Purification and Characterization of the Cytochrome b 6 f Complex from Chlamydomonas reinhardtii *

A protocol has been developed for the purification of the cytochrome b 6 f complex from the unicellular alga Chlamydomonas reinhardtii . It is based on the use of the neutral detergent Hecameg (6- O -( N -heptylcarbamoyl)- methyl- (cid:97) - D -glycopyranoside) and comprises only three steps: selective solubilization from thylakoid mem- branes, sucrose gradient sedimentation, and hydroxylapatite chromatography. The purified complex con- tains two b hemes ( (cid:97) bands, 564 nm; E m ,8 (cid:53) (cid:50) 84 and (cid:50) 158 mV) and one chlorophyll a ( (cid:108) max (cid:53) 667–668 nm) per cytochrome f ( (cid:97) band, 554 nm; E m ,8 (cid:53) (cid:49) 330 mV). It is highly active in transferring electrons from decylplastoquinol to oxidized plastocyanin (turnover number 250–300 s (cid:50) 1 ). The purified complex contains seven subunits, whose identity has been established by N-termi- nal sequencing and/or peptide-specific immunolabeling, namely four high molecular weight subunits (cyto- chrome f , Rieske iron-sulfur protein, cytochrome b 6 , and subunit IV) and three (cid:59) 4-kDa miniproteins (PetG, an electron donor and either 5 (cid:109) M oxidized cytochrome c or 5 (cid:109) M oxidized spinach plastocyanin as an acceptor. The reaction was initiated by adding C 10 -PQH 2 to a solution containing 20 m M Tricine-NaOH buffer pH 8.0, 0.3 m M LM, cytochrome b 6 f (0.5–2 n M ), and the electron acceptor. Reduction of horse cytochrome c was monitored during 2 min at room temperature by the absorbance change at 549 nm, assuming an extinction coefficient of 22,000 M (cid:50) 1 (cid:122) cm (cid:50) 1 (22). Reduction of spinach plastocyanin was monitored under the same conditions by the absorbance change at 600 nm, assum- ing an extinction coefficient of 4,500 M (cid:50) 1 (cid:122) cm (cid:50) 1 (46). We checked that, under the latter conditions, the rate of electron transfer is independent on the concentration of PQH 2 and linearly related to that of plastocyanin. When measuring the oxidoreductase activity of thylakoid mem- branes or of fractions containing partially purified b 6 f complex, using cytochrome c as an acceptor, 1 (cid:109) M antimycin was added to the solution in order to inhibit the oxidoreductase activity of the contaminating bc 1 complex (22).

The photosynthetic electron transfer chain, the device that provides essentially all of the free energy dissipated by living beings, comprises two membrane protein complexes that are homologous to complexes of the respiratory chain, the CF 1 CF 0 (respectively, F 1 F 0 ) proton ATP-synthase and the b 6 f (respectively, bc 1 ) oxidoreductase. The bf/bc complexes transfer electrons from a liposoluble quinol to a hydrosoluble protein, plastocyanin or cytochrome c, and couple the resulting electron free energy drop to setting up a transmembrane proton electrochemical potential. Neither the structure of the b 6 f nor that of the bc 1 complex is known in any detail, and the existence of an unique versus alternative mechanism(s) of electron transfer remains a matter of discussion (see, e.g., Refs. 1-4 and references therein).
Most structural studies of the b 6 f complex to date have been carried out either on higher plants or on cyanobacteria (for reviews, see Refs. 1, 2, and 5-8). Unicellular algae, such as Chlorella sorokiniana and Chlamydomonas reinhardtii, have long been recognized as convenient organisms for combined spectroscopic, biochemical and genetic studies of the b 6 f complex structure and function (9,10). In the present article, we describe the purification of the b 6 f complex from C. reinhardtii. From the point of view of the experimentalist, C. reinhardtii combines a number of advantages: (i) cell suspensions are amenable to in vivo studies of the electron transfer and proton pumping processes using spectroscopic and other biophysical approaches; (ii) C. reinhardtii being a facultative phototroph, mutant strains with defective photosynthetic complexes can be generated and maintained; (iii) large fractions of its chloroplast genome have been sequenced; (iv) techniques for modification of its chloroplast and nuclear genes are being developed; (v) it is amenable to in vivo isotopic labeling; (vi) large scale growth, although not currently resorted to, is not an unrealistic objective (for a general review of C. reinhardtii laboratory use, see Ref. 10).
Previous studies have shown that the b 6 f complex from C. reinhardtii is similar to that of higher plants in its complement of high molecular mass subunits (11,12). It comprises two heme-bearing subunits (cytochrome f and cytochrome b 6 ), a protein carrying a [2Fe-2S] cluster (the so-called Rieske protein), and an integral protein devoid of prosthetic group ("subunit IV") that is homologous to the C-terminal region of mitochondrial cytochrome b (13). More recent studies have shown that additional, very small subunits (ϳ4 kDa) are also present, namely: 1) the product of the petG chloroplast gene, initially identified in higher plants (14,15) and subsequently shown to be present in C. reinhardtii b 6 f (15,16); 2) the PetX protein, identified both in C. reinhardtii (15,16) and in spinach (15) complexes, which is the product of a nuclear gene (16,17); and 3) the product of the chloroplast gene ycf7 (petL), 1 a gene that is also present in higher plant chloroplast genomes.
Obtaining large scale preparations of active integral membrane proteins is a major stumbling block in understanding the function of these proteins at the structural level. Two limitations are the amount of protein produced by laboratory organisms and scale limits on purification procedures. Until now, preparations of b 6 f complex from C. reinhardtii (11,12,15) have suffered from limited purity, low yields, and low or unknown enzymatic activity. In the present article, we describe a rapid (three-step) protocol for the preparation of highly purified C. reinhardtii b 6 f complex that is enzymatically fully active. The neutral detergent used throughout the purification,6-O-(N-heptylcarbamoyl)-methyl-␣-D-glycopyranoside (Hecameg; 2 I-protein A from Amersham, and synthetic peptides coupled to ovalbumin from Neosystem (Strasbourg, France). C 13  ]acetate was added in TAP medium at a final concentration of 37 kBq/ml at the beginning of cell culture. Cells were harvested at 5000 ϫ g for 10 min, and thylakoid membranes prepared as described previously (22,24). Thylakoid membranes were finally resuspended in 10 mM Tricine-NaOH, pH 8.0, containing protease inhibitors (200 M PMSF, 1 mM benzamidine, 5 mM ⑀-aminocaproic acid) at a final chlorophyll concentration of 3 g/liter and stored at Ϫ80°C. 32 P Labeling-Wild-type cells were grown in TAP medium under standard conditions until stationary phase, diluted ϳ10-fold in 2 l of TAP medium containing 1/10 of the usual non-radioactive phosphate concentration and 37 MBq of sodium [ 32 P]phosphate, and further grown under the same conditions until stationary phase. Cells grown under such conditions are in a state intermediate between State 1 and State 2 (see ref. 25). 3 Cells were harvested, thylakoid membranes prepared, and the b 6 f complex purified as described for unlabeled preparations, except that all buffers contained 20 mM EGTA-NaOH, pH 7.5, 200 mM NaF, and 1 M microcystin (26) as phosphatase inhibitors. In addition, the buffer in which the cells were broken contained 0.1 M okadaic acid (27).
Purification of Spinach Plastocyanin-Plastocyanin was prepared according to Ref. 28 from fresh spinach leaves obtained from the local market.
Preparation of Antipeptide Antisera-Subunit-specific antisera were raised against synthetic peptides as described previously (16). All peptides that yielded active antisera were either N-or C-terminal and were coupled to ovalbumin by their extremity internal to the protein (indicated below by an asterisk), one extra residue being generally added to permit coupling (cytochrome f, YPVFAQQNYANPREK* and *YKKK QFEKVQLAEMNF; Rieske protein, AAASSEVPDMNKRNY*; cytochrome b 6 , SKVYDWFEERLEIQC* and *YLMIRKQGISGPL; subunit IV, SVTKKPDLSDPVLKY* and *KGSTFPIDISLTLGLF; PetG, *VTAYLQYLRGDLATY; PetL, *CTLGIYLGLLKVVKLI; PetX, GEAE FIAGTALTMY*. SDS-Polyacrylamide Gel Electrophoresis-Polypeptide composition was analyzed either on 12-18% polyacrylamide gel containing 8 M urea (29) or on three-step gels in the absence of urea (30). Heme staining was performed as in Ref. 31 and silver staining as in Ref. 32. 14 C-or 32 P-labeled polypeptides were detected either by autoradiography of dried gels on Agfa Curix MR4 films or using PhosphorImager plates (Molecular Dynamics).
Immunoblotting-Proteins were electrotransferred onto Immobilon PVDF membranes in a semidry blotting system at 0.8 mA/cm 2 for about 30 min. Immunodetection was carried out either by the enhanced chemiluminescence peroxidase method (Amersham) or by labeling with 125 I-protein A, as described previously (33).
Protein Sequencing-Polypeptide purification by SDS-PAGE and transfer to PVDF membranes were performed as described previously (22). Amino acid sequencing was carried out at the Service Central d'Analyze du CNRS (Vernaison, France) by L. Denoroy.
Analytical Procedures-Protein concentrations were determined by the Bradford method (34). Chlorophyll concentrations were determined according to Ref. 35. Critical micellar concentrations (CMC) were estimated either by the pyrene partitioning method (36) or by counting drops from a Hamilton syringe with a non-beveled needle. The accuracy of quoted values for HG is probably approximately Ϯ1 mM.
Spectroscopy-UV-visible absorbance spectra were recorded either on a home-built spectrophotometer of the Joliot design (37,38) or on a Kontron Uvikon 930 instrument. Cytochromes were oxidized with ferricyanide and reduced either with ascorbate (cytochrome f) or with dithionite or borohydride (cytochromes b h and b l ). The spectrum of oxidized preparations in the UV-visible region was obtained by recording two spectra, the first one following air equilibration and the second one after ascorbate reduction, and subtracting from the former the ascorbate-minus-air difference spectrum, scaled so as to cancel the ␣ peak of cytochrome f. Wavelengths of absorbance maxima and isosbestic points are rounded off to the nearest nanometer.
Redox Potentiometry-Spectroscopic scans were made on a homebuilt Joliot spectrophotometer (39). The monochromator was set to a 2.0 nm wavelength resolution. Purified b 6 f complex was diluted to a concentration of 375 nM in deoxygenated TMK buffer containing 0. Redox potentiometric measurements were made with a platinum electrode, using a standard calomel electrode as a reference. The data were corrected to reflect potentials against the standard hydrogen electrode. Redox poising was accomplished by additions of small quantities of concentrated solutions of sodium ferricyanide and sodium hydrosulfite. At least 3 min were allowed for equilibration after each change in ambient redox potential before spectra were scanned. The pH at the end of the titration was checked to insure that no drift had occurred.
Spectral changes recorded from Ϫ250 to ϩ450 mV were analyzed by taking the peak wavelengths (approximately 564 nm for cytochrome b 6 and 554 nm for cytochrome f) and fitting the absorbance values to Nernst curves. Component spectra were obtained by redox cut analysis; the spectrum of cytochrome f was obtained by taking the average of three spectra recorded at 400 Ϯ 10 mV and subtracting the average of four spectra taken at 230 Ϯ 30 mV; the spectrum of cytochrome b h was obtained by subtracting from a spectrum recorded at Ϫ80 mV a spectrum recorded at 0 mV; the spectrum of cytochrome b l was obtained by subtracting from a spectrum taken at Ϫ250 mV a spectrum recorded at Ϫ132 mV. The ratio of cytochrome b 6 to cytochrome f was estimated from the redox titration curves, assuming extinction coefficients of 18,000 M Ϫ1 ⅐cm Ϫ1 (cytochrome f) and 20,000 M Ϫ1 ⅐cm Ϫ1 (average of the two b-type cytochromes) (peak absorbance minus absorbance at isosbestic point; Ref. 40).
Decylplastoquinone Reduction-A diethylether solution of decylplastoquinone (C 10 -PQ) was reduced to decylplastoquinol (C 10 -PQH 2 ) by an aqueous solution of sodium dithionite and sodium borohydride with strong agitation in the dark until the solution became colorless (see Ref. 41). A small volume of chlorhydric acid was then added to eliminate the sodium borohydride in excess. The diethylether phase was collected and evaporated to dryness under N 2 , and the C 10 -PQH 2 resuspended in 100% ethanol acidified with chlorhydric acid. C 10 -PQH 2 concentration was estimated from the absorption value at 290 nm using an extinction Purification of Cytochrome b 6  buffers contained 200 M PMSF, 1 mM benzamidine, and 5 mM ⑀-aminocaproic acid as protease inhibitors. For selective solubilization, thylakoid membranes of wild type cells were resuspended in ice-cold TMK buffer containing 25 mM HG at a chlorophyll concentration of 1.5 g/liter and incubated at 4°C for 15 min. with occasional agitation. The suspension was centrifuged either at 160,000 ϫ g (80,000 rpm) for 10 min in the TLA 100.3 rotor of a Beckman TL100 ultracentrifuge, or at 200,000 ϫ g (43,000 rpm) for 30 min in the 70 Ti rotor of a Beckman L8 ultracentrifuge, depending on the volume (6 or 30 ml).
The solubilization supernatant was fractionated by centrifugation in 10 -30% (w/w) sucrose density gradients in TMK buffer containing 20 mM HG and 0.1 g/liter egg PC. The presence of egg PC during this and the following step of purification is necessary to prevent the dissociation of the Rieske protein from the complex (43). The gradients, onto which were layered ϳ600 l (Beckman SW41 Ti rotor) or ϳ3 ml (45 Ti rotor) of supernatant, were centrifuged at 270,000 ϫ g (40,000 rpm) for 24 h (SW41 Ti rotor), or at 180,000 ϫ g (38,000 rpm) for 16 h (45 Ti rotor). The yellow-brown band containing the cytochrome b 6 f complex was collected with a syringe.
The band collected from the gradient was layered onto a HA column (0.7 cm ϫ 2 cm, or 1 cm ϫ 3 cm) pre-equilibrated with 20 mM Tricine-NaOH, pH 8.0, 20 mM HG, 0.1 g/liter egg PC. The column was washed with 3 column volumes of 100 -200 mM ammonium phosphate (depending on HA batches), pH 8.0, containing 20 mM HG and 0.1 g/liter egg PC. Pure cytochrome b 6 f complex was eluted with 400 mM ammonium phosphate, pH 8.0, containing 20 mM HG and 0.1 g/liter egg PC. Typical yields are ϳ600 l of pure b 6 f complex at ϳ6 M cytochrome f for one SW41 rotor or ϳ3 ml at ϳ10 M for one 45 Ti rotor, with larger preparations resulting in better overall yields.
Our initial experiments with HG were made somewhat frustrating by irreproducible batch-to-batch results. These appeared to be due to the presence of diheptylurea, a side product from the synthesis. 4 In the presence of traces of diheptylurea, abnormal migration of the b 6 f complex on sucrose gradient was observed, and electron transfer activity was inhibited. The presence of diheptylurea is easily detected by leaving a 20 mM solution of HG in water overnight in the cold room. Diheptylurea precipitates as a fine white powder. This problem has been solved by the manufacturer, and recent HG batches (batches 2001, 2101, WK1-20, and CP1-57) have been free from diheptylurea. From a practical point of view, we note that the CMC of HG is similar at 4°C and at room temperature 5 (see also Refs. 44 and 45), but appears to be sensitive to high salt concentrations, since the CMC at 4°C dropped from ϳ19.5 mM to ϳ15 mM when moving from the low ionic strength TMK buffer to 0.4 M ammonium phosphate (see also Ref. 45).
Assays of Electron Transfer Activity-Cytochrome b 6 f oxidoreductase activity was measured at room temperature in a Kontron Uvikon 930 spectrophotometer, using 15 M C 10 -PQH 2 as an electron donor and either 5 M oxidized cytochrome c or 5 M oxidized spinach plastocyanin as an acceptor. The reaction was initiated by adding C 10 -PQH 2 to a solution containing 20 mM Tricine-NaOH buffer pH 8.0, 0.3 mM LM, cytochrome b 6 f (0.5-2 nM), and the electron acceptor. Reduction of horse cytochrome c was monitored during 2 min at room temperature by the absorbance change at 549 nm, assuming an extinction coefficient of 22,000 M Ϫ1 ⅐cm Ϫ1 (22). Reduction of spinach plastocyanin was monitored under the same conditions by the absorbance change at 600 nm, assuming an extinction coefficient of 4,500 M Ϫ1 ⅐cm Ϫ1 (46). We checked that, under the latter conditions, the rate of electron transfer is independent on the concentration of PQH 2 and linearly related to that of plastocyanin. When measuring the oxidoreductase activity of thylakoid membranes or of fractions containing partially purified b 6 f complex, using cytochrome c as an acceptor, 1 M antimycin was added to the solution in order to inhibit the oxidoreductase activity of the contaminating bc 1 complex (22).

RESULTS
Purification Protocol-Treatment of thylakoid membrane preparations from C. reinhardtii (22,24) with appropriate concentrations of Hecameg (HG; see "Materials and Methods") preferentially solubilizes the CF 0 CF 1 ATP synthase and the cytochrome b 6 f complex, resulting in a ϳ3-fold increase of the b 6 f/protein ratio (Fig. 1, lane 2, and Table I). Distribution of the b 6 f complex in this and subsequent fractionation experi-ments was estimated on the basis of the content in two hemestainable proteins, migrating upon urea/SDS-PAGE with apparent M r 39.5 and 18 kDa, which have been recognized previously as the cytochrome f and cytochrome b 6 holoproteins, respectively (11,12). Fractionation of the supernatant on a 10 -30% sucrose gradient in the presence of HG and lipids revealed that 18.5-, 12.5-, and ϳ4-kDa proteins comigrate with the two cytochromes (Fig. 1, lane 3). Further purification on a hydroxylapatite column resulted in a preparation devoid of any other polypeptides (Fig. 1, lane 4). The overall yield of the purification was close to 50%, a 20-liter growth of wild type C. reinhardtii cells yielding 5-7 mg of pure complex.
Handling the complex under delipidating conditions, e.g. sedimenting it in a sucrose gradient containing 25 mM HG in the absence of lipids and salts, resulted in the loss of the Rieske protein and of the oxidoreductase activity (43). Identified stabilizing factors are (i) working close to the CMC of HG and in the presence of lipids (0.01-0.1 g/liter egg PC) and (ii) adding low concentrations of salt (3 mM KCl, 3 mM MgCl 2 ).
Spectroscopic and Redox Properties-The near UV-visible spectrum of a purified b 6 f preparation is shown in Fig. 2. Absorbance peaks in the Soret, ␣, and ␤ regions of the spectrum are characteristic of the presence of cytochrome f. Additional bands at ϳ460 and 483 nm indicate the presence of carotenoids and a band at 667-668 nm that of chlorophyll a. The molar ratio of chlorophyll a to cytochrome f is close to 1 (47), as previously observed in Synechocystis PCC6803 (48) and spinach (49) b 6 f preparations. The reason for the presence of nonheme pigments in highly purified C. reinhardtii b 6 f preparations will be analyzed in more detail elsewhere.
Upon addition of ascorbate, the contribution of reduced cytochrome f to the spectrum increases more or less strongly, indicating that, in air-equilibrated preparations, cytochrome f is partially to totally reduced. Further reduction with dithionite or borohydride reveals the presence of cytochrome b 6 . Difference spectra (Fig. 2, insets A and B) show absorption maxima at 523 and 554 nm for cytochrome f and 534 and 564 nm for 4 W. Klagba, personal communication. 5  Visible absorbance spectra were recorded over the redox potential range from Ϫ250 to 450 mV. They were analyzed by fitting the absorbance values at peak wavelengths (approximately 564 nm for cytochrome b 6 and 554 nm for cytochrome f) to Nernst curves. A single curve with m ϭ 1 and E m ϭ ϩ 330 mV was sufficient to fit the cytochrome f data (Fig. 3A). Two titration curves with n ϭ 1 and E m ϭ Ϫ84 and Ϫ158 mV were necessary to fit the cytochrome b 6 data (Fig. 3B). These values are similar to those obtained at pH 8.75 in spinach thylakoids for cytochromes b h and b l , respectively (4). The visible spectra of the two b-type cytochromes at room temperature, calculated from the redox titration data, were undistinguishable (Fig. 3C). The concentrations of hemes b h and b l were comparable (Fig.  3C) and the b 6 /f ratio determined from the redox titration curves was ϳ1.9:1, indicating that none of the two b-type hemes had been lost during the purification.
Electron Transfer Activity-The purified preparation actively catalyzed electron transfer from decylplastoquinol (C 10 -PQH 2 ) to spinach plastocyanin (Fig. 4). Under our experimental conditions, one molecule of cytochrome f catalyzed the transfer of 270 Ϯ 60 electrons/s (mean Ϯ S.D. on 5 preparations) when using 5 M plastocyanin as an acceptor. This rate is 20-fold higher than that reported by Schmidt and Malkin (15) for C. reinhardtii b 6 f preparations. It is 3-4-fold higher than the highest rates hitherto reported for purified spinach b 6 f (see Refs. 49 and 50, and references therein) and similar to that estimated in vivo (see Refs. 1 and 5). Electron transfer was totally inhibited by the addition of 1 M C 13 -stigmatellin (Fig.  4). Catalysis of electron transfer from C 10 -PQH 2 to beef heart mitochondrial cytochrome c was observed only when trace amounts (ϳ1 M) of plastocyanin were added to the reaction medium (data not shown). The specific activity of electron transfer to cytochrome c was the same whether measured on the HG supernatant, on the b 6 f fraction from the sucrose gradient or on purified b 6 f preparations, indicating that no inactivation occurred during purification (data not shown).
After 2-week storage at 0°C in the dark, purified preparations retained ϳ75% of their activity and exhibited no obvious change of their polypeptide pattern. They could be frozen without loss of activity.
Polypeptide Composition-The subunit composition of the purified complex was studied by SDS-PAGE using either gradient gels in the presence of urea (29) or a three-step gel system devoid of urea (30). Staining with silver revealed five bands in the first system, with apparent molecular mass values of 39.5, 18.5, 18, 12.5 and ϳ4 kDa (Fig. 1, lane 4). The 4-kDa band often appeared heterogeneous. The five bands comigrated throughout the sucrose gradient and hydroxylapatite purification steps (data not shown). In the discontinuous, three-step gel system, the 4-kDa band was resolved into two, sometimes three or four bands of variable intensity (see Ref. 16). The pattern observed depended on incompletely identified factors that affect the migration and staining of the subunits, and it is not certain that each band represents a distinct polypeptide rather than variable states of denaturation of some of them.
Identification of the bands was based on their apparent M r , heme content as revealed by TMBZ staining, N-terminal sequencing, and cross-reaction with antipeptide antisera (Table  II). These approaches confirmed that the four high M r bands correspond, in this order, to cytochrome f, the Rieske ironsulfur protein, cytochrome b 6 , and subunit IV (11,12). 1) N-terminal sequencing of the TMBZ-stainable 39.5-kDa band yielded a sequence (Table II) that is internal to that of the cytochrome f precursor, as predicted from the sequence of the petA gene (51,52). The band was labeled by antisera raised either against this N-terminal peptide or against a peptide with the predicted C-terminal sequence of cytochrome f (Fig. 1, lane 5Љ, and data not shown). 2) Sequencing of the 18.5-kDa band yielded a sequence (Table II)

that is homologous with TABLE I Purification factors and yields
The amounts of protein and of cytochrome f indicated are normalized to a 20-liter growth. The amount of protein in the thylakoid membranes was estimated from the chlorophyll content, assuming a 5:1 protein/chlorophyll weight ratio (24). Direct determination of cytochrome f in the membrane preparation is not practical; an extraction yield of ϳ70% was estimated by comparing visually dilution series of thylakoid membranes and HG supernatants following SDS-PAGE and TMBZ staining. Other data are based on cytochrome f absorbance in ascorbate-reduced minus ferricyanide-oxidized difference spectra and on direct protein determination (34) and are averages Ϯ S.D. over 3 preparations using SW41 Ti rotors. Larger preparations, using 45 Ti rotors, typically yielded 6 -7 mg of purified complex/20-liter growth. Step ϳ70% 85% 80% Overall ϳ70% ϳ60% ϳ47% FIG. 2. Absorbance spectra of purified b 6 f preparations. Oxidized ( . . . ), ascorbate-reduced (---) and dithionite-reduced (OO) UVvisible spectra of a purified preparation. The spectrum of oxidized b 6 f was calculated by subtracting from the spectrum of air-equilibrated b 6 f the contribution of reduced cytochrome f (see "Materials and Methods"). Inset A, UV-visible difference spectra: ascorbate-reduced minus airoxidized ( . . . ) and dithionite-reduced minus ascorbate-reduced (OO). Inset B, visible difference spectra (from a different b 6 f preparation): ascorbate-reduced minus ferricyanide-oxidized (---) and dithionite-reduced minus ascorbate-reduced (OO).
N-terminal sequences of other Rieske proteins. The 18.5-kDa band was labeled by antisera directed against this N-terminal sequence (Fig. 1, lane 6Љ). Oligonucleotides coding for this sequence have been used to clone and sequence C. reinhardtii petC nuclear gene (53). 3) The diffuse TMBZ-stainable 18-kDa band did not yield any phenylthiohydantoin derivatives upon amino acid sequencing. It was labeled by antisera directed against peptides corresponding to the predicted N-and Cterminal sequences of cytochrome b 6 deduced from that of the petB gene (52) (Fig. 1, lane 7Љ, and data not shown). 4) The 12.5-kDa band did not yield any N-terminal sequence either. This band was labeled by two antisera (Fig. 1, lane 8Љ, and data not shown), raised respectively against the N-terminal and C-terminal sequences predicted from the petD gene, which encodes subunit IV (52).
We have shown previously that the ϳ4-kDa band(s) represent(s) more than one polypeptide (Ref. 16; see also Ref. 15). The presence of the product of the petG gene was inferred from labeling with an antipeptide directed against the C-terminal part of the predicted PetG protein. The presence of a novel subunit, product of an unidentified nuclear gene provisionally named petX, was established by N-terminal sequencing and pulse labeling in the presence of various inhibitors of protein synthesis (Refs. 15 and 16; Table II). Immunolabeling shows that PetX and PetG comigrate during SDS/urea-PAGE and migrate close one to another in three-step gels in the absence of urea (16). The complete sequence of PetX has recently been established by further protein sequencing and by cloning and sequencing of the petX gene (17). Finally, the presence of a third ϳ4-kDa subunit has been recently established by deletion of the chloroplast ycf7 (petL) gene and immunoblot analysis of purified b 6 f preparations obtained from wild-type and ycf7Ϫ strains (47). 1 Immunoblotting of SDS-PAGE gels shows that all of these seven proteins are absent in thylakoid membranes prepared from mutant FuD4, that lacks the b 6 f complex (Fig. 1, lanes  5Ј-8Ј; Ref. 16 and Footnote 1).
Posttranslational Modifications-N-terminal sequencing defines the point of cleavage of addressing and signal sequences in the Rieske protein, in cytochrome f and in PetX (Table II). Antipeptide antisera indicate that most of the predicted Nterminal and C-terminal sequences are present in mature cytochrome b 6 and subunit IV. Spinach cytochrome b 6 has been reported to be a substrate for an endogenous redox-potentialcontrolled kinase (54). In order to search for phosphorylation sites in the purified C. reinhardtii complex, cells were labeled at equilibrium with [ 32 P]phosphate under the conditions indicated under "Materials and Methods." None of the b 6 f polypeptides was found to incorporate a significant amount of radioactivity, under conditions where labeling to an extent of a few percents of each chain or less would have been detected (data FIG. 3. Redox titration curves of cytochromes f (A) and b 6 (B) and absorbance spectra of the three cytochromes in the ␣ band region (C). Redox potential values are given against the standard hydrogen electrode (SHE). Subunit Stoichiometry-In order to examine subunit stoichiometry, b 6 f complex was purified from C. reinhardtii cells grown on [ 14 C 1 ]acetate and the radioactivity associated with polypeptide bands on SDS-PAGE gels quantified using Phos-phorImager plates (Fig. 5). The ratio of 14 C activity in the high M r bands was very close to stoichiometric, when normalized to the number of carbon atoms per protein (Fig. 5B). The relative amount of radioactivity in the five bands suggests that the 4-kDa band contains ϳ3 polypeptides/cytochrome f. These data are compatible with all seven subunits being present in equimolecular amount. The protein content of the purified preparations (112 Ϯ 9 mg/mol of cytochrome f) is close to that expected assuming this stoichiometry (103 mg/mol). However, the dimeric nature of the complex (43), the uncertainty in the determinations, and the small size of the ϳ4-kDa subunits combine to make it difficult to rule out two possible cases of figure, namely (i) the existence of still other unidentified 4-kDa subunits, and/or (ii) a substoichiometric complement of one or more of the 4-kDa subunits, that would be present as one copy per b 6 f dimer rather than one per monomer. A direct, reliable quantitation of each of the 4-kDa subunits has not been possible yet. As noted above, these subunits are not well resolved under our usual SDS/urea-PAGE conditions (Fig. 1, lane 4). Three-step gels (30), which give a better but variable resolution, were found to be unreliable for this purpose, presumably because of the incomplete denaturation of some subunits in the absence of urea. 5

DISCUSSION
Purification Protocol-Our protocol for purification of the b 6 f complex from C. reinhardtii thylakoid-rich membranes is derived from the classical Hauska-Cramer procedures used to purify higher plant b 6 f complexes (49,50,55,56). It presents a number of advantages over current methods (see e.g. Refs. 15 and 49, and references therein). As regards the C. reinhardtii complex, it is much simpler than the 6-step protocol of Schmidt and Malkin (15), and it yields a preparation that is free of polypeptide contaminants and that exhibits a 20-fold higher, native-like electron transfer activity. The greater purity probably reflects the fact that the Schmidt-Malkin protocol has been optimized for spinach rather than for C. reinhardtii b 6 f.
The better preservation of the activity is presumably linked to limiting delipidation. As regards the b 6 f complex in general, our protocol yields preparations whose purity and activity are comparable to or better than those reported for the best higher plant or cyanobacterial preparations (see e.g. Refs. 15,[48][49][50][56][57][58], and references therein). It is nonetheless simpler than the usual spinach protocol (49,50), it does not resort to any heterogeneous detergent mixture, and it presents the added advantage of making use of a facultative phototrophic organism. Finally, this protocol is easily amenable to scaling up. Current limitations on mass production are the size of C. reinhardtii growths (20-liter flasks, yielding 5-7 mg of purified b 6 f) a Apparent M r on 12-18% polyacrylamide/urea gels (29) or on three-step polyacrylamide gels (30); the pattern of migration of the small subunits in the latter system was variable (see text), and no definitive assignments are proposed.
b Calculated from nucleic acid sequence data quoted in the text, taking into account known posttranslational modifications (processing and cofactors). FIG. 5. Subunit stoichiometry in 14 C-labeled b 6 f complex. A, autoradiogram of purified 14 C-labeled b 6 f complex following urea/SDS-PAGE. B, relative radioactivity associated with each subunit or group of subunits; the radioactivity in each band was quantified using Phosphor-Imager plates, divided by the number of carbon atoms contained in the subunit(s) present in the band and normalized to the average activity per carbon atom in the complex (see "Materials and Methods"). Panel shows average Ϯ S.D. of 10 measurements on two different preparations. C, a scan through the middle of the lane shown in panel A; note that because of the variable width and broadness of the bands, peak heights and areas are not proportional to relative activities. and the cost of detergent.
In this latter respect, Hecameg (HG) appears as an economically interesting substitute to the much more expensive octyl-␤-D-glucopyranoside (OG) for membrane protein purification. HG has been developed by Plusquellec and co-workers (19). Its chemical structure, molecular mass, aggregation number, and CMC are similar to those of OG (19,44,45), and it is likely to be able to substitute for it under many circumstances. HG has been hitherto applied to the purification of mycoplasma surface antigens (19,59), to the solubilization of the sarcoplasmic Ca 2ϩ -ATPase (45), and to the crystallization of beef heart cytochrome bc 1 (60). Proteins that are sensitive to OG are not good candidates for purification with HG, as indicated by preliminary experiments with Ca 2ϩ -ATPase (45) 6 and with the nicotinic acetylcholine receptor. 7 C. reinhardtii b 6 f complex, on the other hand, is solubilized by HG somewhat more specifically than by OG, and purified preparations in HG are stable for weeks provided lipids are also present.
Characterization of Redox Components-Knowledge of the spectral and redox properties of the C. reinhardtii b 6 f complex will be essential for conducting experiments on its function. The absorbance spectra of the cytochromes of the b 6 f complex were found to be similar to those of higher plants and of other algae (61)(62)(63). The only minor differences were the absorbance peaks in the ␣ band regions of the hemes; those for cytochrome f, cytochrome b l , and cytochrome b h were found to be 554, 564, and 564 nm, respectively. Within the noise level the peaks from the two cytochrome b 6 hemes were indistinguishable. In contrast, spinach cytochrome b l was found to have an ␣ peak shifted by 1-2 nm to the red (64,65), whereas, in C. sorokiniana, cytochrome b l is shifted by approximately 1 nm to the blue (18). It seems from these data that the cytochrome b 6 ␣ peaks can vary by a few nm from species to species, even though their respective function and redox properties are conserved. The absorbance spectrum of cytochrome f in the purified b 6 f complex is nearly identical to that in intact cells of C. reinhardtii. 8 Anaerobic redox titrations were performed on isolated cytochrome b 6 f complex in order to determine the redox midpoint potential of the components. Cytochrome b l and cytochrome b h were found to have E m,8 values of Ϫ158 and Ϫ80 mV, respectively, similar to those found in spinach cytochrome b 6 f in thylakoids at pH 8.3 (65). Cytochrome f was found to have a slightly lower E m (approximately ϩ340 mV) than reported for other systems (ϩ350 to ϩ370 mV; Refs. 1 and 65).
Polypeptide Composition-Earlier work on C. reinhardtii b 6 f complex had led to the identification of four subunits, cytochromes f and b 6 , the Rieske protein, and subunit IV (11,12). More recent studies have revealed the additional presence of three ϳ4-kDa subunits, products of the petG (15,16), petX (15,16), and ycf7 (petL) (47) 1 genes. We present elsewhere the complete amino acid sequence of PetX, obtained by Edman degradation of the mature polypeptide and by cloning and sequencing of the petX gene (17,66). Sequence examination suggests that each of the three miniproteins spans the thylakoid membrane with a single transmembrane ␣-helix. In keeping with this expectation, the PetG (14, 33), PetL, 1 and PetX (17,66) proteins behaved as integral proteins upon extraction with high pH, high salt, or chaotropic agents.
We show elsewhere that the purified complex is a dimer, whose experimentally determined mass matches well that expected from its protein and detergent composition (43). Equilibrium 14 C labeling is compatible with each subunit being present as two copies per b 6 f dimer. The accuracy of the data, however, does not permit exclusion of the possibility that one or the other of the 4-kDa subunits is present as a single copy per dimer.
Lemaire et al. (11) noted that a 19.5-kDa polypeptide, which was present in partially purified preparations of C. reinhardtii b 6 f complex, was missing from the thylakoid membranes of b 6 f-less strains. They considered it an additional b 6 f subunit, which they called "subunit V." Other authors (1,8,14) subsequently have called "subunit 5" or "subunit V" the product of the petG gene (initially called petE) identified in maize b 6 f preparations (14). In order to avoid confusion while retaining some similarity with the former denomination, we will designate the 19.5-kDa protein hereafter by the italic letter v. Our results show that protein v is absent in purified b 6 f preparations that are highly active in electron transfer. This absence was confirmed by immunodetection using an antiserum raised against a synthetic peptide with the sequence of the N terminus of protein v. 9 From this point of view, protein v does not seem to be an essential component of C. reinhardtii b 6 f complex. On the other hand, the earlier observations indicate that the accumulation of protein v in thylakoid membranes somehow depends on the presence of cytochrome b 6 f. One simple explanation would be that, in situ, protein v physically associates with the b 6 f complex. The absence of protein v in b 6 f-less mutants then might result from its more rapid degradation when not bound to its proper site. While protein v clearly is not required for the b 6 f efficiently to transfer electrons and is not strongly bound to the complex, it cannot be excluded that it carries out an unidentified function associated with it. Further characterization of protein v is in progress in our laboratory.
Conclusion-This work describes a purification protocol that provides for the first time preparations of the b 6 f complex from C. reinhardtii that are both highly pure and highly active in electron transfer. It establishes the polypeptide composition of the complex and some of its physico-chemical properties. These data further assert the great similarity of the b 6 f complex from C. reinhardtii to that from higher plants and, therefore, its usefulness as an experimental model. The purification protocol, as well as a number of other experimental tools established in the course of this work, such as a library of subunit-specific antisera, should be of great use in future studies of the complex. Examination of its non-proteic components and crystallogenesis experiments are currently in progress (47). 10