A stromal region of cytochrome b₆f subunit IV is involved in the activation of the Stt7 kinase in Chlamydomonas

Significance

Sunlight is a powerful but fluctuating energy source used by photosynthetic organisms as a substrate for biochemistry. To optimize photosynthetic efficiency, the distribution of energy input between the two photosystems is regulated through state transitions. This mechanism is common to all organisms of the green lineage and depends on the concerted response of the cytochrome b₆f complex and the Stt7 kinase to changing redox conditions along the electron transport chain. The identification of a cytochrome b₆f stromal motif involved in state transitions gives insight into the mechanism of Stt7 kinase activation.

Abstract

The cytochrome (cyt) b₆f complex and Stt7 kinase regulate the antenna sizes of photosystems I and II through state transitions, which are mediated by a reversible phosphorylation of light harvesting complexes II, depending on the redox state of the plastoquinone pool. When the pool is reduced, the cyt b₆f activates the Stt7 kinase through a mechanism that is still poorly understood. After random mutagenesis of the chloroplast petD gene, coding for subunit IV of the cyt b₆f complex, and complementation of a ΔpetD host strain by chloroplast transformation, we screened for impaired state transitions in vivo by chlorophyll fluorescence imaging. We show that residues Asn122, Tyr124, and Arg125 in the stromal loop linking helices F and G of cyt b₆f subunit IV are crucial for state transitions. In vitro reconstitution experiments with purified cyt b₆f and recombinant Stt7 kinase domain show that cyt b₆f enhances Stt7 autophosphorylation and that the Arg125 residue is directly involved in this process. The peripheral stromal structure of the cyt b₆f complex had, until now, no reported function. Evidence is now provided of a direct interaction with Stt7 on the stromal side of the membrane.

Compared with other biological controls of light harvesting, such as the pH-dependent component of nonphotochemical quenching (q_E), state transitions are energy-conserving: the excess photon energy is not dissipated as heat (1, 2), but redirected from one photosystem (PS) to the other (3). State transitions therefore greatly contribute to the optimization of the maximal quantum yield of photosynthesis (4–6), probably through the poising of cyclic (PSI-dependent) vs. linear (PSI+PSII-dependent) electron flows in limiting light conditions (7, 8). In this process, the relative absorption cross-sections of PSI and PSII are dependent on the redox state of intersystem electron carriers, plastoquinones (PQ) (9), and cyt b₆f complex (10), and on the reversible phosphorylation of light harvesting complexes II (LHCII) (11, 12). Depending on their phosphorylation state, these mobile protein–pigment complexes can migrate between the stacked (grana) and nonstacked (stroma lamellae) regions of the thylakoid membrane, where PSII and PSI localize, respectively (13). LHCII migration and excitonic connectivity to either PSI or PSII therefore balance excitation between photosystems. In Chlamydomonas reinhardtii, when the PQ pool is reduced, the serine-threonine kinase Stt7 (14) is activated and phosphorylates LHCII, causing their migration toward nonappressed thylakoid membranes and attachment to PSI (state II). On the contrary, when the PQ pool is oxidized, LHCII are dephosphorylated by a phosphatase [protein phosphatase 1 (PPH1)/thylakoid-associated phosphatase 38 (TAP 38) in Arabidopsis thaliana (15, 16)] and dock to PSII (state I).

Our knowledge of the mechanism linking Stt7 kinase activation to the redox state of the PQ pool is presently incomplete, apart from certain key elements: cyt b₆f with a functional Q_o site and PQH₂ binding are required (10, 17, 18), the PetO subunit of cyt b₆f is phosphorylated by Stt7 during PQ pool reduction (19), and the lumenal domain of Stt7 interacts directly with the Rieske-ISP subunit of cyt b₆f and contains two conserved cysteine residues (20). However, the Stt7 kinase domain (20) and Stt7-dependent LHCII phosphorylation sites (21) are located on the stromal side of the membrane. Despite the recent characterization of the interaction between these two proteins (20, 22–27), the mechanism linking PQH₂ binding at the lumenal Q_o site to the activation of the stromal kinase domain of Stt7 remains unknown.

In this work, we used the technique of error-prone PCR for the mutagenesis of the chloroplast-encoded petD gene coding for subunit IV of cyt b₆f. We probed the regions of subunit IV putatively interacting with Stt7 (22, 24) by screening in vivo for impaired state transitions and found mutants blocked in state I. Furthermore, we show that cyt b₆f and Stt7 interact directly through their stromal domains and that Arg125^suIV is involved in Stt7 activation.

Results

Random Mutagenesis, Chloroplast Transformation, and Sequencing of petD Variants.

The petD sequence coding for the cyt b₆f subunit IV region going from the PEWY motif (28) to the C-terminal, and comprising helices F and G, was targeted for random mutagenesis (RM), whereas helix E, buried in the cyt b₆f core, remained untouched (see Fig. 5 for a structural view). This petD fragment was randomly mutagenized by error-prone PCR and its variants reconstructed into plasmids (SI Appendix, Text S1). After amplification in Escherichia coli, the plasmid library was used to transform the chloroplast genome of a ΔpetD host strain. RM mutants described in this study were generated through two independent mutagenesis experiments (SI Appendix, Table S1). The diversity introduced in the petD gene was assessed by sequencing the target sequence in both E. coli and C. reinhardtii. Sequencing chromatograms did not show any SNPs or ambiguity in the petD sequences retrieved from C. reinhardtii; that is, when a mutated nucleotide appeared in the chromatogram, the signals corresponding to the other nucleotides stayed within the baseline noise (typically <1% for the sequences reported in this work; see SI Appendix, Fig. S1). Therefore, no petD sequence heterogeneity was observed, suggesting that in a given transformant, only one variant of petD was retained and replicated. In both rounds, no obvious bias was found in the positions of mutations along the sequence; mutations were diverse and covered the entire region of petD targeted for mutagenesis.

Fig. 5.

Simplified drawing of the cyt b₆f monomer from C. reinhardtii and possible interactions with Stt7. The second cyt b₆f monomer would be located on the right-hand side of the figure. Atomic coordinates of cofactors, stromal loops and amino acid side chains shown here are from PDB 1Q90 (33). Helices A, B, and C of cyt b₆ and the transmembrane helix of the Rieske-ISP were removed for the sake of simplicity. The subunit IV (magenta) region targeted for random mutagenesis extends from the PEWY motif of the Q_o site to the C-terminal through helices F and G exposed at the periphery of the complex. The Stt7 kinase interacts with the Rieske-ISP (20), F and G helices of subunit IV (22, 27) and the stromal loop of subunit IV (this work). (Inset) Stromal regions of subunit IV and cyt b₆ (here in their ribbon representations), showing the structural link between Arg125^suIV and the Q_i site through the C-terminal of cyt b₆ and the Arg207^{cyt b6} residue that interacts with heme c_i.

The conditions chosen for error-prone PCR (SI Appendix, Text S1) introduced a higher mutational frequency in trial B than in trial A. This was confirmed by counting the mutations in the petD fragment recovered in E. coli and C. reinhardtii clones (see “Distribution” columns in SI Appendix, Table S1). Expectedly, the relative amount of nonmutated sequences was also smaller in trial B than in trial A for both E. coli and C. reinhardtii (see “% Non-mutated” columns in SI Appendix, Table S1). Interestingly, we observed that in each of the two different RM experiments, C. reinhardtii cells retained mutated copies of petD less often than E. coli cells, and when they did, they contained fewer mutations. This was a result of the difference in selection pressure exerted on the two organisms. Whereas E. coli transformants were selected on ampicillin-containing medium after integration of an intact Amp^R cassette, C. reinhardtii transformants were selected on photoautotrophic growth after integration of a petD variant. Not all versions of randomly mutagenized petD genes are expected to restore a proper synthesis and folding of subunit IV and assembly of functional cyt b₆f complexes. The more mutations are introduced in an essential photosynthetic gene, the more likely they are to be deleterious and lost through phototrophic selection.

Screening for C. reinhardtii Transformants Impaired for State Transitions.

For our screen, chlorophyll fluorescence images were recorded over the course of 10 min of anaerobic adaptation in the dark (SI Appendix, Fig. S2A), using an imaging system described in ref. 29. Anoxic reduction of the PQ pool was detected as an increase of F₀′, and fluorescence quenching associated with the formation of state II (q_T) was measured as a decrease of F_m′ in the WT. In the stt7-1 strain, blocked in state I, no fluorescence quenching was observed after 10 min. All the >2,000 C. reinhardtii clones obtained after transformation were screened for state transitions, and most of them showed a decrease in F_m′ similar to the WT. Sixteen clones, however, showed very stable F_m′ values after anaerobic adaptation, as in stt7-1, suggesting they were affected in state transitions. SI Appendix, Fig. S2A shows a typical state transition kinetics curve for a petD random mutant affected for state transitions compared with the WT and stt7-1 reference strains.

Fig. 1 illustrates some of the growth phenotypes and functional data obtained for three random petD mutants in comparison with the three reference strains [the site-directed (SD) petD mutants are discussed in SD Mutagenesis of the Stromal fg Loop of Subunit IV]. All strains show similar heterotrophic growth (TAP conditions) and maximal quantum efficiency of PSII photochemistry (F_v/F_m ≥ 0.7). The host strain ΔpetD does not grow in photoautotrophic conditions (minimum medium, MIN) and shows a very low photochemical quenching (q_P < 0.05) when grown in heterotrophic conditions (TAP). In all mutants selected on minimal medium, photoautotrophic growth and q_P were similar to WT (q_P ≥ 0.8), suggesting a normal linear electron flow through PSII and downstream electron carriers in the photosynthetic chain, including the cyt b₆f complex. However, the q_T component of nonphotochemical quenching measured on anaerobic adaptation in the dark, high in the WT (q_T ≥ 0.4), was greatly reduced in the two control strains lacking either cyt b₆f or Stt7, as well as in the three random petD mutants shown here (q_T ≤ 0.05).

Fig. 1.

Random and SD petD mutants are impaired for state transitions despite normal photosynthetic growth, electron transfer, and cyt b₆f subunit accumulation. Growth on minimal (MIN, 100 μmol_photons m⁻²⋅s⁻¹, 2% CO₂) and acetate (TAP, 20 μmol_photons m⁻²⋅s⁻¹) media, maximum quantum efficiency of PSII [F_v/F_m = (F_m − F₀)/F_m], photochemical quenching of chlorophyll fluorescence [q_P = (F_m − F)/F_m − F₀)], q_T component of nonphotochemical quenching measured upon anaerobic adaptation in the dark [q_T = (F_m − F_m′)/F_m′, see SI Appendix, Fig. S2] and cyt b₆f subunits accumulation of the three reference strains (WT, ΔpetD, stt7-1) and six petD mutants affected for state transitions. An antibody against AtpB was used as a loading control. SI Appendix, Table S2 mutants 19 ([Asn122Lys; Ile128Phe]), 13 ([Gln121Arg; Tyr124His]), and 28 (Arg125Ser) were obtained by random mutagenesis. Mutants 22 (Asn122Leu), 27 (Tyr124Lys) and 30 (Arg125Glu) were obtained by SD mutagenesis. All false color images and fluorescence parameters were calculated from chlorophyll fluorescence measurements of cells grown on MIN, except for the ΔpetD strain grown on TAP.

Among the 16 selected mutants that showed normal photochemistry but were blocked in state I, nine of them (italicized in SI Appendix, Table S2) were affected in the region coding for the loop linking helices F and G of cyt b₆f subunit IV (hereafter designated as fg loop), exposed at the stromal side of the membrane. These RM mutants also often contained additional mutations on either side of this stromal loop that are reported in columns 2 and 4 of SI Appendix, Table S2. To tease out certain phenotypic ambiguities that emerged from the comparison of the RM mutants, we generated single-point mutations along the loop (from Asn118 to Arg125) by SD mutagenesis.

SD Mutagenesis of the Stromal fg Loop of Subunit IV.

Similar to RM, SD mutants were obtained by chloroplast transformation of the ΔpetD host strain with SD mutagenized petD, followed by photoautotrophic selection of clones. To avoid structural destabilization of the loop, Pro123, Pro127, and residues pointing toward the core of the protein were not touched, whereas residues putatively involved in protein–protein interactions were targeted. For each residue, for example, Arg125, two amino acids with similar steric properties were substituted, bearing either a hydrophobic side chain (e.g., mutant 29 in SI Appendix, Table S2, Arg125Leu) or a side chain of “opposite” chemical property (e.g., mutant 30, Arg125Glu). Similar to the RM mutants, SD mutants grew under photoautotrophic conditions, and some of them, especially those containing mutations at residues Asn122, Tyr124, and Arg125, showed chlorophyll fluorescence parameters identical to the stt7-1 strain, strongly affected in q_T, but not in q_P (Fig. 1 and SI Appendix, Figs. S2B and S4). Obtaining SD point mutants was important to discern the effect of individual mutations on the phenotype of certain RM mutants that contained more than one mutation. For example, for positions 121 and 124, although a marked decrease of q_T was observed in mutant 13 ([Gln121Arg; Tyr124His]), it was not nearly as drastic in mutant 24 ([Tyr124His; Phe149Leu]), and not observed at all in mutant 14 ([Val104Ala; Gln121Arg; Phe149Ser]; SI Appendix, Table S2). The Gln121Arg SD mutant 17 showed that changing Gln121 did not affect q_T, whereas SD mutant 27 (Tyr124Lys) revealed a critical role for Tyr124: substitution to Lys induced a loss of q_T, whereas substitution to the aromatic residue Phe (mutant 26, Tyr124Phe) had no effect on q_T, as observed in the RM mutant 25 ([Tyr124Phe; Ile128Thr; Thr137Ala; Leu159Ser]). Further characterization focused on SD mutants 22, 27, and 30 (Asn122Leu, Tyr124Lys, and Arg125Glu, respectively) that exhibited the most robust “locked-in-state-I” phenotype of all the SD subunit IV mutants.

Characterization of Modified cyt b₆f Complexes.

In ΔpetD, totally devoid of subunit IV, the rate of synthesis of cyt f is strongly decreased, resulting in very little accumulation of cyt f (30), also observed in Fig. 1. In complemented strains, the 34.3-kDa cyt f subunit is tagged with 6-His at its C-terminal and shows a slower electrophoretic mobility on denaturing gels than in the stt7-1 strain, as previously reported for the f₂₉₇C strain (31). Small differences are observed in the migration profiles of the 17.4-kDa subunit IV and can be attributed to slight changes in peptide mobility induced by single-point mutations. In the petD RM and SD mutants, the accumulation of subunit IV and cyt f was similar to WT levels (see SI Appendix, Fig. S5 for 50% and 10% loading controls).

The electron transfer rate through the cyt b₆f complex was assessed in control and mutant strains (SI Appendix, Figs. S6 and S7). The transmembrane electrogenic phase of electron transfer between hemes b_L and b_H, occurring after quinol oxidation at the Q_o site, was measured as an electrochromic shift of carotenoids, giving an absorbance increase at 520 nm. Electron transfer rates in mutants Asn122Leu, Tyr124Lys, and Arg125Glu were very similar to the WT.

Changes in Photosystems Antenna Size Associated with State Transitions.

At 77 K, the light energy absorbed by chlorophyll in the mobile LHCII proteins is not reemitted at 682 nm, as it would be if these protein complexes were disconnected from the photosystems (32), but is transferred either to PSII and reemitted at 686 and 696 nm, or to PSI and reemitted at 713 nm. In the WT, the relative amplitudes of the fluorescence peaks change, with PSII contribution being larger in oxidizing conditions (state I) and PSI fluorescing mostly in reducing conditions (state II). It therefore reflects a change in excitonic connectivity of mobile LHCII to either PSI or PSII that depends on the Stt7 kinase. Similar to that observed in the stt7-1 strain, the Asn122Leu, Tyr124Lys, and Arg125Glu mutants were blocked in state I despite reducing conditions, with LHCII constitutively connected to PSII (Fig. 2).

Fig. 2.

Subunit IV fg loop mutants do not change photosystem antenna sizes between oxidizing and reducing conditions. Low-temperature (77 K) fluorescence emission spectra of the subunit IV mutants compared with the WT and stt7-1 strains. Cells were placed in oxidizing (Ox, gray curve: 10 µM DCMU, moderate light, 30 min) or reducing (Red, black curve: dark anoxia, 30 min) conditions. Spectra were normalized at 685 nm on PSII contribution.

Phosphorylation of Thylakoid Polypeptides.

Incorporation of radiolabeled orthophosphate [³³PO₄³⁻] into proteins during PQ pool reduction was measured in vivo. Intact cells of SD mutants and reference strains were preincubated for 30 min under dark anoxic conditions. During nonphotochemical reduction of the PQ pool, thylakoid polypeptides incorporate radiolabeled phosphate that is detected by autoradiography after membrane preparation and fractionation by PAGE on an 8 M urea 12–18% gradient gel (Fig. 3 and see SI Appendix, Fig. S8 for loading controls). Contrary to the polypeptides that are constitutively phosphorylated in the various strains (CP26, CP29, D2, P11, PsbH), P13 and P17 were phosphorylated in the WT, but not in ΔpetD or in stt7-1, as reported earlier (6, 10). Among the SD mutants, which accumulate WT levels of the Stt7 kinase (SI Appendix, Fig. S9), Asn122Leu and Tyr124Lys showed low phosphorylation of P13 and P17, and Arg125Glu showed none. Similar phosphorylation profiles were observed in another comparative state I/state II ³³P-labeling experiment (SI Appendix, Fig. S10).

Fig. 3.

Subunit IV fg loop mutants show low levels of Stt7-dependent LHCII phosphorylation. Autoradiogram of in vivo ³³P-labeled thylakoid polypeptides, extracted from cells placed in reducing conditions (5 µM FCCP, dark, 30 min) and separated by SDS/PAGE on an 8 M urea 12–18% gradient gel.

Cyt b₆f–Stt7 Interaction Studies.

A GAL4-based yeast two-hybrid assay was carried out to probe the interaction between subunit IV and Stt7 (SI Appendix, Text S2 and Fig. S11). The WT and double-mutant ([Tyr124Lys; Arg125Glu]) subunit IV fg loops (residues 113–128) were expressed as fusions to the Gal4 DNA-binding domain and used as bait proteins. The stromal kinase domain of Stt7 was cut into four fragments that were expressed as fusions to the Gal4 activation domain and used as prey proteins: A, residues 123–238; B, 244–379; C, 397–483; and D, 484–600. Whereas no interaction was detected between the WT fg loop of subunit IV and Stt7 fragments C (397–483) and D (484–600), an interaction was detected between the WT fg loop and Stt7 fragments AB (123–379) and B alone (244–379). As expected, this interaction was lost in the double mutant ([Tyr124Lys; Arg125Glu]) fg loop. We further addressed the question of the biochemical interaction between cyt b₆f and Stt7 in a context in which the folding of subunit IV and assembly into a functional cyt b₆f complex is preserved.

Intact and functional cyt b₆f complexes were purified from C. reinhardtii thylakoid membranes following ref. 33. The recombinant kinase domain of Stt7 (Stt7-KD, residues 139–495) was produced in E. coli and purified as described in the SI Appendix, Text S3 and Fig. S12. Stt7-KD was active in vitro, as seen by its ATP hydrolysis activity (SI Appendix, Table S3). Stt7-KD was incubated for 30 min in the presence of ATP and MgCl₂ with either the WT or Arg125Glu purified cyt b₆f complexes. Whereas Stt7 showed a low level of autophosphorylation when incubated alone with ATP and MgCl₂ (Fig. 4, lane 1), addition of WT cyt b₆f greatly increased its phosphorylation (lane 3). In stark contrast, incubation of Stt7-KD with cyt b₆f containing the Arg125Glu substitution (lane 5) showed only the basal signal of Stt7 autophosphorylation, similar to lane 1. This in vitro reconstitution shows that the cyt b₆f complex and the stromal kinase domain of Stt7 interact directly and that Arg125 of subunit IV is involved in Stt7 autophosphorylation.

Fig. 4.

Purified cyt b₆f complex increases autophosphorylation of the Stt7 kinase domain depending on the presence of Arg125^suIV. Immunoblots of Stt7-KD autophosphorylation reactions in the presence of WT or Arg125Glu cyt b₆f complex. An antibody against phospho-threonine (Cell Signaling) was used to detect Stt7-KD autophosphorylation. Antibodies against Strep-tag (recognizing the C-terminal tag of Stt7-KD) and subunit IV were used as loading and transfer controls.

Discussion

Design of an Optimized Protocol for the Random Mutagenesis of Chloroplast Genes and Screening for Modified Photosynthetic Functions.

We have targeted a ∼300-bp fragment of petD, corresponding to ∼100 amino acids in subunit IV. The number of single amino acid substitutions in a 100-amino acid protein is 19 × 100 = 1,900. In total, among our ∼2,000 transformants, we have obtained ∼1,200 random mutants bearing, on average, ≥2 mutations (rightmost column of SI Appendix, Table S1). When looking for a loss-of-function phenotype, single amino acid substitutions are sufficient, with no need of obtaining all possible combinations of ≥2 substitutions. Therefore, the number of genetic variants generated in this work was adequate for screening for a loss of function.

Previous studies reported the random mutagenesis of chloroplast DNA. Mutagenic agents such as 5-fluorodeoxyuridine can be used to mutagenize the entire chloroplast genome (34), and degenerate oligonucleotides can introduce a given number of mutations in the sequence spanned by the oligonucleotides (35, 36). Although both these methods contain obvious caveats, the technique described here allows for specific targeting of a chloroplast DNA sequence without any size limitation other than the maximal length of amplification by PCR. In our case, considering the rather low efficiency of chloroplast transformation, and because selection on phototrophy discards nonfunctional gene variants, we were able to increase the mutation rate of the error-prone PCR to enrich the transformation plates in putative mutant clones. We did so until we reached quite a high mutational frequency for a protein structure–function study (3.5 mutations/300 bp for RMB; see SI Appendix, Table S1). High mutation rates are used more often for the directed evolution of enzymes, where many substitutions can be introduced (37). As a broader perspective, this approach can be used for the study of many other chloroplast-encoded proteins and opens opportunities for the directed evolution of enzymes in transplastomic cell lines.

Identification of a cyt b₆f Stromal Loop Involved in State Transitions.

After screening for a specific mutant phenotype, it was possible to gain functional insight from the comparison of sequences retrieved from the RM mutants, and we further confirmed the attribution of functional changes to point mutations by using SD mutagenesis. Given that the catalytic site of the Stt7 kinase is found in the stroma (20), we narrowed our target of investigation to the subunit IV fg loop starting from Asn118 and extending to Arg125. Protein sequence alignment showed a high degree of conservation between subunit IV from various organisms, but a strong divergence with cyt b from bacteria and mitochondria (SI Appendix, Fig. S13). We identified NKFQNPxRR as a conserved motif (where x corresponds to aromatic residues tyrosine or phenylalanine). This site, rich in lysine (K), arginine (R), and proline (P), to a lesser extent, strongly resembles a kinase binding site such as that of Stt7 (21), except that it contains no phosphorylatable residue (threonine or serine for Stt7).

Mutations at residues Asn122, Tyr124, and Arg125 produced mutants that are blocked in state I, despite a reduced PQ pool and normal cyt b₆f assembly and electron transfer activity. Upon a switch from oxidizing to reducing conditions, these mutants showed no quenching of PSII fluorescence emission associated with state transitions, no modification of the relative antenna sizes of PSII and PSI, and very low levels of phosphorylation of LHCII antenna proteins. This function was not expected for the peripheral stromal structure of the cyt b₆f complex.

Insight into the Role of cyt b₆f Complex in Stt7 Kinase Activation.

The present work shows that the stromal region of cyt b₆f is involved in the activation of Stt7. Yeast two-hybrid experiments indicated that the stromal fg loop of subunit IV interacts with the kinase domain of Stt7 in the region between Leu244^Stt7 and Pro379^Stt7. Substitutions of fg loop residues Tyr124^suIV and Arg125^suIV abolished this interaction, showing that they are involved in the binding of the kinase to the stromal side of the complex. Furthermore, in vitro reconstitution experiments demonstrated that the autophosphorylation activity of the kinase domain was enhanced in the presence of purified cyt b₆f complex, through an interaction dependent on Arg125^suIV. The Stt7 region between Leu244^Stt7 and Pro379^Stt7 identified by two-hybrid assays contains two important kinase regulatory motifs, HRD and DLG, as well as the activation loop and the α5-hairpin (38). Several acidic Asp and Glu residues are present in this region and could form interactions with Arg125^suIV in vivo. The absence of the lumenal domain [with the two regulatory Cys residues (20)] and the transmembrane domain of Stt7 did not prevent cyt b₆f-mediated autophosphorylation of the kinase. In our attempt to define the minimal functional length of Stt7, and going a step further than the truncation at residue 549 (27), we narrowed the kinase domain down to even before Ser533^Stt7, a phosphorylation site identified by phosphoproteomics but not involved in state transitions (20, 21). In addition to the ^PThr and ^PSer detected in previous phosphoproteomics studies (21, 39), our results suggest there are other phosphorylatable residues in the catalytic domain of Stt7. These residues could be important for the activation of the kinase in vivo.

On the basis of previous studies showing the crucial role of the lumenal Q_o site in state transitions (17, 28) and topology analysis showing that the kinase domain (20) and the LHCII phosphorylation sites (21) are found in the stroma, a signal transduction mechanism across the thylakoid membrane was postulated. The Chl a molecule, bound between helices F and G of subunit IV and protruding into the Q_o site, was proposed to play a role in this signal transduction (22, 24, 25, 40). Redox changes of the lumenal cysteine residues of Stt7, caused by Q_o site catalysis, were also suggested to induce Stt7 activation through transient dimerization (26) or conformational changes of its transmembrane domain (27). The present study identifies the stromal fg loop of subunit IV within the pattern of interactions between the cyt b₆f and Stt7. We propose that the primary activation of Stt7 depends on its interaction with the stromal side of cyt b₆f complex, whereas its release for LHCII phosphorylation could be controlled by plastoquinol occupancy and turnover at the Q_o site, as suggested previously (41).

We found here that Arg125 from the subunit IV fg loop plays a key role in the activation of the state transition kinase Stt7. The electron density map of the cyt b₆f complex from C. reinhardtii shows that the guanidinium group of Arg125 and the C-terminal carboxyl group of cyt b₆ are at a distance suitable to form a salt-bridge or an H-bond (see Fig. 5 and ref. 33). It suggests that Arg125^suIV has two possible interacting partners: either the terminal carboxyl of the cyt b₆ or a negatively charged residue of Stt7 (Glu or Asp) in the region between Leu244^Stt7 and Pro379^Stt7. As shown in Fig. 5, Inset, the cyt b₆ C-terminal end is a short strand away from the conserved Arg207 that interacts with one of the propionates of heme c_i and shapes part of the Q_i pocket. In our previous work on heme c_i, we reported a discrepancy between its redox potentials estimated from in vivo and in vitro measurements and some peculiar electronic properties of this cofactor, depending on the pH and/or the binding of NQNO (42, 43). We attributed these unusual features to conformational heterogeneity at the stromal side of the cyt b₆f complex. It remains to be explored whether the dual interaction of Arg125^suIV with either Stt7 or cyt b₆ could account for such structural heterogeneity.

Materials and Methods

Cell Growth Conditions.

For photoautotrophic growth, C. reinhardtii cells were grown on Tris-phosphate (MIN, pH 7.2) medium in the light (100 μmol_photons m⁻²⋅s⁻¹) in air supplemented with 2% CO₂. For heterotrophic growth, cells were grown on Tris acetate-phosphate (TAP, pH 7.2) medium in very low to moderate light (10–40 μmol_photons m⁻²⋅s⁻¹).

Plasmids and Strains.

Plasmid pWQH₆ (24), carrying the 3′-end sequence of petA with a 6-His tag, the intergenic petA-petD region, and the entire petD sequence, was used for mutagenesis and chloroplast transformation experiments. The mt⁺ acetate-requiring ΔpetD deletion strain (30) was used as the recipient strain for chloroplast transformation. The WT reference strain was obtained by chloroplast transformation of the ΔpetD strain with plasmid pWQH₆ and selection on the restoration of photoautotrophic growth. The petD random and SD mutants were obtained the same way, apart from the use of pWQH₆ plasmids carrying mutations on the petD sequence. The WT reference strain and petD mutant strains therefore all express a petA gene coding for a His-tagged cyt f. The stt7-1 strain (14), containing a deletion of the gene coding for the Stt7 kinase, was used as a control.

Mutagenesis of the petD Gene.

Plasmid pWQH₆ and various primers were used for random and SD mutageneses of petD (SI Appendix, Table S4). For RM, error-prone PCR was performed using kits GeneMorph II EZClone (Agilent Technologies, “RMA” in SI Appendix, Table S1), and Diversify PCR (Clontech, “RMB”; see SI Appendix, Table S1 and see SI Appendix, Text S1 for details on the protocols).

Chloroplast Transformation and Mutant Selection.

Plasmid DNA was precipitated on nanometer-scale gold particles using the Seashell Technology S550d gold DNA protocol. ΔpetD cells on MIN medium were transformed by gold particle bombardment at 7 bars under vacuum. After transformation, cells were left to recover in the dark for up to 24 h and transformants were incubated in air + 2% CO₂ under low light (40–50 μmol_photons m⁻²⋅s⁻¹) for phototrophic selection.

Screening for State Transition Mutants.

Transformation plates were left aerated in the dark for up to 15 min to induce PQ pool oxidation. The plates were then transferred to anaerobic conditions (N₂ flush) and the maximal fluorescence yield of PSII was recorded, using a fluorescence camera similar to the one described in ref. 29, during a 10-min period in the dark with 12 saturating light pulses.

77 K Chlorophyll Fluorescence Emission Spectra.

Chlorophyll was excited with a LED at ∼455 nm, and fluorescence was selected with a red-colored Kodak Wratten filter and collected with an Ocean Optics USB2000 CCD spectrometer. C. reinhardtii cells were placed in an aluminum-made sample holder and snap frozen at 77 K in liquid nitrogen. Sample, LED, and spectrometer were coupled via a Y-shaped fiberoptics.

In Vivo Phosphorylation of Antenna Proteins.

Cells grown to around 2 × 10⁶ cells⋅mL⁻¹ were harvested and resuspended in a phosphate-depleted minimal medium containing 2 μCi mL^{−1 33}P_i. Samples were then treated as described in ref. 44, and thylakoid membranes were isolated as in ref. 45. Phosphorylation patterns of thylakoid membrane polypeptides were then analyzed on a 12–18% polyacrylamide gel containing 8 M urea, which was then dried and autoradiographed.

In Vitro Assays with Purified cyt b₆f Complexes and Recombinant Stt7-KD.

Cyt b₆f complexes were purified as in ref. 33, and recombinant Stt7-KD (residues 139–495) was purified following a protocol adapted from ref. 27, as described in the SI Appendix, Text S3 and Fig. S12. The ADP-Glo Kinase Assay (Promega) was used to measure the in vitro activity of Stt7-KD (see SI Appendix, legend of Table S3). To test the interaction between the two proteins, 4.5 µM purified Stt7-KD and 1.5 µM purified WT or Arg125Glu cyt b₆f complex were incubated together with 1 mM ATP and 20 mM MgCl₂ for 30 min. Samples were denatured in a buffer containing LDS and DTT at 70 °C for 15 min and separated on a 10% Bis-Tris gel by SDS/PAGE. After transfer on a nitrocellulose membrane, antibodies against phospho-threonine (Cell Signaling), Strep-tag (recognizing the C-terminal tag of Stt7-KD), and subunit IV were used to detect the proteins.