A Reaction Center-dependent Photoprotection Mechanism in a Highly Robust Photosystem II from an Extremophilic Red Alga, Cyanidioschyzon merolae*

Background: PSII is a protein complex that captures sunlight to drive water oxidation. Results: Cyanidioschyzon merolae PSII is protected by reversible reaction center-based non-photochemical quenching. Conclusion: C. merolae PSII employs reaction center non-photochemical quenching as the main photoprotective mechanism. Significance: We provide the first direct evidence of the PSII reaction center as the primary locus of non-photochemical quenching in the extremophilic red algae. Members of the rhodophytan order Cyanidiales are unique among phototrophs in their ability to live in extremely low pH levels and moderately high temperatures. The photosynthetic apparatus of the red alga Cyanidioschyzon merolae represents an intermediate type between cyanobacteria and higher plants, suggesting that this alga may provide the evolutionary link between prokaryotic and eukaryotic phototrophs. Although we now have a detailed structural model of photosystem II (PSII) from cyanobacteria at an atomic resolution, no corresponding structure of the eukaryotic PSII complex has been published to date. Here we report the isolation and characterization of a highly active and robust dimeric PSII complex from C. merolae. We show that this complex is highly stable across a range of extreme light, temperature, and pH conditions. By measuring fluorescence quenching properties of the isolated C. merolae PSII complex, we provide the first direct evidence of pH-dependent non-photochemical quenching in the red algal PSII reaction center. This type of quenching, together with high zeaxanthin content, appears to underlie photoprotection mechanisms that are efficiently employed by this robust natural water-splitting complex under excess irradiance. In order to provide structural details of this eukaryotic form of PSII, we have employed electron microscopy and single particle analyses to obtain a 17 Å map of the C. merolae PSII dimer in which we locate the position of the protein mass corresponding to the additional extrinsic protein stabilizing the oxygen-evolving complex, PsbQ′. We conclude that this lumenal subunit is present in the vicinity of the CP43 protein, close to the membrane plane.

Members of the rhodophytan order Cyanidiales are unique among phototrophs in their ability to live in extremely low pH levels and moderately high temperatures. The photosynthetic apparatus of the red alga Cyanidioschyzon merolae represents an intermediate type between cyanobacteria and higher plants, suggesting that this alga may provide the evolutionary link between prokaryotic and eukaryotic phototrophs. Although we now have a detailed structural model of photosystem II (PSII) from cyanobacteria at an atomic resolution, no corresponding structure of the eukaryotic PSII complex has been published to date. Here we report the isolation and characterization of a highly active and robust dimeric PSII complex from C. merolae. We show that this complex is highly stable across a range of extreme light, temperature, and pH conditions. By measuring fluorescence quenching properties of the isolated C. merolae PSII complex, we provide the first direct evidence of pH-dependent non-photochemical quenching in the red algal PSII reaction center. This type of quenching, together with high zeaxanthin content, appears to underlie photoprotection mechanisms that are efficiently employed by this robust natural water-splitting complex under excess irradiance. In order to provide structural details of this eukaryotic form of PSII, we have employed electron microscopy and single particle analyses to obtain a 17 Å map of the C. merolae PSII dimer in which we locate the position of the protein mass corresponding to the additional extrinsic protein stabilizing the oxygen-evolving complex, PsbQ. We conclude that this lumenal subunit is present in the vicinity of the CP43 protein, close to the membrane plane.
Photosystem II (PSII) 3 is a multimeric transmembrane complex present in the thylakoid membranes of cyanobacteria, algae, and higher plants that is able to capture solar energy and use it to drive charge separation and water oxidation catalysis. Recently, the crystal structure of PSII isolated from the thermophilic cyanobacterium Thermosynechococcus vulcanus has been reported at a resolution of 1.9 Å (1). This atomic structure has facilitated formation of the working models for understanding the water-splitting reaction based on the previous crystallographic structures of cyanobacterial PSII (2)(3)(4)(5)(6)(7)(8)(9). The cyanobacterial complex contains 20 protein subunits with a total molecular mass of 350 kDa. Bound within these subunits are over 1,300 water molecules and 85 cofactors: 35 chlorophyll (Chl) molecules, two pheophytins, 11 ␤-carotenes, over 20 lipids, two plastoquinones, two heme irons, one non-heme iron, four manganese atoms, 3-4 Ca 2ϩ ions, three Cl Ϫ ions, and one bicarbonate (1). The catalytic site is composed of a Mn 4 Ca cluster surrounded by a number of highly conserved amino acids, mainly derived from the D1 reaction center (RC) subunit and one from the CP43 inner antenna subunit. The model that places three manganese ions and a calcium ion at the corner of a cubane with oxo-bridges and the fourth manganese linked to the cubane via a bridging oxygen (2) has gained support in the latest atomic structure of PSII (1) as well as from quantum mechanical and x-ray spectroscopic considerations (10 -13).
Although we now have a detailed structural model of PSII from cyanobacteria at an atomic resolution, no corresponding structure of the eukaryotic PSII complex has been published to date. Hence, there is an urgent need to purify a highly stable eukaryotic PSII complex that would prove suitable for high resolution structural analysis. Equally important, PSII as nature's water-splitting enzyme provides a blueprint in terms of the catalytic rate of water oxidation in ambient conditions for the construction of stable synthetic catalysts operating within solar-tofuel nanodevices (14 -17). A logical approach toward these goals is to obtain a eukaryotic form of PSII using an organism that is likely to provide a robust and highly active form of the complex. To this end, we isolated and characterized the dimeric PSII complex from the extremophilic unicellular red alga Cyanidioschyzon merolae.
C. merolae belongs to the rhodophytan order Cyanidiales, whose members thrive in acidic hot springs (18) and are unique among phototrophs in the ability to live at extremely low pH (pH 0. [2][3][4] and moderately high temperatures (40 -56°C). Despite such an extremely acidic environment, the intracellular pH of these algae is most likely neutral due to the active H ϩ efflux across the plasma membrane (19). C. merolae is considered to be one of the most primitive eukaryotic phototrophs because it diverged near the root of the red algal lineage that forms a basal group within the photosynthetic eukaryotes (20). The photosynthetic apparatus of this alga is regarded as the closest equivalent of the prokaryotic ancestor of the present day chloroplast (21). As an evolutionary intermediate between the photosynthetic apparatus of prokaryotic cyanobacteria and that of the eukaryotes in the green lineage, it contains a mixture of prokaryotic and eukaryotic structural traits. Whereas red algal photosystem I (PSI) resembles a higher plant complex, with an crescent-shaped Chla-binding light-harvesting antenna system asymmetrically bound on one side of the core complex (22)(23), PSII of red algae is structurally similar to the cyanobacterial counterpart in that it contains light-harvesting antenna composed of phycobilisomes, large peripheral membrane complexes formed by phycobiliproteins, instead of Chla/ b-binding antenna proteins that constitute the light-harvesting system of green algae and higher plants.
The oxygen-evolving complex (OEC) in C. merolae PSII is stabilized by four extrinsic lumenal subunits: cyanobacterialike PsbV and PsbU, the evolutionarily conserved PsbO subunit, and an additional 20-kDa subunit PsbQЈ distantly related to the higher plant and green algal PsbQ polypeptides (24). The precise localization of this subunit and its role in stabilization of the OEC are currently unknown, although low resolution single particle analysis and in vitro reconstitution experiments suggested that the red algal PsbQЈ may independently associate with the PSII core complex in the close vicinity of PsbV and the PsbU subunits and is required for effective binding of these extrinsic subunits (22,25). In contrast, PsbQ in higher plants functionally associates with PSII via its direct interaction with both PsbO and PsbP extrinsic subunits, stabilizing the OEC (reviewed in Ref. 26). Interestingly, PsbQЈ can functionally replace PsbQ in spinach during cross-reconstitution experiments despite the low amino acid sequence homology between both proteins (25). The binding mode for another OEC subunit, PsbV, also differs between cyanobacteria and red alga. In red algae, PsbV binds via other extrinsic subunits, whereas its cyanobacterial counterpart binds directly with the PSII core (reviewed in Ref. 26). All of these lines of evidence point toward significant structural differences on the lumenal side of the PSII complex that was formed at various evolutionary stages.
It is well established that under conditions of excessive irradiation (i.e. when light energy absorption exceeds the capacity and demands of photosynthesis), several photoprotective and optimizing mechanisms are triggered. These mechanisms include regulation of light absorption capacity between PSI and PSII (state transitions; see Ref. 27), fast D1 protein turnover during photoinhibition of PSII, and photoprotective non-photochemical quenching (see Refs. 28 and 29 for recent reviews). Non-photochemical quenching (NPQ) represents a feedback regulatory mechanism that leads to dissipation of excessive light either in the light-harvesting antennae (29) or in the reaction center of PSII (30). The most flexible and dominant component of NPQ, ⌬pH-dependent high energy quenching (qE) is a major photoprotective strategy that operates on a time scale of seconds to minutes.
Although mechanistic aspects and molecular components of pH-dependent qE have been widely studied in higher plants and green algae, little is known about molecular mechanisms of PSII fluorescence quenching in phycobilisome-containing red algae. In the red algae ancestor, prokaryotic cyanobacteria, excessive light is dissipated in phycobilisomes, aided by the orange carotenoid protein that is absent in red algae (31). The detailed NPQ mechanisms in red algae are largely unknown, and the NPQ properties are mostly derived from in vivo fluorescence measurements in intact cells. Kirilovsky and colleagues (32,33) demonstrated the existence of ⌬pH-dependent quenching in Rhodella violacea and Porphyridium cruentum strains of mesophylic red algae. These authors have suggested that, in contrast to higher plants and green algae, the dominant part of NPQ in red algae might occur in the RC rather than in the antenna (33). Several photochemical mechanisms have been suggested for the RC type of quenching (reviewed in Ref. 34). Although the real importance and extent of RC quenching in photoprotection is still a matter of debate, especially in comparison with qE in antennae (compare Refs. 29 and 30), the existence of this type of quenching has been proved for PSII isolated from various photothrophs (35,36). However, the same direct evidence has been missing for red algae.
Here we report a detailed structural and functional analysis of a highly active and robust dimeric PSII complex isolated from C. merolae. We provide the first direct evidence that the red algal PSII complex is rich in zeaxanthin and employs reversible RC-based non-photochemical quenching that is triggered by low pH. These features provide the basis for the remarkable robustness of this complex across a range of extreme light, temperature, and pH conditions. In the first attempt to provide structural details of this eukaryotic form of PSII, we have employed electron microscopy (EM) and single particle analyses to produce a two-dimensional electron density map in which we locate the position of the additional protein mass over and above that found in cyanobacterial PSII. We attribute this mass to the PsbQЈ subunit, which binds to the OEC in the vicinity of the CP43 protein, close to the membrane plane.

MATERIALS AND METHODS
Cell Culturing and Isolation of Thylakoids-C. merolae strain NIES-1332 was obtained from the Microbial Culture Collection of the National Institute for Environmental Studies (Tsukuba, Japan). Liquid cultures of the C. merolae cells were grown in a Versatile Environmental Test Chamber (Sanyo, Japan) in standard 250-ml tissue flasks oriented in an upright position. Cells were suspended in 50 ml of Allen 2 medium, pH 2.5 (37), at 42°C under continuous 90-E white light illumination with shaking at 115 rpm. Small scale cultures were grown to OD 680 ϳ2.5 and then subcultured into 1-liter interim cultures and finally grown as 10-liter cultures in Allen 2 medium until late log phase (until OD 682 3.5) in the presence of 5% CO 2 administered at a constant flow rate of 3 liters/min under 150-E continuous white light illumination. For thylakoid preparation, cells were harvested at OD 680 ϳ2.5 by centrifugation at 5,000 rpm at 4°C for 10 min. Cell pellets were washed with 50 -100 ml of low ionic strength buffer A (40 mM MES-KOH, pH 6.1, 10 mM CaCl 2 , 5 mM MgCl 2 , 25% (w/v) glycerol) and then resuspended in 50 ml of buffer A supplemented with 50 g/ml DNase I and the Complete TM protease inhibitor mixture (Roche Applied Science). Cells were ruptured by vigorous agitation with 0.1-mm glass beads in a BeadBeater (BioSpec) using 13 cycles of 10 s and interim 4-min cooling off periods. Cell lysate was separated from beads by vacuum filtration. Thylakoids were pelleted by centrifugation at 104,200 ϫ g for 30 min at 4°C and washed once with buffer A. Final thylakoid pellets were resuspended in buffer A at a Chl concentration of 2-2.5 mg/ml, snap-frozen in liquid N 2 , and stored at Ϫ70°C prior to use.
PSII Activity Measurement-Functional activity of the isolated PSII (5 g of Chl) was measured using an oxygen Clarktype electrode (Hansatech). Measurements were performed at 30°C in buffer A in the presence of 0.125 mM 2,6-dichloro-pbenzoquinone (Sigma) and 2.5 mM potassium ferricyanide (POCH, Gliwice, Poland) as the exogenous electron acceptors. Samples (1 g/ml Chl) were illuminated with a standard white light intensity of 6,000 E/m 2 /s or as described, using a KL 2500 LCD white light source (Schott, Mayence, Germany). Activities were calculated from initial rates of oxygen evolution curves. The values of light intensity (600 -25,000 E/m 2 /s), temperature (15-55°C), and buffer pH (3-7.5) were permutated, depending on the experimental settings. The herbicide sensitivity of the Q B site was monitored following preincubation of PSII for 1 min in the dark with 0.001-20 M 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU; Sigma) prior to oxygen evolution measurement under standard conditions. Each measurement was repeated three times for each DCMU concentration.
Carotenoid Content Analysis-Analytical HPLC was performed according to a modified method described before (39), using a maximum flow rate of 1 ml/min and a Nucleosil 100 C18 column (Teknokroma, Barcelona, Spain). Pigments were extracted from cells grown at 90 E/m 2 /s and the corresponding PSII samples (0.5 mg Chl) with a 1-ml acetone/methanol (7:2, v/v) mixture. The extract was concentrated to 1 mg/ml Chl, and samples (20 g of Chl) were analyzed on a C18 column. The content of each carotenoid species was expressed as a ratio of the area under the pigment-corresponding peak to the area under the Chl peak. Pigment molar ratios were calculated using extinction coefficients of 83.2, 91.7, and 125.3 mM Ϫ1 cm Ϫ1 for zeaxanthin, Chla, and ␤-carotene, respectively (40).
LC-MS/MS Identification of PSII Proteins-PSII proteins after acetone precipitation were dissolved in 0.1% (w/v) RapiGest TM surfactant (Waters) in 50 mM ammonium bicarbonate. After reduction and subsequent alkylation of cysteine residues, proteins were digested in sequencing grade trypsin (Sigma-Aldrich) at 30°C overnight. Reaction was stopped by the addition of trifluoroacetic acid to 1% (v/v) final concentration. For small PSII subunits that do not contain trypsin cleavage sites, additional digestion in solution as well as in-gel digestions with chymotrypsin (Sigma-Aldrich) were carried out. Digested peptides were separated using a NanoAcquity Ultra Performance LC system (Waters) connected with a mass spectrometer. Peptides were loaded onto a Symmetry C18 (5 m; l/min in 99% buffer A (0.1% formic acid in water) and 1% buffer B (0.1% formic acid in acetonitrile) for 3 min. Trapped peptides were separated on a BEH 130 C18 (1.7 m; 75 m ϫ 200 mm) analytical column (Waters) equilibrated in 97% buffer A and 3% buffer B. The column was eluted at a constant flow rate of 300 nl/min at 35°C with a linear gradient of buffer B distributed as follows: 3-40% B in the first 145 min; 40 -85% B in 145-151 min; and 85% B in 151-165 min. In the next 5 min, concentration of buffer B was decreased from 85 to 3%, and the column was equilibrated for an additional 10 min before the next injection. Online MS E analyses were performed in a positive ionization mode on the Synapt G2 HDMS mass spectrometer (Waters). Fragmentation spectra were recorded in the range of 50 -2,000 Da (1.0 s/single scan), and the transfer collision energy was ramped in the range of 15-35 V. The mass accuracy of the raw data was corrected using leucine enkephalin (2 ng/l, 1 l/min flow rate, 556.2771 Da/e [M ϩ H] ϩ ) that was infused into the mass spectrometer as a lock mass during sample analysis. Each sample was analyzed at least three times and was mixed with bovine serum albumin tryptic digest (60 fmol) as an internal standard. For protein identification, peak lists were created from the raw data sets and used to search for proteins in a randomized C. merolae protein data bank using ProteinLynx Global Server version 2.4 software (Waters).
77 K Fluorescence Measurement-Steady-state fluorescence spectra were collected using a modified Shimadzu RF-5301PC spectrofluorometer (41) at 77 K and excitation wavelengths of 440 and 580 nm. PSII samples (10 g/ml Chl) in a buffer composed of 40 mM MES, pH 6.1, 10 mM CaCl 2 , 5 mM MgCl 2 , 25% (w/v) glycerol, and 0.03% (w/v) DDM were loaded into prechilled cuvettes and then frozen in liquid nitrogen for 5 min in the dark prior to collecting emission spectra.
Room Temperature Absorbance and Redox Spectroscopy-All measurements were carried out at room temperature using a Shimadzu UV 1800 spectrophotometer. To estimate the content of cytochrome c 550 and cytochrome b 559 , redox difference spectra (oxidized Ϫ reduced) were calculated. For redox spectroscopy, the PSII dimer was diluted to 0.01 mg/ml Chl. 1 ml of the sample was used to prepare the base line in the wavelength range between 600 and 500 nm. Subsequently, a grain of potassium ferricyanide (K 3 [Fe(CN) 6 ]; POCH) was mixed with the sample for 30 s to oxidize the total cytochrome pool prior to collection of the oxidized spectra. For the reduction of cytochromes, a 1-ml sample from the same PSII suspension was mixed with a grain of sodium dithionite (Na 2 S 2 O 4 ; POCH) for 30 s. prior to collection of the reduced absorbance spectra. To ensure complete oxidation or reduction of cytochromes, an additional grain of ferricyanide or dithionite was added before reacquiring the oxidized and reduced spectra, respectively. The cytochrome c 550 concentration was calculated from the maximal absorption at 550 nm and the extinction coefficient of 27.0 mM Ϫ1 cm Ϫ1 (42). For calculation of the cytochrome b 559 content, maximum absorbance at 557 nm and an extinction coefficient of 25.1 mM Ϫ1 cm Ϫ1 were used (42). Each spectrum was recorded three times.

In Vivo Chlorophyll a Fluorescence and Non-photochemical
Quenching-An FL-100 fluorometer (Photon Systems Instruments, Brno, Czech Republic) has been used for measurements of NPQ of maximal fluorescence in vivo. Fluorescence signal was detected in the 690 -750 nm range. Maximal fluorescence of dark-adapted cells (F M ) was measured at a saturating flash (466 nm; ⌬t ϭ 500 ms; ϳ2,000 E/m 2 /s) applied before and after a short period of low intensity blue light (466 nm; duration 60 s; intensity 7 E/m 2 /s). The same setup has been used to exclude the effect of state transitions on the F M value during phycobilisome binding in cyanobacteria (31). NPQ was induced by blue actinic light (466 nm; ⌬, 20 nm; 750 E/m 2 /s, duration 150 s). NPQ value was calculated from quenching of maximal fluorescence in light (F M Ј) at a saturation flash (466 nm; ⌬t ϭ 500 ms; ϳ2,000 E/m 2 /s), according to the Stern-Volmer for- Recovery from the NPQ state was measured in the dark following a given light period. Cells used for in vivo measurements were cultivated in the thermostated bioreactor (38°C) bubbled with air, at a continuous irradiation of 80 E/m 2 /s.
Variable Fluorescence and Effective Antenna Size Measurement in Vitro-Variable fluorescence (F V ) and the effective antennae size of pure PSII ( PSII ) were measured with a custom designed fluorometer FL3500 (Photon Systems Instruments) equipped with a fast repetition rate fluorescence protocol. The single-turnover flash was induced by application of a series of short subsaturating flashes (1.5 s long, ϭ 463 nm) to darkadapted samples (e.g. see Ref. 43). The measured fluorescence rise during the single turnover flash was fitted according to the model described before (44), giving the effective PSII crosssection parameter PSII (43). Variable fluorescence (F V ) was calculated as a difference between maximum (F M ) and minimal (F 0 ) fluorescence measured before (F 0 ) and after (F M ) application of saturating flashes. For measurements at various pH values, purified dimeric PSII samples (0.15-0.2 g Chl/ml) were resuspended in the thermostated (22°C) buffer containing 40 mM MES, 10 mM CaCl 2 , 5 mM MgCl 2 , 25% (w/v) glycerol, and 0.03% (w/v) DDM, in a pH range of 3-7.5. Samples were kept in the dark for 2 min in a buffer of given pH prior to the fluorescence measurement. The reversibility of pH-induced changes was tested by incubating the concentrated sample for 2 min in a buffer of particular pH, followed by transfer to pH 7 by the addition of an appropriate amount of KOH. After a 4-min recovery at pH 7, the sample was used for fluorescence measurements. In all measurements, 1,4-benzoquinone (22.6 M) and potassium ferricyanide (48.8 M) were used as the exogenous electron acceptors of PSII.
SDS-Polyacrylamide Gel Electrophoresis-SDS-PAGE was carried out using the Tris-Tricine system (45,46). Protein bands (5 g of Chl/lane) were resolved overnight on an 18% polyacrylamide gel in the presence of 6 M urea at a constant voltage of 25 V. Proteins were visualized with Coomassie Brilliant Blue R-250 using standard procedures. Prior to SDS-PAGE, samples were treated with 4 volumes of ether/methanol (1:1, v/v) overnight at Ϫ20°C to remove pigments. Total protein was collected by centrifugation at 13,500 rpm at 4°C for 10 min, air-dried at room temperature, and then resuspended in Electron Microscopy-Protein samples were negatively stained with 2% uranyl acetate on glow-discharged carboncoated copper grids. Images were recorded on a Philips CM120 electron microscope with a LaB 6 filament, operating at 120 kV, with a Gatan 4000 SP 4K slow scan CCD camera. The magnification used was ϫ133,000, compatible with a pixel size (after binning the images) of 2.25 Å at the specimen level. GRACE software was used for semiautomated data acquisition (47). From about 2,000 recorded EM images, over 20,000 particle projections were selected. Processing of single particles was performed with the Groningen image-processing (Grip) software package, including multireference alignments, multivariate statistical analysis, and classification. The resolution of the final EM maps was determined with Fourier ring correlation and the 3 criterion (48) and was calculated at 17 Å.

RESULTS
Purification and Compositional Analysis of the C. merolae PSII Dimer-We set out to purify a highly active, intact, and homogenous preparation of the C. merolae dimeric PSII with the aim of examining its capacity to adapt to various extreme conditions as well as obtaining the insight into the eukaryotic PSII molecular structure. Following solubilization of thylakoids with DDM, we separated PSII monomers and dimers using a two-step anion exchange chromatography (AEC) approach (see Fig. 1A). In the second step, we purified a robust and highly active PSII dimer (yield 3.5%) that was stable for up to 120 h of dark incubation at 17°C (see Fig. 3B). The homogeneity of the sample was verified by SEC, which showed a single elution peak with a retention time corresponding to the PSII dimer (Fig. 1B). The purity of the PSII complex was additionally confirmed spectroscopically by identification of characteristic peaks at 673 nm in the room temperature absorbance spectrum and 692 nm in the 77 K emission spectrum (see Fig. 2, A and B), similar to other red algal PSII preparations from Cyanidium caldarium and P. cruentum (22,49). The PSII dimer purity was monitored for each AEC fraction by measuring a ratio of A 673 /A 625 . For the pure dimeric PSII preparation free of phycobilisomes, the A 673 /A 625 ratio was estimated at ϳ5, yielding 99% homogeneous sample.  We then analyzed the subunit composition of the purified C. merolae PSII dimer by biochemical and mass spectrometry approaches. Fig. 2C and supplemental Table S1 show that the reaction center subunits (D1 and D2), inner antenna subunits (CP43 and CP47), four extrinsic subunits stabilizing the OEC (PsbO, PsbU, PsbV, and PsbQЈ), and the majority of small intrinsic subunits were all present in our preparation. In addition, we detected an auxiliary subunit Psb27 and low molecular weight subunit PsbW in our dimer preparations (see supplemental Table S1).
We estimated the stoichiometry of PSII/cytochrome c 550 (PsbV) and PSII/cytochrome b 559 by measuring redox difference spectra in the presence of ferricyanide (oxidized) or dithionite (reduced). The redox difference spectra showed two characteristic peaks at 550 and 557 nm, corresponding to the fully reduced forms of cytochrome c 550 and cytochrome b 559 , respectively (data not shown). The molar ratios of PSII/cytochrome c 550 and PSII/cytochrome b 559 were estimated at 0.954 Ϯ 0.014 and 0.953 Ϯ 0.038, respectively, indicating 1:1 stoichiometry of both cytochromes and PSII (data not shown).
Biochemical Activity and Robustness of the C. merolae PSII-First, we tested the herbicide sensitivity of the Q B site to DCMU. We observed rapid inactivation of oxygen-evolving activity upon the DCMU treatment, with inhibition of more than 95% of PSII activity at a 10 M concentration of the herbicide (data not shown). We estimated the value of the inhibition constant K i at 0.46 Ϯ 0.012 M DCMU, using the Dixon analysis (50). This value agrees well with the previously reported K i value of 0.33 M for DCMU inhibition in spinach thylakoids (51) or in isolated Thermosynechococcus elongatus PSII (52).
We then proceeded with a detailed examination of the catalytical activity of the C. merolae PSII complex across a broad range of extreme conditions (see Fig. 3). The oxygen-evolving activity of PSII varied between 4,000 and 6,000 mol of O 2 /mg of Chl/h for different preparations, with an average activity of 4,500 mol of O 2 /mg of Chl/h under the standard experimental conditions (30°C, pH 6.3, white light of 5,000 -8,000 E/m 2 /s). It was one of the highest recorded oxygen-evolving activities obtained under our standard experimental conditions and exceeded that of PSII isolated from the thermophilic cyanobacteria used to obtain the x-ray structures of the complex (1-3, 53). Moreover, the C. merolae dimeric PSII complex maintained its full activity in a relatively broad range of pH (between 5 and 6.5) when illuminated with white light of 5,000 E/m 2 /s at 30°C (see Fig. 3A). It also exhibited a distinctive temperature optimum, with a maximum activity at 40°C, when measured in standard conditions (see Fig. 3C). Moreover, the C. merolae PSII dimer retained its average oxygen-evolving activity when illuminated with high white light intensities (up to 20,000 E/m 2 /s; see Fig. 3D), indicating that the oxygen-evolving complex remained intact when subjected to light intensities well in excess of the saturation level.
We determined the total pigment composition of C. merolae cells and of PSII dimers, using HPLC analysis (see Fig. 4 and Table 1). Both carotenoids and Chla were detected at 437 nm, whereas exclusive carotenoid peaks were traced at 456 nm, as shown in Fig. 4. The whole cell extract yielded a distinctive peak corresponding to zeaxanthin (Zea) (see Fig. 4, A and B), which nearly equalled the main Chla peak (see Fig. 4A). In contrast, the amount of ␤-carotene was 3-fold lower compared with Chla . PSII activity is sustained in various extreme conditions. A, PSII maintains nearly full activity of 4,500 mol of O 2 /mg of Chl/h in a relatively broad range of pH (between 5 and 6.5) with standard 5,000 E/m 2 /s light intensity and 30°C. B, when stored at 17°C for a period of 5 days, PSII retains nearly 80% of its activity measured in standard conditions. C, PSII has a distinctive temperature optimum, with a maximum activity at 40°C, measured in standard conditions. D, upon increasing light intensity, PSII reaches a maximum activity of ϳ6,000 mol O 2 /mg Chl/h at 8,000 E/m 2 /s, which lowers to 3,000 mol of O 2 /mg of Chl/h at 25,000 E/m 2 /s. Each data point represents an average value from three independent measurements. Error bars, S.D. (Fig. 4A). The level of Zea in the pure PSII dimer was 3-fold lower compared with the whole cells (Fig. 4, compare A and C), whereas the relative ratio of Zea to ␤-carotene was 2-fold lower in the purified PSII dimer compared with the whole cells (Fig. 4, compare B and D; see Table 1). We did not detect any peaks corresponding to violaxanthin or antheraxanthin, either in the whole cells or in the purified PSII.
Investigation into the in Vivo and in Vitro Photoprotection Mechanisms in C. merolae-In order to gain an insight into the photoprotective mechanisms in C. merolae, we set out to dissect the locus of NPQ both in vivo (intact cells) and in vitro (isolated PSII dimers). We investigated the presence and extent of photoprotective NPQ in vivo by measuring the variable Chla fluorescence (F V ) in intact C. merolae cells (Fig. 5). A typical measuring protocol has been applied to exclude the effect of state transitions on the maximal fluorescence in dark (F M ) that is known to be affected by the phycobilisome binding in cyano-bacteria (54). There were no significant changes in the F M value after a short period of illumination with low intensity blue light, excluding the putative influence of state transitions on the measured F M values. The F M value obtained in the dark (see Fig.  5) was then used for calculation of the NPQ kinetics upon transition of the cells to light (Fig. 6). The exposure of dark-adapted C. merolae cells to high intensity blue light resulted in pronounced quenching of maximal fluorescence (F M Ј), in a process that was reversible in the dark (see Fig. 5). These results confirmed the presence of NPQ in C. merolae cells in vivo. Maximal NPQ value was around 0.4 (Fig. 6), and resembled the value determined for cyanobacteria (54), indicating a similar amount of energy that is dissipated during NPQ despite different quenchers employed by both phototrophs. The NPQ was inhibited in vivo after the addition of an uncoupler NH 4 Cl (Fig. 6), FIGURE 4. HPLC pigment analysis of C. merolae cells and PSII dimer. Total pigments were analyzed by HPLC, and their absorbance was measured at 437 nm (for carotenoids and Chla) and at 456 nm (for carotenoids only). Peak identities are as follows. 1, Zea (cis isomer); 2, Zea; 3a, oxidized Chla; 3, Chla; 4, ␤-cryptoxanthin; 5, ChlaЈ; 6, ␤-carotene; 7, ␤-carotene (cis isomer). Peaks were assigned to the corresponding pigments by LC-MS according to Ref. 39. Whole cell extract yielded a distinctive peak corresponding to Zea (A and B) nearly identical to the peak of Chla (A), whereas the ␤-carotene peak was 3-fold lower than that of Chla. The Zea abundance in the pure PSII dimer was significantly lower than in whole cells (C); however, the relative contribution of Zea and ␤-carotene was similar (D) when PSII was compared with the whole cell extract (see Table 1). a.u., arbitrary units.

TABLE 1 Relative abundance of zeaxanthin and ␤-carotene in whole cells and purified PSII
Molar ratios of pigments were calculated by integration of an area underneath the relevant peak and using extinction coefficients, as described under "Materials and Methods." S.D. values were calculated from two independent measurements (n ϭ 2).   demonstrating the role of acidification of the lumen in triggering NPQ in C. merolae cells.

Source
In order to gain the first glimpse into the mechanism of photoprotection in isolated C. merolae dimeric PSII complexes, we measured the effect of various pH levels (pH range of 3-7.5) on variable Chla fluorescence, F V . Photoprotective NPQ that is triggered by low pH can be demonstrated by detection of a decrease of F V (see Ref. 29 for a recent review). As shown in Fig.  7A, variable fluorescence of PSII was progressively reduced at pH below 4.5 and completely disappeared at pH 3. This F V decrease was correlated with the reduction of maximal PSII efficiency, F V /F M (see Fig. 7B), and with a decrease of the oxygen evolution rate at this pH range (compare Figs. 3A and 7). These data indicate that the reduction of photochemical PSII efficiency at low pH is due to stimulation of non-photochemical pathways of de-excitation. In contrast to the pH-dependent F V decrease, there was no quenching of minimal fluorescence (F 0 ) at low pH (see Fig. 7A). Because we have not found any external LHC-type antennae attached to the isolated PSII dimers (see supplemental Table S1), this selective quenching of variable fluorescence (F V ) suggests the presence of the NPQ locus in the C. merolae PSII reaction center. A similar mechanism has been proposed for PSII isolated from higher plants (35,36).
We observed a clear reversibility of low pH-induced changes in maximal PSII efficiency F V /F M , upon transfer of PSII from low to neutral pH. The maximal efficiency of PSII almost fully recovered upon transfer from pH 3.5, with almost 70% recovery of the F V /F M ratio following incubation of PSII at pH 3 (see Fig.  7B, inset), at which no variable fluorescence F V was observed (Fig. 7A). These data clearly confirm the presence of reversible photoprotective NPQ in the dimeric C. merolae PSII exposed to extreme low pH. No NPQ was observed when PSII was exposed to pH below 3 (see Fig. 7B, inset), most likely due to the irreversible damage to the OEC on the donor side and denaturation of the PSII complex. We have tested correlation between low pH-induced changes in F M and inhibition of oxygen evolution rates. We observed linear correlation between F M quenching and reduction in oxygen evolution rates at low pH (pH 3.5-6.5) (see Fig. 8). Conversely, no such correlation was observed at higher pH (pH 7 and 7.5), when the O 2 evolution rate was reduced, whereas F M remained non-quenched (see Figs. 3A and 7A). This is in line with the fact that only low pH can trigger NPQ in vivo.
We further explored the mechanism of NPQ in C. merolae PSII by measuring the effective antennae size of PSII ( PSII ) corresponding to the inner antenna Chls (see Table 2). In intact C. merolae cells, the antennae size of PSII was ϳ30% higher compared with isolated thylakoid membranes or dimeric  Values of F M were normalized at pH 6 and plotted with values of PSII activity at a given pH (numerical values by the data points). A linear correlation with the correlation factor R 2 ϭ 0.89 (p ϭ 0.0014) of both data sets was observed, indicating that within the pH range of 3.5-6.5, the loss of activity is related to quenching of fluorescence and not to irreversible damage to OEC. No such correlation was observed at pH 7 and 7.5. Error bars, S.D.

TABLE 2 Effective PSII antennae size (PSII ) in intact C. merolae cells and isolated PSII
Values represent averages and S.D. from six measurements (n ϭ 6) done with two independent PSII isolations. PSII complexes, indicating a partial loss in delivering excitation energy to PSII upon solubilization of this complex from the membranes. Because the effective antennae size of the C. merolae PSII was similar to that observed for the higher plant complex (see Ref. 36), our PSII dimer was isolated in a fully native and functional form. There was no significant difference in PSII for isolated PSII and thylakoids that contain both PSI and PSII (see Table 2). In isolated PSII, we did not observe any decrease of PSII at pH 3.5, a pH value at which F V was already significantly reduced (Fig. 7, compare A and B). The constant (or slightly increasing) values of PSII at low pH preclude the possibility of de-excitation of excess light occurring within the internal PSII-core antennae CP43 and CP47 because such a mechanism would imply a decrease in both PSII and F V .

Source
Structural Analysis of the C. merolae PSII Dimer-In order to get an impression of subunit organization within the C. merolae PSII dimer, we visualized the complex by negative stain electron microscopy combined with single particle analyses. Single particle averaging of a large set of dimers showed that the particles were mostly present in top lumenal or side view positions. Within the top views, there was little variation, allowing for the final projection map to be obtained at a resolution of 17 Å, as depicted in Fig. 9A. In the class of side view projections, we detected some variation at the site of the extrinsic subunits, which resulted in a partition of the data set into three classes (Fig. 9, B-D). As a main difference between the classes, there is a stronger or fainter presence of the 12-kDa subunit (PsbU) at the top of the other extrinsic subunits (orange arrows in Fig. 9, C and D). In addition, the visibility of the cytochrome c 550 (PsbV) subunit is variable (blue arrows in Fig. 9, B and D), despite the apparent 1:1 stoichiometry of PsbV/PSII (data not shown). The size of a side view C. merolae PSII dimer particle (Fig. 9, B-D) was estimated at 21.2 ϫ 11 nm, which is in a good agreement with the size of the PSII dimers isolated from another red alga, P. cruentum (49), as well as dimensions of the cyanobacterial PSII dimer calculated from the atomic crystal structure (19.9 ϫ 11.6 nm; see Ref. 1).
Comparison with the structural work on cyanobacterial PSII dimers could give us a hint of where PsbQЈ, an extra extrinsic subunit of C. merolae over and above the cyanobacterial homologues (PsbO, PsbU, and PsbV), is located. First, the side views were analyzed because in this position we observed less overlap between the PSII subunits, allowing for the distinctive differences between the maps to be dissected. Indeed, we identified an additional protein mass visible in the map of C. merolae (Fig.  10A) compared with the high resolution structure of PSII from T. vulcanus (Fig. 10B) digitally truncated to 8 Å for comparison of protein densities. The yellow asterisk indicates the position where the C. merolae PSII dimer seems to have protein mass that is absent in the cyanobacterial structure (Fig. 10, compare  A and B). The most obvious candidate for the location in this position is the additional lumenal extrinsic subunit PsbQЈ.

DISCUSSION
In the present work, we isolated and purified the dimeric PSII complex from the red alga C. merolae and demonstrated its remarkable robustness over a range of extreme conditions, including high light illumination, high temperatures, and extreme pH range conditions that usually lead to destablization of the OEC and inhibition of PSII function. The oxygen-evolving activity of the isolated complex is one of the highest reported to date and exceeds the activity of dimeric PSII complexes isolated from, for example, the thermophilic cyanobacteria T. elongatus and T. vulcanus used to obtain medium to high resolution structures of PSII (1,2,53). Although the pH optimum for the activity of the C. merolae PSII complex is at pH 5-5.5, the complex sustains over 40 -60% of its oxygen-evolving activity at pH 3.5 and pH 7, which is unprecendented for other PSII preparations isolated from thermophilic cyanobacteria (55)(56)(57), mesophylic and other thermoacidophilic red algae (38,49), green algae (58), and higher plants (59,60). Moreover, the C. merolae complex increases its activity at relatively high temperatures up to 45°C as well as retaining its intactness over a prolonged period of incubation at room temperatures (see Fig. 3), which may make this complex amenable to application in biometic solar-to-fuel devices often intended to operate at elevated temperatures over extended periods of time (61,62).
One important question arises: what are the molecular mechanisms responsible for the remarkable robustness of C. merolae PSII, especially under extreme illumination conditions? Under the conditions of imbalance between energy capture and utilization (e.g. in high light), accumulation of excitons in external and intrinsic antennae of both photosystems increases the occurrence of singlet Chl-excited states ( 1 Chl*). Under excess light, 1 Chl* species often undergo intersystem crossing to form the Chl triplet state ( 3 Chl*) (63). This species reacts with molecular oxygen to form highly reactive oxygen species, including singlet oxygen ( 1 O 2 ), when excitations are not rapidly quenched. This can occur when the capacity for productive photochemical quenching is exceeded and non- photochemical quenching is insufficient or not yet sufficiently turned on (64 -66).
Our carotenoid and chlorophyll fluorescence quenching analyses point to the involvement of both zeaxanthin and the primary oxidant, P680 ϩ , being responsible for the photoprotection of thylakoids and PSII under high light irradiance. It is well established that carotenoids are essential for photoprotection in oxygenic photosynthesis (reviewed in Refs. 28, 63, 67, and 68). Moreover, they affect assembly and stability of the PSII reaction center and light-harvesting antenna subunits (69,70).
Of all of the carotenoids, Zea plays a particularly important role in photoprotection. High light-induced synthesis of Zea and its binding to the specific proteins enhance photoprotection by decreasing the yield of potentially dangerous Chl-excited states. A recent study using time-resolved differencial spectroscopy demonstrated increased efficiency of this molecule in controlling 3 Chl* species formation (71). Zea is known to be involved in several types of photoprotection of the PSII reaction centers that occur at various time scales: (i) rapid feedback de-excitation quenching (qE) of 1 Chl*, (ii) a slowly inducible quenching (qZ) caused by Zea binding to LHC upon exchange with violaxanthin, and (iii) a long term irreversible quenching associated with PSII photoinhibition (71).
In this work, we have identified a considerable amount of Zea both in intact C. merolae cells and in the isolated PSII complex (see Fig. 4 and Table 1), indicating that this xanthophyll may play an important photoprotective role in reactive oxygen species scavenging in the lipid phase of C. merolae thylakoids. Because the light-harvesting antenna of red algal PSII is formed by phycobilisomes, it is unlikely that Zea would be involved in slowly inducible quenching qZ. Importantly, our pigment quantification suggests that in the C. merolae PSII complex, Zea may also be present inside the PSII core complex, where it may take a direct role in quenching of 3 Chl* species. Taking into account the pigment molar ratios presented in Table 1 and normalizing these values to 35 Chla molecules/PSII, as determined in the latest atomic structure (1), we estimate 4 -5 ␤-carotenes and 8 -10 Zea molecules in our PSII complex. Because we would expect 11 ␤-carotenes/PSII (1), it is possible that Zea may replace some ␤-carotenes in the C. merolae PSII complex, perhaps at the boundary of CP43 and associated low molecular weight subunits. This intriguing possibility needs to be supported by a more detailed study.
Our present data showing insensitivity of the effective PSII antennae size to low pH-induced quenching (Fig. 7B) seems to exclude a direct role of C. merolae PSII inner antennae in quenching. The pH insensitivity of the effective antennae size during NPQ is characteristic of the NPQ locus present within the closed reaction centers upon charge recombination (34,72). It is therefore possible that some Zea molecules putatively bound in the vicinity of the charge separation pathway in C. merolae PSII could be involved in the RC-based quenching observed in the present study. Similarly, Fleming and colleagues (73) observed that the mechanism of non-radiative deactivation of 1 Chl* during excess light occurs by excitation transfer to a Chl-Zea heterodimer, followed by ultrafast Car ϩ⅐ formation. Such a process was observed for Zea bound within the peripheral light-harvesting antenna of PSII. Whether a similar mechanism exists for Zea putatively bound within the RC of red algal PSII remains to be elucidated.
Alternatively, Zea could play a direct antioxidant role in scavenging free radicals and singlet oxygen molecules produced in high light in the lipid phase of the thylakoid membrane in the vicinity of PSII macrodomains (74). In light of the high lipid content of the cyanobacterial PSII dimer shown in its latest atomic structure (1), it is tempting to speculate that free Zea molecules bound within the C. merolae PSII complex-lipid interface may protect these intrinsic lipid molecules from 1 O 2mediated peroxidation. A similar NPQ-independent protective role of Zea in scavenging singlet oxygen has been observed for the ch1 mutant of Arabidopsis thaliana devoid of the LHCII antenna (75). In addition, Zea identified in this study in intact cells of C. merolae might play a role as a thylakoid membranerigidifying molecule (76,77), allowing this extremophilic alga to thrive at high temperatures.
Similarly, Gantt and colleagues (78) showed that in another unicellular red alga, P. cruentum, the cellular content of Zea increased with growth irradiance, confirming a role for this carotenoid in photoprotection. In the latter work, the ratio of cellular Zea/␤-carotene incrementally increased upon high light exposure of Porphyridium cells and was similar to the ratio obtained in this study for similar conditions of illumination (2.6 and 2.8, respectively). The relative content and composition of carotenoids and Chl are also similar in C. merolae (see Table 1 in this study and Ref. 39) and P. cruentum cells (78). Importantly, we did not detect violaxanthin either in whole cells or purified PSII (see Fig. 4) and PSI (not shown), confirming the previous data of Gantt and colleagues (39), who showed that this dieepoxide xanthophyll is absent in C. merolae cells. Instead, in this organism, Zea most likely forms by conversion of ␤-carotene via intermediate ␤-cryptoxanthin (79), a process catalyzed by ␤-carotene 3-hydroxylase enzymes, such as the CrtR cyanobacterial type ␤-carotene hydroxylase or cytochrome P450 (39).
In addition to the Zea-mediated photoprotection, our fluorescence measurements clearly point to the pH-dependent reaction center quenching as the main photoprotective mechanism in the C. merolae PSII. This type of quenching seems to sustain a high oxygen-evolving activity of C. merolae PSII even at extremely high irradiation (see Fig. 3). Kirilovsky and colleagues (32,33) have shown that non-photochemical quenching in mesophilic red algae is triggered by low pH and may occur in the PSII reaction center rather than the inner antenna. However, a direct proof for this type of photoprotective mechanism and dissection of its exact locus were missing because their work was performed exclusively in intact cells. Due to this limited experimental approach, the authors could only suggest the reaction center-based mechanism of NPQ in red algae based on indirect measurements of F 0 pH insensitivity during stimulation of non-photochemical quenching (32,33).
Here, we provide the first direct evidence of the PSII reaction center as the primary locus of pH-dependent NPQ in the extremophilic red algae. This type of quenching can occur through (i) energy dissipation via alternative electron transport after primary radical pair production, (ii) charge recombination between Pheo Ϫ and P680 ϩ (80), (iii) rapid charge recombination between Q A Ϫ and P680 ϩ , and (iiii) accumulation of P680 ϩ that can act as a direct quencher (30,36). All four types of RC-based quenching are consistent with pH insensitivity of effective antenna size during quenching. Indeed, our measurements of the absorbance cross-section of isolated C. merolae PSII revealed no decrease in the functional antenna size at low pH (see Fig. 7B), which supports a reaction center-based quenching mechanism in this extremophile when exposed to saturating light levels. Importantly, we detected significantly decreased F V values at low pH, a phenomenon that was rapidly reversible upon exposure of PSII to physiological pH (see Fig.  7B). We also observed a significant correlation between F M (and F V ) quenching and inhibition of oxygen evolution in isolated PSII at low pH (see Fig. 8) and a lack thereof at higher pH. These data exclude the possibility of the OEC inactivation as the primary cause for fluorescence quenching in the isolated C. merolae PSII and confirm the presence of reversible low pHinduced NPQ in this complex.
We previously obtained three-dimensional crystals of the C. merolae PSII dimer (81) with the aim of solving the structure of this complex. Because we were unable to significantly improve the resolution of x-ray diffraction patterns obtained from these crystals, we applied electron microscopy coupled with single particle analyses of the isolated PSII dimer to gain an insight into the structure of this eukaryotic complex, in particular its lumenal side. Comparison with structural work on cyanobacterial PSII dimers allowed us to suggest an assignment of the position for the PsbQЈ, an extra extrinsic subunit over and above the cyanobacterial-like PsbO, PsbU, and PsbV subunits stabilizing the OEC in the C. merolae complex (24 -26). In our difference map in which the atomic structure of cyanobacterial PSII was modeled onto the averaged side and lumenal top view projections of the C. merolae PSII dimer particles, the additional density attributed to PsbQЈ is present close to the lumenal side of the membrane, in the vicinity of the CP43 inner antenna subunit (see Fig. 10). Our assignment of the PsbQЈ position is supported by previous EM studies of cyanobacterial PSII (82,83), whereby the side view dimer projection lacks the additional mass identified in our C. merolae PSII dimer particles (Fig. 10E). Another position would be in the center, because the EM and x-ray data also differ here, such that in the EM map, the extrinsic protein densities for both monomers appear to be more compact. However, this location is unlikely, because the cyanobacterial map of T. elongatus is essentially the same in this position (Fig. 10E).
Our assignment of PsbQЈ differs somewhat from that presented for the dimeric PSII particles from a closely related red alga, C. caldarium (22), in that it places this additional OEC subunit closer to the membrane in the vicinity of the CP43 intrinsic antenna subunit. Vacha and colleagues proposed positioning of PsbQЈ more toward the lumenal space, between the PsbU and PsbV subunits (22). However, in contrast to the present work, they were unable to reveal any additional protein densities in the projection maps of C. caldarium compared with the equivalent maps of P. cruentum (49) and T. elongatus (82) PSII dimer particles.
The comparison with the cyanobacterial data further shows a surprising similarity between the top lumenal views of both C. merolae (Fig. 10C) and T. elongatus (Fig. 10D). There are no significant differences in surface densities, indicating that the extra subunit of C. merolae is not visible, most likely due to an overlap with the membrane-integrated moiety of the PSII dimer, possibly the CP43 intrinsic antenna subunit. The close similarity in the surface and shape points to a highly similar composition of the membrane-bound subunits. The map, as depicted in Fig. 10D, was obtained from a PSII particle lacking the PsbZ protein. This protein was, however, present in the map shown in Fig. 10F (83). Thus, it seems that our C. merolae PSII particles do not contain a subunit homologous to the cyanobacterial PsbZ subunit. Interestingly, in the side view projection map of the C. merolae PSII dimer, the position of the PsbO subunit seems to be slightly shifted toward the pseudosymmetry axis of the dimer, compared with the crystal structure of the cyanobacterial counterpart (see Fig. 10A). A similar phenomenon was observed for subpopulations of the dimeric PSII from a green alga Chlamydomonas reinhardtii with altered organization of the LHCII antenna, 4 suggesting a certain level of structural flexibility within the OEC region.
Interestingly, our mass spectrometry analysis showed the presence of the PSII auxiliary subunit Psb27 and a low molecular weight subunit PsbW implied in the regulation of the PSII repair cycle (84) and stabilization of the PSII dimers in higher plants (85). Both proteins were detected in all rounds of MS/MS analysis from three independent preparations, suggesting their stable association with the population of C. merolae PSII (prob-ably in its inactive form), which most likely undergoes a repair cycle. However, the stoichiometry of their binding to PSII as well as the turnover rate for the C. merolae PSII are presently unknown and will be a subject of future studies.