Reconstitution of barley photosystem I with modified PSI-C allows identification of domains interacting with PSI-D and PSI-A/B.

The PSI-C subunit of photosystem I shows similarity to soluble 2[4Fe-4S] ferredoxins. Alignment analysis clearly shows that PSI-C contains an 8-residue internal loop and a 15-residue C-terminal extension that are absent in the ferredoxins. The remaining residues in PSI-C are likely to be folded in a way similar to the soluble 2[4Fe-4S] ferredoxins. Two modified PSI-C subunits lacking either the 8-residue loop or 10 residues of the C terminus were expressed in Escherichia coli and used to reconstitute a barley P700-F core prepared to specifically lack PSI-C, PSI-D, and PSI-E. As shown by EPR spectroscopy, the modified proteins carry two [4Fe-4S] clusters with characteristics similar to those of native PSI-C. Western blot analysis of the reconstituted photosystem I complexes showed that the modified PSI-C proteins bind to the P700-F core. Flash photolysis revealed that in photosystem I complexes reconstituted in the presence of PSI-D with the C-terminally deleted PSI-C, the F/F back-reaction was less efficiently restored than with wild-type PSI-C. The loop-deleted PSI-C was even less efficient. We attribute these differences to altered binding properties of the modified proteins. Comparison of reconstitutions performed in the presence and absence of PSI-D shows that the loop-deleted PSI-C is unable to bind without PSI-D, whereas the C-terminally deleted PSI-C binds only weakly with PSI-D. These results imply that the internal loop of PSI-C interacts with the PSI-A/B heterodimer and that the C terminus of PSI-C interacts with PSI-D.

The photosystem I (PS I) 1 reaction center complex mediates electron transport from plastocyanin to ferredoxin in oxygenic photosynthesis. PS I contains the primary electron donor P700 (a Chl a dimer) and the electron acceptors A 0 (Chl a), A 1 (phylloquinone), and three [4Fe-4S] centers F X , F A , and F B (1)(2)(3). The terminal acceptors F A and F B are bound to the PSI-C subunit (4,5), while the remaining electron acceptors are bound to the PSI-A⅐PSI-B heterodimer (6,7). A major unresolved question is the precise pathway of electron flow through the terminal acceptors F X and F A /F B . F B has been shown to be essential for electron transfer to ferredoxin but not for low temperature photoreduction of F A , suggesting F B to be the final acceptor in the electron transfer chain (8 -10).
The amino acid sequence of PSI-C is highly conserved among species and contains a repeated CXXCXXCXXXCP motif that is characteristic of 2[4Fe-4S] proteins. Alignment analysis of these proteins shows that, apart from an internal segment of 8 residues and 15 residues in the C terminus that are not present in the ferredoxins, PSI-C is similar to the bacterial 2[4Fe-4S] ferredoxins (5,11,12). The folding of PSI-C must be very similar to that of the soluble ferredoxins, indicating that the 8 internal residues form an extra loop. Since the ferredoxins are soluble and the PSI-C polypeptide is membrane-bound, the additional amino acids in the loop and in the C terminus may be responsible for the binding of PSI-C to the P700-F X core (12). EPR data obtained with PS I crystals and comparison with the known structure of Peptococcus aerogenes ferredoxin allow the orientation of PSI-C within the PS I complex to be limited to two possibilities by positioning the two clusters and two highly conserved helices along a fixed axis (13). Ligands for F A are provided by Cys 20 , Cys 47 , Cys 50 , and Cys 53 , while Cys 10 , Cys 13 , Cys 16 and Cys 57 provide ligands for F B (14). The positions of two [4Fe-4S] clusters within the complex are known from the crystal structure of PS I (15), but which cluster represents F A and which represents F B cannot be distinguished at 6-Å resolution. From alignment with P. aerogenes ferredoxin, the internal loop and the C terminus are predicted to extend from opposite sides of PSI-C. Therefore, determination of the orientation of the loop and the C terminus relative to PS I would yield further information on the positions of F A and F B .
We characterized the interaction and binding of PSI-C within the PS I complex by producing two PSI-C deletion mutants that lack either the 8-residue internal loop or 10 residues of the C terminus. Using a modification of the method of Parrett et al. (16), we recently established that urea treatment of barley PS I followed by a combined detergent, salt, and urea wash leads to specific dissociation of the PSI-C, PSI-D, and PSI-E subunits without affecting the remaining polypeptides (17). The resulting P700-F X core can be used for reconstitution by the addition of Escherichia coli expressed PSI-C and PSI-D in the presence of the reagents Na 2 S, FeCl 3 , and 2-mercaptoethanol, which serve to rebuild the FeS clusters of PSI-C (17)(18)(19). Using this reconstitution system, we show that the PSI-C protein lacking the internal loop interacts inefficiently with the P700-F X core, while the protein lacking the C terminus interacts inefficiently with PSI-D.

EXPERIMENTAL PROCEDURES
Plasmid Construction and Protein Isolation-The PSI-D expression clone from barley was made as described by Scott et al. (20) using the cDNA clone of Kjaerulff and Okkels (21). The PSI-C expression clone from barley has been described previously (17). The design of the two PSI-C deletion mutants is outlined in Fig. 1, where the amino acid sequence of the ferredoxin from P. aerogenes is compared with that of PSI-C from barley and Synechococcus sp. PCC 6301. To produce barley PSI-C Hor lacking the 8 residues Glu 26 -Cys 33 , appropriate oligonucleotides were used as primers in a polymerase chain reaction with barley PSI-C Hor to yield two fragments, one on each side of the loop. Ligation of the fragments using an EcoRI site resulted in a psaC Hor ⌬L gene lacking the Glu 26 -Cys 33 fragment. The psaC Hor ⌬L gene was cloned into an NcoI-and BamHI-digested Pet3d expression vector (Novagen, Stockholm, Sweden). For expression of Synechococcus sp. PCC 6301 PSI-C (PSI-C Syn ) the clone of Herman et al. (22) was transferred to Pet3d. For expression of the PSI-C Syn without the 10 C-terminal residues (PSI-C Syn ⌬C), oligonucleotides were used as primers in a polymerase chain reaction, and the resulting psaC Syn ⌬C was cloned into the Pet3d vector. Overexpression of the proteins was carried out in E. coli strain BL21 (DE3). Inclusion body disruption and purification were performed as previously reported (20,23). Sequencing the plasmid constructs by the dideoxy reaction method (24) confirmed the expected sequence of the psaC Hor ⌬L and psaC Syn ⌬C (data not shown). The term "PSI-C polypeptides" is used for the total of PSI-C Hor , PSI-C Hor ⌬L, PSI-C Syn , and PSI-C Syn ⌬C. Barley plastocyanin, ferredoxin, and ferredoxin:NADP ϩ oxidoreductase were isolated as described (25)(26)(27).
Preparation of the P700-F X Core-The PS I complex from barley (Hordeum vulgare cv. Svalöfs Bonus) was isolated by a modification of the procedure of Andersen et al. (28) as described previously (17). The plant material was homogenized in 50 mM Tricine (pH 7.5), 0.3 M sucrose, 10 mM NaCl, 5 mM MgCl 2 , and the thylakoids were isolated as described (28). After the initial solubilization with 15% decylmaltoside at 2 mg/ml Chl and subsequent centrifugation, the solubilized material was applied to a 0.4 -1 M sucrose gradient with a 2 M cushion. The sucrose solutions contained 20 mM Tricine (pH 7.5), 0.1% decylmaltoside. The gradients were centrifuged at 130,000 ϫ g for 24 h, and the lower green band containing PS I was collected. The PS I preparation was treated with 7 M urea at 4°C for approximately 2 h until only center F X of the three iron-sulfur clusters was intact as monitored by flashinduced absorbance changes. The urea-treated PS I preparation was applied to a chromatofocusing PBE column (Pharmacia, Uppsala, Sweden) equilibrated with 6 M urea, 100 mM NaCl, 25 mM imidazole (pH 7.6). Dissociated PSI-C, PSI-D, and PSI-E subunits were eluted from the column by washing for 16 h with the same buffer. A second wash was carried out using the same buffer without urea. The remainder of the PSI-C, PSI-D, and PSI-E was eluted in a third wash with 100 mM NaCl, 25 mM imidazole (pH 7.6), 0.3% decylmaltoside. As a final step, the PS I core (0.4 -0.5 mg/ml Chl) was eluted with 1 M NaCl and dialyzed against 50 mM Tris-HCl (pH 8.3). The Chl:P700 ratio of the core ranged between 150 and 200. All steps in the preparation were carried out at 4°C in green light.
Rebinding of the PSI-C and the PSI-D to the P700-F X Core-Reconstitutions were performed in 2-ml reaction volumes containing 10 g/ml Chl according to Parrett et al. (29) and Naver et al. (17). The molar ratios of the PS I subunits to P700 were determined by comparing Coomassie Brilliant Blue staining of sodium dodecyl sulfate-polyacrylamide gels with dilution series of isolated subunits and known amounts of PS I. After incubation for 16 h at 4°C the reconstituted complexes were concentrated to approximately 300 l by ultrafiltration using a YM 100 membrane (Amicon). The concentrated samples were used for the first set of analyses. Subsequently, the samples were washed by ultrafiltration five times over a YM 100 membrane (Amicon) using 50 mM Tris (pH 8.3), 0.1% Triton X-100 and used for the second set of analyses.
Electron Paramagnetic Resonance Spectroscopy-EPR spectra were acquired using a Bruker ECS 106 EPR spectrometer fitted with an ER/4102 ST resonator. Sample temperature was maintained using a liquid helium cryostat and monitored at the helium outlet 2 mm below the sample tube with a thermocouple using liquid nitrogen as a reference. Samples were chemically reduced in an anaerobic chamber by first adjusting the pH of the sample with 1 M glycine to pH 10.0 and then adding about 1 mg of solid sodium dithionite. Illumination in the EPR cavity was performed by directing the beam of a 150-W xenon arc lamp to the frozen sample through the slotted opening in the resonator. Photoaccumulation of electrons in the terminal acceptors was accomplished by passing the thawed sample through the focused beam of a 300-W tungsten lamp into a container with liquid nitrogen over the course of 1 min.
Flash-induced Absorption Change-Flash-induced absorption transients at 834 nm were measured in a reaction mixture of 200 l containing 50 mM Tris (pH 8.3), 2 mM sodium ascorbate, 60 M 2,6dichlorophenolindophenol (DCPIP), and 20 g/ml Chl. A diode laser (model 06GIC108, Melles Griot, Irvine, CA) in combination with a laser driver (model 06DCD201, Melles Griot) provided the measuring beam. The measuring beam was passed through a beam splitter with one exiting beam passing through the sample while the other beam was passed to a reference detector identical to the sample detector. The detectors were 100-mm 2 photodiodes (model 12DSI011, Melles Griot) operated with 9-V reverse bias and 500-⍀ parallel resistors. The signals from the two detectors were amplified with a Tektronix ADA400A differential preamplifier, and the difference signal was passed to a TDS 420 oscilloscope (Tektronix, Wilsonville, OR). The sample was repeatedly flashed with a xenon flash lamp (1.6-s FWHM, model L4634, Hamamatsu, Shizuoka-ken, Japan) at a frequency of 1 Hz for a total of 16 averages. The recorded absorbance changes were transferred to a computer and resolved into exponential decays by a Levenberg-Marquart nonlinear regression procedure.
NADP ϩ Photoreduction Measurements-NADP ϩ photoreduction activity was determined from the absorbance change at 340 nm in a 500-l reaction mixture containing 20 mM Tricine (pH 7.5), 40 mM NaCl, 7 mM MgCl 2 , 2 mM sodium ascorbate, 60 M DCPIP, 0.5 mM NADP ϩ , 2 M barley ferredoxin, 2 M barley plastocyanin, 0.05 M barley ferredoxin-NADP ϩ -reductase, and PS I equivalent to 2 g of chlorophyll. The samples were kept in an anaerobic cuvette and were flushed with argon before measurements to prevent electron donation from ferredoxin to oxygen. The production of NADPH was measured in an Aminco DW 2000 spectrophotometer operated in the split beam mode as described (30).
Additional Analytical Procedures-Chlorophyll was determined according to Arnon (31). SDS-polyacrylamide gel electrophoresis was carried out in 8 -25% gradient gels according to Fling and Gregerson (32). Prior to electrophoresis, the reconstituted PS I complexes were pelleted by ultracentrifugation. Western blot analysis was carried out by transferring electrophoresed proteins to nitrocellulose membranes followed by incubation with monospecific rabbit antibodies (28) and visualization using an alkaline phosphatase-conjugated secondary antibody (DAKO, Copenhagen, Denmark).

RESULTS
Characterization of the Rebinding of PSI-C to the P700-F X Core by Western Blot Analysis-Treatment with urea leads to dissociation of the PSI-C subunit from barley PS I. Reconstitution using a molar ratio of 20:20:1 of PSI-C/PSI-D/P700 in the presence of FeCl 3 , Na 2 S, and 2-mercaptoethanol resulted in rebinding of the PSI-C Hor as shown by Western blot analysis of the washed complexes (Fig. 2). Reconstitution with PSI-C Syn and both of the deletion proteins PSI-C Syn ⌬C and PSI-C Hor ⌬L in a similar experiment resulted in binding of the proteins to the P700-F X core (Fig. 2). As expected, PSI-C Syn ⌬C migrates faster in SDS-polyacrylamide gel electrophoresis than PSI-C Syn . However, the PSI-C Hor ⌬L migrates at the same rate as the full-length PSI-C Hor .
EPR Analysis of the Reconstituted PS I Complexes-The ironsulfur clusters can be reconstituted in both PSI-C Syn ⌬C and PSI-C Hor ⌬L as shown by optical and EPR spectroscopy (23). Having determined that both modified proteins bind to PS I, we set out to determine if F A and F B were photoreducible in the reconstituted PS I complexes. The complex reconstituted with PSI-C Syn ⌬C was illuminated at 15 K in the EPR cavity, and the resulting spectrum was nearly identical to that of PS I reconstituted with PSI-C Hor (Fig. 3). Features of F A were apparent at g ϭ 2.047, 1.945, and 1.867, and features of F B were found at g ϭ 2.064, 1.940, and 1.931. The two clusters are reduced in the same ratio in PSI-C Syn ⌬C as they are in wild-type PSI-C Hor . From this we conclude that the C-terminal region is not required for electron transfer from P700 to F A /F B .
Similar experiments were performed with complexes reconstituted with PSI-C Hor ⌬L (Fig. 4). When illuminated at 15 K, the low-field peaks of F A and F B were clearly visible but small in amplitude, while the midfield and high-field features were not well resolved. We next froze the sample during illumination, conditions allowing two or more electrons to accumulate in the electron acceptor system. The resulting spectrum had peaks with much increased amplitude and was very similar to that of the wild-type reconstitution, containing the features characteristic of an F A /F B interaction spectrum. The data show that PSI-C Hor ⌬L contains two functional [4Fe-4S] clusters with properties similar to F A and F B .
Flash-induced Absorption Changes-The back-reaction of the PS I complex was studied after incubation of the P700-F X core with purified PSI-D and one of the PSI-C polypeptides (PSI-C Hor /-C Syn ⌬C/-C Hor ⌬L) in the presence of FeCl 3 , Na 2 S, and 2-mercaptoethanol. The back-reaction was observed to be an [F A /F B ] back-reaction after a 5-min incubation at room temperature (data not shown) when each of the PSI-C polypeptides were present in excess (see below), but all samples were left for 16 h at 5°C before the measurements. The reconstituted PS I complexes were characterized by flash-induced ab-sorption changes at 834 nm in the millisecond time range. The extent of the reconstitution was monitored at room temperature by calculating the absorbance change from the 1-ms backreaction from F X Ϫ to P700 ϩ versus the slower back-reaction from [F A /F B ] Ϫ . The time course of the flash-induced absorption changes was determined at various molar ratios of PSI-C to P700 (Fig. 5). In all experiments, PSI-D was added in a molar ratio of 20 (PSI-D:P700). With the wild-type PSI-C Hor the reconstitution was 77% complete at a PSI-C:P700 ratio of 4. Higher ratios of PSI-C to P700 did not result in more complete reconstitution. PSI-C Syn ⌬C was saturating at a molar ratio of 7.5 (PSI-C Syn ⌬C:P700) and led to the same level of reconstitution as the full-length PSI-C Hor . PSI-C Hor ⌬L was saturating at a molar ratio of 10 (PSI-C Hor ⌬L:P700), but the loop deletion led to less complete restoration of the PS I complex (55%) than PSI-C Hor . Thus, greater amounts of PSI-C Syn ⌬C and PSI-C Hor ⌬L than PSI-C Hor are required in the reconstitution reaction relative to P700. This could be a result of a different folding FIG. 3. EPR analysis of a P700-F X core rebuilt with PSI-C Syn ⌬C and PSI-C Hor . A, reconstituted with PSI-C Syn ⌬C and PSI-D; B, reconstituted with PSI-C Hor and PSI-D; C, mock reconstitution of the P700-F X core without added proteins. The free radical signal at g ϭ 2.0 was deleted for clarity. FIG. 4. EPR analysis of a P700-F X core reconstituted with PSI-C Hor ⌬L and PSI-C Hor . A, reconstitution with PSI-C Hor ⌬L and PSI-D. The sample was illuminated at 15 K, and the spectrum was acquired. B, subsequently the sample in panel A was thawed and refrozen during illumination to allow photoaccumulation. C, reconstitution with PSI-C Hor ⌬L and PSI-D, photoaccumulated. D, mock reconstitution of the P700-F X core without added proteins, photoaccumulated. The free radical signal at g ϭ 2.0 was deleted for clarity.

FIG. 5. Flash-induced absorption changes of P700 at 834 nm in PS I complexes reconstituted with varying amounts of PSI-C.
The rebinding was analyzed by the addition of different molar ratios of the PSI-C Hor (E), PSI-C Syn ⌬C (q), and PSI-C Hor ⌬L (Ç). PSI-D was added in a molar ratio of 20 PSI-D/1 P700 in all experiments. The samples (20 g/ml Chl) were resuspended in 50 mM Tris-HCl, (pH 8.3), 2 mM sodium ascorbate, 60 M DCPIP. The extent of reconstitution was determined by fitting the traces to biphasic exponential decay curves containing phases with time constants of ϳ1 ms and Ͼ16 ms. The fraction constituting the slow decay phase was calculated in percentage of the total absorbance change.
FIG. 2. Western blot analysis before and after reconstitution of the P700-F X core with the PSI-C polypeptides. The PS I preparation before the urea treatment was included. The samples were pelleted by ultracentrifugation, and equal amounts of P700 were applied to each lane of the gel. The blots were incubated with polyclonal antibodies raised against the PSI-C Hor subunit. Chl. indicates the position of pigments released from PS I. geometry of the modified subunits. All subsequent experiments were carried out with a 20-fold molar excess of PSI-C. Under these experimental conditions, any differences in behavior between the PSI-C polypeptides is presumed to be due to differences in binding efficiency.
The PS I complex was reconstituted using molar ratios of 20:20:1 (PSI-C:PSI-D:P700). The PS I complex reconstituted with PSI-D and PSI-C Hor , PSI-C Syn , or PSI-C Syn ⌬C resulted in 88, 84, and 82% recovery of the slow decay phase, respectively (Fig. 6A, Table I). Rebuilding with PSI-C Hor ⌬L and PSI-D resulted in an incomplete reconstitution, with only 52% recovery of the PS I activity (Fig. 6A, Table I). To assay stability, the reconstituted complexes were washed with 0.1% Triton X-100 over an ultrafiltration membrane. The PS I complex reconstituted with PSI-C Hor and PSI-C Syn remained stable, showing 84 and 88% reconstitution, respectively, whereas the PS I complex reconstituted with PSI-C Syn ⌬C was unstable as shown by a reduction in the magnitude of the slow ⌬A 834 decay from 82 to 53% (Fig. 6B, Table I). The PS I complex reconstituted with PSI-C Hor ⌬L was also affected by the wash (52% before to 38% after). It should be noted that mock reconstitution of the P700-F X core in the absence of added PSI-C resulted in about 17% recovery of a slow absorbance decay. This background value has not been subtracted from the data presented in Table  I. Subtraction of the background value would obviously accentuate the differences reported.
Rebuilding the PS I complex with the PSI-C Hor and PSI-C Syn in the absence of PSI-D resulted in retention of 64 and 66% of the slow decay phases, respectively (Fig. 6C, Table I), which is less than when PSI-D was included for reconstitution (88 and 84%). The PSI-C Syn ⌬C had an enhanced ability to bind to the P700-F X core, retaining 85% of the slow phase. PSI-C Hor ⌬L did not bind to the P700-F X core in the absence of PSI-D. The [F A /F B ] Ϫ P700 ϩ back-reaction in all the PS I complexes reconstituted without PSI-D was greatly reduced by washing (Fig.  6D, Table I), indicating that PSI-C was removed by this procedure.
NADP ϩ Photoreduction Measurements-The extent to which complexes provided with PSI-C Hor , PSI-C Syn , or PSI-C Syn ⌬C and PSI-D were reconstituted as measured by NADP ϩ photoreduction after the wash protocol described above corresponded to that measured by flash analysis (Table II). PSI-C Hor ⌬L was not able to support NADP ϩ reduction after the wash. However, when the reconstituting agents were removed using a desalting column the sample exhibited NADP ϩ -reducing activity in correspondence with the flash analysis. Flash analysis of a P700-F X core reconstituted with PSI-C Hor ⌬L and PSI-D confirmed that purification of the sample using a desalting column did not alter the extent of F A /F B back-reaction as compared with the unwashed samples (data not shown). Complexes reconstituted in the absence of PSI-D and washed by ultrafiltration were unable to mediate NADP ϩ reduction (data not shown).

DISCUSSION
Deletion of neither the internal 8-residue loop nor the 10 C-terminal residues of PSI-C affected the iron-sulfur clusters F A and F B in the unbound proteins. Therefore, any differences in reconstitution efficiency exhibited by these proteins can be attributed to altered binding efficiency of the modified proteins PSI-C Syn ⌬C and PSI-C Hor ⌬L to the P700-F X core. PSI-C is a highly conserved subunit, and only nine amino acid residues are different between PSI-C from barley and Synechococcus sp. PCC 6301 (Fig. 1). In all experiments, the wild-type proteins PSI-C Hor and PSI-C Syn have given identical results.
Western blot (Fig. 2) and EPR analysis (Figs. 3 and 4) show that the P700-F X core does not contain detectable amounts of PSI-C. Nevertheless, the core exhibited 17% recovery of the slow phase of the flash-induced absorbance change at 834 nm upon mock reconstitution. This suggests that a small fraction of the slow back-reaction is due to the presence of reconstitution reagents that may function as electron acceptors. This would be in agreement with the previous observation that electron transport to NADP ϩ cannot be accomplished in the presence of the reagents employed for F A /F B reconstitution (17). EPR analysis at 15 K of the P700-F X core reconstituted with the PSI-C Hor ⌬L showed features of F A and F B , but the electron transport was inefficient at 15 K (Fig. 4). However, when the sample was subjected to photoaccumulation, the spectrum was  Table I  nearly identical to that of PS I reconstituted with PSI-C Hor and contained resonances typical of an interaction spectrum of F A and F B . This is in contrast to the PSI-C Syn ⌬C reconstitution, which gives a spectrum similar to wild-type reconstitution when illuminated at 15 K. In order to measure NADP ϩ reduction, it is essential to remove the reconstitution reagents, and this was generally accomplished by ultrafiltration. However, in the case of PSI-C Hor ⌬L significant loss of reconstitution efficiency occurred as determined by flash photolysis, and no NADP ϩ reduction could be detected. When a milder desalting procedure was employed, there was no impairment of the reconstitution, and NADP ϩ reduction was detected. This shows that PSI-C Hor ⌬L does bind to PS I in the presence of PSI-D, but the binding is weak. The measured NADP ϩ -reducing rates are low compared with native PS I but correspond well to previously observed values in reconstitution experiments (17). Washing in the presence of Triton X-100 causes a loss of activity possibly due to loss of PSI-F and thereby an inefficient interaction with plastocyanin (33,34). This effect of Triton washing is also seen with native PS I (17). In all cases, the extent of NADP ϩ reduction mirrored the extent of reconstitution observed by flash analysis. Therefore, the data indicate that PS I reconstituted with the mutant PSI-C polypeptides functions identically to wild-type proteins in ferredoxin reduction. Further flash experiments to determine the kinetic constants of ferredoxin reduction will be needed to address this question in detail.
The C-terminally truncated PSI-C Syn ⌬C restored the [F A / F B ] Ϫ P700 ϩ back-reaction nearly as well as PSI-C Hor and PSI-C Syn , but the complex was unstable. When the PS I complex was washed intensively, the recovery of reconstituted PS I decreased to 53%. Using PSI-C Hor and PSI-C Syn the reconstituted PS I complexes were resistant to the wash procedure. Thus the C terminus is not essential for correct positioning of PSI-C on the P700-F X core, but it is important for the stability of the resulting PS I complex. In reactions with and without PSI-D, the reconstitution behavior of PSI-C Syn ⌬C was similar, in contrast to the significant effect of PSI-D observed with wild-type PSI-C. This suggests that the C-terminal domain that was removed in PSI-C Syn ⌬C includes the major site of interaction with PSI-D. PSI-C Hor , PSI-C Syn , and PSI-C Syn ⌬C were almost completely removed by washing of the PS I complexes rebuilt in the absence of PSI-D. This is in agreement with previous results showing PSI-D as essential for the stabilization of PSI-C in PS I (17,19). However, since the activity of the PS I complex rebuilt with PSI-C Syn ⌬C in the presence of PSI-D was not entirely destroyed by the extensive wash, there must be some additional sites of interaction with PSI-D that remain on PSI-C Syn ⌬C.
Rebuilding the PS I complex with the loop deletion mutant in the presence of PSI-D was achieved, but it was incomplete. Furthermore, while the other proteins tested interacted weakly with PS I in the absence of PSI-D, PSI-C Hor ⌬L was unable to interact functionally with PS I under these conditions. This suggests that the PSI-D interaction provides the only functional link between the PS I complex and PSI-C Hor ⌬L. The 8-residue loop, which was deleted in this mutant, must therefore contain sites of interaction between PSI-C and the P700-F X core. The sites of interaction between PSI-C and the PS I polypeptides have been shown to be limited to PSI-A, PSI-B, PSI-D, and PSI-E (35, 36) 2 in the native PS I complex. The loop was shown not to be involved in binding of PSI-D, and reconstitution can be performed completely in the absence of PSI-E (this paper and Refs. 17 and 19); therefore, the binding activity of the loop must involve interaction with PSI-A and PSI-B. Studies of Rodday et al. (36,37) suggest that arginine residues in the conserved domain in PSI-A and PSI-B that binds F X are important for the interaction with PSI-C. The 8-residue internal loop contains two negative charges that may be responsible for the specific interaction with the arginines. To interact with PSI-A and PSI-B and most likely with residues very close to the ligands of F X , we propose that the loop of PSI-C must face the thylakoid membrane, whereas the C terminus providing the binding site for PSI-D is likely to be oriented toward the stroma. None a 0 PSI-C Hor ϩ PSI-D a 52/48 PSI-C Syn ϩ PSI-D a 52/53 PSI-C Syn ⌬C ϩ PSI-D a 24/25 PSI-C Hor ⌬L ϩ PSI-D b 23/25 PSI-C Hor ⌬L ϩ PSI-D a 0/0 a Reconstitutions subjected to ultrafiltration. b Reconstitution desalted on a column.