A new subunit of cytochrome b6f complex undergoes reversible phosphorylation upon state transition.

A 15.2-kDa polypeptide, encoded by the nuclear gene PETO, was identified as a novel cytochrome b(6)f subunit in Chlamydomonas reinhardtii. The PETO gene product is a bona fide subunit, subunit V, of the cytochrome b(6)f complex, because (i) it copurifies with the other cytochrome b(6)f subunits in the early stages of the purification procedure, (ii) it is deficient in cytochrome b(6)f mutants accumulating little of the complex, and (iii) it colocalizes with cytochrome f, which migrates between stacked and unstacked membrane regions upon state transition. Sequence analysis and biochemical characterization of subunit V shows that it has a one transmembrane alpha-helix topology with two large hydrophilic domains extending on the stromal and lumenal side of the thylakoid membranes, with a lumenal location of the N terminus. Subunit V is reversibly phosphorylated upon state transition, a unique feature that, together with its topological organization, points to the possible role of subunit V in signal transduction during redox-controlled short term and long term adaptation of the photosynthetic apparatus in eukaryotes.

The cytochrome bc complexes such as bc 1 and b 6 f are ubiquitous in energy-transducing membrane systems; their major metabolic function is to couple the oxidation of quinols to the translocation of protons across the membrane, resulting in the establishment of a transmembrane electrochemical proton gradient required to drive the synthesis of ATP (1). Beside its role in photosynthetic electron transport, the cytb 6 f 1 complex is also recruited for redox sensing and signal transduction in chloroplasts (2,3). The cytb 6 f complex plays a key role in the supramolecular reorganization of the photosynthetic apparatus upon state transitions: changes in the redox state of plastoquinones (PQs) that bind to cytb 6 f complexes allow a photosynthetic cell to adapt to changes in light quality as well as to changes in intracellular ATP levels (4). At the molecular level, plastoquinol binding at the Q o site of cytb 6 f complexes on the lumen side of the thylakoid membranes activates a kinase that phosphorylates the light-harvesting antenna on the stromal side of the membranes (5)(6)(7)(8)(9). How the plastoquinol binding signal is transduced across the membrane for kinase activation remains unknown. This reversible protein phosphorylation process controls the migration of the antenna and the cytb 6 f complex between the stacked membrane regions containing photosystem II (PSII) and the unstacked membrane regions containing PSI (10). Studies with Chlamydomonas reinhardtii in vivo have shown that state I corresponds to a low phosphorylation of the peripheral antenna proteins that are mainly associated with PSII, with the photosynthetic apparatus being set for a linear electron flow from PSII to PSI aimed at carbon fixation; in state II, the peripheral antenna proteins become heavily phosphorylated, are mainly associated with PSI and the photosynthetic apparatus is set for cyclic electron flow and supplies high ATP levels (4,11).
C. reinhardtii displays the same photosynthetic apparatus as that of vascular plants but shows dispensable photosynthesis. Its cytb 6 f complex has been purified and comprises at least seven subunits (12). cytf, cytb 6 , and subunit (su)IV are encoded, respectively, by the chloroplast genes petA, petB, and petD (13), whereas the Fe 2 S 2 Rieske protein is encoded by the nuclear gene PETC (14). In addition, there are three small 4-kDa hydrophobic subunits, each forming a single transmembrane ␣-helix, products of the chloroplast genes petG (15) and petL (16), and the nuclear gene product PETM (17). Recently, a putative fourth small subunit has been identified as a product of a chloroplast gene petN in higher plants (18). This gene is located in the nucleus in Volvox (19) and in C. reinhardtii as suggested by homologies to expressed sequence tags (20).
Last, a nucleus-encoded 19-kDa polypeptide, termed suV, had been proposed to associate with the rest of the cytb 6 f subunits in C. reinhardtii, based on its presence in cytb 6 fenriched fractions and its absence or low representation in cytb 6 f-deficient mutants (21).
We report here the biochemical, molecular, and topological characterization of suV, encoded by the gene termed PETO and identified just after PETN, and provide evidence that it is conserved in other photosynthetic eukaryotes. The potential roles played by this cytb 6 f phosphoprotein in photosynthetic electron transport, redox sensing, and signal transduction are discussed.
suV Protein Microsequencing-Thylakoid membranes (1.5 mg of chlorophyll/ml) from the F54 mutant, deficient in ATP synthase, were solubilized 15 min at 4°C in the presence of 2.2% Mega-8 (w/v), 20 mM Tricine, pH 8.0, 3 mM KCl, and 3 mM MgCl 2 . The solubilized supernatant, which is enriched in cytb 6 f complex was separated on 12-18% SDS-polyacrylamide gels in the presence of 8 M urea, and electrotransferred onto polyvinylidene difluoride membranes in a semidry system as described previously (24). Sequencing of the band containing suV and of a tryptic fragment was performed according to Edman degradation by J. d'Alayer (Laboratoire de Microséquençage des Protéines, Institut Pasteur, Paris).
DNA and RNA Analyses-C. reinhardtii cDNA library in phage gt10 (25) was generously provided by L.-G. Franzén (Department of Plant Physiology, Botanical Institute, Göteborg University, Sweden). Phage DNA of the cDNA library was prepared out of 10 13 phage particles from liquid Escherichia coli culture as described previously (26). Desalted oligonucleotides were purchased from Oligo Express (Paris, France). Polymerase chain reaction (PCR) procedure followed was as in Ref. 6, and the annealing temperature was 55°C. Two phage-specific primers (5Ј-TGAGCAAGTTCAGCCTGGTTAAGTC-3Ј; 5Ј-GCTTAT-GAGTATTTCTTCCAGGGTA-3Ј) and a degenerate sense primer deduced from the N-terminal sequence of suV 5Ј-CTGCAGCAGCC(G/T/ C)GT(A/G/T/C)CTGAAGAAGGC(G/T/C)TTCCAGGA-3Ј were designed to amplify the 3Ј part of the PETO cDNA. Either of the two phage primers in combination with the sense degenerate primer led to a 1.1-kilobase (kb) PCR amplification product when 0.5-1 g of library DNA was used as a template. An antisense primer, 5Ј-CCCATCACGC-CCCAGCTCCC-3Ј, chosen from the sequence of the 1.1-kb PCR product was used to amplify, with either phage-specific primer, a 0.7-kb PCR product corresponding to the 5Ј part of the PETO cDNA. 1.3-kb full length cDNA was reconstituted using megaprime PCR by mixing the 1.1-and 0.7-kb products (obtained with different phage primers) and both phage primers. Sequencing was performed on both strands by Genome Express (Paris, France). Total RNA analysis was performed according to a previous study (27).
Protein Isolation, Separation, and Analysis-Biochemical analyses were carried out on cells grown to a density of about 2 ϫ 10 6 cells per milliliter. For polypeptide analysis, samples were resuspended in 100 mM 1,4-dithiothreitol and 100 mM Na 2 CO 3 and solubilized in the presence of 2% SDS at 100°C for 50 s. Polypeptides were then separated on 12-18% SDS-polyacrylamide gels in the presence of 8 M urea, cells, and thylakoid membranes were loaded at chlorophyll constant. Heme staining was detected by peroxidase activity of heme-binding subunits using 3,3Ј,5,5Ј-tetramethylbenzidine as described previously (21). Immunodetection using antisera raised against cytf, suIV, Rieske protein in combination with 125 I-labeled protein A, or an enhanced chemiluminescence method was carried out as in Ref. 6. Thylakoid membrane proteins were purified as done previously (24). Cytb 6 f complexes were purified by 6-O-(N-heptylcarbamoyl)-methyl-␣-D-glucopyranoside (Hecameg; Vegatec, Villejuif, France) solubilization of thylakoid membranes (12). Polypeptide extraction from thylakoids were performed by two cycles of 10-min incubation with various dissociating agents at room temperature followed by freeze/thaw as described previously (28). After Hecameg solubilization, the supernatant was subjected to sucrose gradient centrifugation and gradient fractions were then assayed for kinase activity using [␥-32 P]ATP and casein as substrate according to a previous study (5).
Antibodies Preparation-Anti-PSII PsbB, anti-Rieske, and anti-cytf antibodies were raised against the entire polypeptides. Anti-suV was raised against a synthetic peptide containing the sequence of the N terminus of mature suV (LQQPVLKKAFQDDTP) coupled to ovalbumin through an extra COOH-tyrosine and prepared by Neosystem (Strasbourg, France). Anti-cytf monoclonal antibody was kindly provided by O. Vallon (Institut de Biologie Physico-Chimique, Paris). Anti-PSII OEE2 preparation was described previously (29).
Proteolysis of Thylakoid Membranes-Thylakoid membranes were washed twice with 20 mM ammonium phosphate, pH 7.4, without protease inhibitors. Various concentrations of protease V8 or of trypsin were added to thylakoid membranes at a final chlorophyll concentration of 1 mg/ml. Half of each sample was sonicated for 10 s on ice. Trypsin and V8 treatments were performed at room temperature for 20 min, and the reaction was arrested by adding 2 mM phenylmethylsulfonyl fluoride as a protease inhibitor. The samples were analyzed by urea/ SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotting.
Electron Microscopy-Immunocytochemistry on thin sections was performed according to a previous study (29). Immunogold labeling of thylakoid membrane vesicles was performed according to earlier work (30). Anti-suV antibody raised against a synthetic peptide designed from the sequence of the N terminus of mature suV and anti-cytf monoclonal antibody recognizing the lumenal domain cytf were used at dilutions of 200 and 10, respectively, for single or double immunogold labeling of the vesicles.
Protein Phosphorylation in Vivo-Cells grown at 3 ϫ 10 6 cells/ml were incubated for 120 min, under dim light (5 mol of photons m Ϫ2 s Ϫ1 ), in a phosphate-depleted medium containing 1 Ci/ml of [ 32 P]orthophosphate or 2 Ci/ml of [ 33 P]orthophosphate. After resuspension in 20 mM Hepes, pH 7.5, 5 mM MgCl 2 , 10 mM NaCl, the prelabeled cells were placed for 30 min in either state I or state II conditions. For state I, cells were illuminated at 50 mol of photons m Ϫ2 s Ϫ1 in the presence of 10 Ϫ5 M 3,4-dichlorophenyl-1,1-dimethylurea to get oxidation of the PQ pool. For state II, the PQ pool was reduced by placing dark-adapted cells in anaerobic conditions using 20 mM glucose and 2 mg/ml glucose-oxidase as described previously (31). State I and II cells were rapidly broken at 4°C in a French pressure cell at 4000 psi after adding 10 mM NaF, 10 mM EDTA, 0.1 M sucrose, and protease inhibitors to the suspension medium. Thylakoid membrane proteins were prepared as in Ref. 25 with 10 mM NaF and 10 mM EDTA added in all buffers.
suV Immunoprecipitation-In vivo 33 P-labeled thylakoid membrane proteins were solubilized in the presence of 2% SDS at 100°C for 50 s and diluted 10-fold in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1 mM EDTA, 2% Triton X-100, 10 mM NaF, 200 M phenylmethylsulfonyl fluoride (buffer A). Protein A-Sepharose CL4B (Amersham Pharmacia Biotech) was incubated 30 min in 1 ml of distilled water, and washed three times in 1 ml of buffer A. To 1 ml of protein A-Sepharose CL4B in buffer A 30 l of suV antiserum was added or none as a control and samples were incubated 1 h at 4°C and washed once in 1 ml of buffer A. The diluted solubilized thylakoid membrane proteins were mixed with the pellets of protein A-Sepharose, which had or not been in presence of suV antibodies; samples were incubated 1 h at 4°C, and washed three times in 1 ml of buffer A. Samples were resuspended in 100 mM 1,4-dithiothreitol and 100 mM Na 2 CO 3 and solubilized in the presence of 2% SDS at 100°C for 50 s for polypeptide analysis.

RESULTS
Sequence and Topology of the PETO Gene Product-We have previously identified a nucleus-encoded polypeptide of 19-kDa apparent molecular mass that was absent from the thylakoid membranes of cytb 6 f mutants (21). To avoid its copurification with some ATP synthase subunits that display a similar apparent molecular mass, we recovered suV (PetO) from the ATP synthase-deficient mutant F54, using a thylakoid membrane solubilization with Mega-8, which selectively extracts cytb 6 f complex. After gel electrophoresis and transfer onto nitrocellulose membranes, we were able to microsequence 31 residues of the N terminus and one internal tryptic fragment of purified suV (Fig. 1).
The cDNA for suV was cloned by PCR amplification of a C. reinhardtii cDNA library with a degenerate primer derived from the N-terminal microsequence and phage-specific primers. The nucleotide and amino acid sequences (GenBank™ accession number AF222893) are shown in Fig. 1 and are aligned with the sequence obtained from N terminus sequencing. These sequences are homologous to ones found within a library of recently released expressed sequence tags (cDNA sequences) of C. reinhardtii (20).
The methionine codon, CACCAUGGCC, has features of an initiation codon with a consensus sequence similar to that of higher plants, i.e. AACAAUGGCC (14). The bipartite transit peptide is 51 residues long and shows features typical of a lumenal-targeting sequence. The first domain is analogous to stroma-targeting peptides: it comprises an alanine in the second position, a short uncharged N-terminal region, a central region rich in residues with basic (R, K) or small (A, S) side chains, a consensus sequence VXA just before the putative intermediate processing site. This latter feature is observed in many transit peptides for nuclear-encoded chloroplast proteins of C. reinhardtii such as the Rieske protein, PSI subunits (P28, P30, and P37), ATPase subunit ␥, heat shock protein HSP70B, phosphoribulokinase and ADP-glucose pyrophosphorylase large subunit. The second domain is homologous to a thylakoid transfer domain, with a central hydrophobic region, and a consensus cleavage site AXA for the lumenal-processing peptidase. C. reinhardtii suV transit peptide was not recognized as a chloroplast transit peptide by most prediction programs, because C. reinhardtii chloroplast peptides share more features with mitochondrial than do higher plant chloroplast presequences (32).
The mature protein is 147 residues long and has a calculated molecular mass of 15.2 kDa. A central hydrophobic region is long enough to span the membrane once in an ␣-helical conformation (Fig. 2). The lateral amphipathy of this putative ␣-helix is low.
Northern blots probed with the entire open reading frame of suV revealed a 1.3-kb transcript that corresponds to the size of the insert amplified from the cDNA library (data not shown). We also detected a few longer transcripts in low amounts, which might correspond to splicing intermediates of the primary PETO transcript.
The PETO Gene Product, suV, Behaves as a Genuine Subunit of cytb 6 f Complex-The antiserum raised against the N terminus of suV recognized only suV in the thylakoid membrane of C. reinhardtii (Fig. 3A). The steady-state level of suV was greatly diminished in whole cell protein extracts and in purified thylakoid membrane proteins from cytb 6 f-deficient mutants that do not accumulate significant levels of any of the cytb 6 f complex subunits (see ⌬petB or ⌬petD mutants, Fig. 3, A and B, which lack, respectively, cytb 6 and suIV). In contrast, suV still accumulated together with the other transmembrane subunits of cytb 6 f in mutants lacking specifically the Rieske protein, such as the petC-⌬1 mutant (Fig. 3C).
suV was recovered together with the other subunits of the cytb 6 f complex in the supernatant after solubilization with Hecameg of the thylakoid membranes (Fig. 4). Upon centrifugation of the supernatant on a sucrose gradient, part of suV remained associated with the cytb 6 f complex. However, some suV spread over fractions of higher density, most likely due to aggregated forms of the isolated subunit. suV fully dissociated from the other cytb 6 f subunits during the next purification steps that consisted of a chromatography on a hydroxyapatite column followed by a desorption step in the presence of high concentrations of ammonium phosphate.
suV Is a Transmembrane Protein with Its N Terminus Facing the Lumen of the Thylakoid Membranes-We investigated the binding of suV to the thylakoid membranes using various dissociating treatments (Table I). Membranes were treated with chaotropic agents or incubated at high ionic strength or alkaline pH. Neither cytf, cytb 6 , suIV, nor PetG were extracted by these treatments, as expected from their transmembrane anchoring. In contrast, typical peripheral membrane polypeptides (OEE1) were released in the supernatant. As reported earlier, the Rieske protein showed an intermediate susceptibility to dissociating conditions (28), although it has been clearly identified as a transmembrane protein in the three-dimensional structure of bc 1 complexes (33,34). suV behaved similarly to the Rieske protein: Although it was not released from the membranes at high ionic strength, it readily dissociated with chaotropic agents such as KSCN (see also Fig. 8) or at alkaline pH. Biochemical evidence for a transmembrane orientation of suV, with the N terminus facing the lumen, was provided by analysis of proteolysis experiments. Thylakoid membrane vesicles were incubated in the presence of protease V8 (endoproteinase Glu-C) and then subjected or not to sonication. Before sonication, most of the thylakoid vesicles are in a right-side-out position and exogenous proteases have no access to the lumen side of the membranes. In contrast, upon sonication, the vesicles burst and exogenous proteases have access to both sides of the membranes. suV was degraded by proteases in the absence of sonication, in contrast to the OEE2 protein used here as a control for a lumen resident protein (Fig. 5). Thus, suV exposes some protein motifs to the stromal side of the thylakoid membranes. However, the N-terminal-directed antibody detected a partly protease-protected suV fragment in unsonicated vesicles. This fragment was no longer detected after sonication. Similar observations were made using trypsin as an exogenous protease (experiments not shown). The selective detection of this fragment in unsonicated thylakoid vesicles shows that the N-terminal domain of suV extends in the lumen. The V8-produced fragment of 11-kDa apparent molecular mass should correspond to a truncated product of suV at its stromal C terminus, at E96, E109, or E115, leaving together the transmembrane and N terminus domains that have a predicted size of 10 -12 kDa. However, its limited accumulation argues for the high susceptibility of the truncated product to endogenous proteases, whether they are membrane-associated or lumen-located. This high protease susceptibility of suV is also substantiated by in vivo experiments, which show a drastic decrease in suV content in strains that do not accumulate cytb 6 f complexes (Fig. 3). This behavior is in marked contrast with that of OEE2, which is stable in the thylakoid lumen in the absence of PSII assembly (28). Therefore, we conclude from these assays with exogenous proteases that, in agreement with the sequence data, suV behaves as a transmembrane protein with the N terminus facing the thylakoid lumen and the C terminus extending into the stroma.
We then compared the immunogold labeling of thylakoid membrane vesicles that were exposed either to the anti-suV antibody, which recognizes the N-terminal sequence of the mature protein, or to a monoclonal antibody that recognizes a lumen-located motif of cytf. Evidence for the lumen location of this cytf epitope came from its immunodetection in mutant cells expressing only a soluble form of cytf that lacks both the C-terminal transmembrane ␣-helix and the stromal stretch of the polypeptide chain (data not shown). Immunogold labeling showed an unambiguous colocalization of these two antibodies on the same inside-out vesicle (Fig. 6), giving further support to the localization of the N terminus of suV on the membrane lumenal side.
A Protein Phosphorylated in State II-When the intersystem electron carriers of the photosynthetic electron transport chain switch from an oxidized (state I) to a reduced (state II) steady state, the photosynthetic apparatus undergoes a change in its supramolecular organization such that most of light-harvesting complex II (LHCII) and some cytb 6 f complex gather next to PSI in the stromal lamellae membrane regions (10). The supramolecular reorganization of the thylakoid membranes can be detected by immunocytochemistry (Table II). In this experiment, thin sections of broken cells of C. reinhardtii, pretreated in either of the two states, were incubated with colloidal goldcoupled antibodies specific for PSII, LHCII, or cytb 6 f antigens. The movement of LHCII or cytb 6 f complexes toward PSI-enriched unstacked membrane domains in state II is demonstrated in Table II by the increase in the ratio of immunogold labeling of unstacked versus stacked membrane region in state II as compared with state I. In contrast PSII remained in the stacked membrane regions in the two states. suV displayed the same behavior as cytf, being enriched in the unstacked regions in state II as compared with state I. This change in lateral distribution between the two states further supports the association of suV with the rest of the cytb 6 f complex, which shows lateral displacement upon state transitions.
State transitions are accompanied by reversible changes in the phosphorylation of several thylakoid membrane polypeptides. The major phosphoproteins that were previously identified in C. reinhardtii (31) belong either to the light-harvesting complex proteins (LHCP) family or to PSII (Fig. 7A). In particular, phosphorylation of the LHC polypeptides increases in state II conditions, this process being dependent upon the presence of the cytb 6 f complex (5). Therefore, this increased phosphorylation is not observed in mutants lacking cytb 6 f complexes (Fig. 7A). The pattern of phosphoproteins detected in state II showed the presence of an additional component that has the same electrophoretic mobility as suV. This phosphorylated protein is absent in cytb 6 f mutants that show very little accumulation of the cytb 6 f complex subunits and do not undergo state transition (Fig. 7A). Fig. 7B shows that this change in phosphorylation was not accompanied by a change in the steady-state level of membrane-bound suV between states I and II. That this phosphoprotein is indeed a phosphorylated form of suV was further confirmed by its behavior upon extraction with chaotropic agents or detergents (Fig. 8A and data not  shown). Furthermore, the phosphorylated protein can be immunoprecipitated by the anti-suV antiserum (Fig. 8B). Thus suV is a cytb 6 f subunit that can be reversibly phosphorylated upon state transitions.
Conservation of suV in Other Photosynthetic Eukaryotes-We found no homologues of suV within the fully sequenced cya- nobacterial genomes available in data banks. However, a homologue to suV (GenBank™ accession number AF110791) was identified among cDNAs from Volvox (19), which is a another green alga closely related to C. reinhardtii. A BLASTP 2.0 alignment (35) showed 147 identities out of the 198 residues of suV from C. reinhardtii. In particular, the consensus processing sites, hydrophobic segments, and putative phosphorylation motifs are conserved.
Because the N-terminal peptide sequence that we have used to prepare antibodies to suV is well conserved between C. reinhardtii and Volvox, with 12 identities out of 15 residues, we attempted to see whether the antipeptide would crossreact with suV candidates from other photosynthetic eukaryotes. A protein of similar apparent molecular weight as suV was indeed recognized in thylakoid membranes from spinach (Fig. 9). This cross-reaction argues for the presence of suV in higher plant cytb 6 f complexes.

DISCUSSION
A Transmembrane Protein Associated with the cytb 6 f Complex-We first identified suV as a nucleus-encoded polypeptide of apparent molecular mass of 19 kDa that was present in cytb 6 f-enriched fractions and deficient in thylakoid membranes of C. reinhardtii mutants lacking the cytb 6 f complex (21). We therefore proposed that suV was a genuine subunit of the cytb 6 f complex. This conclusion was subsequently challenged on the basis that suV was not recovered in highly purified and active cytb 6 f preparations (12,36). Indeed, we show here that suV dissociates from cytb 6 f complex during the purification steps. Thus suV behaves as a loosely bound partner of the protein complex and is not required for plastoquinol/plastocyanin oxidoreductase activity. However, it should be considered as a   FIG. 4. suV through purification steps of cytb 6 f complex. Thylakoid membranes were solubilized with Hecameg and centrifuged; S, supernatant; P, pellet. The supernatant was centrifuged on a 10 -30% sucrose gradient, and fractions were collected. The peak of cytb 6 f complex is indicated by b 6 f max. The complex purified after hydroxyapatite column chromatography is designated b 6 f HA. All fractions from sucrose gradient and hydroxyapatite purification were analyzed by urea/SDS-PAGE and levels cytf, Rieske protein, and suV were revealed by immunodetection. Sucrose fractions were assayed for kinase activity using [␥-32 P]ATP and casein as substrate.  5. Analysis of the transmembrane topology of suV by proteolysis of thylakoid membrane vesicles. Thylakoid membrane vesicles were incubated with increasing protease V8 concentrations. Vesicles in the presence of the protease were sonicated (ϩ) or not (Ϫ). Samples were analyzed by urea/SDS-PAGE and immunodecorated using antisera against the lumenal protein OEE2 and against a peptide of the N terminus of suV. Nonstromal suV fragments protected in nonsonicated vesicles and degraded in sonicated vesicles are indicated by the arrow .  FIG. 6. Lumenal localization of the N terminus of suV by immunogold labeling of thylakoid membrane vesicles. Inside-out membrane vesicles were colabeled with an antibody against the N terminus peptide of suV (large gold beads) and with an antiserum against a lumenal segment of cytf (small gold beads) indicating that suV has a transmembrane orientation.
subunit of cytb 6 f complexes in situ, because (i) it is found in association with the cytb 6 f complex during the first purification steps, (ii) it follows the same lateral redistribution along the stacked and unstacked thylakoid regions as cytf upon state transition, and (iii) it cannot accumulate in a protease-resistant form in the membranes in the absence of the other cytb 6 f subunits.
Analysis of the sequence of the PETO cDNA allowed us to predict the topology of the polypeptide chain in thylakoid membranes. Because it displayed a typical bipartite transit sequence with an ANA motif before the N terminus of the mature protein, we expected suV would have its N terminus domain located on the lumen side of the membrane. Indeed, the N terminus location of suV in the lumen was supported by proteolysis experiments and immunogold labeling. The presence of a hydrophobic stretch of 17-19 residues suggested that suV formed one transmembrane ␣-helix connecting two hydrophilic domains of about 65 residues each. The hydrophobicity of the transmembrane ␣-helix, 1.9 kcal/residue on the Goldman-Engelman-Steitz scale, is in the range of the other predicted transmembrane ␣-helices of cytb 6 f subunits (17). The release of suV from the membranes with chaotropic agents or in alkaline pH, when the Rieske protein but not the rest of the cytb 6 f complex is extracted, suggests a peripheral location of its transmembrane span with respect to the helix bundle of suIV and cytb 6 . This distal position of the ␣-helix of suV within the cytb 6 f complex is consistent with its progressive release during cytb 6 f purification.
What is the Function of suV?-suV is required neither for electron transfer nor for the dimerization of the isolated cytb 6 f complex, because a Hecameg-based purification procedure yields suV-depleted protein complexes that are fully active and dimeric (12). However, we cannot exclude a role of suV in the electron transfer in vivo or in the supramolecular organization of the photosynthetic apparatus. For example, PufX has only an indirect role in electron transfer in Rhodobacter sphaeroides: this small transmembrane polypeptide absence retards quinone exchange between the reaction center and the bc 1 complex probably due to the PufX role in the supramolecular organization of the photosynthetic apparatus (37).
Searches of the GenBank showed no sequence homologies between suV and proteins of known function, and no specific motif speak for any obvious biochemical activity. Therefore, the function of suV can only be speculated upon. There are neither histidines nor cysteines in the sequence of the mature protein.  Thus suV has no heme-binding motif and cannot correspond to cytG, a protein that interacts with cytb 6 in green algae (38,39).
The most conspicuous feature of suV is its phosphorylation in state II. suV probably corresponds to a phosphoprotein of similar apparent molecular weight that has been detected in higher plant thylakoids but not in cytb 6 f mutants (40 -42). Indeed, suV has a homologue in higher plants, as shown here by the specific immunological cross-reaction we observed with spinach thylakoids. Because the putative higher plant homologue is phosphorylated on threonine residues (43), we dismissed two putative serine phosphorylation sites in the stromal domain of suV (indicated in Fig. 1 and conserved in Volvox): S 82 KID, which matches the consensus site for the casein kinase II phosphorylation site ((S/T)XX(D/E)), and S 135 KK, which matches the consensus site for the protein kinase C phosphorylation site ((S/T)X(R/K)). The most likely residues for suV phosphorylation are then two threonine phosphorylation sites T 109 LK, T 126 KK, which both match consensus sites for protein kinase C. These putative phosphorylation sites are located in the C-terminal, stromal-exposed, domain in a sequence context that differs from the other sites of protein phosphorylation previously identified in the thylakoid membranes (44).
suV is the only known cytb 6 f-associated subunit that is reversibly phosphorylated upon state transitions in C. reinhardtii. It suggests that suV is a possible key partner of the phosphorylation-mediated state transition process in photosynthetic eukaryotes. That the PETO gene is absent from cyanobacterial chromosomes is consistent with such a function because cyanobacteria use a different, although presently poorly understood, mechanism to perform state transitions (45). State transitions in photosynthetic eukaryotes encompass the redox-controlled activation of a kinase and the subsequent redistribution of LHCII and cytb 6 f next to PSI, once LHCII proteins are phosphorylated. We found no evidence that suV could act as a kinase by itself, because there are no known kinase motifs in its amino acid sequence. Moreover, the peaks of kinase activity that are detected along the sucrose gradient loaded with the cytb 6 f-enriched supernatant (5) do not match suV distribution (Fig. 4). The available data rather suggest that suV could play a role in the activation of the LHCII kinase. There is a need for signal transduction upon kinase activation by reduced plastoquinol: the redox sensor for kinase activation is the Q o site of the cytb 6 f complex, which is located on the lumen side of the membranes (6 -9), whereas the catalytic site of the kinase is on the stromal face of the thylakoid membranes where reside all of the target sites for protein phosphorylation. The redox activation of the kinase via changes at the Q o site of cytb 6 f complexes is most likely accompanied by significant conformational changes of the Rieske protein in the thylakoid lumen (7)(8)(9). suV transmembrane topology with two large hydrophilic domains extending on both sides of the membrane has the features of a typical signal-transducing protein. The extended N-terminal domain of suV in the lumen is well suited to sense the structural changes of the Rieske protein. The single transmembrane helix of suV, whose flexibility is supported by the suV sensitivity to chaotropic agents, could transduce conformational changes to the phosphorylatable stromal C-terminal domain that interacts with the LHCII kinase. It would then remain phosphorylated as long as the kinase is activated. The need for some signal-transducing protein in thylakoid membranes also stems from the recent finding that an immunophilin-like lumenal membrane protein (46) regulates the stromal activity of a membrane-bound phosphatase (47). The transducer protein has not yet been identified.
Redox sensing at the plastoquinone pool level has also been advocated for long term adaptation of the photosynthesis ap-paratus involving changes in the stoichiometry of the reaction centers and/or antenna proteins (48). In particular, the ratio of PSII to PSI centers has been proposed to be controlled by the plastoquinone pool redox state (49,50). cytb 6 f is also thought to be the sensor for the up-regulation of the nuclear chlorophyll a/b binding protein genes in oxidizing conditions (51,52). The cytb 6 f-associated suV is a reasonable candidate for these various signal transduction processes that control gene expression. In particular, its stromal C-terminal domain is highly basic and might interact with negatively charged mRNAs or proteins. A search for mutants showing altered expression of the PETO gene products should provide an answer as to the possible role of suV in signal transduction.