Soluble Variants of Rhodobacter capsulatus Membrane-anchored Cytochrome cy Are Efficient Photosynthetic Electron Carriers*

Photosynthetic (Ps) electron transport pathways often contain multiple electron carriers with overlapping functions. Here we focus on two c-type cytochromes (cyt) in facultative phototrophic bacteria of the Rhodobacter genus: the diffusible cyt c2 and the membrane-anchored cyt cy. In species like R. capsulatus, cyt cy functions in both Ps and respiratory electron transport chains, whereas in other species like R. sphaeroides, it does so only in respiration. The molecular bases of this difference was investigated by producing a soluble variant of cyt cy (S-cy), by fusing genetically the cyt c2 signal sequence to the cyt c domain of cyt cy. This novel electron carrier was unable to support the Ps growth of R. capsulatus. However, strains harboring cyt S-cy regained Ps growth ability by acquiring mutations in its cyt c domain. They produced cyt S-cy variants at amounts comparable with that of cyt c2, and conferred Ps growth. Chemical titration indicated that the redox midpoint potential of cyt S-cy was about 340 mV, similar to that of cyts c2 or cy. Remarkably, electron transfer kinetics from the cyt bc1 complex to the photochemical reaction center (RC) mediated by cyt S-cy was distinct from those seen with the cyt c2 or cyt cy. The kinetics exhibited a pronounced slow phase, suggesting that cyt S-cy interacted with the RC less tightly than cyt c2. Comparison of structural models of cyts c2 and S-cy revealed that several of the amino acid residues implicated in long-range electrostatic interactions promoting binding of cyt c2 to the RC are not conserved in cyt cy, whereas those supporting short-range hydrophobic interactions are conserved. These findings indicated that attaching electron carrier cytochromes to the membrane allowed them to weaken their interactions with their partners so that they could accommodate more rapid multiple turnovers.

Biological energy transduction systems contain multiple electron carrier molecules connecting various chromophore-bearing membrane proteins (1). Often, these electron carriers are freely diffusible in inter-membrane spaces to link efficiently large multisubunit membrane-embedded complexes. For example, Photosystem I and Photosystem II are interconnected via the copper containing plastocyanin in chloroplasts. An iron bearing c-type cytochrome (cyt) 3 plays a similar role in photosynthetic (Ps) microbes under appropriate environmental conditions (2). In the case of the widely studied anoxygenic, purple non-sulfur, facultative phototrophs of the genus Rhodobacter, Ps growth relies on cyclic electron transfer (ET) between the photochemical reaction center (RC) and the cyt bc 1 complex (3,4). In Rhodobacter capsulatus, either a soluble, freely diffusible cyt c 2 (5) or a membrane-anchored cyt c y (6, 7) reduces the photooxidized RC (8,9). On the other hand, in R. sphaeroides only the cyt c 2 fulfills this role, as the cyt c y homologue of this species is unable to support Ps growth (10). The c y cyts are structurally distinct from the c 2 cyts as they have a linker domain of varying lengths in different species, attaching a highly conserved cyt c domain to a transmembrane anchor (7,11). In Rhodobacter species, both cyts c 2 and c y can act as efficient electron carriers between the cyt bc 1 complex and the respiratory (Res) cyt c oxidases (C ox ) (12,13). Moreover, both Rhodobacter species also contain a hydroquinone oxidase (Q ox )dependent alternate Res pathway (14,15) (Fig. 1). R. capsulatus mutants lacking cyts c 2 and c y are unable to grow by Ps. They can be complemented to Ps ϩ by either R. capsulatus or R. sphaeroides cyt c 2 , but only by R. capsulatus (and not by R. sphaeroides) cyt c y (10,16). On the other hand, R. sphaeroides mutants lacking only the cyt c 2 are Ps Ϫ (17), and they can be complemented to Ps ϩ by cyt c 2 of either species, or by R. capsulatus (and not R. sphaeroides) cyt c y (10,18).
The soluble cyt c 2 and membrane-anchored cyt c y interact differently with their physiological redox partners, which are the RC, the cyt bc 1 complex, and the C ox (13,19). During multiple turnovers, cyt c y transfers electrons from the cyt bc 1 complex to the RC more efficiently than cyt c 2 (19). Furthermore, whereas cyt c 2 interconnects a large number of RCs, cyt c y inter-acts with a limited number of them (8,19). In addition, in the absence of the cyt bc 1 complex, the steady-state amounts of cyt c y become lower, and the cyt c y kinetics change (9), suggesting a higher order of organization between the physiological partners (20,21). Available data indicate that both the differing lengths of the linker portions of cyt c y (47), and the amino acid sequences of their cyt c domains are important for the ET properties (10). Although cyt c 2 can be fused to the linker portion of R. capsulatus cyt c y to yield a functional membrane-anchored cyt c 2 variant (cyt MA-c 2 ) (11), the cyt c domain of R. sphaeroides cyt c y does not substitute readily for its R. capsulatus counterpart (10).
To investigate the structural and functional features of cyt c y as an efficient electron carrier, and its interactions with its redox partners, genetically modified versions of membrane-anchored cyt c y were sought. Here, the production and properties of soluble, freely diffusible variants of cyt c y are described. Various c y cyts with differing linker lengths were fused to the cyt bc 1 complex to prohibit their diffusion 4 and yielded novel cyt bc 1 -c y fusion complexes (21). We found that cyt S-c y variants support native-like Ps growth provided that they are present at sufficient amounts in vivo. Remarkably, the RC-mediated oxidation kinetics of cyt S-c y are distinct from those exhibited by cyts c 2 and c y . The findings suggest that, unlike the freely diffusing electron carriers (e.g. cyt c 2 ), which depend heavily on electrostatic contacts to recognize their physiological partners, membrane-anchored electron carriers (e.g. cyt c y ) are evolved to minimize binding interactions to maximize rapid multiple turnovers.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-The bacterial strains and plasmids used are described in Table 1. MPYE enriched medium (6) or Sistrom's minimal medium A (22) supplemented with kanamycin or tetracycline as appropriate (10 or 2.5 g/ml, respectively) were used for growth of R. capsulatus strains at 35°C. Ps cultures were incubated under saturating light intensity in anaerobic jars containing H 2 and CO 2 generating gas packs (BBL 270304, BD Biosciences) (7). Escherichia coli strains were grown on Luria-Bertani medium containing ampicillin, kanamycin, or tetracycline (at 100, 50, or 12.5 g/ml, respectively) as appropriate (6).
Molecular Genetic Techniques-R. capsulatus cyt S-c y was constructed by ligating the 3.4-kb EcoRI fragment of plasmid pHM5 containing the signal sequence of the cycA (structural gene of cyt c 2 ) (6) to the 0.8-kb EcoRI fragment of pRKE12 ϩ containing the cyt c domain of cycY* (structural gene of cyt c y with a silent EcoRI site) to yield plasmid pYO12 ( Table 1). The 1.2-kb KpnI-BamHI fragment of pYO12 carrying cyt S-c y was transferred to plasmid pRK415 using the same sites, yielding pYO100.
Biochemical and Biophysical Techniques-Intra-cytoplasmic membrane vesicles (chromatophores), and chromatophore supernatant fractions (containing cytoplasmic and periplasmic proteins), were prepared from R. capsulatus cells grown under Res conditions in MPYE by using a French pressure cell (at 18,000 p.s.i.) as described earlier (23). Chromatophore membranes used for SDS-PAGE/3,3Ј,5,5Ј-tetramethylbenzidine (TMBZ) or light-activated kinetic spectroscopy analyses were prepared in 50 mM MOPS buffer, pH 7.0, 1 mM phenylmethylsulfonyl fluoride dissolved in dimethyl sulfoxide and containing either 1 or 100 mM KCl, respectively (9). Protein concentrations were determined by the method of Lowry et al. (24), and 16.5% SDS-PAGE analyses were performed as described by Schägger and von Jagow (25). Samples were denatured in SDS-loading buffer for 5 min at 37°C, and following electrophoresis, the c-type cytochromes were revealed by their peroxidase activities using TMBZ and H 2 O 2 (26).
Optical spectra were recorded on a Perkin-Elmer UV-visible spectrophotometer (Lambda 20) fitted with an anaerobic redox cuvette, as needed. The difference spectra for c-type cytochromes were obtained with samples that were first oxidized by the addition of 20 M potassium ferricyanide and then reduced by using either 1 mM sodium ascorbate, or a small amount of FIGURE 1. Photosynthetic and respiratory electron transport pathways of R. capsulatus. In R. capsulatus, the soluble cyt c 2 and membrane-attached cyt c y operate between the photochemical RC and the cyt bc 1 complex during photosynthesis, and between the cyt bc 1 complex and cyt c oxidase during respiration. The alternative respiratory pathway that rely on the hydroquione oxidase is also shown.
Light-activated, millisecond time scale kinetic spectroscopy was performed using chromatophore membranes reduced with sodium ascorbate, as described earlier (8,9). Following a flash of actinic light, cyt c re-reduction kinetics in the absence or presence of the cyt bc 1 complex inhibitor stigmatellin (2.5 M), and as needed, in the presence of valinomycin (2.5 M) were monitored at 550 -540 nm. The carotenoid band shift was monitored at 490 -475 nm to follow the generation of the transmembrane potential by cyclic ET (8,9).

RESULTS
Soluble Variants of R. capsulatus Cyt c y -A water-soluble, freely diffusible form of the membrane-anchored cyt c y from R. capsulatus (cyt S-c y ) was obtained by genetically fusing the signal sequence of cyt c 2 to the cyt c domain of cyt c y , as described under "Experimental Procedures" (Fig. 2). Initially, this construct did not complement the R. capsulatus strain FJ2 lacking both cyts c 2 and c y for Ps growth (Table 1), possibly because of the absence of adequate amounts of a soluble electron carrier. However, FJ2 derivatives harboring cyt S-c y reverted frequently to Ps ϩ (at a frequency of 10 Ϫ6 to 10 Ϫ7 on MPYE medium), and three such revertants (R3, R4, and R5) were retained for further studies. Molecular genetic analyses described under "Experimental Procedures" revealed that revertants R3 and R5 contained plasmid-borne Cys to Thr and Ala to Gly base pair substitutions in cycY, respectively, and produced cyt S-c y R3 (His at position 53 substituted with Tyr) and cyt S-c y R5 (Lys at position 19 substituted with Arg) variants of cyt S-c y . Subsequently, these two mutations were combined via site-directed mutagenesis to yield a double mutant (cyt S-c y R35) ( Table 1), which complemented fully FJ2 for Ps ϩ growth (Fig. 3). On the other hand, it was found that Ps ϩ revertant R4 had a plasmid-borne wild type copy of cycY, but contained a chromosomal mutation (X). Although the molecular nature of X was not determined, revertant R4 was cured of its plasmid by successive subcultures on MPYE medium without antibiotic selection to yield FJ2-R4 (cyt c 2 Ϫ , cyt c y Ϫ , X Ϫ ) ( Table 1), used for some of the subsequent work due to its ability to enhance cyt c production, as described below. Ps ϩ growth abilities of cyt S-c y revertants were estimated by measuring their average colony sizes on MPYE medium in both FJ2 and FJ2-R4 backgrounds ( Table 2). In the FJ2 background, cyt S-c y , cyt S-c y R3 (H53Y), and S-c y R5 (K19R) conferred Ps Ϫ , Ps ϩ/Ϫ (slow Ps), and Ps ϩ phenotypes, respectively (Fig. 3A), whereas the double mutant cyt S-c y R35 exhibited a Ps ϩ phenotype as vigorous as that of a wild type R. capsulatus. Moreover, the Ps growth phenotypes of all three cyt S-c y variants were further improved in the FJ2-R4 background to the point that even cyt S-c y that was Ps Ϫ in FJ2 conferred a weak Ps ϩ/Ϫ growth in this strain (Fig. 3B).
Amounts and Properties of Cyt S-c y Produced by Various Ps ϩ Revertants-The c-type cyt profiles of Ps ϩ revertants, grown under Res ϩ conditions in MPYE medium, were examined by both SDS-PAGE/TMBZ analysis and reduced minus oxidized optical difference spectra. As expected, the membrane-bound cyt c y was absent in chromatophore membranes of FJ2 (cyt , whereas the amounts of the cyts c 1 , c o , and c p were unchanged (data not shown). In chromatophore membrane supernatants prepared using FJ2 derivatives cyt S-c y was undetectable, whereas cyt S-c y R5 (K19R), cyt S-c y R3 (H53Y), and cyt S-c y R35 (K19R and H53Y) were barely visible (Fig. 4A). In contrast, in the FJ2-R4 background the amounts of these c-type cytochromes, especially those of cyt S-c y R5 and cyt S-c y R35, increased and became readily detectable. As expected, the cyts S-c y are slightly smaller (about 10 kDa) than cyt c 2 and cyt cЈ (about 12 kDa) (Fig. 4B). Apparently, the effect(s) of the mutation(s) in FJ2-R4 is specific to cyt S-c y variants, because the amounts of native cyt c 2 or cyt cЈ were not affected significantly. The strain FJ2-R4 background was used for subsequent work instead of FJ2, as it enhanced production of the cyt S-c y variants (Table 2), and improved their Ps ϩ growth phenotypes (Fig. 3). We note that two additional TMBZ/H 2 O 2 -stained bands (about 19 and 34 kDa) were also detected in the supernatant fractions (Fig. 4). The higher molecular weight protein was identified by nano-LC-MS/MS analyses to correspond to R. capsulatus cyt c peroxidase (RRC03475). 4 Semi-quantitative estimation of the amounts of cyt S-c y derivatives by using reduced minus oxidized optical difference spectra of chromatophore supernatants ( Table 2) indicated that cyt S-c y R5 and cyt S-c y R35 were produced in FJ2-R4 at approximately one-third and equal amounts, respectively, to that of cyt c 2 found in a wild type R. capsulatus strain like MT1131 or its cyt c y Ϫ derivative FJ1 (Fig. 5A). Furthermore, the availability of a mutant producing high amounts of cyt S-c y allowed the determination of its redox midpoint potential (E m,7 ) using chromatophore membrane supernatants without further purification, as it is the main c-type cytochrome that  absorbs around 550 nm in these fractions. Both oxidative and reductive chemical dark titrations of cyt S-c y variants yielded E m,7 values around ϩ340 mV, which is similar (ϩ365 mV) to that of the membrane-attached, native cyt c y (Fig. 5B). Thus, neither rendering cyt c y soluble, nor the K19R and H53Y amino acid substitutions, which apparently increased the steady-state amounts of cyt S-c y in the FJ2-R4 background, affected appreciably its E m,7 value. Multiple Turnover ET Kinetics of Cyt S-c y Variants-ET kinetics exhibited by various cyt S-c y variants to the RC was examined by light-activated, time-resolved, optical difference spectroscopy (8). As expected, cyt c oxidation and re-reduction kinetics monitored at 550 -540 nm wavelengths indicated no detectable ET activity to the RC either in FJ2 (not shown) (9) or FJ2-R4 backgrounds, although the previously established ET features of cyt c y were readily observed with control strains MT1131 (cyt c 2 , cyt c y ), FJ1 (cyt c 2 ), and pFJ631/FJ2 (cyt c y ). In addition, pHM14/FJ2-R4 overproducing cyt c 2 exhibited kinetics similar to, but more amplified than, those seen with MT1131, in agreement with the increased cyt c 2 pool size. Strains pYO135/FJ2 or pYO135/FJ2-R4 harboring different amounts of cyt S-c y exhibited various levels of cyt c re-reduction activities consistent with their Ps ϩ phenotypes (Fig. 6, left column). Indeed, in all cases the full extent of cyt c oxidation was revealed by addition of the cyt bc 1 complex inhibitor stigmatellin (29,30), which blocked cyt c re-reduction (Fig. 6, middle column). Of a train of eight flashes used in each case, complete cyt c oxidation required a different number of flashes in agreement with the different cyt c pool sizes in different strains (compare e.g. FJ1 containing wild type levels of cyt c 2 and pHM14/FJ2-R4 overproducing cyt c 2 ). In addition, it was noted that the amounts of cyt c oxidation seen with cyt c 2 were larger than those seen with cyt S-c y , possibly indicating that the amounts of the latter electron carrier associated with RC were lower in vivo. In any case, the data established clearly that cyt S-c y was able to transfer electrons from the cyt bc 1 complex to the RC upon its light activation. Furthermore, carotenoid band shifts monitored at 475-490 nm wavelengths also confirmed the ET events between the physiological partners (Fig. 6, right column) (8). Indeed, the slower phases of the carotenoid band shifts were proportional to the extents of the ET activities in each case, demonstrating that transmembrane charge separation occurred upon establishment of the cyclic ET pathway between the cyt bc 1 complex and the RC via the cyt S-c y (Fig. 6, compare e.g. pYO135/FJ2-R4 to FJ2-R4). Remarkably, the modes of ET exhibited by the cyt S-c y was distinct from those seen with either cyt c y or cyt c 2 (8,18,19). However, the differences observed between the amounts of the carotenoid band shifts by different strains could not reflect any relationship between cyt c y and its efficiency in photosynthesis.

Comparison of Single Turnover Cytochrome c Oxidation Kinetics Exhibited by Different Electron Carrier Cytochromes-
Kinetic differences between different electron carrier cytochromes were further examined by monitoring single turnover cyt c oxidations observed in the presence of stigmatellin  a Ps ϩ and Res ϩ indicate the ability to grow under photosynthetic and respiratory (Res) growth conditions, respectively. Ps ϩ/Ϫ and Ps Ϫ refer to slow and no photosynthetic growth, respectively. b Average colony diameters (mm) determined after three days of incubation under the Ps growth conditions. c Amplitude of 550-535 nm absorption pick of ascorbate reduced minus ferricyanide-oxidized spectra obtained using approximately 1.52 mg/ml of chromatophore supernatant proteins. Under similar conditions A 550-535 values of 50.8 and 44.2 mOD were observed for the wild type strain MT1131 and the cyt c y Ϫ mutant FJ1, respectively. All strains are described in Table I. ( Fig. 7). As expected, whereas the oxidation of cyt c y was very fast and had no readily detectable slower phase at the millisecond time scale, that of cyt c 2 exhibited an equally fast oxidation phase followed by a slower phase of oxidation. The differences between the ET kinetics exhibited by the membrane-anchored cyt c y and the water-soluble cyt c 2 were attributed previously to their restricted and unrestricted diffusion abilities, respectively (9,19,31). On the other hand, cyt c oxidation kinetics exhibited by the cyt S-c y was composed of a barely perceptible fast oxidation phase, followed by a very prominent slower phase of oxidation (Fig. 7). This striking difference in the kinetic behavior of cyt S-c y compared with cyts c 2 and c y revealed that a noteworthy consequence of converting the diffusion-restricted membrane anchored cyt c y to a freely diffusible cyt S-c y was to slow down its ET abilities to the RC. Furthermore, the quasi-ab-sence of the fast oxidation phase in the cyt S-c y kinetics suggested that this cytochrome does not associate with the RC as tightly as does cyt c 2 .

DISCUSSION
To shed further light to the molecular basis of the Ps ET competence of cyt c y from some species in comparison with those from others (7, 10), a soluble variant of this naturally membrane-attached protein was sought by fusing genetically its cyt c domain to the signal sequence of cyt c 2 (6). It was anticipated that a soluble, freely diffusible version of cyt c y would not be constrained by its membrane attachment, and reveal more readily the electron carrier properties of its cyt c domain. However, the cyt S-c y thus constructed was produced at very low amounts in vivo, and was unable to support the Ps growth of a R. capsulatus strain lacking both cyts c 2 and c y . Whether these very low amounts reflected the poor biosynthesis or rapid degradation of cyt S-c y is unknown. Earlier, it was observed that in the absence of the cyt bc 1 complex, cyt c y is not detectable when cells are grown in enriched medium (9). Moreover, when the cyt c domains are exchanged between R. capsulatus and R. sphaeroides cyts c y (10), the latter chimeras were not stable in vivo. Fortunately, upon growth on enriched medium, strains carrying cyt S-c y yielded Ps ϩ revertants that produced more soluble cytochromes. Molecular analyses of these revertants indicated that they contained either single amino acid substitutions (e.g. K19R and H53Y) in the cyt c domain of cyt S-c y that increased their stability in vivo without changing their kinetic behaviors or had cyt S-c y unrelated chromosomal mutation(s) (e.g. FJ2-R4). Enhanced production of cyt S-c y derivatives in the latter revertants was reminiscent of protective associations between the cyt c y and its redox partners (9). However, the molecular basis of this chromosomal mutation(s) is unknown, and is beyond the scope of this work.
Dark, equilibrium titrations indicated that the E m,7 of the cyt S-c y is very similar to that of cyt c y or cyt c 2 , suggesting that rendering cyt c y soluble has not changed drastically its thermodynamic properties, even though its ET properties are modified. The three-dimensional structure of R. capsulatus cyt c y is not available, but its amino acid sequence is highly similar (63% identity and 75% similarity over 95 amino acids) to that of cyt c 552 of Paracocus denitrificans of known structure (32,33). A structural model for the cyt c domain of R. capsulatus cyt c y built by homology modeling based on that of cyt c 552 indicated that Lys-19 is on the surface of the protein and solvent exposed, whereas His-53 is slightly buried into the protein (Fig. 8). Thus, the larger side chain of Arg, substituting Lys at position 19, and more hydrophobic Tyr, substituting His at position 53, appear to have limited, if any, effects on the interactions of cyt S-c y with the RC (34), while increasing its stability and restricting its degradation.
A notable finding is the mode of ET kinetics exhibited by cyt S-c y . Light-activated RC-coupled cyt c oxidation kinetics indicated clearly that electron donation by R. capsulatus cyt S-c y is much slower than that mediated by either cyts c y or c 2 (35)(36)(37). The rapid kinetics observed with cyt c y have been attributed to its close proximity to the RC, and to its inability to diffuse freely (19,36). In the case of the cyt c 2 , the biphasic kinetics observed FIGURE 5. Amounts of soluble cyt S-c y produced in various R. capsulatus strains. A, ascorbate-reduced minus ferricyanide-oxidized optical difference spectra were recorded between 500 and 600 nm using in each case 3 mg of chromatophore supernatant proteins. Note that the amount of cyt S-c y R35 produced in FJ2-R4 is similar to that of cyt c 2 found in the wild type MT1131 or strain FJ1 (cyt c y Ϫ ). B, the redox midpoint potential of the supernatants containing the cyt S-c y R35 was determined in 50 mM MOPS, 100 mM KCl buffer at pH 7 as described under "Experimental Procedures." The data were fit to a Nernst equation for a one-electron couple, and the E m,7 of the cyt S-c y was estimated to be ϩ340 mV.
are thought to arise from two distinct populations: the fast phase reflecting the subpopulation of cyt c 2 already bound, and the slower phase that is to be bound, to the RC prior to light activation (37,38). Thus, the rate of the fast phase is independent from the cyt c 2 concentration, but the amplitude of this phase depends on both its concentration and binding affinity to the RC (38). The rate of the slower phase then reflects the fraction of cyt c 2 that needs to diffuse to reach an oxidized RC (36, 38 -40). In the case of cyt S-c y kinetics, the quasiabsence of a fast phase, and the prominence of a slower phase together indicate that almost no cyt S-c y is associated with the RC prior to light activation. If this is the case, then comparison of appropriate strains with similar amounts of soluble electron carrier cytochromes suggests that the binding affinity of cyt S-c y to the RC is apparently lower than that of cyt c 2 (Fig. 6) (35,38,40,41). In agreement with this deduction, it was noted that cyt S-c y , unlike cyt c 2 , does not remain membrane-associated when chromatophores are prepared at low (e.g. 1 mM KCl) salt concentrations, and that increased steady-state concentrations of cyt S-c y sustain more vigorous Ps ϩ growth in various mutants.
Docking of freely diffusible electron carriers (like cyt c 2 ) to their redox partners (like the RC) is thought to be governed by protein-protein interactions mediated at least partly by complementary surface charges (34,38,41,42). Different charge distributions on the docking surfaces of redox partners affect their mode of interactions, and mutations that affect the binding affinities of the partners also change the ET rates (e.g. the second-order rate constant, k 2 ) between them. Upon docking, additional hydrophobic interactions are involved for precise orientation of the redox partners and subsequent optimal ET between them, as illustrated by extensive RC-cyt c 2 studies (37,38), or the cyt bc 1 -cyt c cocrystal structures (43). The kinetic data available for the cyt c y and S-c y and comparison of their three-dimensional structures suggest that the hydrophobic interactions between the cyt c domain of cyt c y and the RC are efficient enough to mediate fast ET, despite their apparently weak electrostatic interactions, as revealed by rendering cyt c y soluble (Fig. 8). Indeed, structural examinations indicate that most of the amino acid residues implicated in electrostatic connections are not conserved, whereas those involved in hydrophobic interactions are preserved between cyt c y and cyt c 2 . Conceivably, membrane anchoring of electron carriers counterbalances their weaker electrostatic interactions in recogniz- In each case, chromatophore membranes were prepared in 50 mM MOPS, 100 mM KCl buffer at pH 7, resuspended in the same buffer at the appropriate concentrations, and the ambient redox potential was poised at 100 mV, as described under "Experimental Procedures." Light induced (a train of eight flashes) transient cyt c oxidation and re-reduction kinetics exhibited by various strains were monitored at 550 -540 nm in the absence (left column) and presence (middle column) of the cyt bc 1 complex inhibitor stigmatellin, which blocks re-reduction of cytochromes. Right column shows the 490 -475 nm traces obtained using the same strains to monitor the carotenoid band shifts in response to the cyclic photosynthetic electron transport reactions in the absence of inhibitor and in response to a single flash. FIGURE 7. Single turnover cytochrome c oxidation kinetics observed in various cyt S-c y producing strains. Samples were prepared, and traces (in the presence of stigmatellin) were recorded as described in the legend to Fig.  6, except that 2.5 M valinomycin was present and data obtained by one hundred flashes were averaged to visualize the cyt c oxidation (550 -540 nm) kinetics of strains harboring only the cyt c y (pFJ631/FJ2, top row), cyt c 2 (FJ1, middle row), or cyt S-c y R35 (pYO135/FJ2-R4, bottom row). Note that the cyt c oxidation kinetics of the cyt S-c y are highly distinct from those seen with both the cyts c y and c 2 .
ing their redox partners (44). Thus, a lack of tight binding of the cyt c domain of cyt c y to the RC, along with its efficient interaction with the cyt bc 1 complex, is consistent with both its rapid oxidation and re-reduction kinetics during multiple turnovers, and with the slower oxidation of its soluble derivative cyt S-c y (18,19). It remains to be seen whether in the resting state, cyt c y remains associated mainly with the cyt c 1 subunit of the cyt bc 1 complex (21), rather than the RC, to ensure rapid and transient interactions upon light activation during multiple turnovers and Ps growth. If so, then limited numbers of RC-cyt c y -cyt bc 1 complex can form "hardwired" photosynthetic units (21). These units then turn over efficiently to sustain vigorous Ps growth as long as the length of the linker portion of cyt c y is adequate (47).
In summary, this work demonstrates that the cyt c domain of the membrane-anchored electron carriers (e.g. cyt c y ) apparently does not bind tightly to the RC due to the lack of strong protein-protein interactions between the partners. The findings suggest that attaching an electron carrier to the membrane, while restricting spatial diffusion, allows weaker binding to its partners to ensure rapid multiple turnovers. Unlike the freely diffusible electron carriers (e.g. cyt c 2 ) that rely on complemen-tary electrostatic interactions and large pool sizes, a small number of membrane-attached electron carrier cytochromes then suffices to support efficient Ps growth via enhanced multiple turnover rates, as observed with cyt c y .