ICM from the PSA-CSC mutant have no ubiquinol-cytochrome c reductase activity. This was expected, since bc 1 complex is required for photosynthetic growth and this mutant does not grow photosynthetically.
To determine whether the loss (or decrease) of the cytochrome bc 1 complex activity in the mutant membranes results from a lack of or improper assembly of ISP protein in the membrane, the amount of ISP and its EPR characteristics in mutant and complement membranes were compared. Western blot analysis with antibodies against R. sphaeroides cytochrome b, cytochrome c 1 , ISP, and subunit IV revealed that the amount of these four subunits in the six mutant membranes is the same as that in the complement membrane (Fig. 1, lanes  2-8). Absorption spectral analysis shows that the content of cytochrome b and c 1 /c 2 in all of these mutant membranes is the same as that in complement membrane. These results indicate that the mutations did not affect the assembly of ISP protein into the membrane. The [2Fe-2S] cluster in all of these mutant membranes has an EPR spectrum identical to that observed in complement chromatophores, with resonance at g x ϭ 1.80 and g y ϭ 1.9 (Fig. 2). Thus the mutations did not change the microenvironment of the iron-sulfur cluster.
Effect of Mutation on the Disulfide Bond Formation in the Neck Region of the Iron-Sulfur Protein-The bc 1 complex activity decreased by 23, 32, 14, and 100% in the VQA-CQC, ADV-CDC, ADVQA-CDVQC, and PSA-CSC mutant membranes, respectively, with no change in the amount of ISP incorporated into the membrane and the EPR characteristics of the [2Fe-2S] cluster. Did this activity loss result from increased neck region rigidity due to the formation of a disulfide bond? We addressed this question by comparing cytochrome bc 1 complex activity in mutant and complement membranes with and without ␤-ME treatment. Also, we used molecular modeling to examine the feasibility of disulfide bond formation in the PSA-CSC mutant.
When ICM from the PSA-CSC mutant, which has no bc 1 activity, was treated with ␤-ME, about 60% of the activity found in complement ICM was restored (Table II). It should be emphasized that the observed activity restoration is not due to nonenzymatic reduction of cytochrome c by ␤-ME, because less than 0.05 M ␤-ME is present in the assay mixture, and the restored activity is sensitive to antimycin. This restored cytochrome bc 1 complex activity is diminished when ␤-ME is re-moved by repeated centrifugation and suspension. However, the restored bc 1 activity is retained if the ␤-ME-treated membrane is reacted with sulfhydryl modifying reagents, such as NEM or PCMB, at 5-fold molar excess to ␤-ME, followed by repeated centrifugation and suspension (Table II). Activity restoration is not observed when the PSA-CSC mutant membrane is treated with NEM or PCMB without prior reduction with  (Cyt c 1 ), ISP, and subunit IV (Sub IV). After transfer, membrane was cut into two pieces so that the upper part contained proteins with a molecular mass greater than 23 kDa, including cytochromes b and c 1 , and the bottom part contained the low molecular mass proteins, including ISP and subunit IV. Since antibodies against R. sphaeroides cytochrome bc 1 complex available in our laboratory have antibody titers for cytochromes b and c 1 much higher than those for ISP and subunit IV, the top section of the membrane was developed with a horseradish peroxidase system, whereas the bottom section was developed with an alkaline phosphatase system. Lane 1, prestained molecular mass standards; lane 2, the complement chromatophores; lanes 3-7, the A42C, V44C, VQA-CQC, ADV-CDC, and ADVQA-CDVQC chromatophores; lane 8, the PSA-CSC ICM; lanes 9 -12, effluent from the Ni 2ϩ -NTA column, which is the unbound portion from detergent (1% DM)-solubilized membrane fraction of the VQA-CQC, ADV-CDC, PSA-CSC, and ADVQA-CDVQC mutants.

FIG. 2. EPR spectra of the [2Fe-2S] cluster of the Rieske iron sulfur protein in mutant and complement membranes.
Chromatophore membranes from the A42C, V44C, VQA-CQC, ADV-CDC, and ADVQA-CDVQC mutants and complement cells and ICM from the PSA-CSC mutant were incubated with a small amount of ascorbate on ice for about 10 min and frozen in liquid nitrogen. EPR spectra were recorded at 77 K with the following instrument settings: microwave frequency, 9.37 Hz; microwave power, 20 milliwatts; modulation amplitudes, 20 G; modulation frequency, 100 kHz; time constant, 0.1 s; scan rate, 20 G/s. ␤-ME. These results indicate that the complete lack of the bc 1 activity in the PSA-CSC mutant membrane results from decreased mobility of the ISP head domain, due to a neck region made rigid by the formation of a disulfide bond between Cys 40 and Cys 42 . The addition of ␤-ME reduces this disulfide bond, restoring flexibility to the neck region. Modification of ␤-ME released SH-groups with NEM or PCMB has no effect on the flexibility of the neck region but blocks reformation of the disulfide bond, thus guaranteeing retention of the restored activity.
Formation of a disulfide bond between Cys 40 and Cys 42 in the ISP neck of the PSA-CSC mutant is also supported by molecular modeling using the peptide sequence of NCSCQVQALA, corresponding to the neck sequence of the PSA-CSC mutant. In the bovine bc 1 structure (4), the distance between C-␤ atoms of Ala 64 and Ala 66 (corresponding to residues 40 and 42 of R. spheroides ISP) is 5.2 Å, and the side chains of the two residues point in the same direction relative to the backbone of this stretch. Molecular modeling, based upon this structural information, indicates that a disulfide bond can be formed. This would result in an 11-membered ring structure. A distance of 5.2 Å for disulfide bond formation is slightly longer than the 4.1 Å observed for the two C-␤ atoms of the disulfide bond, between Cys 160 and Cys 144 of ISP, in bovine bc 1 crystals (4).
The possibility that the lack of bc 1 activity in the PSA-CSC mutant ICM is due to the formation of a intermonomer disulfide bond between one of the two cysteines at positions 40 or 42 and a cysteine in the other monomer is ruled out because bc 1 complex activity in the A42C mutant chromatophore or ICM is the same as that in complement membranes and is not activated by the addition of ␤-ME. Furthermore, the distance between the two cysteines at position 40 and 40Ј in ISP from different monomers appears to be over 33 Å, as estimated from the distance between Ala 64 of one ISP monomer and Ala 64Ј in the symmetrically related monomer in the bovine bc 1 structure. Intermonomer disulfide bonding over such a long distance is impossible.
When the VQA-CQC and ADVQA-CDVQA mutant chromatophores, which have, respectively, 77 and 86% of the cytochrome bc 1 complex activity of the complement chromatophore, were treated with ␤-ME, the bc 1 activities were unchanged (Table II). Also, when ␤-ME-pretreated mutant chromatophores were reacted with NEM or PCMB, no change in activity was observed (Table II). These results proved that no disulfide bonds are formed between the cysteines at positions 44 and 46 or positions 42 and 46. The lack of disulfide bond formation in the ISP neck region of these two mutant chromatophores is also supported by the bovine bc 1 crystal structure, which shows that the distances between C-␤ atoms of Val 68 and Ala 70 of ISP (Val 44 and Ala 46 in R. sphaeroides) and between C-␤ atoms of Ala 66 and Ala 70 (Ala 42 and Ala 46 in R. sphaeroides) are 7.35 and 12 Å, respectively. These exceed the distance that permits disulfide bond formation. Since no disulfide bond is formed, the flexibility of the ISP neck should not change in these mutant chromatophores. Thus, the decreased bc 1 activity in these two mutant chromatophores cannot be attributed to decreased mobility of the head domain of ISP.
The finding that the ADV-CDC mutant ICM has only 20% as much bc 1 complex activity as does its chromatophore membrane is rather surprising (Table I), because bc 1 complex activities in chromatophore and ICM are comparable in all of the other cysteine-substituted mutants and in complement cells (Table I). The addition of ␤-ME to ADV-CDC mutant membrane has no effect on bc 1 complex activity in chromatophores, while increasing the activity in ICM to the level observed in untreated chromatophores (Table II). The restored activity in the mutant ICM is diminished by removal of ␤-ME. The addition of 5-fold molar excess NEM or PCMB to the ␤-ME-treated mutant ICM preserves the restored activity in the membrane. These results indicate that the loss of bc 1 activity in the ADV-CDC mutant ICM, compared with that in its chromatophore, results from the decreased mobility of the head domain of ISP, due to formation of a disulfide bond between Cys 42 and Cys 44 in the neck region of ISP. However, a similar explanation does not account for the lower (32% less than complement) bc 1 complex activity in ADV-CDC mutant chromatophore or in ␤-MEtreated ICM, because neither contain disulfide bonds. Apparently, some other effect of cysteine substitution decreased the bc 1 activity.
It should be noted that the loss of bc 1 complex activity in the ADV-CDC mutant ICM is not due to the formation of an intermonomer disulfide bond, because the distance between potential cysteine pairs in the two monomers is over 30 Å. Recall that the bc 1 activity in the single cysteine substitution mutants A42C and V44C was not affected by the addition of ␤-ME. In the bovine bc 1 complex structure, the distance between C-␤ atoms of Ala 66 and Val 68 (corresponding to residues Ala 42 and Val 44 in R. sphaeroides) is 7 Å, and the orientations of the side chains are opposite each other. However, since the neck region is flexible, a different crystal form might result in a different conformation. This comparison is at the "noise" level of homologous sequence similarities, e.g. bovine versus R. sphaeroides. Therefore, formation of a disulfide bond between Cys 42 and Cys 44 in ADV-CDC mutant ICM is not impossible.
Since disulfide bonds occur in ICM, but not in the chromatophore, of the ADV-CDC mutant, probably disulfide bond formation in this mutant ISP neck region depends on the oxygen supply in the cell culture and is related to the functionality of the bc 1 complex in the membranes. When these mutant cells are grown in the dark with an ample oxygen supply, no bc 1 activity is detected in the cytoplasmic membrane due to the formation of a disulfide bond in the ISP neck region. The addition of ␤-ME to the cytoplasmic membrane can restore bc 1 activity to the level observed in the chromatophore membrane. This is consistent with the fact that bc 1 complex is not required for dark aerobic growth of this organism because the cells can utilize quinol oxidases. When the O 2 supply is limited, the cells adjust their protein expression to prepare for photosynthetic growth, as indicated by the appearance of chlorophyll. The bc 1 activity in ICM increases slightly (0.3 mol of cytochrome c reduced per min per nmol of b; Table I), because either some part of the bc 1 complexes become active by disulfide bond cleavage or the newly synthesized complexes stay in a conformation that prevents disulfide bond formation between Cys 42 and Cys 44 . When oxygen is completely depleted from the culture, the cells undergo photosynthetic growth and require active bc 1 complex. Apparently, no disulfide bonds are formed in the mutant chromatophores, and bc 1 activity increases greatly (1.5 mol of cytochrome c reduced per min per nmol of b; Table I). 1 Complex-Although chromatophores from mutants with a double cysteine mutation in the ISP neck region (VQA-CQC, ADV-CDC, and ADVQA-CDVQC) have ISP amounts, EPR properties, and ISP neck flexibility similar to those in complement chromatophores, the bc 1 activities are 23, 32, and 14% less, respectively. Is this activity loss due to a decreased binding affinity of ISP to the whole complex? Assuming that ISP binding is sensitive to detergent treatment, any difference in subunit composition of purified His-tagged bc 1 complexes from mutant and complement strains will indicate detergent lability and binding affinity differences between normal and mutant ISPs.

Effect of Mutation on Binding Affinity of the Iron-Sulfur Protein in the bc
When chromatophores from these three mutant cells, VQA-CQC, ADV-CDC, and ADVQA-CDVQC, at a cytochrome b concentration of 25 M were mixed with various amounts of dodecylmaltoside (up to 0.44 mg/nmol of cytochrome b), the cytochrome bc 1 complex activities decreased as the concentration of dodecylmaltoside (DM) increased (Fig. 3A). More than 95% of the bc 1 complex activity in all of these mutant chromatophores was inactivated by 0.44 mg of DM/nmol of cytochrome b. Under identical conditions, complement chromatophores did not lose activity (Fig. 3A), indicating that the mutant cytochrome bc 1 complexes (two-cysteine substitution) are more labile to detergent.
Since DM solubilizes the bc 1 complex from R. sphaeroides chromatophores, it is important to establish that DM denaturation of the bc 1 complex in mutant chromatophores is due to ISP dissociation from the complex and not due to a decreased affinity for membrane. This is done by centrifuging DM-treated chromatophores from mutants and complement cells at 100,000 ϫ g for 30 min to separate the solubilized from the unsolubilized fractions and measuring cytochrome content, ISP amount, and bc 1 activity in both fractions. Since the amount of cytochrome bc 1 complex protein (cytochromes b and c 1 , ISP, and subunit IV) in the supernatant fractions of the three mutant chromatophores is comparable with that in the supernatant fraction from complement chromatophores at a given concentration of detergent, it is apparent that DM solubilization of the cytochrome bc 1 complex is not affected by mutation (data not shown). However, the specific activity of cytochrome bc 1 complex, based on b content in the supernatant fractions from mutant chromatophores, decreased as the detergent concentration increased (Fig. 3B). Under identical conditions, the bc 1 activity in supernatant fractions from complement chromatophores does not drastically decrease. The extraction efficiency, based on cytochrome b content, is the same for all mutants and complement membranes (Fig. 3C). To confirm that the loss of bc 1 activity in the DM-solubilized fractions resulted from dissociation of ISP from the complex, the supernatant fractions, extracted by 0.5 or 1% of DM, were passed through a Ni 2ϩ -NTA column. The column effluents and eluates were examined for bc 1 complex subunit composition. Two subunits, corresponding to cytochromes b and c 1 , were found in the Ni 2ϩ -NTA column eluates (Fig. 4, lanes 5-7). ISP and subunit IV were detected in the column effluents by Western blot analysis (Fig. 1, lanes 9 -11). On the other hand, four subunits (cytochrome b, cytochrome c 1 , ISP, and subunit IV) in unit stoichiometry are in the Ni 2ϩ -NTA column eluates from detergent-solubilized complement chromatophores (Fig. 4, lane  2). These results indicate that a double cysteine substitution in the ISP neck region decreases the binding affinity of ISP for the bc 1 complex. Therefore, detergents dissociate ISP from the complex. Since no ISP is found in the Ni 2ϩ -NTA column eluates from the DM (0.5%)-solubilized mutant chromatophore fractions having 40% of the untreated cytochrome bc 1 complex activity, the activity detected in this fraction reflects the ability of dissociated ISP to reconstitute into a functionally active bc 1 complex after dilution of detergent in the assay mixture. Perhaps the activity decrease in untreated mutant chromatophores indicates the extent of ISP dissociation from the complex. However, the 14% decrease in bc 1 complex activity in the A42C or V44C mutant chromatophores is probably due to a deviation in the activity assay and not to a mutational effect, since their activities are not labile to detergents. Indeed, all four subunits of the bc 1 complex are absorbed on the Ni 2ϩ -NTA column after the detergent solubilization (Fig. 4, lanes 2 and 3). The purified complexes from these two mutants have specific activities comparable with that of the complement (Table I).
Although the restored bc 1 complex activity in the PSA-CSC mutant ICM treated with ␤-ME is more labile to DM treatment than are the bc 1 activities in VQA-CQC, ADV-CDC, and AD-VQA-CDVQC mutant chromatophores (Fig. 3), the effectiveness of dodecylmaltoside in solubilizing bc 1 complex from PSA-CSC mutant ICM and the bc 1 complex subunit composition of the Ni 2ϩ -NTA column effluents and eluates (Fig. 1, lane 11, and Fig. 4, lane 8) are the same as that for the three double cysteine-substituted mutant chromatophores. Probably the binding affinity of ISP for the bc 1 complex in PSA-CSC is weaker than in the other double cysteine mutant chromatophores. Thus, ISP is more easily dissociated from the complex by detergent. Since only 60% of the bc 1 activity found in complement chromatophore is restored to the PSA-CSC mutant ICM upon the addition of ␤-ME, it is likely that 40% of the ISP in this mutant ICM is dissociated from the bc 1 complex.
The ability of detergent-dissociated ISP to reconstitute into functionally active bc 1 complex depends on the detergent concentration in the reconstituting mixture (or in the assay mixture) and the status of the dissociated ISP. It has been reported (23) that including reducing reagents, such as dithiothreitol or ␤-ME, during isolation of ISP from the complex protects free ISP from oxidation by oxygen and that the addition of phospholipid to the detergent-containing solution reduces effective detergent concentration. Therefore, we examined the effect of ␤-ME and phospholipid on the reassociation of ISP into functionally active bc 1 complex in the DM-solubilized chromatophore fraction from the ADV-CDC mutant cells. As described above, ␤-ME has no effect on the bc 1 activity in this mutant chromatophore or on the solubilization efficiency of DM. ␤-ME preserves the activity of mutant ISP after its dissociation from bc 1 complex. In the presence of ␤-ME, when the mutant chromatophore is treated with 1.2% dodecylmaltoside, the cytochrome bc 1 complex activity in the supernatant fraction is 10-fold higher than that in the supernatant fraction obtained from mutant chromatophore without pretreatment with ␤-ME. The activity of the supernatant with ␤-ME is about 50% of that in the chromatophore before detergent solubilization. When the detergent-solubilized fraction obtained from the ␤-ME-treated mutant chromatophore is incubated with phospholipid (10 mg/ ml), the cytochrome bc 1 complex activity increases to 95% of that in the untreated chromatophore. Without prior addition of ␤-ME, the supernatant has minimal activity (2%), which only increases slightly after the addition of phospholipid. These results indicate that the addition of ␤-ME to the ADV-CDC chromatophore prior to detergent solubilization prevents denaturation of dissociated ISP and subunit IV and thus enables them to reconstitute with cytochromes b and c 1 to form functionally active bc 1 complex after excess detergent is removed by phospholipid and dilution with the assay mixture.