Strains of synechocystis sp. PCC 6803 with altered PsaC. I. Mutations incorporated in the cysteine ligands of the two [4Fe-4S] clusters FA and FB of photosystem I.

Two [4Fe-4S] clusters, FA and FB, function as terminal electron carriers in Photosystem I (PS I), a thylakoid membrane-bound protein-pigment complex. To probe the function of these two clusters in photosynthetic electron transport, site-directed mutants were created in the transformable cyanobacterium Synechocystis sp. PCC 6803. Cysteine ligands in positions 14 or 51 to FB and FA, respectively, were replaced with aspartate, serine, or alanine, and the effect on the genetic, physiological, and biochemical characteristics of PS I complexes from the mutant strains were studied. All mutant strains were unable to grow photoautotrophically, and compared with wild type, mixotrophic growth was inhibited under normal light intensity. The mutant cells supported lower rates of whole-chain photosynthetic electron transport. Thylakoids isolated from the aspartate and serine mutants have lower levels of PS I subunits PsaC, PsaD, and PsaE and lower rates of PS I-mediated substrate photoreduction compared with the wild type. The alanine and double aspartate mutants have no detectable levels PsaC, PsaD, and PsaE. Electron transfer rates, measured by cytochrome c6-mediated NADP+ photoreduction, were lower in purified PS I complexes from the aspartate and serine mutants. By measuring the P700+ kinetics after a single turnover flash, a large percentage of the backreaction in the aspartate and serine mutants was found to be derived from A1 and FX, indicating an inefficiency at the FX → FA/FB electron transfer step. The alanine and double aspartate mutants failed to show any backreaction from [FA/FB]−. These results indicate that the various mutations of the cysteine 14 and 51 ligands to FB and FA affect biogenesis and electron transfer differently depending on the type of substitution, and that the effects of mutations on biogenesis and function can be biochemically separated and analyzed.

Photosystem I (PS I) 1 functions as a plastocyanin:ferredoxin oxidoreductase in the thylakoid membranes of chloroplasts. In cyanobacteria, cytochrome c 6 and flavodoxin serve as alternate donors and acceptors under conditions of low copper and low iron. The PS I complex contains the photosynthetic pigments, the primary donor P700, and five electron transfer centers (A 0 , A 1 , F X , F A , and F B ) that are bound to the PsaA, PsaB, and PsaC proteins. In cyanobacteria, the PS I complex contains at least eight other polypeptides that are ancillary to electron transfer in their function (1). The cofactors of PS I participate in electron transport across the membrane, oxidizing plastocyanin, and reducing ferredoxin according to the following sequence: plastocyanin (copper) or cytochrome c 6 (heme) 3 P700 (Chl a dimer) 3 A 0 (Chl a)3 A 1 (phylloquinone) 3 F X (a [4Fe-4S] cluster) 3 F A or F B ([4Fe-4S] clusters) 3 ferredoxin ([2Fe-2S] cluster) or flavodoxin (flavin).
The PsaC subunit, encoded by the psaC gene, provides the ligands for two [4Fe-4S] clusters, F A and F B . Previous studies showed that the introduction of aspartic acid in position 14 (C14D PsaC ) and in position 51 (C51D PsaC ) led to the introduction of [3Fe-4S] and mixed-ligand [4Fe-4S] clusters in the modified F B and F A sites, respectively, of Escherichia coli-expressed PsaC proteins (2). However, when the mutant PsaC proteins were rebound to P700-F X cores, only mixed-ligand [4Fe-4S] clusters were found in the modified sites of the reconstituted C14D PsaC -PS I (3) and C51D PsaC -PS I (4) complexes. In both mutant PS I complexes, electrons could be transferred to the mixed-ligand iron-sulfur cluster at 15 K, and room temperature NADP ϩ photoreduction was supported at rates similar to the wild type.
The formation of mixed-ligand [4Fe-4S] clusters with altered spectral and redox properties in vitro provided a rationale for probing the functions of F A and F B in vivo. The ability to support [4Fe-4S] clusters that are able to transfer electrons at both 15 and 298 K suggests that the putative oxygen-ligated iron-sulfur clusters should be functional in living organisms. The step taken in this work is to move mutations in these ligands into a genetic system such as Synechocystis sp. PCC 6803 and to study the in vivo consequences on growth and electron transfer. Unlike other cyanobacteria, two different psaC genes have been reported in Synechocystis sp. PCC 6803. One (psaC1) (5) has a deduced amino acid sequence identical to that of tobacco, while the other (psaC2) (6) has a deduced amino acid sequence similar to those reported for other cyanobacteria. The psaC1 gene is not involved in PS I, nor can it substitute for psaC2 when the latter is insertionally inactivated. The amino acid sequence of Synechocystis sp. PCC 6803 PsaC matched that predicted from psaC2. Insertional inactivation of psaC2 prevented the formation of PsaC, thus demonstrating that this gene encodes the PS I-bound polypeptide. Further work showed that the PsaC polypeptide is necessary for stable assembly of PsaD and PsaE into PS I complexes in vivo and that PsaC, PsaD, and PsaE are not needed for assembly of PsaA/PsaB dimer and electron transport from P700 to F X (7). In this paper the term psaC refers to the gene psaC2.
Site-directed mutagenesis and transformation of Synechocystis sp. PCC 6803 has been successfully used in the study of PS I biogenesis and function (10,11). In the work presented here, a set of strains with mutations in PsaC has been created in which a cysteine ligand to F A and/or F B is substituted by aspartate, serine, or alanine in positions 14 and/or 51. The genetic, physiological, and biochemical characterization of the PS I mutants will be presented. A companion paper (12) describes the EPR and optical kinetic properties of the mixedligand F A and F B clusters in site-modified PS I complexes.

MATERIALS AND METHODS
Strains and Growth Conditions-Experiments were performed using a glucose-tolerant strain of Synechocystis sp. PCC 6803, which was acclimated for growth on solid medium in the dark. Except for tests for photoautotrophic and mixotrophic growth, cells were grown at 30°C under light-activated heterotrophic growth (LAHG) conditions, as described previously (13). Antibiotics were added in the following concentrations: kanamycin (Km), 5 mg/liter; gentamicin (Gm), 1 mg/liter. Transformations were performed essentially as described (15), except in the case of strain ⌬C-RCPT, which was carefully maintained in dim light (5 mol m Ϫ2 s Ϫ1 during 2-h incubation) throughout the procedure, since it is light-sensitive (its mixotrophic growth with added glucose is inhibited at light intensities greater than 10 mol m Ϫ2 s Ϫ1 ). Tests for photoautotrophic and mixotrophic growth were performed using solid media with or without supplemental glucose in a chamber providing continuous light. The light intensity was varied by covering plates with layers of cheese cloth and were monitored using a L1-185A photometer (LICOR, Lincoln, NE). Large cultures were grown in carboys (15 liters) under LAHG conditions and were bubbled with air.
DNA Manipulations-Nucleic acids were manipulated using standard methodology (16), unless otherwise stated. Site-directed mutagenesis was performed using an oligonucleotide-directed in vitro mutagenesis kit as directed by the manufacturer (Amersham Corp.). For amplification of psaC from cyanobacterial strains, cells picked from a medium sized colony or equivalent amount of cells collected from liquid culture were washed once with water and used as template. Amplification products were purified using a PCR purification kit (Promega Corp., Madison, WI). Procedure for preparation of cyanobacterial DNA was adapted from Ref. 17, with two "loopfuls" of cells scraped from plates or cells from 10 ml of liquid culture being used to extract DNA.
Transformation of Synechocystis 6803 and Selection Conditions-Synechocystis sp. PCC 6803 that had been maintained under LAHG conditions for at least two subcultures (cells were subcultured once a week) was transformed with plasmids containing resistance genes to kanamycin or gentamicin. Selection for antibiotic-resistant colonies was performed under LAHG conditions. Resistant colonies were restreaked to single colonies with at least five serial transfers to obtain full segregation of the mutation, as verified by restriction enzyme analysis of PCR products, direct sequencing of PCR products, Southern hybridizations, and growth tests.
Western Blot Analysis of Thylakoid Membrane Proteins-Thylakoid membranes were isolated and SDS-polyacrylamide gel electrophoresis and immunoblotting were performed as described (18). To resolve the PsaA/PsaB proteins, 10% SDS-polyacrylamide gel electrophoresis gels were used; 17% gels were used to resolve PsaC, PsaD, PsaE, and PsaF. D2 protein in Photosystem II (PS II) was resolved with a 17% gel to serve as a control. Protein assays were performed using the method of (19). Equal amounts of chlorophyll (4 g for PS I complexes) were loaded in each lane. Equal amounts of protein (150 g) was loaded in each lane for thylakoids preparations. Rabbit antiserum to PsaC or PsaD were raised using protein purified from strains of E. coli expressing, respectively, the psaC gene from Synechococcus sp. PCC 7002 or the psaD gene from Nostoc sp. PCC 8009 (20). Antibodies against PsaA/B proteins from Synechococcus were raised in rabbits as described previously (21). Antibodies against PsaE and PsaF (gifts from Dr. Parag Chitnis, Kansas State University) were raised in rabbits immunized with PsaE and PsaF purified from Synechocystis sp. PCC 6803. Rabbit antibodies to the D2 polypeptide from spinach (gifts from Dr. Wim Vermaas, Arizona State University) were raised as described (22). Chlorophyll was extracted with methanol and quantified using published extinction coefficients (23).
Oxygen Evolution-LAHG-grown cells were washed once in 40 mM HEPES buffer, pH 7.0 and cells containing 10 g Chl were resuspended in 1 ml of the same buffer and illuminated by saturating light at 25°C. Rates of oxygen evolution were determined with a Rank-type oxygen electrode unit. Whole-chain electron transport was measured in the presence of 10 mM NaHCO 3 . PS II electron transport was measured in the presence of 1 mM 2,6-dichloro-p-benzoquinone.
Membrane Isolation and Purification of PS I Complex-Thylakoid membranes from Synechocystis sp. PCC 6803 were isolated using a modification of the procedure described in Ref. 7. The cells were suspended in 0.8 M sucrose, 50 mM HEPES buffer, pH 7.8, and pelleted by centrifugation. The cells were washed twice and suspended in the same buffer containing 0.8 M sucrose and protease inhibitors. The cells were broken in 10 cycles of a prechilled bead-beater (Biospec Products, Bartelsville, OK); one cycle consisted of a 45-s "on phase" and a 10-min "off phase." The cell solution was removed from the beads by vacuum suction and centrifuged at r av of 5,000 ϫ g to remove unbroken cells. The supernatant was pelleted by high speed ultracentrifugation at 100,000 ϫ g with repetitive washing. The thylakoid membranes were suspended in 50 mM Tris, pH 8.3, frozen in liquid nitrogen, and stored at Ϫ95°C. n-Dodecyl ␤-D-maltoside-PS I (DM-PS I) complexes were isolated using the protocol described in Ref. 24 with minor modifications. The membranes were solubilized in 1% n-dodecyl ␤-D-maltoside (DM, Calbiochem) at 4°C for 1 h at the Chl concentration of 0.5 mg ml Ϫ1 . DM-PS I complexes were isolated from the lower green zone of PS I trimers which appeared after centrifugation of the solubilized membrane suspension loaded in a sucrose density gradient (0.1-1.0 M sucrose) for 24 h at 4°C. In the alanine and double aspartate mutants, a mixture of PS I monomers and trimers was used, because it was difficult to completely separate the lower and the upper bands. The problem is that PsaL is easily lost in the presence of 1% DM at 0.4 mg Chl/ml due to the absence of PsaD (9). Even though the lower part of the broad band containing mostly PS I trimers was used (containing Ͻ10% of the total Chl), there was nonetheless some contamination by PS I monomers. Isolated PS I complexes were dialyzed in 50 mM Tris, pH 8.3, resuspended with the same buffer containing 15% glycerol and 0.03% DM, frozen as small aliquots in the liquid nitrogen, and stored at Ϫ95°C.
Steady State Reductase Activity-The steady state reductase activities of PS I complexes were measured according to Ref. 25. The absorbance kinetics were measured using a Cary 219 spectrophotometer with the photomultiplier shielded by appropriate narrow band and interference filters. The sample was illuminated from both sides using two banks of high intensity, LEDs emitting at ϳ660 nm (LS1, Hansatech Ltd.). The light intensity was saturating at the chlorophyll concentration used.
Rates of flavodoxin photoreduction were measured in a 1.3-ml volume using 15 M flavodoxin and DM-PS I at 5 g ml Ϫ1 of Chl in 50 mM Tricine, pH 8.0, 50 mM MgCl 2 , 15 M cytochrome c 6 from Spirulina maxima, 6 mM sodium ascorbate, 0.05% DM. The measurement was made by monitoring the rate of change in the absorption of flavodoxin at 467 nm.
Time-resolved Optical Absorption Spectroscopy-Transient absorbance changes of P700 at 820 nm (⌬A 820 ) were measured from the microseconds to tens-of-seconds time domain with a laboratory-built double-beam spectrometer as described in Ref. 28 upon excitation with a frequency-doubled Nd-YAG laser with a flash energy of 250 mJ. The decay transients were fitted to "sum of several exponentials with base line" using the Marquardt algorithm in Igor Pro. The user-defined fit function enabled a fit up to 7 exponentials with all amplitudes and rate constants set free during the fit. In most cases the fit comprised a base-line component accounting for long term phases and/or possible drift of signal zero during long time scale acquisition. The quality of the fit was estimated using standard techniques, including analyses of the residuals plots and comparison of the ⌾ 2 values and standard errors of the fit parameters between different fits.

Genetic Characteristics of the PS I Mutants
Construction and Characterization of ⌬C-RCPT-To allow rapid segregation of mutations in the psaC gene, a recipient strain of Synechocystis sp. PCC 6803 was engineered with the psaC coding region deleted and replaced by a kanamycin-resistant (Km R ) cassette (Fig. 1A). PCR-amplified psaC upstream (454 bp) and downstream (208 bp) flanking regions, with linker sequences for BamHI and EcoRI at either end, were cloned into pUC118 (29). A Km R cassette excised from pUC4K (30) using EcoRI was inserted into the EcoRI site that separated the upstream and downstream flanking regions to form the plasmid pUC118-⌬C. Glucose-tolerant but otherwise wild type Synechocystis sp. PCC 6803 was transformed with pUC118-⌬C, and Km R colonies were selected under LAHG conditions and were genetically verified by Southern hybridization using a 1.5-kb EcoRI-NcoI fragment containing psaC as probe and by PCR. Complete segregation of the deletion mutation was confirmed (data not shown).
PsaC Mutagenesis and Genetic Characterization of the Mutants-Plasmids for site-directed mutagenesis were constructed in vitro and manipulated in E. coli. A 1.5-kb EcoRI-NcoI fragment containing psaC was cut out of the plasmid p6.1S3.5I (6) and cloned into pUC119 (29) at the SmaI site to create plasmid pC. A 2.0-kb gentamicin resistance (Gm R ) cassette, cut with BamHI from pUC119-gen (31), was inserted at BbsI site downstream of psaC gene on pC to create plasmid pCG. Single-stranded DNA of pC was used as template for site-directed mutagenesis. Oligonucleotides were designed to effect the desired changes in the coding sequence, while also destroying an endonuclease restriction site (RsaI for the Cys-14 site and BbvI for the Cys-51 site). The resulting change in digestion pattern serves as an effective and simple means of screening for the desired mutation. After verification of the mutations on pC using restriction mapping and DNA sequencing, a 953-bp BglII-PstI fragment containing the psaC mutation was cut out of pC and ligated with a 5.7-kb partial digestion product from pCG devoid of the corresponding fragment to form pCG with mutated psaC gene. The mutated pCG plasmids were checked by restriction mapping and DNA sequencing to verify the presence of the desired mutations and the proper sequences. Taking advantage of a XbaI site in between Cys-14 site and Cys-51 site, pCG with a double mutation C14D/ C51D PsaC was created by ligating a 758-bp XbaI-XbaI fragment from pCG with a single mutation C14D PsaC and a 5.9-kb XbaI-XbaI fragment from pCG with a single mutation C51D PsaC . Plasmid pCG with wild type psaC and its mutated variants (shown with asterisk on psaC) were then used to transform the strain ⌬C-RCPT (Fig. 1B). Gm R colonies were selected under LAHG conditions and were genetically verified by three different methods: (i) Southern hybridization of genomic DNA using a 1.5-kb EcoRI-NcoI fragment containing psaC as probe; (ii) restriction mapping and direct DNA sequencing of PCR product amplified from single colonies; and (iii) growth in the presence of kanamycin. Gene replacement by double crossover and complete segregation of all the mutations was confirmed (data not shown).

Physiological Characterization of the PS I Mutants
Growth Analysis of the Mutants-As shown in Table I, all of the PsaC mutants were unable to grow autotrophically under light intensities ranging from 2.2 to 22 mol m Ϫ2 s Ϫ1 . They were able to grow mixotrophically under 2.2 mol m Ϫ2 s Ϫ1 but not under 22 mol m Ϫ2 s Ϫ1 . In addition, C51D and C14S were not able to grow photoautotrophically or mixotrophically under 60 mol m Ϫ2 s Ϫ1 . All PsaC mutants showed similar growth rates to wild type when growing under LAHG or mixotrophically under 2.2 mol m Ϫ2 s Ϫ1 . These results show that the mutants are deficient in photosynthesis, that they are lightsensitive, and that their mixotrophic growth is inhibited by moderate light intensity of 22 mol m Ϫ2 s Ϫ1 . The likely source of light inhibition is the overreduction of the quinone pool by PSII activity. This interpretation arises from results (data now shown) where C51D PsaC was found to grow mixotrophically under 22 mol m Ϫ2 s Ϫ1 light in the presence of 5 mM DCMU, a PS II inhibitor. Transformation of C14S PsaC and C51D PsaC with wild type psaC DNA restored photoautotrophic growth and relieved light inhibition under mixotrophic conditions, demonstrating that the only lesion in these mutants is on psaC (data not shown).
Oxygen Evolution of the Mutant Cells-Whole-chain electron transport (H 2 O 3 CO 2 ) and PS II electron transport (H 2 O 3 2,6-dichloro-p-benzoquinone) were measured using LAHGgrown cells. As shown in Table I, there was essentially no O 2 evolution in ⌬C-RCPT, indicating that PsaC is required for whole-chain photosynthetic electron transport. This is consistent with previous observations (7). Compared with wild type control, all the PsaC mutants had lowered whole-chain O 2 evolution rates to varying degrees, with Asp and Ser mutants showing the highest rates, and Ala mutants and the double Asp mutant showing the lowest rates. However, all the mutants, including ⌬C-RCPT, showed near-wild type levels of PS II O 2 evolution rates, indicating electron transport in PS II was not affected in the short term in the mutants. Variation of these rates most probably represents physiological variations of the whole cells of several independent cultures of each mutant tested over a period of several months.
Rates of NADP ϩ Photoreduction- Table I shows

Biochemical Characterization of the PS I Mutants
PS I Subunit Composition in the Mutants-Immunoblotting of thylakoid membrane proteins was performed on an equal protein basis, and PS I complexes on an equal chlorophyll basis, using antibodies against the proteins PsaA/B, PsaC, PsaD, PsaE, and PsaF. Compared with wild type thylakoid membranes, the levels of PsaC, PsaD, and PsaE were lower to varying degrees in all mutants (Fig. 2). The Asp and Ser mutants demonstrated moderately low levels of these proteins, while Ala mutants and the double Asp mutant did not contain detectable amounts of these three subunits. All mutants contained near-wild type levels of PsaF and PsaA/PsaB in all samples (Fig. 2). It is difficult to quantitate PsaA/PsaB with immunoblots (10) or by enzyme-linked immunosorbent assay 2 ; however, it is assumed that the levels of PsaA/PsaB in the PsaC mutant thylakoids are similar to that in wild type, since it has been demonstrated that assembly of the PsaA/PsaB dimer does not require PsaC, PsaD, or PsaE (7,9).
In all mutants the levels of PsaC, PsaD, and PsaE appeared to be closely related, an observation that is consistent with the finding that the stable binding of PsaD and PsaE to PS I complex is dependent on the presence of PsaC (7,8). The levels of PsaA/B and PsaF are representative of the amounts of PS I core in thylakoids (32). The levels of PsaC, PsaD, and PsaE represent the amounts of these three subunits bound on PS I core to form a complete PS I complex. These results show that the amount of complete PS I complexes is lower to varying degrees in all the mutants compared with the wild type. In thylakoids from the Asp and Ser mutants, a minor portion of PS I contains PsaC, PsaD, and PsaE bound to form a complete PS I complex, while the majority of PS I cores are lacking these subunits (it is difficult to specify the precise amount from the Western blots and efforts to do this by enzyme-linked immunosorbent assay techniques were unsuccessful). In thylakoids from Ala mutants and the double Asp mutant, no complete PS I complexes are found. To our knowledge this is the first report that the population of PS I complexes is found to be heterogeneous in vivo as a result of mutagenesis.
Heterogeneous PS I populations, assayed by immunoblot analysis, were also observed for purified PS I complexes from the Asp and Ser mutants after solubilizing the membranes with 1% DM at 1 mg/ml Chl for 20 min followed by a single TABLE I Physiological characterization of PsaC mutants (a) Growth tests: BG11 plates supplemented with appropriate antibiotic and glucose, when applicable, were incubated at 30°C; ϩ, growth; Ϫ, no growth in 50 days. (b) O 2 evolution: LAHG-grown cells containing 10 g of Chl were resuspended in 1 ml of 40 mM HEPES buffer, pH 7.0, and illuminated by saturating light at 25°C. Rates of oxygen evolution were determined with a Rank type oxygen electrode unit. Whole-chain electron transport was measured in the presence of 10 mM NaHCO 3 . PS II electron transport was measured in the presence of 1 mM 2,6-dichloro-pbenzoquinone. (c) PS I-mediated substrate reduction: See Refs. 24 and 25 for measurement conditions. (d) Efficiency of electron donation from P700 to F A /F B shown as the percent backreaction with a lifetime Ն 7 ms. sucrose density centrifugation step (data not shown). Heterogeneity complicates functional analysis of the mutant PS I complexes, making it difficult to distinguish between (i) inefficient electron transfer from A 1 to F X to F A /F B and (ii) a mixed population of P700-F X cores and P700-F A /F B complexes. To solve this problem, a purification procedure with sucrose density gradient centrifugation was adopted to separate the PS I core subpopulation (i.e. those devoid of PsaC, PsaD, and PsaE proteins) from the integral PS I complex. The separation principle is based on the finding that at high ratios of DM to Chl, PsaL is readily lost in PS I complexes that lack PsaC, PsaD, and PsaE (9). On the other hand, the PsaL-less P700-F X core was shown to be present in the upper, monomeric band, while the intact PS I complex is present in the lower, trimeric band (33). As the result nearly homogeneous PS I holocomplexes, as analyzed by immunobloting, were isolated from the Asp and Ser mutants. Compared with wild type PS I complexes, all the mutants had similar levels of PsaB and PsaF; and the single Asp and Ser mutants had similar levels of PsaC, while the Ala mutants and the double Asp mutant had no detectable levels of PsaC (Fig. 3).

Single Turnover Flash Studies of Mutant PS I Complexes-
The effects of different ligand substitutions on P700 ϩ re-reduction kinetics were determined in homogeneous DM-PS I complexes after a single turnover, saturating flash. In wild type PS I complexes, the vast majority of the backreaction is derived from [F A /F B ] Ϫ . As shown in Fig. 3A, 60% of the recombination kinetics are derived from the 25-and 112-ms components attributed to P700 ϩ [F A /F B ] Ϫ recombination. An additional 26% are derived from slower phases with lifetimes of 221 ms and 2.2 s, leading to an 86% efficient electron transfer to F A /F B . The slowest kinetic phases are due to exogenous donors undergoing redox reactions with P700 ϩ and come about in reaction centers where [F A /F B ] Ϫ has become oxidized by exogenous electron acceptors in the medium. The sum contribution of earlier acceptors, including F X Ϫ and A1 Ϫ , is 14% of the total absorption change.
The major contribution to the absorbance change in the Synechocystis sp. PCC 6803 PS I complex with chemically reduced terminal iron-sulfur clusters and in the P700-F X PS I core isolated by urea treatment of the PS I complex is brought about by the components with lifetimes of ϳ400 and 1.5 ms, which result from back reaction of P700 ϩ and F X Ϫ (34). The faster components with lifetimes of ϳ10 and 100 s appearing in these preparations may result from recombination of P700 ϩ and A 1 Ϫ (35) in a fraction of centers which either have a lower quantum efficiency of electron transfer between A1 and F X due to some changes of F X microenvironment or have the F X cluster missing or chemically reduced. There is also an evidence that decay of 3 Chl formed upon laser flash excitation in the antenna may contribute to ⌬A decay in the tens-of-microseconds time domain.
The kinetics of the C14D/C51D PsaC -PS I (Fig. 4B), C14A PsaC -PS I (Fig. 4C), and C51A PsaC -PS I (Fig. 4D) complexes are similar to those for the P700-F X PS I core isolated by urea treatment of the PS I complex (data not shown; see Ref. 34). The majority of the backreaction kinetics for all three mutant complexes is derived from tens-of-microseconds to milliseconds decay phases, with little or no measurable backreaction in the tens-to-hundreds of milliseconds time scale. Since the PsaC protein is missing in these mutants (Figs. 2 and 3), and therefore the F A and F B clusters, which govern the tensof-milliseconds decay kinetics are lacking, this is the expected result. The larger ⌬A in these three mutants, which is due to a significant microsecond component, may be derived from additional absorption changes from the decay of 3 Chl in the antenna bed. The optical kinetic data are in agreement with the steady state rate data (Table I) in which no significant NADP ϩ reduction occurs in the absence of F A and F B .
The kinetics of the C14D PsaC -PS I, C14S PsaC -PS I, C51D PsaC -PS I, and C51S PsaC -PS I complexes are mixed "core"and "complex-type," with each mutant showing a slightly different fractions of the backreaction derived from tens-of-microseconds to milliseconds decay phases (Fig. 5). The C14D PsaC -PS I complex (Fig. 5A) has the largest, and the C51S PsaC -PS I mutant (Fig. 5D) has the smallest, percentage of electrons which arrive at F A /F B . The [F A /F B ] Ϫ backreaction in C14D PsaC -PS I (Fig. 5A) has lifetime components of 33 and 335 ms which contribute 26% to the total absorption change; an additional 15% is contributed by the slower donation to P700 ϩ by external donors, leading to a 41% efficient transfer to F A /F B . The remainder of the backreaction occurs from components with lifetimes of 712 s (19%) and 2.4 ms (20%) derived from F X , and 19% occurs from a component with a lifetime of 45 s. The [F A /F B ] Ϫ backreaction in C51D PsaC -PS I (Fig. 5B) has lifetimes of 9 and 106 ms which contribute 15% to the total absorption change; an additional 13% is contributed by the slower donation to P700 ϩ by external donors, leading to a 28% efficient electron transfer to F A /F B . The remainder of the backreaction occurs from components with lifetimes of 206 s (20%) and 1.0 ms (12%) derived from F X , and over 39% occurs from a component with a lifetime of 34 s.
The C14S PsaC -PS I complex shows relatively similar kinetic behavior to the above two mutants. The [F A /F B ] Ϫ backreaction in C14S PsaC -PS I (Fig. 5C) has lifetime components of 51 and 148 ms which contribute 26% to the total absorption change; an additional 6% is contributed by the slower donation to P700 ϩ by external donors, leading to a 32% efficient transfer to F A /F B . The remainder of the backreaction occurs from components with lifetimes of 1.13 ms (25%) and 2.81 ms (13%) derived from F X , and over 31% occurs from components with lifetimes of 32 and 198 s. The [F A /F B ] Ϫ backreaction in C51S PsaC -PS I (Fig.  5D) has lifetime components of 28 and 89 ms which contribute only 9% to the total absorption change; an additional 12% is contributed by the slower donation to P700 ϩ by external donors, leading to a 21% efficient electron transfer to F A /F B . The remainder of the backreaction occurs from components with lifetimes of 580 s (41%) and 2.3 ms (22%) derived from F X , and only 16% occurs from a component with a lifetime of 36 s.

DISCUSSION
Multiple site-specific mutations to individual cysteine ligands for each of the two PS I [4Fe-4S] centers F A and F B were used to produce seven mutant strains of PsaC in Synechocystis sp. PCC 6803. Cysteines 14 and 51 were changed to alanine (C14A PsaC , C51A PsaC ), aspartic acid (C14D PsaC , C51D PsaC ), and serine (C14S PsaC , C51S PsaC ), and the results were compared with in vitro reconstitution studies (3,4,36). In each instance, the PsaC mutant strains could not grow under standard photoautotrophic growth conditions, but could grow mixotrophically under weak light, indicating a light-sensitive lesion in PS I that had become limiting for growth. The characteristics of these separate lines were dependent on the specific ligand substitution for the cysteines in two separable ways. First, some of the mutations resulted in PS I reaction centers where stable incorporation of PsaC was precluded by the specific mutation. These strains included the substitutions C51A PsaC , C14A PsaC , and the double aspartate substitution C14D/ C51D PsaC . Second, some of the mutations resulted in PS I reaction centers incorporating lower-than-wild type levels of PsaC. These strains included the substitutions C51D PsaC , C51S PsaC , C14D PsaC , and C14S PsaC . In spite of the heterogeneity of PS I found in the membranes, near-homogeneous PS I complexes with bound PsaC, PsaD, and PsaE as judged by protein blotting were isolated from these four mutant strains by detergent fractionation.
These results vary from those recently published for two mutant strains constructed in Anabaena variabilis ATCC 29413 (37). In A. variabilis, the C13D and C50D mutants (the reaction media is 25 mM Tris buffer, pH 8. same functional cysteines in mutants C14D PsaC and C51D PsaC in Synechocystis sp. PCC 6803), grew photoautotrophically, and electron transport rates, measured using ascorbate/DCPIP as a donor, were similar to the wild type. Species-specific differences may be invoked to account for these differences. For example, A. variabilis 29413 is a filamentous organism and a natural heterotroph and does not require special light-pulse treatment as does Synechocystis sp. PCC 6803 (5) to grow heterotrophically.
PsaC Mutations and PS I Biogenesis-The biogenesis and redox properties of membrane complexes containing bound iron-sulfur clusters is dependent upon the associated protein structure; however, protein structure can also be modified by incorporation of mutations to ligands of iron-sulfur clusters (38). The in vivo experiments with Synechocystis sp. PCC 6803 demonstrate that some mutations give rise to altered forms of PsaC that are not stable in the PS I complex: C51A PsaC , C14A PsaC , and the double mutant C14D/C51D PsaC . In these cases little PsaC is seen in thylakoid membranes, and none is detected in purified PS I complexes (Figs. 2 and 3).
The Ala mutants, which contain an aliphatic side group, are not capable of providing a ligand to an iron in the modified site of the cluster. These substitutions are only capable of supporting [3Fe-4S] clusters (39). The implication is that PsaC proteins containing [3Fe-4S] clusters are unable to bind to P700-F X cores. We suspect that PsaC hosting a [4Fe-4S] cluster and a [3Fe-4S] cluster, as has been found in the unbound PsaC mutants C14D PsaC , C51D PsaC , C14S PsaC , and C51S PsaC (2, 39) has a structure sufficiently different from that of wild type PsaC to preclude its incorporation into PS I. The single substitutions with either aspartic acid (C14D PsaC , C51D PsaC ) or serine (C14S PsaC , C51S PsaC ) lead to lower-than-wild type levels of PsaC incorporation into PS I complexes in thylakoids ( Fig. 2A). The Asp and Ser mutants, which contain carboxylate and hydroxy side groups respectively, are potentially capable of providing an oxygen ligand to an iron in the modified site of the cluster. These substitutions are capable of supporting mixedligand [4Fe-4S] clusters. The implication is that two [4Fe-4S] clusters must be present in PsaC to be incorporated into PS I complexes. It is likely that the C51D PsaC , C51S PsaC , C14D PsaC , and C14S PsaC proteins also have altered structures, resulting in less-than-perfect interactions with PsaA/PsaB. Yet, size and charge considerations are only one possibility. If the iron-sulfur center insertion requires an efficient ligand exchange reaction at the Cys-14 or Cys-51 positions (this is not provided by the Ala mutations and may be altered to some extent in the Ser and Asp changes), then PsaC biogenesis could be suppressed with the resulting phenotype. Consequently, the mutant PsaC proteins may not bind tightly, or they may dissociate easily, or they may be degraded in the cell, leading to a decreased amount of fully assembled PS I complex.
PsaC Mutations and Electron Transport-The C51D PsaC , C51S PsaC , C14D PsaC , and C14S PsaC mutant strains demonstrate whole chain oxygen evolution of modest but significant levels along with at least 50% or better wild type PS I electron transport capacity. Characterization of isolated PS I complexes  (Fig. 3) demonstrates that electron transport is less efficient when incorporating a modified PsaC subunit. Assuming that the levels of PsaC revealed by Western blots in the isolated PS I complexes of the above four mutants are similar to the wild type, the ⌬A 820 kinetics results imply that the lower rates of electron transport are due to a qualitative functional alteration of PS I. The single turnover flash data are in an agreement with the NADP ϩ reduction data and show that the inefficient electron transfer step occurs between F X and F A /F B regardless of whether the mutation is in cysteine 14, which is associated with the F B site, or in cysteine 51, which is associated with the F A site. Particular values of lifetimes and contributions of the ⌬A 820 decay phases vary slightly from preparation to preparation even in the wild type samples. However, analysis of decay kinetics over several orders of time scale provides a rationale to distinguish between the "integral-complex" and "core-type" kinetics signature and detect a decrease in efficiency of electron transport to F A and F B (34). We have tabulated the overall amplitude of the ⌬A 820 kinetic components with lifetimes longer than 7 ms as an indicator of efficient photoreduction of F A and F B . As shown in Table I, electron transfer to F A /F B in the serine and aspartate mutants roughly correlates with the rates of ferredoxin-mediated NADP ϩ photoreduction.
Comparison of some of these mutants (C14S PsaC , C51S PsaC ) with analogous in vitro PS I mutants (36) indicates that F A /F B photoreduction in the in vivo mutants samples occurs with a lower apparent quantum efficiency. One possible explanation is that the external thiolate of ␤-mercaptoethanol, used in the iron-sulfur insertion protocol, provides the ligand to the [4Fe-4S] cluster. The PsaC conformation would be rendered closer to that of the wild type, which would then confer higher efficiency of forward electron transfer to the PsaC-bound clusters. On the other hand, the higher contribution of the fast kinetic phases (in the microseconds to milliseconds time domain) in the in vivo PS I mutants may occur due to decrease of the amount of PsaC per reaction center, which would not be resolved in the Western blots of the PsaC protein. In any event, one important question regarding the functional properties of PsaC in the in vivo mutants is whether only one or both of the PsaC-bound [4Fe-4S] clusters are functioning as the electron acceptors. The goal of the companion paper (12) is to probe this question using low-temperature EPR spectroscopy and absorbance difference kinetics measurements using repetitive flash excitation.
Given the proposed redox equilibrium between A 1 and F X (40), a reasonable rationale for the function of F A and F B is to draw the equilibrium completely away from A 1 , thereby ensuring a high quantum yield in PS I. The single substitutions with either aspartic acid (C14D PsaC , C51D PsaC ) or serine (C14S PsaC , C51S PsaC ) are sufficiently similar in structure and charge for the mutant PsaC subunit to function in this capacity. For the Ala substitutions, C14A PsaC and C51A PsaC , and the double Asp mutant, C14D/C51D PsaC , the modified PsaC did not stably incorporate into PS I thylakoids in vivo. These mutants also demonstrated greater impairment in whole chain oxygen evolution than the other mutants, with rates similar to the ⌬C-RCPT recipient strain entirely lacking a functional PsaC. PS I electron transport measured in three different assays also displayed baseline activity, indicating that no functional PS I was present.
Decreased amounts of complete PS I complexes and reduced electron transport efficiencies in complete complexes could both lead to total reduced PS I function, making PS I activity the rate-limiting step in photosynthetic electron transport. The reduced PS I activity apparently causes dramatically different phenotypes in growth of the PsaC mutants by two different mechanisms. Under low light the reduced PS I activity limits photosynthetic conversion of light energy to the extent that cells cannot grow photoautotrophically; under high light (22 mol m Ϫ2 s Ϫ1 ) reduced PS I activity causes light inhibition on growth by a yet unknown mechanism. The light sensitivity issue has already been addressed in cyanobacterial PS I-less mutants (41); the major problem is that the electron transfer chain stays reduced, leading to the sensitivity to light.
Mutations in PsaC may therefore alter the structure of the protein, resulting in altered interaction with PS I core and thus decreased amount of PS I holocomplex in the thylakoids. Suppressor mutations on PsaA or PsaB should alter the structure of the PS I core sufficiently to restore this interaction and have the potential to partially restore PS I function. Spontaneous pseudorevertants have recently been isolated from the sitedirected mutants based on the restoration of photoautotrophic growth. Characterization of the suppressor mutations is likely to provide new insights into protein-protein interactions in PS I and into the function of PS I in energy metabolism.