On the Presence and Role of a Molecule of Chlorophylla in the Cytochromeb 6  f Complex*

Highly purified preparations of cytochromeb 6  f complex from the unicellar freshwater alga Chlamydomonas reinhardtii contain about 1 molecule of chlorophyll a/cytochrome f. Several lines of evidence indicate that the chlorophyll is an authentic component of the complex rather than a contaminant. In particular, (i) the stoichiometry is constant; (ii) the chlorophyll is associated with the complex at a specific binding site, as evidenced by resonance Raman spectroscopy; (iii) it does not originate from free chlorophyll released from thylakoid membranes upon solubilization; and (iv) its rate of exchange with free, radioactive chlorophyll a is extremely slow (weeks). Some of the putative functional roles for a chlorophyll in the b 6 f complex are experimentally ruled out, and its possible evolutionary origin is briefly discussed.

In the present article, we present further evidence in support of our earlier conclusion that the native b 6 f complex from Chlamydomonas reinhardtii comprises 1 molecule of chlorophyll a (Chla) per monomer as an authentic component (14). Namely: (i) free [ 3 H]Chla added to C. reinhardtii thylakoid membranes at the time of solubilization does not associate with the b 6 f complex; 2 (ii) the rate of exchange of b 6 f-associated Chla for free [ 3 H]Chla is extremely slow; and (iii) Chla is bound to the b 6 f complex at a single, specific site. Putative functional roles for a chlorophyll in the b 6 f complex are examined and some of them are experimentally ruled out.
Strains and Growth Conditions-Wild-type strain (WT) and mutant strain LDS (lacking chlorophyll synthesis when grown in the dark) were kindly provided by J. Girard-Bascou and P. Bennoun (CNRS UPR 9072, Institut de Biologie Physico-Chimique). C. reinhardtii was grown in Tris acetate-phosphate medium (TAP) (15) at 25°C under an illumination of 300 -400 lux (WT) or in the dark (LDS) on a rotary shaker until stationary phase (ϳ10 7 cells/ml). Cells were harvested at 5,000 ϫ g for 10 min. Thylakoid membranes were prepared as described previously (WT, Ref. 16;LDS,Ref. 17), resuspended in 10 mM Tricine-NaOH, pH 8.0, containing protease inhibitors (200 M PMSF, 1 mM benzamidine, 5 mM ⑀-aminocaproic acid), and stored at Ϫ80°C. The final concentration of WT membranes was adjusted at 3 mg of Chl/ml. The concentration of LDS etioplast membranes was estimated from their optical density at 460 nm.
Preparative and Analytical Techniques-Cytochrome b 6 f complex was purified and analyzed, and its PQH 2 -plastocyanin oxidoreductase activity was determined as described previously (13). UV-visible absorbance spectra were recorded either on a Kontron Uvikon 930, a Varian Cary 2300, or a Joliot-type homemade spectrophotometer (18), as specified in the legends to Figs. 2, 3, and 6. Cytf concentrations were determined from the ascorbate-reduced minus ferricyanide-oxidized spectra, using ⑀ 554 ϭ 18,000 M Ϫ1 ⅐cm Ϫ1 (19); A 554 was measured by reference to a line joining isosbestic points at 545 and 575 nm. Chl concentrations were determined from the absorption at 668 nm, using ⑀ 668 ϭ 75,000 M Ϫ1 ⅐cm Ϫ1 (20; we checked that the extinction coefficient of Chla bound to the complex is identical to that in acetone).
Pigment Analysis-Pigments were extracted from thylakoid membranes or from cytochrome b 6 f preparations by 10 volumes of ice-cold 100% acetone under vigorous stirring. Precipitated proteins were spun down at 5,000 ϫ g for 10 min. The supernatant was collected, evaporated to dryness in a glass flask under a flow of N 2 and stored at Ϫ80°C. Pigments were first separated by chromatography on thin-layer silica gel plates according to Eichenberger and Grob (21). After methanol extraction from the silica powder, each fraction was further purified by reversed-phase HPLC on a Zorbax-ODS column (Rockland Technologies, Inc.; 4.6 ϫ 250 mm, 5 m granulometry). Elution proceeded in the following three phases: (i) during 8 min, 0.1% methylene chloride in acetonitrile/methanol (70:30 v/v); (ii) during 4 min, a 0.1-40% (v/v) gradient of methylene chloride in the same solvent mixture; and (iii) a constant concentration of 40% methylene chloride in the same mixture. The absorption spectrum of the eluted fractions was continuously monitored with a Hewlett-Packard 1040 A diode array detector (wavelength range 230 -600 nm). The detector response was calibrated with standards, using extinction coefficients given by Lichtenthaler (22) for pheophytins and chlorophylls, by Britton (23) for carotenoids, and by Barr and Crane (24) for quinones. Preparation of [ 3 H]Chlorophyll a-Wild-type C. reinhardtii cells were grown in TAP medium under standard conditions until stationary phase, diluted 10 times into 200 ml of TAP medium containing 3.7 GBq of sodium [ 3 H]acetate, and further grown under about 1000 lux until stationary phase. Cells were harvested, thylakoid membranes were prepared, and 3 H-labeled pigments were separated as described above (Fig. 1). Their specific activity was determined by liquid scintillation counting in Aqualuma in a LS1801 counter (Beckman) and spectrometry.
Purification of Cytochrome b 6 f Solubilized in the Presence of Radioactive Chlorophyll-A 500-l sample of C. reinhardtii thylakoid membranes (containing 1.5 mg Chl) in 10 mM Tricine-NaOH buffer, pH 8.0, plus protease inhibitors, was solubilized by addition of an equal volume of HG 50 mM in 2ϫ concentrated TMK buffer supplemented with 37 g of [ 3 H]Chla (3.25 ϫ 10 7 cpm). After 15 min of incubation at 4°C in the dark and 10 min of centrifugation at 80,000 rpm (160,000 ϫ g) in the TLA 100.3 rotor of a TL100 ultracentrifuge (Beckman), the supernatant was layered on top of an 11-ml 10 -30% (w/w) sucrose gradient in TMK-HP buffer and centrifuged for 24 h at 40,000 rpm (270,000 ϫ g) in the SW41 rotor of an L8 ultracentrifuge (Beckman). Fractions of 400 l were collected. The top fractions were analyzed for Chl and radioactivity. The fractions containing the b 6 f complex were pooled, and the complex was purified by HA chromatography as described (13). The specific activity of free Chl was taken to be that of the Chl present in the uppermost three fractions of the gradient (Fig. 5).
Determination of the Rate of Exchange of b 6 f-bound for Free 3 Hlabeled Chlorophyll-An acetone solution containing 2 nmol of [ 3 H]Chla (ϳ480,000 cpm) was evaporated to dryness under N 2 . The [ 3 H]Chla was redissolved in 0.5 ml of a 5 M solution of purified b 6 f complex in AP-HP buffer and incubated in the dark under N 2 at 4°C. At time intervals, 0.1-ml aliquots were layered onto 2-ml 10 -30% (w/w) sucrose gradients in TMK-HP buffer and centrifuged for 3 h at 55,000 rpm (260,000 ϫ g) in the TLS 55 rotor of a TL100 ultracentrifuge (Beckman) to separate free from b 6 f-bound Chl. The brownish band containing the complex was collected, and the specific radioactivity of b 6 f-bound Chl was determined and compared with that of the free Chl present at the top of the gradient.
Fluorescence Measurements-Low-temperature (77 K) fluorescence spectra were recorded on a homemade instrument (see Ref. 25). The exciting beam and the fluorescence emission were passed through a Y-shaped light guide. The sample (ϳ5 M in AP-HP buffer) was placed in a flat quartz cuvette (0.1-mm light path), immersed in liquid nitrogen, and held at the common end of the guide. The exciting beam wavelength was selected by a monochromator, with a bandwidth set at 3 nm for excitation spectra and 12 nm for emission spectra. The fluorescence emitted by the sample was monitored by a photomultiplier through a monochromator with a bandwidth of 12 nm for excitation spectra and 3 nm for emission spectra. Spectra were recorded as uncorrected responses of the photomultiplier.
Resonance Raman Spectroscopy-Resonance Raman spectra were recorded with a U1000 Raman spectrometer equipped with a chargecoupled device camera (Jobin-Yvon, France) on pellets of oxidized cytochrome b 6 f obtained by ultracentrifugation after exposure to 1.5 mM ferricyanide and dilution under the critical micellar concentration of HG. The 441.6 nm excitation light (less than 15 milliwatts on the sample) was provided by a HeCd continuous laser (Model 4270N, Liconix, CA). To prevent photodegradation of Chla during the experiments, samples were cooled at 77 K in a gas flow cryostat (TBT, France).
Photobleaching of b 6 f-bound Chlorophyll a-Purified b 6 f complex in AP-HP buffer was diluted 10 times into 20 mM HG, 0.1 g/liter PC, to a final concentration of 0.5 M Cytf, 40 mM AP. A 300-l sample was irradiated at 4°C under gentle stirring by the two light beams produced by a KL 1500 lamp (Schott; power set at 3). The white light was filtered by a red Wratten low-pass filter No. 92 (Kodak; cut-off wavelength 620 nm) and by 1-cm plastic cuvettes filled with water.

RESULTS
Highly Purified Preparations of b 6 f Complex from C. reinhardtii Contain One Molecule of Chlorophyll a/Cytochrome f-The b 6 f complex from C. reinhardtii contains seven subunits in stoichiometric ratio and four identified redox carriers, one c-type heme, two b-type hemes, and a [2Fe-2S] cluster (13). In addition to the three cytochromes, UV-visible spectra of even the most highly purified preparations reveal the presence of carotenoids (absorbance peaks at ϳ460 and 483 nm) and of Chla (peak at 667-668 nm) (13). Within experimental accuracy, the visible spectrum of the Chl does not depend on the redox state of the complex (Ref. 13; see Fig. 3A). Using the in situ extinction coefficient of b 6 f-associated Chla (⑀ 668 ϭ 75,000 M Ϫ1 ⅐ cm Ϫ1 ; cf. "Experimental Procedures") and an extinction coefficient ⑀ 554 ϭ 18,000 M Ϫ1 ⅐ cm Ϫ1 for Cytf (19), the Chla/Cytf ratio was found to be 0.93 Ϯ 0.18 (mean Ϯ S.D. over 26 preparations). Chemical analysis confirmed that b 6 f preparations contain essentially pure Chla; Chlb, which makes up to ϳ30% of Chl in thylakoid membranes from WT C. reinhardtii, represents less than 10% of Chl in purified b 6 f (Table I). Carotenoids are present in substoichiometric ratio with respect to Chla, while other pigments and quinones either are totally absent or are present in trace amounts (Table I).
The approximate 1:1 molar ratio of Chla to Cytf, the excess of Chla over Chlb, as compared with thylakoid membranes, and the retention of Chla throughout the purification procedure suggest that there exists, on the b 6 f complex, one binding site with high affinity and specificity for Chla. However, the average stoichiometry is somewhat smaller than 1:1, and its variation from one preparation to the next tends to be larger than the uncertainty on the measurements would lead one to expect. Several factors may explain the dispersion of the data. (i) Traces of Chl collected from the sucrose gradient and incompletely washed from the hydroxylapatite column may contaminate some preparations; (ii) the b 6 f-associated Chl is easily bleached (see below); and (iii) exposure of the complex to de- tergent micelles tends to release the Chl (26).
The Spectrum of the b 6 f-associated Chlorophyll a Is Affected by Its Interactions with the Complex-Exposing the complex to an excess of laurylmaltoside (LM) micelles induced a bathochromic shift of the Chla peak by ϳ2 nm, from 667-668 to 669 -670 nm ( Fig. 2A). Similar shifts were observed following denaturing treatments, such as heating the preparation at 50°C or adding 8 M urea, and occurred whether the b 6 f complex was in its detergent-solubilized state or reconstituted into lipid vesicles (Fig. 2B).
The spectrum of Chla in b 6 f preparations treated with an excess of detergent resembles that of pure Chla dissolved in LM micelles ( Fig. 2A), suggesting that this treatment releases Chl from the complex. Delipidation by detergents indeed induces dissociation of the complex into chlorophyll-free monomers (26). However, closer examination reveals that the spectral shift actually precedes Chl dissociation. We show elsewhere that mild treatment of the complex with detergent first generates a dimeric form that has lost the Rieske protein and retains the Chl, while a harsher treatment is required for the complex to release the Chl and break down into monomers (26). Analysis of the visible spectrum of the Chl bound to purified, Rieskedepleted b 6 f dimer revealed a red-shifted (and broadened) absorption peak (Fig. 2C), indicating that the environment of the Chl has been affected even though it is still bound to the complex and co-purifies with it (26).
Fluorescence Characteristics of b 6 f-associated Chlorophyll-Interactions of the Chl with its environment in the b 6 f complex were further examined by low temperature fluorescence measurements. Cytochrome b 6 f exhibits, in the Soret region, several absorption bands due to the Chl, the carotenoids, and cytochromes f and b 6 , the latter bands being modulated by the redox potential (Ref. 13 and Fig. 3A). The possible occurrence of energy transfer between hemes and Chla was examined by analyzing the fluorescence characteristics of the Chl under various redox conditions. At 77 K, excitation at 440 nm in the Chl band of cytochrome b 6 f produced an emission of fluorescence with a maximum at 673 nm (Fig. 3B). Excitation spectra of the fluorescence emitted at 673 nm and emission spectra of the fluorescence excited at 440 nm were recorded in the presence or absence of 5 mM ascorbate or ϳ5 mM sodium dithionite. The effects of these additions on the redox state of the cytochromes were checked by absorption spectroscopy (Fig. 3A).
Regardless of the addition, there was no significant change in the fluorescence excitation and emission spectra (Fig. 3C, and data not shown).
Incubation at room temperature for 1 h with 100 mM HG, which is known to induce partial dissociation of the cytochrome b 6 f complex (26), did not modify the shape of the fluorescence spectra, but enhanced fluorescence intensity by a factor of ϳ2 (Fig. 3B). A more limited increase in fluorescence intensity was observed after freezing and thawing the b 6 f solution (Table II). A comparison of these fluorescence intensities with that of free Chla in the same buffer is shown in Fig. 3B and Table II. These observations indicate that association of Chla with the b 6 f complex results in a redox-independent quenching of its fluorescence (by a factor of ϳ4), which is partially relieved following detergent treatment.
Chlorophyll a Is Bound to the b 6 f Complex at a Specific Site-The mode of binding of Chla to the protein was further investigated using resonance Raman spectroscopy. To detect TABLE I Prosthetic group composition (mol/mol ratios) of thylakoid membranes and cytochrome b 6 f complexes purified from either wild-type or LDS C. reinhardtii strains Most determinations were performed on two different wild-type preparations. All other pigments (antheraxanthin, lutein-5,6-epoxide, zeaxanthin, ␣-carotene, ␤-carotene-5,6-epoxide) were either absent or present in trace amounts (Յ1% of total pigment mass). WT LDS  selective contributions from the Chl molecules present in the samples, experiments were performed at 441.6 nm excitation wavelength on oxidized cytochrome b 6 f. This laser line is located on the red side of the Soret electronic transition of Chla, more than 1500 cm Ϫ1 away from the Soret band of the oxidized cytochromes (ϳ413 nm). As expected, resonance Raman spectra recorded under these conditions led to barely detectable signals from cytochrome b 6 . Nevertheless, under these conditions of excitation, intense contributions typical of carotenoid molecules partially masked the middle frequency modes of Chla (data not shown). Analysis, therefore, was focused on the high-frequency region of the spectrum (Fig. 4). Below 1600 cm Ϫ1 , resonance Raman spectra of Chla molecules typically feature an intense band at ϳ1550 cm Ϫ1 , which has been attributed to complex vibrational modes of the chlorin ring (27). Between ϳ1600 and 1710 cm Ϫ1 , two or three bands may be observed: (i) a band between 1595 and 1615 cm Ϫ1 , arising from the stretching modes of the methine bridges of the molecule (27), the frequency of which depends on the conformation of the chlorin ring (28) and is thus sensitive to the coordination state of the central Mg 2ϩ ion (27); (ii) a band at ϳ1620 cm Ϫ1 , arising from the stretching mode of the conjugated vinyl group in position C 2 (29), and often just appearing as a weak shoulder on the high frequency side of the methine stretching band (29); and (iii) between 1640 and 1710 cm Ϫ1 , bands arising from the stretching modes of the conjugated 9-keto carbonyl group, the frequency of which is extremely sensitive to the H-bonding state and to the environment of this group (27,30).
In the resonance Raman spectrum of the b 6 f-bound Chl (Fig.  4), the band arising from the methine bridge stretching modes is observed at 1606 cm Ϫ1 , and is, as expected, asymmetric, because of the presence of the weak contribution of the vinyl stretching modes. At higher frequencies, a single band is observed, at a frequency of 1676 cm Ϫ1 . The full width at halfmaximum of these two bands (14 and ϳ11 cm Ϫ1 , respectively) is similar to that observed in resonance Raman spectra of isolated, monomeric Chla molecules (ϳ12 cm Ϫ1 ; see Ref. 27). Since the frequency of the 9-keto carbonyl mode is extremely sensitive to the environment of this group, this indicates that the binding sites of all Chla molecules share very similar, if not identical physicochemical properties. The frequency of this band is as high as 1696 cm Ϫ1 when the keto group is free from intermolecular interactions (27,30). A 1676 cm Ϫ1 frequency unambiguously indicates that this group is involved in a medium-strength hydrogen bond.
The 1606 cm Ϫ1 frequency observed for the methine bridge stretching mode is somewhat ambiguous with respect to the liganding of the Mg 2ϩ ion. It could originate either from a 5-coordinated Chla molecule with an unusually planar conjugated system or from a 6-coordinated molecule slightly distorted by its proteic environment (28,31). Imidazole side chains of histidine residues are known to be particularly strong ligands for the central magnesium atom of (bacterio)chlorophyll molecules. In most of the well-documented cases where a magnesium atom interacts with a histidine residue, the methine bridge band is observed at 1612-1615 cm Ϫ1 at low temperature (see Ref. 31, and references therein). A 1606 cm Ϫ1 frequency makes it extremely unlikely that the Chla of the b 6 f complex  interacts with such a strong ligand. Chlorophyll a Does Not Become Artifactually Bound to the Complex during Solubilization-The existence of a single Chla binding site per b 6 f complex does not in itself exclude the possibility that this site is normally empty, or occupied by another ligand, and that Chla binding occurs artifactually when the thylakoid membrane is disrupted by the detergent. Two indirect arguments militate against this view, but do not strictly rule it out. (i) The presence of Chl in b 6 f complex preparations has been reported in several species, which implies evolutionary conservation of the site; and (ii) it is not clear where a Chl molecule artifactually picked up by the b 6 f complex would originate from since very little Chl is actually set free by the solubilization process. Most of the Chl found in the supernatant from HG solubilization indeed is associated to proteins. Upon fractionating the supernatant on a sucrose gradient, it entered the gradient while free radiolabeled Chl added as a tracer stayed at the top (Fig. 5). Those few Chl molecules that did not enter the gradient, and were presumably free in detergent micelles, represented less than one-tenth of the amount associated with the b 6 f. Should one accept the hypothesis that free Chla becomes artifactually bound to the b 6 f complex upon solubilization, one would have to contend with the curious coincidence that the solubilization process releases only the amount of Chl that is needed to saturate the complex, and no more.
To more directly rule out this possibility, we have purified C. reinhardtii b 6 f from membranes solubilized using detergent micelles preloaded with 3 H-labeled Chla and compared the specific radioactivity of free versus b 6 f-associated Chl. In a first experiment (experiment 1 in Table III), the regular protocol for b 6 f purification (13) was followed, except for the addition of trace amounts of [ 3 H]Chla to the HG stock solution used to solubilize the membranes (see "Experimental Procedures"). The specific radioactivity of the Chl associated with purified b 6 f was found to be ϳ10-fold lower than that of the free Chl collected from the top of the gradient. In a second set of experiments, we checked on the possibility that [ 3 H]Chla initially bound to cytochrome b 6 f upon solubilization might back-exchange with the unlabeled Chla bound to antenna proteins that comigrate with the complex during the first hours of the centrifugation, thus lowering the final specific activity. Immediately following solubilization, half of the sample was adsorbed onto a HA column, washed, and eluted as described (13). This resulted in the rapid removal (within ϳ15 min) of most of the antenna proteins. This sample (experiment 3) was then layered on a sucrose gradient along with the other half of the supernatant (experiment 2), and the two samples were purified by ultracentrifugation and HA chromatography. As shown in Table III, early separation of the freshly solubilized b 6 f from Chl-containing proteins does not increase the specific activity of the Chl molecule in the final purified complex; it actually decreases it, because this protocol prevents the limited exchange of b 6 f-bound for free Chl that can take place at the top of the gradient during the first hours of centrifugation. The different specific activities of b 6 f-bound Chl in experiments 1 and 2 probably stem from a higher level of contamination in experiment 1 by free, radioactive Chl.

Rate of Exchange of Bound and Free
Chlorophyll-In keeping with these observations, the rate of exchange between b 6 f-bound Chl and free Chl was found to be very slow. Purified b 6 f was incubated at 4°C with [ 3 H]Chla in AP-HP buffer. At intervals, aliquots were removed and free Chl separated from the complex by sucrose gradient centrifugation. The specific radioactivity of b 6 f-bound Chl increased slowly, until after 10 days it reached approximately one-third that of the free Chl recovered from the top of the gradient (Table IV).
Cytochrome b 6 f Accumulates in a Chlorophyll-poor C. reinhardtii Mutant-Accumulation of the b 6 f complex was examined in C. reinhardtii LDS mutant. When grown in the dark, cells from this strain etiolate due to the almost complete shutdown of Chl synthesis. 3 The Chl content of thylakoid membranes purified from dark-grown LDS cells was severely depleted, as reflected in the fact that the Chla/␤-carotene ratio 3 P. Bennoun, personal communication.  (Table III, experiment 1). dropped by a factor of 20 -30 (Table I). Such membranes contained almost no photosystem I, photosystem II, and lightharvesting complex proteins (not shown), but they did contain cytochrome b 6 f (Fig. 6). No cytochrome b 6 f accumulated, on the other hand, in mutant strain ⌬Gid (kindly provided by P. Bennoun, IBPC, Paris) in which Chl synthesis is totally blocked (not shown).
Chlorophyll a Is Not Essential to in Vitro Electron Transfer by Cytochrome b 6 f-To examine whether Chl plays a role in the electron-transfer function of cytochrome b 6 f, purified samples were exposed to red light, and their enzymatic activity was followed as a function of time, in parallel with the bleaching of Chla. We have shown previously that, under our experimental conditions, the rate of electron transfer from PQH 2 to plastocyanin is limited by the rate of Cytf/plastocyanin collisions and is strictly proportional to the concentration of active b 6 f complex (13). As shown in Fig. 7, bleaching up to 70% of the Chla in a b 6 f preparation had no effect on its PQH 2 -plastocyanin oxidoreductase activity. DISCUSSION Our conclusion that a molecule of chlorophyll a is a genuine component of the cytochrome b 6 f complex is based on a central observation and backed by a number of corroborative experiments. The most direct observation is the lack of radioactivity in b 6 f complex preparations that have been solubilized and purified in the presence of [ 3 H]Chla. Depending on the exact way the experiment was performed, the radioactivity of the Chl associated with the complex varied between ϳ10 and Ͻ1% that of the free Chl present in the supernatant, demonstrating that b 6 f-bound Chl does not originate from free Chl artifactually picked up by the complex. Our data rule out the (far-fetched) possibility that free, radioactive Chla becomes initially bound upon solubilization but is replaced in the course of purification by back-exchange with the non-radioactive Chla present on the small fraction of antenna proteins co-solubilized with the b 6 f complex.
More circumstantial evidence runs as follows: (i) the mole ratio of Chla to Cytf in highly purified preparations is always close to 1:1 (0.93 Ϯ 0.18 over 26 preparations); (ii) very little Chl is actually set free by the solubilization process (Ͻ10% of that recovered in the b 6 f complex); (iii) spectral changes upon mild treatment of the complex with detergents show that the Chl, while still bound to the complex, is sensitive to its confor-mational state and/or to the presence of the Rieske subunit; (iv) resonance Raman data indicate that all Chl molecules in a b 6 f preparation experience an identical, specific environment; and (v) there is spectroscopic evidence for the presence of Chla in the b 6 f complex in vivo (32; see below).
While these observations leave little doubt that the native b 6 f dimer contains two molecules of Chla, they raise a number of questions. One of them is how the complex is protected from oxidation by the triplet state of Chla generated upon illumination. From this point of view, it is probably significant that purified b 6 f preparations also contain carotenoids, albeit at a substoichiometric level. Experiments are in progress to determine their origin and the extent of protection they confer against photooxidation.
Another question is the localization of the chlorophyll a molecule in the complex and the identity of the subunit(s) it interacts with. Spectroscopic evidence indicates that a molecule of Chla is located close to the Q 0 site in vivo in both C. reinhardtii and Chlorella sorokiniana (32). This conclusion is based on the observation of a Chl spectral shift that correlates with proton release at this site. Our observations indicate that, while the Chla molecule remains bound to cytochrome b 6 f dimer depleted of the Rieske protein (26), its spectrum shifts to the red (this work). The latter effect is compatible with a localization of the Chl close to Q 0 , where it would become exposed to a more polar environment upon removal of the Rieske protein. It cannot be excluded that part of the red shift of the Chl spectrum is a consequence of the delipidation that is used to trigger the dissociation of the Rieske protein (26), but it is notable that a similar shift also occurred following heat treatment of purified b 6 f reconstituted into lipid vesicles. Also compatible with a localization close to Q 0 is the recent observation that chlorophyll comigrates with cytochrome b 6 upon SDS-polyacrylamide gel electrophoresis analysis of Synechocystis PCC 6803 thylakoid membranes. 4 We have not been able, unfortunately, to corroborate this experiment using purified C. reinhardtii b 6 f. Resonance Raman data indicate that the 9-keto group of the Chl is, in the native complex, involved in a medium strength hydrogen bond. They are more ambiguous in regard to the number of ligands to the Mg 2ϩ ion, but they do not support 5-fold liganding with a strong electron donor such as 4 R. Barbato, personal communication.  a The reason for which the specific radioactivity of free Chl␣ determined at time zereo was off by a factor of ϳ2 has not been identified.
histidine as the fifth ligand. Experiments are in progress to determine the orientation of the chlorin ring with respect to the membrane plane and the accessibility of the Chl from the lipid and aqueous phases. More fundamentally, the presence of Chla in the b 6 f complex raises the question of its eventual function. The plastoquinolplastocyanin oxidoreductase activity of the b 6 f complex is not driven by the energy of light, and the redox potentials of either ground state or excited state Chla (33) are too different from those of either of the redox carriers in the b 6 f for electron transfer to the Chl to be contemplated. The spectrum of the b 6 f-associated Chl, indeed, does not depend on the oxidized or reduced state of the cytochromes. The homologous ubiquinolcytochrome c oxidoreductase, the cytochrome bc 1 complex, does not require any Chl for its function. We have examined a number of possibilities-none of which, admittedly, is very compelling-and tried to rule out some of them.
Light Protection-A role of the Chl in deactivating the hemes appears a priori unlikely, given the high efficiency of quenching of heme fluorescence by the iron atom. Indeed, fluorescence energy transfer measurements show no change of the excitation spectrum of the Chl upon heme reduction.
Facilitation of Electron Transfer-This also is not a priori very likely given that conjugated double bond systems such as that of the chlorin ring are thought to hamper rather than facilitate electron tunneling (34). Bleaching of the b 6 f Chl with red light resulted in no loss of the PQH 2 -plastocyanin oxidoreductase activity. It should be noted, however, that this experiment rules out a complete shut-down of electron transfer upon Chla bleaching, but not a slowing down, since electron tunneling is not the rate-limiting step under our experimental conditions. This point is under more detailed examination.
Regulation of b 6 f Assembly-In this hypothesis, binding of Chl would regulate the stability of the complex, preventing accumulation of b 6 f in the absence of light-energy transduction. This idea holds little appeal given that the complex accumulates in cells of mutant LDS grown in the dark, even though their low content of Chl prevents accumulation of both reaction centers and light-harvesting complexes. The substoichiometric ratio of Chl to cytochrome f in b 6 f preparations purified from etioplasts of this mutant suggests that, while the complex may bind Chl even under these conditions, a full complement of 2 Chl/b 6 f dimer is not necessary to its accumulation. We have not, on the other hand, observed any accumulation of b 6 f in mutant ⌬Gid, which does not synthesize any Chl at all. Whether this is a direct or an indirect consequence of the absence of Chl is not known.
A Structural Role-There are precedents for prosthetic groups that serve no catalytic function or that are used in atypical manners. The inactive electron transfer branch in purple bacteria reaction centers appears to fulfill a primarily or purely structural role (see e.g. Refs. 35 and 36). Pyridoxal 5Ј-phosphate, whose role in enzymatic catalysis usually depends on Schiff base formation between its aldehyde function and amino acid amino groups, is used by glycogen phosphorylase in a totally different way. Its 5Ј-phosphate group serves as a proton donor-acceptor shuttle while the Schiff base that associates it to the protein can be reduced without loss of activity (see Refs. 37 and 38). Chla being freely available in thylakoid membranes could have been recruited by the b 6 f either as a mere building block in the assembly of the complex or in a catalytic function not necessarily related to its usual roles as a light harvester and an exciton or electron carrier. The hypothesis of a purely structural role, of course, is very difficult to rule out. In the present case, it is made particularly unappealing by the fact that incorporation of Chla implies the simultaneous development of photoprotecting devices, such as the additional recruitment of carotenoids.
An Evolutionary Relic-One possible view would be that some of the small subunits of the b 6 f complex, which have no equivalent in bc 1 complexes, originated from Chl-binding proteins that were recruited by the complex at a late stage in FIG. 6. Visible absorption spectra of b 6 f complexes purified from wildtype and LDS C. reinhardtii strains.
Cytochrome b 6 f was purified from WT and LDS strains as described under "Experimental Procedures," and the preparations were analyzed for Cytf (left), following addition of ascorbate, and for Chl (right). Spectra were recorded on a Kontron Uvikon 930 spectrophotometer normalized to the same absorbance at 554 nm and vertically displaced. Purified b 6 f complex was irradiated with red light ( Ͼ620 nm) at 4°C. Chla absorbance at 667 nm (f) and PQH 2 -plastocyanin oxidoreductase activity (ϩ) are normalized to their initial value. evolution and that have lost most, but not all, of their Chlbinding sites. There is indeed a very low level of sequence identity between b 6 f subunits PetG and PetM and the ␣ subunit of purple bacteria light-harvesting proteins. 5 There are several arguments against this hypothesis, among which are the following. (i) Chloroplasts originate from cyanobacteria with quite different light-harvesting systems (however, see Ref. 39); (ii) in Synechocystis PCC 6803, Chl has been observed 4 to be associated with apocytochrome b 6 , even though (iii) the genome of Synechocystis PCC6803 contains genes homologous to petG and PetM (40).
An alternative view would hold that Chla, similar to the inactive electron transfer branch in reaction centers, is a relic from an earlier evolutionary stage of the b 6 f complex. This could be understood, for instance, if quinol oxidoreductases and reaction centers shared a common, photochemically active ancestor. Such a hypothesis is relatively straightforward if, as is often thought, photosynthesis predates respiration (see e.g. Refs. 36 and 41). It becomes more involved if, as proposed by some recent evolutionary schemes, reaction centers evolved in the context of electron transfer chains where b-type cytochromes already operated (7,39,42,43). Further studies of the b 6 f Chl-binding site and of the arrangement of the two Chl in the b 6 f dimer, examination of the distribution of bf/bc-associated (bacterio)chlorophylls among phyla, and comparison of the upcoming three-dimensional structure of mitochondrial bc 1 complex with those of reaction centers should contribute to shedding light on this puzzling question.