Modified ligands to FA and FB in photosystem I. II. Characterization of a mixed ligand [4Fe-4S] cluster in the C51D mutant of PsaC upon rebinding to P700-Fx cores.

A Photosystem I (PS I) complex reconstituted with PsaC-C51D (aspartate in lieu of cysteine in position 51) shows light-induced EPR signals with g values, line widths, and photoreduction behavior characteristic of FB. Contrary to an earlier report, a [3Fe-4S] cluster was not located in the reconstituted PS I complex. Instead, a second set of resonances with g values of 2.044, 1.942, and 1.853 becomes EPR-visible when the C51D-PS I complex is measured at 4.2 K. This fast relaxing center, termed FA' is likely to represent a [4Fe-4S] cluster in the mixed ligand (3Cys.1Asp) site. Redox studies show that the Em of FA' and FB are -630 mV and -575 mV, respectively. Room temperature optical studies support the presence of two functioning electron acceptors subsequent to Fx, and NADP+ photoreduction rates mediated by ferredoxin and flavodoxin are nearly equivalent to the wild type. In addition to [3Fe-4S] clusters and S = 1/2 ground state [4Fe-4S] clusters, the free PsaC-C51D protein shows resonances near g = 5.5, which may represent a population of high spin (S = 3/2) [4Fe-4S] clusters in the mixed ligand FA' site. Similar to the C14D-PS I mutant complex, it is proposed that the P700-Fx core selectively rebinds those free PsaC-C51D proteins that contain two [4Fe-4S] clusters. These studies show that primary photochemistry and electron transfer rates in PS I are relatively unaffected by the presence of a highly reducing, mixed ligand cluster in the FA' site.

Photosystem I (PS I) 1 is a multisubunit pigment-protein complex located in the thylakoids of chloroplasts and inner membranes of cyanobacteria (reviewed recently in Refs. 1 and 2). This enzyme functions primarily as a light-driven cytochrome c 6 :ferredoxin oxidoreductase, transforming a photon into stable chemical free energy. An indispensible feature of PS I is a series of bound electron acceptors that function to suppress the back reaction with P700 ϩ by providing a kinetically favorable pathway to ferredoxin. One strategy for studying the function of the bound electron carriers is to modify their structure and redox potentials, with the expectation that this will lead to changes in function. This approach is currently being applied to individual components of the PS I reaction center, where it is anticipated that subtle structural alterations will influence electron transfer rates and photosynthetic efficiency.
The majority of the bound electron carriers in PS I are located on the PsaA and PsaB polypeptides, including P700, A 0 , A 1 , and F X . The terminal electron carriers of PS I, designated F A and F B , are two [4Fe-4S] clusters located on the stromal, extrinsic PsaC polypeptide. This 9-kDa protein is potentially more amenable to chemical and genetic modification than are the larger subunits of the complex because it can be expressed as the apoprotein in Escherichia coli (3,4). Further, this polypeptide can be removed from the PS I complex by chemical treatment, and either wild-type PsaC or site-directed mutants of PsaC can be rebound onto a purified P700-F X core (reviewed in Ref. 5). Since there is no three-dimensional structure for PsaC (a crystal structure of PS I is published at a resolution of 6 Å (6)), one challenge is to identify changes in amino acids that will lead to insights concerning the function of these two ironsulfur clusters.
One obvious target for study is the ligands to the F A , F B , and F X iron-sulfur clusters. Site-directed mutagenesis has been used recently to alter the cysteine residues in PsaC with the expectation that the substitute amino acid will affect the spectral, thermodynamic, and kinetic properties of the F A and F B clusters. Based on precedent with naturally occurring ferredoxins, the second cysteine in each CXXCXXCXXXCP binding motif was replaced with aspartic acid (7). The free PsaC-C14D and PsaC-C51D proteins were found to contain [3Fe-4S] and [4Fe-4S] clusters (8). When PsaC-C51D and PsaC-C14D were rebound to a P700-F X core in the presence of PsaD, the [4Fe-4S] cluster in the PsaC-C51D-PS I complex assumed g values and photoreduction behavior that identified that cluster as F B . The [4Fe-4S] cluster in the rebound C14D mutant assumed g values and photoreduction behavior that identified that cluster as F A (7).
One complication was that, although a [3Fe-4S] cluster was found in the PsaC-C51D-PS I complex, it was not reduced by light (7). This was troubling, especially since the midpoint potential of the [3Fe-4S] cluster in the free PsaC-C51D protein (E m of Ϫ98 mV) should have allowed it to function as an electron acceptor (8). A further complication was that a [3Fe-4S] cluster was not found in the mirror image PsaC-C14D-PS I complex; instead, an unusual set of resonances around g ϭ 2.0 were found that were also unresponsive to light. In this paper, the results of a comprehensive search for the [3Fe-4S] clusters in a PS I complex reconstituted with PsaC-C51D are reported. The "missing" [3Fe-4S] clusters were not located, and it is suspected that in the earlier work they represented a population of unbound PsaC-C51D that was incom-pletely removed from the complex. Instead, a spin system was found that showed characteristics of a [4Fe-4S] cluster in the mixed ligand F A site. This new center, termed F A Ј, showed photochemical behavior, spin relaxation properties, and a midpoint potential different from F A in the wild-type PS I complex. The consequence of this change on the low temperature photoreduction of F A Ј and F B and on the room temperature photoreduction of NADP ϩ are reported.

MATERIALS AND METHODS
Preparation of the P700-F X Core, Purified Recombinant Proteins, and Reconstitution Procedures-PS I complexes were isolated as described previously (9), and P700-F X cores were prepared as described (10). PsaC and PsaC-C51D were overproduced in E. coli and purified as described previously (7). The mutation in the PsaC-C51D protein was confirmed by sequencing of a tryptic peptide as described (11). Nostoc sp. PCC 8009 PsaD and Synechococcus sp. PCC 7002 PsaE were prepared as described (4).
The iron-sulfur clusters were reinserted into the unbound C51D mutant according to a modification of published procedures (10). Briefly, the purified apoproteins were added to a final concentration of 0.5 mg ml Ϫ1 to a solution of 50 mM Tris-HCl (pH 8.3) containing 1% (v/v) ␤-mercaptoethanol. After 5 min of stirring at room temperature, a solution of 30 mM FeCl 3 was added slowly to a final concentration of 0.15 mM. After an additional 5 min of stirring, a solution of 30 mM Na 2 S was added slowly to a final concentration of 0.15 mM. The reaction mixture was incubated at 4°C for 12 h. The reconstituted iron-sulfur protein was purified free of inorganic reagents by gel filtration over Sephadex G-25 in an anaerobic chamber (Coy Laboratory Products Inc.) and concentrated under anaerobic conditions to a final protein concentration of 20 mg ml Ϫ1 .
Reconstitution of the PsaC-C51D mutant holoprotein with P700-F X cores from Synechococcus sp. PCC 6301 was performed as described (4,10,12). Unbound PsaC and PsaD were removed by gel exclusion chromatography over Sephadex G-75.
Electron Paramagnetic Resonance Spectroscopy and Electrochemical Redox Titrations-EPR spectroscopic studies were performed as described (11). Redox tritrations were carried out by electrochemical poising followed by EPR detection (Ref. 8 and references therein). Spin quantitations were determined at a power level at least 1 order of magnitude below the P1 ⁄2 . Simulations of the EPR spectra were performed using the "Xpow" version of "Qpow" (13) after recompiling the source code in MacFortran (Absoft Corp., Rochester Hills, MI) to run on a Motorola/IBM 601 RISC processor. The spectra were ported to a Macintosh 7100/80 computer, and all data manipulations were performed using IGOR Pro 2.02 (Wavemetrics, Lake Oswego, OR).
Isolation of P700-F X Cores for NADP ϩ Rate Studies-Freshly harvested cells of the PsaC-less mutant (14) of Synechocystis sp. strain PCC 6803 (the mutant also does not contain PsaD and PsaE) were broken in buffer containing 20 mM MES, pH 7.2, 0.8 M sucrose, and protease inhibitors using a Bead beater (Biospec Products, Bartelsville, OK). The membranes were precipitated with 40 mM CaCl 2 , pelleted by centrifugation, and stored at Ϫ95°C until use. Dodecyl maltoside PS I core complexes (DM-PS I) were prepared using a protocol described earlier (15) with little modification. Membranes were solubilized in 1% ␤-Ddodecyl maltoside (Calbiochem) at 4°C for 1 h at a chlorophyll concentration of 1 mg ml Ϫ1 . DM-PS I cores were isolated from the lower green band appearing after centrifugation of the solubilized membrane suspension in a sucrose density gradient (0.1-1.0 M) for 24 h at 4°C. The isolated DM-PS I cores were dialyzed in buffer containing 50 mM Tris, pH 8.3, resuspended with the same buffer containing 15% glycerol and 0.03% ␤-D-dodecyl maltoside, frozen as small aliquots in liquid nitrogen, and stored at Ϫ95°C.
Room Temperature Optical Spectroscopy-Flash-induced absorbance changes were measured in cooperation with Klaus Brettel at the Commissariat à l'Energie Atomique (Saclay, France), in a manner similar to that described in Bottin et al. (16). The apparatus consisted of an 811-nm measuring beam provided by a collimated CW laser diode and a silicon photodetector. The output of the photodiode was fed into an AM502 wide band amplifier (Tektronix, Beaverton, OR; gain 100; bandwidth DC to 1 MHz) and digitized with a DSA 602A oscilloscope (11A52 plug-in). Excitation was provided by two simultaneously fired xenon flashes; the duration of each flash was 3 s. The optical path length for the measuring light was 1 cm. The PS I complexes were incubated with 100 M phenazine methosulfate (PMS) and approximately 3 mg ml Ϫ1 sodium dithionite for 7 s in the dark, and the absorbance changes were measured in a three-flash sequence separated by 100 ms. The concentration of PMS was chosen to assure re-reduction of P700 ϩ prior to recombination with F A Ϫ and/or F B Ϫ . Sodium dithionite was used to reduce the PMS completely to prevent the oxidized form from functioning as an electron acceptor.
Low Temperature Optical Studies-Optical experiments were carried out at 77 K using an Oxford Instruments DN 1704 cryostat and a TC1 temperature controller. The samples were suspended under anaerobic conditions in 25 mM Tris buffer (pH 8.3), 4 mM 1,6-dichlorophenolindophenol (DCPIP), 10 mM sodium ascorbate and 60% (v/v) glycerol to a chlorophyll a concentration of 50 g ml Ϫ1 and transferred to a 10 ϫ 5-mm polystyrene cuvette with a stopper. Transient absorbance changes of P700 (⌬A 820 ) were measured at 820 nm with a split measuring beam provided by a DC 25 F semiconductor diode laser (Spindler and Hoyer). The beam was divided with an optical beam splitter; the sample beam passed through the optical windows of the cryostat, and the reference beam passed through a variable attenuator. The beams were converted to photocurrents with reverse biased planar diffused silicon photodiodes (United Detector Technology model PIN-10D) and into voltages with 1000-ohm resistors. The sample and reference signals were subtracted in real time using a wide bandwidth differential comparator (Tektronix 11A33) and digitized with a Tektronix DSA 601 oscilloscope (Tektronix) interfaced to a Macintosh Quadra computer. Data were collected at 8-bit vertical resolution. The excitation beam was provided by a frequency-doubled, Q-switched Nd-YAG laser (DCR-11, Spectra-Physics) operating at 532 nm at the pulse width at halfmaximum of 10 ns and flash energy of 135 mJ. The ⌬A 820 decay curves were fitted to "sum of several exponentials with base line" using the Marquardt algorithm in Igor Pro 2.02 (WaveMetrics, Lake Oswego, OR).
Rates of NADP ϩ Photoreduction-Rates of NADP ϩ photoreduction mediated by flavodoxin and ferredoxin were measured in a 1.3-ml volume using reconstituted PS I complexes at 5 g of chlorophyll/ml in 50 mM Tricine, pH 8.0, 50 mM MgCl 2 , 15 M Synechococcus sp. PCC 7002 flavodoxin, 15 M Spirulina maxima cytochrome c 6 , 6 mM sodium ascorbate, and 0.05% dodecyl maltoside. P700-F X cores from Synechocystis sp. PCC 6803 (14) were used instead of urea-isolated cores because the former sustained higher rates of NADP ϩ photoreduction when reconstituted with wild-type PsaC. In this strain, PsaC, PsaD, and PsaE are missing, but electron transfer from P700 to F X remains intact. Rates of flavodoxin-mediated NADP ϩ photoreduction were measured with the addition of 0.5 mM NADP ϩ , 0.8 M spinach ferredoxin:NADP ϩ oxidoreductase (Sigma), 0.1% ␤-mecaptoethanol, and 10 mM MgCl 2 . Rates of ferredoxin-mediated NADP ϩ photoreduction were measured under the same conditions except for the substitution of 5 M spinach ferredoxin. Both measurements were made by monitoring the rate of change in the absorption of NADPH at 340 nm using a Cary 219 spectrophotometer fitted with appropriate narrow band and interference filters on the surface of the photomultiplier. The 4-sided (clear) cuvette was illuminated from the both sides using two banks of high intensity, red light-emitting diodes (LS1, Hansatech Ltd., Norfolk, UK). The light intensity was saturating at the chlorophyll concentration used. Fig. 1 shows the EPR spectrum of a PS I complex assembled by binding the PsaC-C51D holoprotein in the presence of excess PsaD to a P700-F X core. Contrary to an earlier report (7), no resonances were found at 15 K in a dark-frozen sample characteristic of an oxidized [3Fe-4S] cluster (Fig. 1A). When the sample was illuminated at 15 K, a weak set of resonances at g ϭ 2.067, 1.934, and 1.882 were present, which are characteristic of F B (Fig. 1C). When the sample was thawed and frozen under continuous illumination to photoaccumulate the electron acceptors, the F B resonances became about 5 times more intense (Fig. 1E). These results show that the F B cluster in the PsaC-C51D-PS I complex is capable of undergoing photochemical reduction at room temperature, but only to a small degree at cryogenic temperatures. In contrast, the sharp radical at g ϭ 2.002 (Fig. 1C), formed irreversibly in the light at 15 K, implies that charge separation to F A has also occurred; hence, it is curious that F A is absent. Except for the lack of resonances from F A , this behavior had been noted (7) and resembles the response of F B in a wild-type PS I complex.

Absence of a [3Fe-4S] Cluster in the C51D-PS I Complex-
To eliminate the possibility that [3Fe-4S] clusters are present in the F A site but are reduced under these conditions, the PsaC-C51D-PS I complex was reexamined in the presence of 5 mM potasssium ferricyanide and after electrochemical poising to a solution potential of ϩ400 mV. No resonances were found at or near g ϭ 2.02 that would indicate an oxidized [3Fe-4S] cluster (data not shown). A reduced [3Fe-4S] cluster is paramagnetic with an integer spin state of S ϭ 2 and can usually be observed as a broad, asymmetric resonance at g values of 10 -12 (the zero-field splitting parameter must be smaller than the X-band microwave quantum). The complex was examined after reduction with sodium dithionite at pH 10.0 and after electrochemical poising at a solution potential of Ϫ650 mV. No resonances were found near g ϭ 10 -12 that would indicate the presence of a reduced [3Fe-4S] cluster (data not shown). The complex was also examined using parallel mode EPR spectroscopy, for which the transition probability of an S ϭ 2 spin system is maximum when the alternating magnetic field is polarized parallel to the static magnetic field. Provided this "⌬M S ϭ 2" transition is allowed, this mode provides an en-hancement in the sensitivity of integer spin systems as characterized by reduced [3Fe-4S] clusters. As shown in Fig. 1B, a dark-frozen sample showed no significant resonances at temperatures of 4.2-30 K in any field position (the transition at 140 mT is due to contaminating ferric iron). Likewise, there was no evidence for a reduced [3Fe-4S] cluster around g ϭ 10 -12 when the sample was illuminated at 15 K (Fig. 1D) or when the sample was thawed and photoaccumulated by freezing during illumination ( Fig. 1F; the resonance at g ϭ 2.002 is due to the strong P700 ϩ radical, which is observed as a weak resonance due to imperfections in the dual mode cavity). The absence of a signal under both oxidizing and reducing conditions demonstrates that a [3Fe-4S] cluster is not present in the F A site of the PsaC-C51D-PS I complex.
Redox Titration of the F B Cluster in the Unmodified Site of the Rebound C51D Mutant-The midpoint potential of the F B cluster in the unmodified site of the PsaC-C51D-PS I complex was determined by electrochemical poising and EPR detection. As shown in Fig. 2, the intensity of the F B resonances increases when the potential is lowered from Ϫ530 mV ( Fig. 2A) to Ϫ580 mV (Fig. 2B) to Ϫ640 mV (Fig. 2C). The most reduced sample was subsequently thawed and frozen under illumination; the signals did not change, indicating that the end point of the titration had been reached. As shown in the bottom half of Fig.  2, the g ϭ 2.067, 1.934, and 1.882 resonances titrate according to a Nernstian response with a midpoint potential of Ϫ575 mV (n ϭ 1.0). This E m value is nearly identical to the midpoint potential of Ϫ580 mV measured for F B in the unbound PsaC-C51D protein. It is also similar to the midpoint potential of ϳϪ575 mV measured for F B in a wild-type spinach PS I complex (17), implying that the mutation at cysteine 51 does not affect the thermodynamic properties of the F B cluster (E m ϭ Ϫ575 mV) in the unmodified site of the PsaC-C51D-PS I complex.
Identification of the F A Cluster in the C51D-PS I Complex- Fig. 3 shows the EPR spectrum of the PsaC-C51D-PS I complex when the sample is frozen in darkness and illuminated at low temperatures. As expected, the amplitude of the F B resonances at g ϭ 2.067, 1.934, and 1.882 decreases as the temperature is lowered from 18 to 6 K due to the onset of microwave power saturation. However, the weak resonances at g ϭ 2.044 and 1.853 begin to increase in intensity with lower temperature, becoming most prominent at 6 K. In the midfield region, the g ϭ 1.942 resonance appears as a slight bend on the g ϭ 1.932 midfield resonance of F B , attaining maximum intensity also at 6 K. These are the g values typically associated with the F A cluster in the wild-type PS I complex, although the spin concentration is lower than expected. The temperature dependence of this new cluster, termed F A Ј, indicates a spin relaxation rate faster than either F A in the wild-type complex, indicating an environment different from the wild-type cluster. Further, the irreversible behavior with respect to light agrees with its assignment as F A , but not F X . A small amount of the F X cluster, shown by the highfield resonance, is visible at g ϭ 1.769, either because of inefficient reconstitution or an altered equilibrium constant between F X and F B /F A . Fig. 4 shows the EPR spectrum of the PsaC-C51D-PS I complex when the sample was frozen during illumination to photoaccumulate the electron acceptor system. The amplitude of the F B resonances at g ϭ 2.067, 1.934, and 1.882 increase as the temperature is lowered from 32 to 18 K. This is followed by a decrease in amplitude as the temperature is lowered further from 15 to 6 K due to the onset of microwave power saturation. The F X cluster, seen most clearly in the high field resonance at g ϭ 1.769, achieves a maximum intensity at a temperature of 6 -8 K. A closer examination of the EPR spectrum of the PsaC-C51D-PS I complex at very low temperatures revealed several new features. A shallow peak at g ϭ 2.089 and a broad highfield trough at g ϭ 1.847 begins to develop at a temperatures of 12 K, becoming clearly visible only at a temperatures around 6 K. In the midfield region, the g ϭ 1.934 resonance from F B decreases in amplitude between 12 and 6 K, and a new derivativeshaped resonance at g ϭ 1.922 begins to appear as a slight bend on this signal. These new features are most evident at 6 K, where the contribution of the F B cluster is diminished.
A plot of signal intensity versus temperature for the g ϭ 1.934 and 1.882 resonances of F B and for the g ϭ 1.847 and 1.922 resonances of the new spin system indicates that the relaxation behavior of F B , inferred from the temperature dependence and half-saturation parameter P1 ⁄2 (data not shown), are similar to that of F B in the wild-type PS I complex. The amplitude of the g ϭ 1.769 resonance of F X is maximum at 6 -8 K, similar to its behavior in the wild-type complex. In contrast, the amplitude of the signals derived from the new spin system increases as the temperature is lowered from 12 to 4.2 K. It is surprising that the F B cluster has not changed under these conditions, i.e. that it does not contribute to an interaction spectrum as seen in the wild-type complex when F A and F B are reduced together in the same reaction center. Instead, the pattern of reduction implies that two spin systems are represented, one by the F B cluster, with a temperature optimum of 18 K and the second by the presumed spectrum of F A Ј.

Redox Titration of the F A Ј Cluster in the Modified Site of the PsaC-C51D-PS I Complex-
The midpoint potential of the F A Ј cluster in the modified site of the PsaC-C51D-PS I complex was determined by electrochemical poising and EPR detection. When the PsaC-C51D-PS I complex is poised at a solution potential of Ϫ580 mV, only F B is reduced, but when the potential is lowered to Ϫ690 mV, F A Ј is partially reduced, as shown by the shoulder at g ϭ 1.847 along with a small amount of F X at g ϭ 1.769 (data not shown). One practical issue is that the g ϭ 1.922 midfield resonance of the presumed F A Ј spectrum is merged with the g ϭ 1.938 midfield resonance of F B , making it difficult to separate the two separate spin systems. Although the narrower line widths of F B compared with F A Ј lead to a change of slope of the midfield resonance, the point of inflection is difficult to determine accurately. To isolate F A Ј, a spectrum of F B was recorded at a redox potential of Ϫ580 mV and subtracted after suitable scaling from spectra measured at lower redox potentials (the intensity of the F B resonances at a given potential was calculated using the Nernst equation). This allowed the midfield resonance in the spectrum of the F B cluster to be removed from the midfield resonance in the spectrum of F A Ј. Fig. 5A shows the midfield spectral region between 340 and 356 mT, where the intensity of the g ϭ 1.922 resonance can be measured in the absence of reduced F B . The most prominent feature is the relatively shallow slope of the derivative resonance, denoting a broader line width, and the lack of the prominent subtraction artifact. As depicted in Fig. 5A, the intensity of the midfield F A Ј resonance increases when the potential is lowered from Ϫ595 to Ϫ645 and to Ϫ690 mV. The end point of the titration was determined by thawing the most reducing sample and photoaccumulating the acceptors by illuminating the sample during freezing. The highfield resonance of F A Ј at g ϭ 1.847 is in an undisturbed spectral region, and its intensity was determined directly (the low field resonance is in a region too cluttered to allow an accurate determination). As shown in Fig. 5B the midfield feature at g ϭ 1.922 and the highfield trough at g ϭ1.847 titrate at a midpoint potential of Ϫ630 mV (n ϭ 1). This value is considerably more reducing than the midpoint potential of the F A cluster in a wild-type spinach PS I complex (E m of Ϫ520 mV).
Optical Absorbance Studies of the PsaC-C51D-PS I Complex-Electron transfer kinetics were also studied using optical absorption spectroscopy at 77 K. It was shown previously that the extent of irreversibly photoreduced F A and F B clusters strongly increases upon cooling from room temperature to 25 K (18,19), while F X and A 1 maintain reversible photoreduction at cryogenic temperatures (20). To check for possible alterations of electron transfer efficiencies on its path from P700 to terminal clusters, F A Ј and F B , photoinduced changes of P700 absorbance were determined in the near infrared upon illumination of the samples with a train of eight saturating flashes spaced at 10-s intervals. The difference in the base line prior to the first and second flashes shows that the reduction of P700 ϩ after the first flash was not complete in the wild-type PS I (Fig. 6A). Increasing the interval between the flashes did not lead to more complete reduction due to irreversible charge separation in a fraction of the centers at 77 K. The increase in the irreversible portion of the signal continued to be lower on subsequent flashes; this behavior is consistent with electron transfer at a relatively low quantum yield to F A and/or F B . In reconsti- tuted PsaC-C51D-PS I complexes, most of the irreversible signal is also generated on the first flash, and each subsequent flash results in a less significant increase of the irreversible ⌬820 nm absorption change (Fig. 6B). Independent EPR experiments showed that the illumination of the PsaC-C51D-PSI complex at 77 K followed by measurement at 9 -15 K led to no apparent F A Ј reduction but the same relative insignificant reduction of F B as a sample illuminated at 9 K (see Fig. 1). From these experiments it is inferred that the pathway of electrons in PsaC-C51D-PS I is essentially the same as in the wild-type complex. It is likely that the F A Ј cluster functions as the electron acceptor at cryogenic temperatures but that it does so in a manner that is largely EPR invisible (i.e. as an S ϭ 1 ⁄2 system with resonances around g av ϭ 1.94).
Optical absorption studies were also carried out at room temperature to determine whether F A Ј and F B function as efficient electron acceptors during single turnover flashes. The number of acceptors functioning subsequent to F X can be determined by the number of electrons that can be stabilized for more than 1 ms on the acceptor side of PS I during a series of saturating flashes (18). The pattern of the flash-induced absorbance changes in response to the first and the second flashes arises from the donation of reduced PMS to P700 ϩ (Fig. 6C). In the wild-type PS I complexes, about 80 -90% of the total amplitude is derived from a component with a lifetime of 4 -5 ms. In the PsaC-C51D-PS I complex the lifetime is longer, perhaps because of deleterious effects on the donor side due to the urea treatment. In response to the third and fourth flashes, the decay of the absorption change becomes considerably faster. In the wild-type, the components with lifetimes of 230 s (43% on the third and 52% on the fourth flash) and 2-2.5 ms (54% on the third and 43% on the fourth flash) become dominant. These components are most likely ascribed to back reactions from A 1 Ϫ and/or F X Ϫ . The remaining slower part of the decay is approximated by a base line that probably corresponds to electron donation from PMS to P700 ϩ . In the PsaC-C51D-PS I complex (Fig. 6D), the components with lifetimes of 200 s (29%) and 750 s (41%) are dominant on the third flash, and components with lifetimes of 65 s (22%) and 410 s (46%) are dominant on the fourth flash. The two fast components are assumed to result from back reactions of A 1 Ϫ and/or F X Ϫ , although the F X Ϫ back reaction is somewhat faster in the PsaC-C51D-PS I complex than in the wild-type PS I complex. The remainder of the decay is represented by a 5-ms component (ϳ20% for both the third and fourth flashes) and a base line. Were the F A Ј cluster photochemically inactive at room temperature, the back reaction of F X Ϫ would override the PMS forward reaction on the second flash. Instead, comparison of wild-type and PsaC-C51D sets of kinetic data shows that both F A Ј and F B function as electron acceptors at room temperature.
Rates of NADP ϩ Photoreduction in the C51D-PS I Complex- Table I shows rates of NADP ϩ photoreduction mediated by flavodoxin and ferredoxin in the wild-type PS I reaction center, a P700-F X core used for reconstitutions, and in PS I complexes reconstituted with PsaC, PsaC-C51D, and PsaC-C34S. The P700-F X core used in these studies, a DM-PS I particle from a PsaC-less mutant of Synechocystis sp. PCC 6803 that lacks PsaC, PsaD, and PsaE (14), showed only 13 and 22% of the rates of NADP ϩ photoreduction of a reconstituted PsaC-PS I complex when ferredoxin and flavodoxin were used as accep-tors. The PsaC-PS I complex (reconstituted with E. coli-expressed PsaC, PsaD, and PsaE) supported rates of NADP ϩ photoreduction of 300 mol/mg of chlorophyll/h, compared with 670 mol/mg of chlorophyll/h with the wild-type PS I complex. This rate is very similar to the rate supported by the PsaC-C34S-PS I complex, rebuilt with a mutant of PsaC in which the free cysteine was replaced with serine (see Ref. 11). The PsaC-C51D-PS I complex supported rates of NADP ϩ photoreduction of 283 mol/mg of chlorophyll/h and 312 mol/mg of chlorophyll/h when mediated by flavodoxin and ferredoxin, corresponding to 91 and 104% of the rates of the reconstituted PsaC-PS I complex. These values show that the mixed ligand [4Fe-4S] cluster in the F A Ј site has a minimal influence on the electron throughput from cytochrome c 6 to ferredoxin and flavodoxin. The shift of the midpoint potential of F A Ј to a value 55 mV more reducing than F B does not preclude electron transfer to physiologically relevant electron acceptors.
Tentative Identification of a High Spin [4Fe-4S] Cluster in the Free PsaC-C51D Protein-Assuming F A Ј is a low potential cluster in the aspartate site of the PsaC-C51D-PS I complex, the question of how a [4Fe-4S] cluster arises from a [3Fe-4S] cluster in the free PsaC-C51D protein becomes relevant. As shown in Fig. 7, the spectral region between 300 and 400 mT shows the resonances of the S ϭ 1 ⁄2, [4Fe-4S] 1ϩ cluster in the unmodified F A site. Just downfield of the g ϭ 4.3 resonance derived from adventitiously bound iron between 120 and 160 mT is a broad, fast relaxing spin system that is EPR-visible only at very low temperatures and high microwave powers. When the [3Fe-4S] cluster is reduced in the presence of additional iron, this resonance increases in intensity at the expense of the g ϭ 12 resonance, which is no longer present in this sample. The g ϳ 5.5 resonance may represent one or both of two low field peaks of an S ϭ 3 ⁄2, [4Fe-4S] cluster. Assuming an E/D value of 0.3, the two Kramers doublets would result in low field resonances at g eff ϭ 5.37 and 5.55. While the intensities of the g ϭ 5.5 resonances superficially appear to be low, the anisotropy of the g tensor is distributed over 300 mT for a highly rhombic system, and the low field resonance will appear small when the resonance is depicted as a first derivative signal (20  tion was made that PsaC can only be reconstituted when both clusters are present (see accompanying paper (11)). The S ϭ 1 ⁄2, [3Fe-4S] 1ϩ cluster in the modified site under oxidizing conditions and the S ϭ 1 ⁄2, [4Fe-4S] 1ϩ cluster in the unmodified site under reducing conditions were present at a ratio of 0.2, indicating that up to 80% of the spin population in the modified site may be represented by the S ϭ 3 ⁄2 [4Fe-4S] cluster in the free C51D protein. DISCUSSION In an earlier paper (8), it was shown that substitution of aspartic acid in position 51 leads to the incorporation of a high potential [3Fe-4S] cluster (E m ϭ Ϫ98 mV) in the modified site and a low potential [4Fe-4S] cluster (E m ϭ Ϫ580 mV) in the unmodified site of PsaC-C51D. On rebinding PsaC-C51D to a P700-F X core, the g values of 2.063, 1.934, and 1.880, the relatively narrow line widths, and the differential response to light at 15 K and at 298 K allowed us to identify the [4Fe-4S] cluster in the PsaC-C51D-PS I complex as F B (7). In the present work, the [3Fe-4S] cluster could not be located in the reconstituted PsaC-C51D-PS I complex. Instead, a set of resonances characteristic of F A were found that were EPR-visible at 6 K when the sample was frozen in darkness and illuminated at low temperatures. Although the highfield resonance at g ϭ 1.847, is similar to that of the reduced F A cluster in the wildtype PS I complex (under conditions where there is no magnetic interaction with F B ), the line widths are somewhat broader. Also, the temperature and power dependences imply a spin relaxation rate more rapid than F A alone in the wild-type PS I complex (and even more rapid than the F X cluster). It is therefore likely that the [3Fe-4S] cluster observed in Zhao et al. (7) was due to a population of unbound PsaC-C51D that was incompletely removed from the complex. The most significant difference in protocol is that in this study, the PsaC-C51D-PS I complex was purified by anaerobic gel filtration chromatography to remove free PsaC-C51D; in the earlier study, the sample was purified by repeated ultrafiltration over a 100-kDa cut-off membrane.
The tacit assumption is that F A Ј is a fast relaxing, S ϭ 1 ⁄2, [4Fe-4S] cluster in the aspartate-modified site. This is reasonable since there the insertion of a rubredoxin-like single iron center or a ferredoxin-like [2Fe-2S] cluster into a [4Fe-4S] or a 2[4Fe-4S] protein is probably unlikely. Also, the spin state of a rubredoxin-like [Fe] center would likely be S ϭ 5/2, and the temperature dependence of a ferredoxin-like [2Fe-2S] cluster should allow EPR detection at temperatures approaching 77 K. There is precedent for the insertion of an iron into a [3Fe-4S] cluster; for example, the [3Fe-4S] and [4Fe-4S] forms of Azotobacter vinelandii and Azotobacter chroococcum ferredoxins are interconvertible (21). The reduced [3Fe-4S] cluster in the double cluster ferredoxin III from Desulfovibrio africanus and in the single cluster ferredoxin from Pyrococcus furiosus contains an aspartic acid in the second cysteine position of the CXX-DXXCXXXCP motif and can incorporate an iron to form an S ϭ 3 ⁄2, [4Fe-4S] cluster (22). Evidence for the presence of a S ϭ 3 ⁄2, [4Fe-4S] cluster that may likewise have been derived from the insertion of iron into the [3Fe-4S] cluster resident in the modified site in the free PsaC-C14D protein has been obtained (see also Refs. 11 and 19). Our hypothesis is that the [3Fe-4S] and S ϭ 3 ⁄2, [4Fe-4S] clusters in the modified site of PsaC-C51D co-exist in solution but that the P700-F X core selectively binds only those proteins that contain two cubane clusters (see Ref. 11). Further, the spin state of the majority population of the cluster must change from S ϭ 3 ⁄2 to S ϭ 1 ⁄2 on binding and/or on reduction of F B , since the resonances of F A and F B are visible in the g ϭ 2 region.
In summary, the EPR and optical data show that electron transfer can occur to a mixed ligand [4Fe-4S] cluster at both cryogenic and room temperatures. The EPR resonances of the F A Ј cluster are visible in the g ϭ 2 region when the measurement is made at temperatures less than 6 K. The optical absorption measurements made at room temperature indicate that there are two electron acceptors functioning after F X , and the measurements made at low temperature show that the same amount of P700 ϩ is generated irreversibly as in the wild-type PS I complex under these conditions. The room temperature optical data indicate that the efficiency of electron transfer is largely unaffected by the change in the reduction potential of F A Ј (E m ϭ Ϫ630 mV) to a value 55 mV more reducing than F B (E m ϭ Ϫ575 mV). Rates of NADP ϩ photoreduction mediated by ferredoxin or flavodoxin are also relatively unaffected by an F A that is considerably more reducing than in wild-type spinach PS I (E m ϭ Ϫ520 mV). The overall large Gibbs free energy change between F X (E m ϳ Ϫ710 mV) and ferredoxin (E m ϳ Ϫ420 mV) is most likely be responsible for driving the reaction against the changed reduction potential of F A Ј.