Mutational Analysis of Photosystem I Polypeptides ROLE OF PsaD AND THE LYSYL 106 RESIDUE IN THE REDUCTASE ACTIVITY OF PHOTOSYSTEM I

Abstract The ADC4 mutant of the cyanobacterium Synechocystis sp. PCC 6803 was studied to determine the structural and functional consequences of the absence of PsaD in photosystem I. Isolated ADC4 membranes were shown to be deficient in ferredoxin-mediated NADP reduction, even though charge separation between P700 and F/F occurred with high efficiency. Unlike the wild type, F became preferentially photoreduced when ADC4 membranes were illuminated at 15 K, and the EPR line shapes were relatively broad. Membrane fragments oriented in two dimensions on thin mylar films showed that the g tensor axes of F and F were identical in the ADC4 and wild type strains, implying that PsaC is oriented similarly on the reaction center. PsaC and the F/F iron-sulfur clusters are lost more readily from the ADC4 membranes after treatment with Triton X-100 or chaotropic agents, implying a stabilizing role for PsaD. The specific role of Lys of PsaD, which can be cross-linked to Glu of ferredoxin (Lelong et al.(1994) J. Biol. Chem. 269, 10034-10039), was probed by site-directed mutagenesis. Chemical cross-linking and protease treatment experiments did not reveal any drastic alterations in the conformation of the mutant PsaD proteins. The EPR spectra of F and F in membranes of the Lys mutants were similar to those of the wild type. Membranes of all Lys mutants showed wild type rates of flavodoxin reduction and flavodoxin-mediated NADP reduction, but had 10-54% decrease in the ferredoxin-mediated NADP reduction rates. This implies that Lys is a dispensable component of the docking site on the reducing side of photosystem I and an ionic interaction between Lys of PsaD and Glu of ferredoxin is not essential for electron transfer to ferredoxin. These results demonstrate that PsaD serves distinct roles in modulating the EPR spectral characteristics of F and F, in stabilizing PsaC on the reaction center, and in facilitating ferredoxin-mediated NADP photoreduction on the reducing side of photosystem I.

The ADC4 mutant of the cyanobacterium Synechocystis sp. PCC 6803 was studied to determine the structural and functional consequences of the absence of PsaD in photosystem I. Isolated ADC4 membranes were shown to be deficient in ferredoxin-mediated NADP ؉ reduction, even though charge separation between P700 and F A /F B occurred with high efficiency. Unlike the wild type, F B became preferentially photoreduced when ADC4 membranes were illuminated at 15 K, and the EPR line shapes were relatively broad. Membrane fragments oriented in two dimensions on thin mylar films showed that the g tensor axes of F A ؊ and F B ؊ were identical in the ADC4 and wild type strains, implying that PsaC is oriented similarly on the reaction center. PsaC and the F A /F B iron-sulfur clusters are lost more readily from the ADC4 membranes after treatment with Triton X-100 or chaotropic agents, implying a stabilizing role for PsaD. The specific role of Lys 106 of PsaD, which can be crosslinked to Glu 93 of ferredoxin (Lelong et al. (1994) J. Biol. Chem. 269, 10034 -10039), was probed by site-directed mutagenesis. Chemical cross-linking and protease treatment experiments did not reveal any drastic alterations in the conformation of the mutant PsaD proteins. The EPR spectra of F A and F B in membranes of the Lys 106 mutants were similar to those of the wild type. Membranes of all Lys 106 mutants showed wild type rates of flavodoxin reduction and flavodoxin-mediated NADP ؉ reduction, but had 10 -54% decrease in the ferredoxinmediated NADP ؉ reduction rates. This implies that Lys 106 is a dispensable component of the docking site on the reducing side of photosystem I and an ionic interaction between Lys 106 of PsaD and Glu 93 of ferredoxin is not essential for electron transfer to ferredoxin. These results demonstrate that PsaD serves distinct roles in modulating the EPR spectral characteristics of F A and F B , in stabilizing PsaC on the reaction center, and in facilitating ferredoxin-mediated NADP ؉ photoreduction on the reducing side of photosystem I.
Photosystem I (PS I) 1 in cyanobacteria and chloroplasts is a multisubunit membrane-protein complex that catalyzes electron transfer from reduced plastocyanin (or cytochrome c 6 ) to oxidized ferredoxin (or flavodoxin) (see Refs. 1 and 2 for recent reviews). The PsaA and PsaB subunits form a heterodimeric core that harbors ϳ100 antenna Chl a molecules, the primary electron donor, P700, and a chain of electron acceptors A 0 , A 1 , and F X . PsaC, PsaD, and PsaE are peripheral subunits that are located on the n-side (stromal in chloroplasts and cytoplasmic in cyanobacteria) of the photosynthetic membrane. PsaC binds the terminal electron acceptors, F A and F B , each a [4Fe-4S] cluster. PsaE may be involved in ferredoxin reduction (3)(4)(5)(6) and cyclic electron flow around PS I (7). The nature of interactions between PsaE and ferredoxin are not well understood. The remaining proteins of cyanobacterial PS I are integral membrane proteins. Among them, PsaL is required for the formation of PS I trimers (8,9). PsaF can be chemically crosslinked in vitro to plastocyanin (10,11). PsaJ and PsaI may be involved in proper assembly of PsaF and PsaL, respectively (12,13). PsaM has been implicated in cyclic electron flow around PS I. 2 Role of PsaK has not been identified.
The extrinsic protein, PsaD, has two reported functions in the PS I complex of cyanobacteria, algae and higher plants. The first function, deduced from in vitro reconstitution experiments, is to stabilize PsaC on the PS I reaction center (14). When PsaC is rebound to a P700-F X core in the absence of PsaD, F B rather than F A becomes photoreduced at 15 K, and the EPR line widths of F A Ϫ and F B Ϫ are significantly broader than the wild type. These experiments did not investigate the ability of PsaE to bind independently of PsaD. The second function, inferred from cross-linking studies, is to serve as a "docking" protein to facilitate interaction of soluble ferredoxin with the PS I complex (15)(16)(17). Recent cross-linking experiments have shown that Lys 106 of PsaD from Synechocystis sp. PCC 6803 can be cross-linked to Glu 93 in ferredoxin (18). Therefore these two residues come in physical proximity with each other during at least one stage of electron transfer from PS I to ferredoxin. These results indicate a ferredoxin-docking function of PsaD, but do not illustrate a functional requirement of PsaD for NADP ϩ photoreduction. The latter issue was addressed with membranes of a PsaD-less cyanobacterial mutant, where it was shown that ferredoxin-mediated NADP ϩ photoreduction was severely inhibited (3). This defect may be due to either the lack of ferredoxin-docking site in the PsaD-less membranes or the absence of PsaD may cause alterations in PsaC and its F A /F B redox centers. In this paper, we further charac-terized ADC4, a PsaD-less cyanobacterial mutant strain, to investigate proposed roles of PsaD in ferredoxin-mediated NADP ϩ photoreduction and in stabilizing PsaC on the reaction center. We also used site-directed mutagenesis to alter Lys 106 in PsaD to Arg, Glu, Asp, Asn, Ser, and Gly and examined functional significance of the physical proximity between Lys 106 of PsaD and Glu 93 of ferredoxin during electron transfer from PS I.

EXPERIMENTAL PROCEDURES
Generation of PsaD Mutants-The ADC4 strain of Synechocystis sp. PCC 6803 (3), generated by replacing psaD with a cassette for chloramphenicol resistance (Fig. 1), was used as the PsaD-less strain. To introduce the wild type and mutant psaD into the ADC4 strain, the pKD1 recombinant plasmid was generated by cloning the wild type psaD gene along with 370-base pair 5Ј-flanking region (EcoRI-PstI fragment), a kanamycin resistance cartridge (PstI fragment), and 1040base pair 3Ј-flanking region (PstI-HindIII fragment) into the polylinker of pGem3z (Fig. 1). Appropriate restriction endonuclease recognition sites were introduced by incorporating them in the oligonucleotide primers that were used for DNA amplification. The kanamycin resistance cartridge was isolated by digesting the plasmid pUC4K with PstI (19). ADC4 cells were transformed with pKD1, kanamycin-resistant transformants were selected, segregated for two generations, and replica-plated to confirm the absence of chloramphenicol resistance gene. The resulting strain KD1 is referred as K106K in this paper and contains an active, wild type psaD gene, as demonstrated by PCR analysis of genomic DNA (data not shown) and Western blotting of thylakoid proteins using an anti-PsaD antibody (Fig. 1B).
The codon for Lys 106 in PsaD was changed using PCR-mediated mutagenesis (20). The sequences of mutagenic oligonucleotides are given in Table I. The DNA fragments containing mutations were cloned into pKD1 by replacing its EcoRI-SacI fragment. The amplified fragment was completely sequenced to confirm the presence of desired mutation and to ensure the fidelity of Taq polymerase. The ADC4 strain was transformed with the recombinant DNAs containing mutations, and mutant strains were obtained according to the procedure described earlier for the KD1 strain. The presence of mutant genes in the resulting strains was confirmed by amplification and sequencing of psaD.
Isolation of Photosynthetic Membranes and PS I Complexes-Previously published methods were used for culture of Synechocystis sp. PCC 6803 (3), for isolation of thylakoid membrane (8), for isolation of PS I trimers using dodecyl-␤-D-maltoside (8), and for purification of photochemically active PS I complexes using Triton X-100 (21).
Oriented Thylakoid Membranes-Membrane fragments were oriented in two dimensions on mylar film using the method described in Ref. 36. The membranes were washed five times with 10 mM Tris, pH 8.3, and suspended with 5 mM Tris, pH 8.3, at a Chl concentration of 2.5 mg/ml. This suspension was spread on mylar strips which had been prewashed with detergent and rinsed with deionized, distilled water. The membranes were slowly dried for 3-4 days in darkness at 4°C in an inert atmosphere maintained at 90% relative humidity.
Removal of PsaC from Synechocystis sp. PCC 6803 Thylakoid Membranes-Thylakoid membranes (25 g of Chl in a total of 200 l volume) of wild type and ADC4 strains of Synechocystis sp. PCC 6803 were incubated with 0.05% Triton X-100, 2 M NaCl, 2 M NaBr, and different concentrations of NaI for 15 min at ice temperature. The thylakoids were diluted to 1.2 ml with STN (0.4 M sucrose, 100 mM Tris, pH 8.0, 10 mM NaCl) buffer and centrifuged at 20,000 ϫ g for 30 min. Each pellet was suspended in 500 l of STN buffer, centrifuged again, and used for Western blotting.
Treatment of PS I Complexes with Thermolysin, Glutaraldehyde, or NHS-biotin-Purified wild type PS I complexes at a concentration of 100 g of Chl/ml were treated with 10 mM glutaraldehyde (Sigma) in the presence of 10 mM Mops-NaOH, pH 7.0, 0.05% Triton X-100 for 30 min on ice. The cross-linking reactions were quenched with 100 mM glycine. To probe topography of the wild type or mutant PsaD, PS I complexes (100 g Chl/mL) were incubated with 50 g of thermolysin (Sigma)/mg of Chl and with 5 mM CaCl 2 at 37°C for 30 min. The reactions were terminated with 20 mM EDTA. For NHS-biotinylation, PS I complexes at 100 g of Chl/ml were incubated with 50 M NHSbiotin (Sigma), 10 mM Mops, pH 7.0, 0.05% Triton X-100, 0.05% Me 2 SO for 30 min at 25°C. The reaction was quenched with 50 mM ammonium bicarbonate, pH 7.8.
Analytical Gel Electrophoresis and Immunodetection-Isolated PS I complexes and thylakoids were solubilized with 1% SDS and 0.1% ␤-mercaptoethanol at 25°C for 1 h. Proteins were fractionated by Tricine-urea-SDS-PAGE (3), and proteins were electrotransferred to Immobilon-P membranes. Immunodetection was performed using enhanced chemiluminescence (Amersham Corp.). In some experiments, the relative amounts of these subunits in the membranes were estimated by laser densitometry. The PsaB antibodies were obtained from James Guikema, Kansas State University. The PsaD and PsaC antibodies were generated at the University of Nebraska Polyclonal Antibody Core Facility using recombinant overexpressed proteins of Synechococcus sp. PCC 7002. The PsaL, PsaF, and PsaE antibodies were raised against respective proteins from Synechocystis sp. PCC 6803 (12,22). In biotinylation experiments, Immobilon-P membranes were incubated with an avidin-peroxidase conjugate, and labeled proteins were detected with hydrogen peroxide and 4-chloro-1-naphthol (23).
Rebinding of Recombinant Peripheral PS I Proteins to Thylakoid Membranes or to the P700-F X Core-Nostoc sp. PCC 8009 PsaD and Synechococcus sp. PCC 7002 PsaE were prepared as described previously (14,24). For overexpressing PsaD of Synechocystis sp. PCC 6803, the gene was amplified using primers that added NdeI site at the 5Ј end of the coding region and EcoRI site at the end of the coding region. The amplified fragment was digested simultaneously with NdeI and EcoRI and cloned into pET21a that had been digested with the same enzymes. The plasmid containing psaD was sequenced to ensure fidelity of Taq polymerase and introduced into BL21(DE3) strain of Escherichia coli. Protein overexpression, isolation of inclusion bodies, and PsaD purification were performed by the same methods that have been used for PsaD from Nostoc sp. PCC 8009 (14). PsaD was rebound to the ADC4 thylakoid membranes by adding concentrated PsaD at a molar ratio of 20:1 with P700 and incubating 60 min at 4°C. The PsaD-less thylakoids and the PsaD-reconstituted thylakoids were separately incubated with 0.05% Triton X-100 for 5 min at 20°C and pelleted. The supernatant was concentrated by ultrafiltration and used to resuspend the pellet prior to EPR analysis.
The P700-F A /F B complex and the P700-F X core were prepared as described previously (26,27). Reconstitution of the F A /F B iron sulfur clusters and rebinding of PsaC, PsaD, and PsaE to the PS I core were performed at ratios of 15 PsaC:20 PsaE:1 P700-F X core and 7 PsaC:5 PsaD:10 PsaE:1 P700-F X core according to Ref. 14. Dissolved oxygen in the cuvette, as measured in an oxygen electrode, decreased to near-zero levels shortly after addition of ␤-mercaptoethanol.
Biochemical and Spectroscopic Assays-Chl a was determined in 80% acetone (25). Protein concentrations were determined using a dye binding method (26) after applying correction factors (0.70 for PsaC, 0.54 for PsaD, and 0.50 for PsaE) determined by quantitative amino acid analysis (14). EPR studies were performed with a Bruker ECS-106 X-band spectrometer. All samples contained 1 mM ascorbate and 0.03 mM DCPIP.
Rates of flavodoxin photoreduction by photosynthetic membranes were measured by monitoring the change in the absorption of flavodoxin at 467 nm under saturating actinic light (3,27). Rates of flavodoxin-or ferredoxin-mediated NADP ϩ photoreduction were measured using ferredoxin:NADP ϩ oxidoreductase (Sigma) and by monitoring the change in the absorption of NADPH at 340 nm (3,27). Spinach ferredoxin was purchased from Sigma. A strain of E. coli with a plasmid containing the flavodoxin gene of Synechococcus sp. PCC 7002 was provided by Dr. D. Bryant (Pennsylvania State University). The expressed flavodoxin was purified by DEAE-Sepharose and gel filtration chromatography.

RESULTS
Targeted mutagenesis was used to generate the ADC4 mutant strain of Synechocystis sp. PCC 6803. This strain lacks PsaD in its membranes, but contains all other PS I subunits, including PsaE and PsaC (3). The wild type thylakoids reduce NADP ϩ at the rate of 658 mol⅐h Ϫ1 ⅐mg Chl Ϫ1 when ferredoxin is used as the electron acceptor (Table II). The ADC4 thylakoids are severely deficient in reducing NADP ϩ using ferredoxin as electron acceptor, showing less than 10% of wild type rates. When overexpressed Synechocystis sp. PCC 6803 PsaD was reconstituted with the PsaD-less thylakoids, 72% the ferredoxin-mediated NADP ϩ reductase activity was restored. These results confirm our previous observation that PsaD is essential for electron transfer using ferredoxin as mediator (3). In vitro reconstitution experiments have shown that PsaD influences the binding of PsaC to the reaction center and the EPR spectral characteristics of the F A /F B clusters (14). The defect in electron transfer to ferredoxin may result from the changes in PsaC and its redox centers that are caused by the absence of PsaD. We therefore examined PsaC in the PsaD-less mutant strain using biochemical and spectroscopic techniques.
Removal of PsaC from the ADC4 Membranes-Immunoblot analyses indicated that the relative amounts of PsaC in untreated membranes from the wild type and ADC4 strains were equivalent (Fig. 2). PsaC was not removed from either the wild type or ADC4 membranes by 2 M NaCl, but 2 M NaBr removed a small amount of PsaC from the ADC4 membranes. Depending on the presence or absence of PsaD, the stronger chaotrope NaI showed a pronounced differential effect on the removal of PsaC. Hence, PsaC in the ADC4 membranes was more suscep-tible to removal by chaotropic agents than in the wild type membranes.
When wild type membranes were incubated with 0.05% Triton X-100, PsaC was not significantly extracted, but when the ADC4 membranes were incubated similarly for 10 min, about 50% of the PsaC protein was removed (Fig. 2). EPR studies of a similarly treated sample illuminated during freezing showed no removal of F A or F B from wild type, but the loss of more than 50% of the F A and F B acceptors from the ADC4 membranes (data not shown). The concurrent loss of the F A and F B clusters is consistent with the detergent-mediated removal of the PsaC protein rather than the in situ destruction of the iron-sulfur clusters. When the ADC4 membranes were incubated with Nostoc sp. PCC 8009 PsaD and subsequently treated with 0.05% Triton X-100, there was no loss of F A and F B . There was no change in the appearance of the EPR spectrum when the concentration of Triton X-100 was increased to 1% (data not shown).
EPR Studies of the F A and F B Clusters in the PsaD-less Membranes-F A , characterized by resonances at g ϭ 2.046, 1.943, and 1.855, is the iron-sulfur center primarily photoreduced when wild type thylakoids are frozen in darkness and illuminated at 15 K (28). Under these conditions, usually less than 20% of F B , characterized by resonances at g ϭ 2.066, 1.929, and 1.878, is also photoreduced. As shown in Fig. 3A, however, F B is the iron-sulfur center primarily photoreduced when membranes from the ADC4 mutant are frozen in darkness and illuminated at 15 K. There is also a small amount of F A photoreduced, as shown by the low-field resonance at g ϭ 2.044 and the mid-field resonance at g ϭ 1.949 (the high-field resonance of F A is the shoulder on the broad, asymmetrical resonance centered at g ϭ 1.878). When the ADC4 thylakoids are frozen during illumination, both F A and F B are photoreduced, but the line widths of the resonances are broader than those in wild type thylakoids (Fig. 3C). The low-field and highfield resonances of F A and F B are merged in the wild type thylakoids at g ϭ 2.046 and 1.885, but in the ADC4 mutant there is a distinct low-field shoulder at g ϭ 2.062 due to F B and a high-field shoulder at g ϭ 1.872 due to F A . The midfield resonances are similar to those of the wild type, but the line widths are not as narrow or as clearly defined.
When recombinant Nostoc sp. PCC 8009 PsaD is added to the ADC4 thylakoids, resonances characteristic of F A at g ϭ 2.045, 1.941, and 1.857 predominate when the sample is frozen in darkness and illuminated at 15 K (Fig. 3B). There is also a small amount of F B photoreduced, as shown by the resonances at g ϭ 2.066, 1.930, and 1.882. The line widths and g values of the F A and F B resonances, and the pattern of photoreduction at   (Fig. 3D).

Studies of F A and F B in PS I Complexes Reconstituted with PsaC and PsaE-
The presence of PsaE as well as PsaC in the ADC4 thylakoids implies that PsaE cannot substitute for PsaD in stabilizing PsaC on the PS I core. To test this implication in an in vitro reconstitution system, we rebound Synechococcus sp. PCC 7002 PsaC to a P700-F X core in the presence of Synechococcus sp. PCC 7002 PsaE. When the reconstituted PS I complex is frozen in darkness and illuminated at 15 K, F B is still preferentially reduced, as shown by the characteristic resonances at g ϭ 2.073, 1.930, and 1.875 (Fig. 4A). There is also a small amount of F A photoreduced, as shown by the low-field resonance at g ϭ 2.045 and the mid-field resonance at g ϭ 1.952 (the high-field resonance of F A is most likely the high-field shoulder on the broad, asymmetrical resonance centered at g ϭ 1.875). When the reconstituted PS I complex is frozen during illumination, both F A and F B are reduced, but the broad line widths of the F A and F B resonances are even more apparent (Fig. 4C). Unlike wild type PS I, the complex reconstituted without PsaD shows two noninteracting low-field resonances at g ϭ 2.067 (F B ) and g ϭ 2.046 (F A ). The slope of the mid-field "derivative" resonance at g ϭ 1.935 is less sharp than for either F A or F B in the wild type membranes and probably results from a merging of the two individual resonances. The high-field resonance at g ϭ 1.883 is broad and is probably an envelope of the two noninteracting resonances derived from F A and F B . These results are similar to those of reconstitution with PsaC alone (not shown; see Ref. 14), and confirm that PsaE cannot substitute for PsaD in stabilizing PsaC on the PS I reaction center core.
When Nostoc sp. PCC 8009 PsaD is added to the PsaC and PsaE-reconstituted PS I complex, resonances characteristic of F A appear at g ϭ 2.046, 1.943, and 1.857 after illumination at 15 K (Fig. 4B). A small amount of F B is reduced, as shown by the resonances at g ϭ 2.070, 1.931, and 1.884. In the presence of PsaD, the ratio of F A to F B reduced by illumination at 15 K is about 4:1. The line widths and g values of the F A and F B resonances, as well as the pattern of photoreduction at 15 K, are very similar to those observed in wild type Synechocystis sp. PCC 6803 complexes (Fig. 4B). When the reconstituted sample is illuminated during freezing, the F A and F B clusters undergo magnetic interaction, resulting in a set of resonances at g ϭ 2.047, 1.939, 1.918, and 1.884 and thus appear similar to wild type complexes isolated from Synechocystis sp. PCC 6803 thylakoids (Fig. 4D).
Oriented Thylakoid Membranes-The three-dimensional structure of PsaC is modeled on Peptococcus aerogenes ferre-FIG. 2. Relative PsaC contents of the wild type and PsaD-less thylakoids after treatment with Triton X-100 or chaotropic agents. Thylakoid membranes from the wild type (WT) and PsaD-less mutant (ADC4) were treated with detergent and different chaotropes as described under "Experimental Procedures." The relative amounts of PsaC were determined by immunoblotting and laser densitometry of x-ray films exposed to different intensities and normalized from the levels of PsaB. Densitometric values of PsaC in the wild type membranes without any treatment (control) were considered as 100%. Results from three independent experiments are averaged. Bars indicate the standard deviation. Representative immunoblots are shown in the insets. doxin, where the crystal structure shows an approximate 2-fold axis of symmetry which runs through the two iron-sulfur clusters. One explanation for the preferential photoreduction of F B is that PsaC may be bound differently in the absence of PsaD, leading to a reorientation of the F A and F B clusters relative to F X (it is not certain which cluster is proximal to F X in wild type membranes). The role of PsaD may be to lock the PsaC into a fixed conformation, orienting the F A and F B clusters in a manner that makes the photoreduction of F A more likely than that of F B . This hypothesis was tested by comparing directions of the g tensor axes of F A and F B in wild type and in ADC4 membranes after orienting membrane fragments on thin mylar films. The EPR signal intensities of F A Ϫ and F B Ϫ were measured as a function of angle in the static B 0 magnetic field (0°o rientation is for the magnetic field parallel to the normal to the film surface). If PsaC is misoriented, the g tensor axes of F A and F B will not match those of the wild type unless the two clusters are reoriented precisely 180 degrees about the axis of symmetry. The results are summarized in Table III. The g z intensity of F A Ϫ in ADC4 is minimum at 0°, grows to a maximum at 64°and begin to fall for the field magnetic angle greater than 64°. In the same manner, g y (1.94) of F A Ϫ has maximum intensity at 50°and then falls away in both directions. It is difficult to determine the angle of g x experimentally; that angle was calculated to be 51°from the equation, cos 2 w x ϩ cos 2 w y ϩ cos 2 w z ϭ 1, where w i is the angle at which a particular tensor has the maximum intensity in the EPR signal. As for F B , the angles that have maximal intensity in g z and g y are 29 and 69°, respectively. Within experimental error, the orientation of principal g tensor axes of F A Ϫ and F B Ϫ in ADC4 is therefore identical with wild type. It is unlikely that the polypeptide can be reoriented precisely 180 degrees about the symmetry axis simply because the binding site for the PS I core is likely to be localized on a unique region of the polypeptide backbone. Hence, this result would not be not consistent with a function of PsaD in reorienting PsaC on the reaction center core.
Generation of Lys 106 Mutations in PsaD-The absence of PsaD in the ADC4 thylakoids impairs their ability to reduce NADP ϩ using ferredoxin. It also affects association of PsaC with the PS I core and the EPR spectra of F A and F B clusters. These findings raise the issue that the role of PsaD in ferredoxin-mediated electron transfer is an indirect consequence of its function in stabilizing the association of PsaC with the reaction center. To probe this possibility, we generated sitedirected mutants in a residue that has been proposed to interact with ferredoxin: the lysyl 106 residue of PsaD. A PCR-based method was used to replace AAA nucleotide sequence for Lys 106 in psaD with various nucleotide combinations (Table IV). The resulting mutations included a conservative replacement (K106R), two potentially disruptive replacements (K106E, K106D), two replacements to polar amino acids (K106S, K106N), and a replacement to glycine, the amino acid without a side chain (K106G).
All psaD mutant strains could grow under photoautotrophic conditions, with doubling times comparable with K106K strain. In contrast, the ADC4 strain grew significantly slower than the wild type and K106K strains. We examined the accumulation of mutant PsaD proteins in the membranes using Western analysis (Fig. 5). On an equal Chl basis, all strains contained similar levels of PsaD in the membranes. Within the PS I complex, PsaD interacts with PsaA-PsaB, PsaE, PsaC, and PsaL. When the accumulation of these PS I subunits was examined by Western analysis, membranes of the mutant and K106K strains contained similar levels of PsaB, PsaC, PsaE, and PsaL (Fig. 5). Therefore Lys 106 mutations in PsaD do not affect accumulation of PsaD and PsaD-interacting subunits in the membrane.
Topography and Subunit Interactions of the Lys 106 Mutant PsaD Proteins-Effects of mutations on the topography and  subunit interactions of PsaD were investigated using biochemical techniques. When the wild type PS I complexes were treated with glutaraldehyde, the PsaD antibody recognized three protein species corresponding to native PsaD, a 29-kDa species minimally containing PsaD and PsaL and a 25-kDa species containing the PsaD-PsaC and PsaD-PsaE cross-linked products. Glutaraldehyde treatment of PS I complexes from all mutant strains also resulted in 25-and 29-kDa cross-linked products. Therefore, mutations in Lys 106 do not affect interaction of PsaD with PsaL and PsaC/PsaE. The PsaD-PsaL interaction may be crucial in the formation of PS I trimers; the absence of PsaD affects the ability of PS I to form trimers (8).
Upon dodecyl-␤-D-maltoside treatment and sucrose gradient ultracentrifugation, approximately 7% Chl in ADC4 thylakoids, but 70% Chl in wild type thylakoids, could be obtained in PS I trimers. None of the Lys 106 mutations affected the ability of PS I to form trimers (Table IV), again indicating that PsaD-PsaL interactions remain unaffected in these mutants.
To investigate the surface-exposed domains in PsaD, purified PS I complexes from different strains were treated with thermolysin (Fig. 6). Previously, we have reported that incubation of wild type PS I complexes with thermolysin resulted in accumulation of distinct fragments from proteolysis at sites in the C-terminal ϳ3 kDa of PsaD. Proteolytic patterns of Lys 106 mutant proteins could be classified into three types. First, K106R and K106D proteins showed fragments that were similar in size and abundance as the wild type. Second, level of K106G protein declined more rapidly during proteolysis than in the wild type PsaD. However, the number and sizes of the fragments that were immunoreactive with anti-PsaD antibody were remarkably similar to those in wild type, suggesting that these mutations did not expose new proteolytic sites on the major fragments of PsaD. Third, K106E, K106N, K106S, and K106G mutant proteins degraded to one additional fragment that was 0.9 kDa smaller than the smallest fragment generated from wild type PsaD. Therefore, these mutations at best altered only slightly the conformation of the surface-exposed domain of PsaD.
Modification with NHS-biotin has been used to investigate exposure of protein surfaces to the aqueous phase. NHS-biotin reacts with the N terminus and the ⑀-amino group of lysyl residues (29). We have shown previously that one or more of four lysyl residues in the C-terminal 6-kDa region of PsaD are present on the aqueous surface of PS I (34). To examine if lysyl residues other than Lys 106 are exposed on the PS I surface, the PS I complexes from the K106K and K106G strains were biotinylated and protein modification was detected using avidin peroxidase and a chromogenic substrate for peroxidase. In both complexes, PsaA-PsaB, PsaF, PsaL, and PsaE showed significant amounts of biotin incorporation (Fig. 7). PsaD was biotinylated in PS I complexes from K106K and to a lesser extent in K106G PS I complexes. Therefore, besides Lys 106 , additional lysyl residues in PsaD are exposed on the surface and may be involved in ionic interactions observed between PS I and ferredoxin.
Function of PS I Containing Mutant PsaD Proteins-The consequence of mutations at Lys 106 on the photoreduction of the F A /F B iron-sulfur clusters were investigated by EPR spectroscopy. The spectra of F A and F B in membranes isolated from the K106D and K106E strains are shown in Fig. 8. When thylakoids of two mutants are frozen in darkness and illumi-  K106K  AAA  100  940  525  490  ADC4  10  440  225  65  K106R  AGA  98  960  450  325  K106E  GAG  98  1020  450  225  K106D  GAU  98  1060  450  280  K106N  AAC  100  1060  500  335  K106S  AGC  103  980  500  350  K106G  GGC  96  1140  525  440 FIG. 5. Accumulation of PS I subunits in the membranes of Lys 106 mutants cells. Photosynthetic membranes were isolated from different cyanobacterial strains, proteins were separated by Tricineurea-SDS-PAGE, and transferred to Immobilon membrane. Immunodetection was performed using horseradish peroxidase-conjugated secondary antibody and enhanced chemiluminescence reagents.

FIG. 6. Thermolysin accessibility and nearest neighbors of PsaD in mutant and wild type PS I complexes.
The PS I complexes were purified from different cyanobacterial strains and incubated with thermolysin at 50 g of protease/mg of Chl for 30 min at 37°C. For chemical cross-linking of PS I subunits, PS I complexes were incubated with 10 mM glutaraldehyde on ice. The reactions were terminated by addition of EDTA (20 mM) and glycine (100 mM). The control (lanes C), thermolysin-treated (lanes T), and glutaraldehyde-treated (lanes G) samples (5 g of Chl/lane) were analyzed using Tricine-urea-SDS-PAGE and Western blotting. Immunodetection was performed using anti-PsaD antibody. nated (Fig. 8, A and C), only one electron is transferred from P700 to the terminal iron-sulfur cluster. As indicated before, about 80% of F A becomes photoreduced with g values of 2.046, 1.943, and 1.855, and 20% of F B becomes photoreduced with g values of 2.068, 1.928, and 1.880 in wild type PS I complexes. When the thylakoids of the K106D and K106E strains are illuminated during freezing (Fig. 8, B and D), more than one electron is transferred, resulting in a quantitative reduction of both F A and F B . The g values of 2.049, 1.940, 1.922, and 1.885 are characteristic of an interaction spectrum of F A and F B and appear identical to the wild type. A similar analysis of the remaining mutants showed identical behavior (data not shown). In summary, the preference for F A reduction over F B and the normal magnetic interaction between the two clusters indicate that mutations at Lys106 do not affect the magnetic properties of the iron-sulfur clusters in PsaC.
Thylakoids from the Lys 106 mutant strains were used to estimate the reductase activity of PS I in three different assays: direct measurement of the rates of flavodoxin reduction, NADP ϩ reduction mediated by flavodoxin, and NADP ϩ reduction mediated by ferredoxin (Table IV). The K106K membranes reduced flavodoxin or NADP ϩ mediated by flavodoxin at rates of 940 and 525 mol/mg of Chl/h, respectively. PsaD-less membranes were able to reduce flavodoxin and NADP ϩ via flavodoxin at rates 40 -50% that of wild type membranes. All Lys 106 mutations showed normal electron transfer rates to flavodoxin. This suggests that although PsaD may enhance the interaction of flavodoxin with PS I, it is not an essential component when flavodoxin functions as an electron acceptor of PS I. These results also show that Lys 106 is not involved in flavodoxin-PS I interactions. The K106K membranes reduced NADP ϩ at the rate of 490 mol/mg/Chl/h when ferredoxin was used as a mediator of electron transfer to ferredoxin-NADP ϩ reductase. The PsaD-less membranes showed a drastic decrease in their ability to reduce NADP ϩ via ferredoxin. The replacement of Lys 106 in K106R, K106G, K106S, and K106N mutants reduced the rates of electron transfer via ferredoxin by 10 -34%. Disruptive replacement of Lys 106 by aspartate or glutamate led to 43 and 54% reduction in the ferredoxin-mediated NADP ϩ photoreduction rates, respectively, compared with the K106K membranes. These results imply that Lys 106 is important, but not essential, for the interaction between PS I and ferredoxin.

DISCUSSION
The function of PsaD has been studied previously using chemical cross-linking, biochemical resolution, and reconstitution and targeted mutagenesis of the cyanobacterium Synechocystis sp. PCC 6803. These studies have indicated that PsaD has multiple roles: it can be cross-linked to ferredoxin, its absence in the ADC4 cyanobacterial mutant results in loss of NADP ϩ photoreduction, and its absence affects the EPR spectral properties of the F A /F B clusters of PsaC. In the present study, we investigated further the different roles of PsaD and examined whether they were correlated using a mutant of Synechocystis sp. PCC 6803. The PsaD-less mutant organism can synthesize and assemble a PS I reaction center containing PsaC and PsaE (3). This result agrees with in vitro reconstitution experiments, which showed that PsaC was able to bind to a P700-F X core in the absence of added PsaD (14). We found that low concentrations of Triton X-100 led to release of PsaC FIG. 7. Mutant PsaD can be labeled with NHS-biotin. The biotinylated PS I from KD1 (K106K) and K106G equivalent to 10 g of Chl were solubilized, and the proteins were separated by Tricine-urea-SDS-PAGE, electroblotted on Immobilon-P membranes, and stained with Coomassie Blue. A replica of the blot was probed with avidin-peroxidase conjugate, and the immunoreaction was visualized using 4-chloro-1naphthol and hydrogen peroxide.
FIG. 8. EPR spectra of thylakoids from the Lys 106 mutant strains, K106D and K106E. The spectra in A and C were taken after freezing in darkness and illumination at 15 K. The spectra in B and D were taken after illuminating the sample during freezing to 15 K. The resonances were resolved by subtracting the light-off (before light-on) from the light-on spectrum. Samples were suspended in STN, containing 1 mM sodium ascorbate and 30 M DCPIP at 1 mg ml Ϫ1 Chl. Spectrometer conditions: temperature, 15 K; microwave power, 20 milliwatts; microwave frequency, 9.647 GHz; receiver gain, 2.0 ϫ 10 4 ; modulation amplitude, 10 G at 100 kHz. from membranes isolated from the PsaD-less strain of Synechocystis sp. PCC 6803. This result is also consistent with those from in vitro reconstitution experiments, where it was shown that Triton X-100 removed a large fraction of the added recombinant PsaC from a P700-F X core (14). Addition of PsaD to the ADC4 thylakoids (this work) and to the PsaC-reconstituted P700-F X core (14) led equally to an enhanced resistance of PsaC to Triton X-100 extraction.
Low temperature EPR studies of ADC4 membranes showed that F B was predominantly reduced at 15 K, and the line widths of the reduced F A and F B clusters were broader than to the control. The addition of PsaD to the ADC4 membranes resulted in the photoreduction of F A when the sample was illuminated 15 K and in a sharpening of the EPR line widths. This is also in close agreement with the in vitro reconstitution experiments (14) and shows that Triton X-100 is not responsible for the altered orientation of added PsaC. The presence of PsaE on the ADC4 membranes indicates that this protein does not require PsaD to assemble on the PS I core. PsaE also does not influence the stability of PsaC or the pattern of photoreduction of F A or F B , a result which agrees with in vitro reconstitution experiments performed on a P700-F X core in the presence of added recombinant PsaE (this study).
The similarity of PsaC to P. aerogenes ferredoxin (30) has led to the prediction that the former contains a 2-fold symmetry axis related to the two iron-sulfur binding sites (31, 32). The altered pattern of photoreduction of F A and F B in the ADC4 membranes could be explained if one function of PsaD were to orient PsaC on the reaction center core. In the PsaD-less membranes, PsaC would be misoriented; however, EPR studies on the membrane fragments oriented on the thin mylar films have shown that the orientation of principal g tensors of F A and F B in ADC4 membranes are very similar with those of wild type. This makes it unlikely that PsaC is oriented differently in ADC4; the explanation of the altered pattern of photoreduction of F A and F B must hence be sought elsewhere.
The interprotein electron transfer on the reducing side of PS I has been studied using spectroscopy, chemical cross-linking, and subunit-deficient cyanobacterial mutants. At least three proteins (PsaC, PsaD, and PsaE) in PS I must interact with ferredoxin for efficient electron transfer to occur (3,4). Recently, kinetics of reduction of soluble cyanobacterial ferredoxin by cyanobacterial PS I were investigated by flash absorption spectroscopy (33). This study revealed the existence of three different first order components with t1 ⁄2 of ϳ500 ns, 20 s, and 100 s. The 500-ns phase corresponds to electron transfer from F A /F B to ferredoxin. These analyses also pointed to the presence of at least two types of PS I-ferredoxin complexes, all competent in electron transfer. Ferredoxin accepts electrons from F A /F B of PsaC, implying that these proteins should be in intimate contact with each other, yet physical association between PsaC and ferredoxin has not been demonstrated. The major obstacles in the association of PsaC and ferredoxin are their unfavorable electrostatic interactions; both PsaC and ferredoxin have strongly electronegative surfaces at the physiological pH. Docking proteins may be required to facilitate the interaction by providing clusters of amino acids with opposite charges. PsaD and PsaE may fulfill this role, since they are required for the interaction and reduction of ferredoxin (3,4).
Current evidence suggests that the interactions between ferredoxin and PS I are electrostatic in nature. Spectroscopic study of ferredoxin reduction suggests that complex formation precedes electron transfer and the rate constants for complex formation depend on ion, especially magnesium, concentrations (33,36). Lys 106 of PsaD and Glu 93 of ferredoxin can be cross-linked and thus provide one possible pair for ionic interaction. Changing Lys 106 to uncharged amino acids, however, does not hinder ferredoxin-mediated electron transfer. Clearly, an ionic interaction between Lys 106 of PsaD and Glu 93 of ferredoxin is not essential for docking of ferredoxin on the reaction center. Our results show that one and more of the three lysyl residues (at positions 117, 131, and/or 135) may also be exposed on the reducing side of PS I. These residues along with three arginyl residues in the C-terminal domain of PsaD could also provide ionic interactions during docking of ferredoxin on PsaD. Alternatively, Lys 106 may participate in ionic interactions between the wild type PS I and ferredoxin, but its function can be replaced by other surface-exposed lysines. Since Lys 106 of PsaD can be cross-linked to Glu 93 of ferredoxin (18), these residues come in vicinity of each other. Close proximity between these residues may facilitate docking, as the site-directed replacement of Lys 106 with negatively charged amino acids drastically reduces rates of electron transfer to ferredoxin. Interestingly, a nonconservative mutation in Anabaena ferredoxin at Glu 94 , equivalent of Glu 93 in ferredoxin from Synechocystis sp. PCC 6803, does not affect its interaction with spinach PS I (37).
The absence of PsaD has a profound effect on PsaC and its redox centers, which may be partly responsible for the inability of PsaD-less membranes to carry out ferredoxin-mediated NADP ϩ reduction. However, phenotypes of Lys 106 mutants show that PsaD has a distinct docking function that may be its primary role in ferredoxin-mediated photoreduction. K106D and K106E mutants contain normal amounts of PsaC and their F A /F B clusters show EPR spectral characteristics similar to similar to the wild type. These membranes also exhibit wild type rates of flavodoxin reduction, again indicating normal electron transfer within the PS I complex. Yet, membranes of these strains are unable to support wild type rates of NADP ϩ photoreduction using ferredoxin. Therefore, wild type PsaC and its redox centers are not sufficient for obtaining wild type rates of NADP ϩ photoreduction. These results demonstrate the requirement of PsaD in ferredoxin-mediated electron transfer and provide a strong functional evidence for a ferredoxin-docking role of PsaD.