In Vitro Random Mutagenesis of the D1 Protein of the Photosystem II Reaction Center Confers Phototolerance on the Cyanobacterium Synechocystis sp. PCC 6803*

The D1 protein of the photosystem II reaction center is thought to be the most light-sensitive component of the photosynthetic machinery. To understand the mechanisms underlying the light sensitivity of D1, we performed in vitro random mutagenesis of the psbA gene that codes for D1, transformed the unicellular cyanobacterium Synechocystis sp. PCC 6803 with mutated psbA, and selected phototolerant transformants that did not bleach in high intensity light. A region ofpsbA2 coding for 178 amino acids of the carboxyl-terminal portion of the peptide was subjected to random mutagenesis by low fidelity polymerase chain reaction amplification or by hydroxylamine treatment. This region contains the binding sites for QB, D2 (through Fe), and P680. Eighteen phototolerant mutants with single and multiple amino acid substitutions were selected from a half million transformants exposed to white light at 320 μmol m−2s−1. A strain transformed with non-mutagenizedpsbA2 became bleached under the same conditions. Site-directed mutagenesis has confirmed that one or more substitutions of amino acids at residues 234, 254, 260, 267, 322, 326, and 328 confers phototolerance. The rate of degradation of D1 protein was not appreciably affected by the mutations. Reduced bleaching of mutant cyanobacterial cells may result from continued buildup of photosynthetic pigment systems caused by changes in redox signals originating from D1.

Although photosynthetic organisms require light for growth, they can suffer damage from high light intensity, especially at low temperatures and reduced concentrations of CO 2 (1). The loss of photosynthetic productivity that occurs when organisms are exposed to visible light quanta above the level required for saturating photosynthetic electron flow is known as "photoinhibition" (2). Under photoinhibitory conditions, the reaction center of photosystem (PS) 1 II, which consists of D1 and D2 proteins (3), is specifically inactivated (4,5). Although the precise mechanisms of photoinactivation have not been fully elucidated, the process involves several steps, including an initial reversible reduction of electron flow and irreversible damage to the D1 protein (6 -8). Recovery of PSII activity can occur when irreversibly photodamaged D1 is replaced by newly synthesized protein (7).
Site-directed or deletion mutagenesis has been employed to investigate the basic mechanisms of the photosensitivity, including the relationships between the structure and function of D1 (9 -12). Although site-specific mutagenesis seldom confers new function, this technique has produced some mutants in which PSII and D1 gained partial resistance to high irradiance (9,10,12). In vitro random mutagenesis, ideally saturation mutagenesis, coupled with appropriate methods for screening mutants, is a useful method to obtain new species of D1 protein that are phototolerant. For these experiments, we have chosen the unicellular cyanobacterium Synechocystis sp. PCC 6803. In addition to having plant-like photosynthetic activity, Synechocystis can be genetically transformed at high efficiency and can be easily screened as colonies under defined conditions (13).

EXPERIMENTAL PROCEDURES
Strains and Growth Conditions-Synechocystis sp. PCC 6803 strain Cm4⌬-1 provided by R. J. Debus (University of California, Riverside, CA) was employed as a host of mutated psbA2. In this strain, both psbA1 and psbA3 genes (14) have been inactivated by insertion mutagenesis with antibiotic resistance cassettes (15,16). Cultures were grown on BG-11 agar medium in the presence of 20 g/ml spectinomycin and 5 g/ml chloramphenicol at 30°C under a photon flux density of 50 mol m Ϫ2 s Ϫ1 (13). The Cm4⌬-1 strain was transformed as reported (13) with plasmid pPSBA2-KM (Fig. 1). This plasmid consists of native psbA2 with a kanamycin resistance cassette (Kan r ) integrated at the StuI site by blunt-end ligation on the backbone of pUC118. The Kan r was excised from pCR1000 (Invitrogen) by digestion with AvaI and SacII, followed by end-blunting. The resulting transformant designated "KC" contained the psbA2-Kan r construct integrated in the chromosome by homologous recombination. Strain KC was grown in the presence of 20 g/ml spectinomycin, 5 g/ml chloramphenicol, and 20 g/ml kanamycin under the same conditions as strain Cm4⌬-1. Liquid cultures were grown under the same conditions as cultures on agar plates, except that air mixed with 1% CO 2 was continuously supplied.
In Vitro Random Mutagenesis-Mutations were randomly introduced into the 3Ј half of psbA2 by polymerase chain reaction (PCR) under low fidelity conditions (17). This region occurs between nucleotide numbers 538 (KpnI) and 1071 (HincII) in the coding frame (534 base pairs) (Fig. 1) and corresponds to amino acid residues Phe-180 to Ala-357 of D1. PCR was performed using 1 ng of the DNA template and * This work was supported by Grants-in-aid for Scientific Research for Priority Areas (04273101) and General Research (06404003 and others) from the Ministry of Education, Science and Culture of Japan (Monbusho); and by grants from the Human Frontier Science Program (to K. S.), the New Energy and Industrial Technology Development Organization/Research Institute of Innovative Technology for the Earth (to H. K.), and the Toray Science Foundation (to H. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ The first two authors contributed equally to this work. ** To whom reprint requests should be addressed: Laboratory of Plant Cell Technology, Graduate School of Nutritional and Environmental Sciences, University of Shizuoka, 52-1 Yada, Shizuoka 422-8526, Japan. E-mail: hirokazu@u-shizuoka-ken.ac.jp. primers enclosing the entire coding region (5Ј-GCCGAAGCTTAAG-GAATTATAACCAAATGACAACGACTC-3Ј and 5Ј-AGGGAAGCTTAC-CAAGGAATTAACCGTTGACAGCAGGAGC-3Ј). The reaction was carried out in presence of 1 mM each of dCTP, dGTP, and dTTP, 0.2 mM dATP, 6.1 mM MgCl 2 , and 0.5 mM MnCl 2 for 25 cycles to induce nucleotide transition from A to G and T to C as reported (17). The PCR products were digested with KpnI and HincII and ligated into pPSBA2-KM. A mutant library of D1 proteins was produced by transforming Synechocystis sp. PCC 6803 strain Cm4⌬-1 with this construct as described (13) except that ligated DNA was used directly without amplification in Escherichia coli since constructs harboring psbA likely interfere with the growth of E. coli cells.
Random mutagenesis with hydroxylamine as reported, to introduce the mutations C 3 T and G 3 A (18), was also performed. Fifty g of the KpnI-HincII fragment of psbA2 ( Fig. 1) was incubated in a reaction mixture (200 l) containing 0.5 M hydroxylamine HCl and 5 mM EDTA in 0.1 M sodium phosphate (pH 6.0) at 70°C for different periods. An aliquot (20 l) of the mixture was then diluted 10 times with TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA) to terminate the reaction. The treated DNA was precipitated with ethanol two times and resuspended in 20 l of TE buffer. The DNA fragments treated with hydroxylamine for 90 min were inserted into pPSBA2-KM and used to transform strain Cm4⌬-1 as described for PCR mutagenesis.
Site-directed Mutagenesis-Point mutations were introduced into wild-type psbA2 by site-directed mutagenesis using a modification of the PCR method (19). To achieve higher fidelity nucleotide incorporation, Pfu DNA polymerase (Stratagene) was used instead of Taq DNA polymerase. Mutagenesis experiments were performed on a KpnI-Hin-dIII fragment covering the 3Ј portion of psbA2 and its 3Ј downstream region subcloned into pUC118. Mutagenized DNA species were amplified by PCR using the primer 5Ј-GCTTGATATCGCATTCAGCGTC-3Ј (the italicized sequence for disrupted EcoRI site in the vector adjacent to the HindIII site on the fragment described above). The mutated KpnI-EcoRI fragment was ligated with corresponding sites in pBluescript KS Ϫ (Stratagene) and cloned into E. coli JM109 to confirm the mutation by DNA sequencing. The mutagenized KpnI-HincII fragment was returned into pPSBA2-KM and used to transform strain Cm4⌬-1.
DNA Sequencing-The full-length psbA2 (1.1 kilobase pairs) was amplified from a single colony of Synechocystis or E. coli by PCR with the same primers used for PCR random mutagenesis. The nucleotide sequences of the PCR products were determined using a 373A DNA Sequencer (Applied Biosystems, Perkin-Elmer). The samples were prepared by the cycle sequence method (20) using Taq DNA polymerase and one of four kinds of primer labeled with fluorescent markers according to the manufacturer's instructions (5Ј-Aminolink2CAGGACC-ATTGGTCAAGGCTCCTTC-3Ј, 5Ј-Aminolink2-CAGGACCTACAACAT-CGTTGCCGCC-3Ј, 5Ј-Aminolink2-CAGGAGCATTGCGTTCGTGCAT-TAC-3Ј, and 5Ј-Aminolink2-CAGGAATGCCGATTACAGGCCAAGC-3Ј). Samples were separated on sequencing gels containing 8 M urea and 6% acrylamide. Both strands of DNA were sequenced to confirm the nucleotide sequences.
Turnover of D1 Protein-Pulse-chase labeling of cyanobacterial cells with [ 35 S]Met was performed essentially as reported (12). Washed cells were resuspended in sulfur-depleted BG-11 medium (BG-11-S) in which MgSO 4 , ZnSO 4 , and CuSO 4 were replaced with MgCl 2 , ZnCl 2 , and Cu(NO 3 ) 2 , respectively. The chlorophyll concentration of each suspension was adjusted to 25 g/ml, and the cells were exposed to 50 mol m Ϫ2 s Ϫ1 white light for 30 min at 30°C. L-[ 35 S]Met (Ͼ37 MBq/mmol, ICN) was added to achieve a final activity of 37 kBq/ml, and light treatment continued for another 30 min. Radiolabeled cells were collected by centrifugation and resuspended in the same volume of BG-11-S medium containing 2 mM non-radioactive Met. Cells were subjected to high irradiance (500 mol m Ϫ2 s Ϫ1 ), and 10-ml aliquots were withdrawn at appropriate intervals for preparation of thylakoid membranes (12). Thylakoid membrane proteins were electrophoresed in 15% polyacrylamide gels containing 6 M urea and 0.1% SDS (21). Thylakoid membranes containing 5 g of chlorophyll were loaded in each lane. Autoradiograms were recorded using a BAS2000 Bio-Imaging Analyzer (Fujix, Tokyo).

Screening of Phototolerant Mutants-
To determine suitable conditions for screening phototolerant mutants, Synechocystis sp. PCC 6803, its progeny strain Cm4⌬-1 in which psbA1 and psbA3 were disrupted (15,16), and a derivative of Cm4⌬-1 transformed with the psbA2-Kan r construct (strain KC), were cultured under 50 -640 mol m Ϫ2 s Ϫ1 white light. Although recent information compiled at CyanoBase 2 indicates that the StuI site where Kan r was introduced lies within one of two copies of speA encoding arginine decarboxylase present in the chromosome (Fig. 1), the phenotype of strain KC was indistinguishable from wild-type PCC 6803 and strain Cm4⌬-1. When exposed to 320 to 640 mol m Ϫ2 s Ϫ1 light, the pigment contents of all three strains decreased dramatically with incubation time, changing from deep blue-green to yellow-green (data not shown). Spectroscopic analysis of pigments demonstrated that, under high irradiance, the levels of phycobiliproteins decreased rapidly, followed by a decrease in the level of chlorophyll a. In contrast, carotenoids were stable under the same conditions (data not shown). Therefore, the screening of phototolerant transformants with mutagenized psbA2 was conducted on agar plates exposed to continuous irradiance at 320 mol m Ϫ2 s Ϫ1 for 10 days. A region of psbA2 coding for 178 amino acids from the carboxyl-terminal portion of D1 was subjected to random mutagenesis by PCR under low fidelity conditions and by hydroxylamine treatment. The peptide encoded by this DNA fragment, includes the binding sites for Q B , D2 (through Fe), and P680, as well as the proposed cleavage sites affected by photodegradation. Thirty-five colonies that did not bleach at this light intensity were selected from approximately one half million transformants harboring randomly mutagenized psbA2 species. Southern blot analysis confirmed that the endogenous psbA2 was replaced with one copy of psbA2-Kan r per chromosome in each mutant (data not shown).
Nucleotide Substitutions-Nucleotide sequences of the mutagenized region of psbA2 from all the phototolerant colonies were determined, and amino acid sequences were deduced from them (Fig. 2). A total of 35 phototolerant colonies were assigned to 18 different species of mutated D1 protein. Some mutants were not unique because some mutagenized cells apparently divided during the 5-h transformation before being spread on agar plates. Two mutants designated "H2" and "H7" were generated by hydroxylamine treatment; the others were produced by PCR random mutagenesis. In the mutagenized DNA regions of the 18 unique mutated protein species, there were 63 transitions and 26 transversions. These mutations covered all possible conversions among A, C, G, and T with a strong bias toward A 3 G (27%) and T 3 C (34%) transitions, and resulted in 56 missense mutations and 33 silent ones.
Amino Acid Substitutions-Amino acid substitutions occurred at 34 of the 178 amino acid positions exposed to mutagenesis. Mutagenesis resulted in 18 unique mutated D1 protein species, with one to five amino acid replacements per species (Fig. 2). Changes at 21 of the 34 sites occurred at residues that are conserved in D1 of photosynthetic organisms ranging from prokaryotic cyanobacteria to higher plants (25). Of the 56 amino acid conversions that occurred in the phototolerant mutants, 18 were to Ser and 8 were to Asp. The conversion of Phe to Ser was especially common, occurring in 12 of 18 cases. Conversions tended to substitute polar or charged amino acids for nonpolar ones. Although mutations at multiple positions interfered with the identification of single amino acid residues responsible for phototolerance, single mutations at 254 or 267 in mutants H7 or G6 were also obtained. Amino acid substitutions generally occurred in three regions: (i) the PEST-like sequence (26) and its vicinity, (ii) the region between the amphipathic DE-helix and the beginning of the transmembrane E-helix, and (iii) the carboxyl-terminal extrusion from the E-helix, based on the putative organization of D1 in the PSII reaction center (22)(23)(24).
We examined whether phototolerance of mutants A1, A2, A8, A9, B2, G6, H7, and I6 was caused by mutation in the targeted region or by spontaneous mutations or DNA rearrangements at other loci. The KpnI-HincII fragment of psbA2 in each mutant was prepared by PCR and restriction enzyme digestion, ligated with corresponding sites of pPSBA2-KM, and introduced into strain Cm4⌬-1. The resulting transformants tolerated 320 mol m Ϫ2 s Ϫ1 light just as the original mutants did (data not shown), indicating that the phenotypes were due to mutation of psbA2.
Site-directed Mutagenesis-To identify the residues responsible for the observed phototolerance, site-specific mutations were introduced into amino acid residues having a high frequency of displacements by random mutagenesis (Asn-234, Tyr-254, Phe-260, Asn-266, Asn-267, Phe-273, Phe-274, Asn-322, Ile-326, and Phe-328) ( Table I). Site-directed mutagenesis has revealed that a single mutation at Tyr-254 or Asn-267 is involved in phototolerance and that mutations at Asn-234 or Phe-260 also contribute to phototolerance. Single mutations at other positions have less effect on the phenotype. In contrast, a double mutation at Asn-234 and Phe-260 has a dramatic cumulative effect on phototolerance (Table I). It is remarkable that Asn-234 is located in the area of a "PEST-like sequence," which is proposed to be involved in photodamage (26), and that is the most frequently mutated residue in the 18 phototolerant mutants (Fig. 2). A single mutation at Asn-322 or a double mutation at Ile-326 and Phe-328 on the lumenal extrusion near the COOH terminus did not contribute to phototolerance, but a triple mutation at these residues conferred high phototolerance (Table I). Similarly, mutations at Phe-260 or at Ile-326 and Phe-328 did not affect phototolerance, but a triple mutation at these positions produced phototolerant transformants ( Table  I).
The overall results led us to conclude that substitution of one or more amino acids at residues 234, 254, 260, 267, 322, 326, or

FIG. 2. Amino acid substitutions in D1 proteins generated by in vitro random mutagenesis and selection under high irradiance.
Amino acids in large capitals and nucleotide sequences of codons in small capitals are presented along amino acid residue numbers: Wild indicates amino acids of wild-type D1 protein of Synechocystis sp. PCC 6803; A1, A2, and other designations on the left are mutated D1 protein species in selected transformants. Two mutants designated H2 and H7 were those generated by hydroxylamine treatment. All other mutants were generated by PCR random mutagenesis. Numbers of transformants possessing the same amino acid substitutions are shown on the right. Defined secondary structures are indicated along with the amino acid residues of wild-type D1 as reported (22)(23)(24). Underlined residue numbers are those subjected to site-directed mutagenesis. Italicized designations of mutants on the left, G6, H7, and I6, were mutagenized randomly but resulted in mutations at a specific site or region. Amino acids underlined in the row of wild-type D1 are those conserved among photosynthetic organisms so far examined (25). 328 plays a role in phototolerance. A single mutation at Tyr-254 or Asn-267, double mutation at Asn-234 and Phe-260, and triple mutation at Asn-322, Ile-326, and Phe-328 confer the highest level of phototolerance.
Visible Characteristics of Phototolerant Mutants- Fig. 3 shows examples of mutant A9 on agar medium (panels A and B) and A2 in liquid culture (panel D). There was no significant difference in pigmentation between the control strain KC and mutant A9 grown under 50 mol m Ϫ2 s Ϫ1 light (Fig. 3A). However, under 320 mol m Ϫ2 s Ϫ1 light, the control was bleached in 1 week, while the A9 mutant retained its deep green color (Fig. 3B). In liquid culture, mutant A2 retained its color during 36-h exposure to 750 mol m Ϫ2 s Ϫ1 light (Fig. 3D), but the control strain KC was completely bleached under these conditions (Fig. 3C). All mutants generated by random mutagenesis exhibited pigmentation profiles equivalent to those of A9 and A2 (data not shown).
Turnover of D1 Protein-The rapid turnover of D1 in vivo may be due to a PEST-like sequence located between residues 225 and 238 (26). Pulse-chase experiments using [ 35 S]Met were performed to determine the turnover rate of D1 in phototolerant mutants. D1 protein bands were identified in separate experiments by immunoblotting with an antibody against D1 (data not shown). At 500 mol m Ϫ2 s Ϫ1 light, the turnover rate (both synthesis and degradation) was similar in KC and mutants N234D-F260S (NDFS) and I6 (Fig. 4). Similar results were obtained with mutants G3, G6, and H7 (data not shown). Therefore, it is concluded that the mechanism of phototolerance observed in the mutants is distinguished from that proposed as "damage-repair cycle" (7). DISCUSSION What events occur prior to bleaching in cyanobacterial cells under irradiance? The cell-doubling times of the control strain KC and all mutants were nearly identical at approximately 24 h under irradiance at 500 mol m Ϫ2 s Ϫ1 (data not shown), when the control strain gradually bleached. Therefore, bleaching is an event primarily independent of cell growth. Superoxide anion radical known as the primary product of oxygen photoreduction in thylakoids (27) may degrade photosynthetic pigments. Absorption spectra of the control strain KC and mutant G3 as determined time-sequentially after exposure to irradiance indicate that chlorophyll a content diminishes specifically during bleaching in the control strain (28). This observation is contrary to the phenomenon of sodB Ϫ mutant of the cyanobacterium Synechococcus sp PCC 7942, which is deficient in functional iron superoxide dismutase (29), where the carotenoid-to-chlorophyll ratio strikingly decreased when cultured in the presence of methyl viologen. Active oxygen species have also been reported to degrade D1 protein (30), in relation to "damage-repair cycle of D1 protein" (26,31,32). In the present investigation, the rate of degradation of D1 protein did not change in the wild-type and mutant strains under irradiance (Fig. 4). Therefore, the possibility that generation of oxygen radicals is reduced in the mutants may be excluded.
Phototolerance associated with PSII in photosynthetic organisms has been suggested to occur by one or more mechanisms: (i) an efficient energy dissipation or leakage at the stage of excitation energy transfer in the pigment system (33)(34)(35)(36) capable of being detected by the fluorescence emission spectra at low temperatures (37)(38)(39), and (ii) an efficient electron leakage via some kind of cyclic electron flow around the PSII in the electron transport system due to instability or redox potential change of the primary or secondary reactants as detected by shifts in thermoluminescence bands (40,41). It is hypothesized that redox signaling can regulate gene expression (42,43) and physical linkage that facilitates energy transfer between light-harvesting complex II and PSII/PSI (44 -46) through protein phosphorylation. The redox sensor is postulated to be an electron carrier located between PSII and PSI (42), and a kinase is activated by a reduced cytochrome b/f complex interacting with a plastoquinol at the Q B site (46).
Our preliminary analysis of the thermoluminescence profiles indicates that changes by the mutagenesis of D1 in the equilibrium between Q A and Q B due to (i) changes in their stability including redox-potential and (ii) low efficiency of electron transport in PSII, could explain the observed phototolerance (47). Analysis of the fluorescence measurements indicates that these mutants perform low quantum-yield electron flow in PSII (47). Therefore, it is possible that mutated D1 proteins influence redox signaling and in this way interfere with downregulation of the biogenesis of the photosynthetic apparatus. This may explain the retention of cell pigmentation in these mutants under high irradiance. These hypotheses in turn suggest the existence of an intrinsic protection mechanism in wild-type cells that copes with photoinhibition by down-regulating phycobilisome formation.
In phototolerant mutant species of D1, most amino acid conversions involved changes from nonpolar species to polar or charged ones, i.e. to Ser (32%), Asp (14%), Arg (7%), and Tyr (5%); the conversion of Phe to Ser occurred most often. This may indicate the increase in the hydrophilic or hydrogen-bonding interactions in the D1 protein is responsible for the enhanced phototolerance of the organism. Interestingly, Ser 3 Phe substitution in the lumenal loop of the D2 protein (48) appears to endow PSII with photosensitivity. Further biophysical analysis of cyanobacterial cells harboring the mutated species of D1 is needed to prove this proposal. Aliquots were taken at times indicated during pulsechase experiments, and thylakoid membranes were isolated as described under "Experimental Procedures." Thylakoid membranes containing 5 g of chlorophyll were loaded in each lane, electrophoresed in 15% polyacrylamide gels containing 6 M urea and 0.1% SDS (21), and subjected to autoradiography.