Biochemical and Biophysical Characterization of Photosystem I from Phytoene Desaturase and -Carotene Desaturase Deletion Mutants of Synechocystis Sp. PCC 6803 EVIDENCE FOR PsaA- AND PsaB-SIDE ELECTRON TRANSPORT IN CYANOBACTERIA*

From the ‡Central Research and Development, Experimental Station, E. I. du Pont de Nemours & Co., Wilmington, Delaware 19880-0173, ¶UMR 7141 CNRS/University Paris 6, Institut de Biologie Physico-Chimique, 13 rue Pierre et Marie Curie, 75005 Paris, France, the Department of Biochemistry and Molecular Biology and the **Department of Chemistry, Pennsylvania State University, University Park, Pennsylvania 16802


In photosystem I, oxidation of reduced acceptor A 1
؊ through iron-sulfur cluster F X is biphasic with halftimes of ϳ5-30 ns ("fast" phase) and ϳ150 -300 ns ("slow" phase).Whether these biphasic kinetics reflect unidirectional electron transfer, involving only the PsaA-side phylloquinone or bi-directional electron transfer, involving both the PsaA-and PsaB-side phylloquinones, has been the source of some controversy.Brettel (Brettel, K. (1988) FEBS Lett.239, 93-98) and Joliot and Joliot (Joliot, P., and Joliot, A. (1999) Biochemistry 38, 11130 -11136) have attributed to nearby carotenoids electrochromic band shifts, accompanying A 1 reduction, centered at ϳ450 and 500 -510 nm.As a test of these assignments, we separately deleted in Synechocystis sp.PCC 6803 the genes that encode phytoene desaturase (encoded by crtP (pds)) and -carotene desaturase (encoded by crtQ (zds)).The pds ؊ and zds ؊ strains synthesize phytoene and -carotene, respectively, both of which absorb to shorter wavelength than ␤-carotene.Compared with wild type, the mutant A 1 ؊ (FeS) ؊ A 1 (FeS) ؊ difference spectra, measured in cells and photosystem I complexes, retain the electrochromic band shift centered at 450 nm but show a complete loss of the electrochromic band shifts centered at 500 -510 nm.Thus, the latter clearly arise from ␤-carotene.In the wild type, the electrochromic band shift of the slow phase (centered at 500 nm) is shifted by 6 nm to the blue compared with the fast phase (centered at 506 nm).Thus, the carotenoid pigments acting as electrochromic markers during the fast and slow phases of A 1 ؊ oxidation are different, indicating the involvement of both the PsaA-and the PsaB-side phylloquinones in photosystem I electron transport.
The photosystem I (PS I) 1 reaction center is a multisubunit intrinsic membrane protein complex that functions as a lightdriven plastocyanin/ferredoxin oxidoreductase.At the core of PS I is a heterodimer of two highly homologous subunits, PsaA and PsaB, which contain the redox components involved in the first steps of photoinduced electron transfer.Light energy initiates primary electron transfer such that the oxidizing equivalent resides on P700, a heterodimer of chlorophyll a (Chl a) and its C13 2 -epimer Chl aЈ (1)(2)(3).In chronological order, the electron acceptors are as follows: A 0 (Chl a), A 1 (phylloquinone or vitamin K 1 ), F X (a [4Fe-4S] cluster), and the terminal electron acceptors F A and F B (both [4Fe-4S] clusters).All are bound by PsaA and PsaB except for F A and F B , which are bound to subunit PsaC.The PS I complex also includes two additional small extrinsic polypeptides, PsaD and PsaE, which mediate the docking of ferredoxin (4), the soluble electron carrier responsible for the reduction of NADP ϩ , and PsaF, which facilitates the binding of plastocyanin, the mobile electron carrier responsible for the rereduction of the oxidized donor P700 ϩ (5-7).
In addition to these redox cofactors, carotenoids and other Chls a are also bound within the PS I complex.A 2.5 Å x-ray crystallographic structure of the trimeric PS I reaction center from the thermophilic cyanobacterium Thermosynechococcus elongatus was recently reported (2).Resolved in each monomer of the crystal structure were 22 carotenoids, 96 chlorophylls a, 2 phylloquinones, and 3 [4Fe-4S] clusters.The two phylloquinone molecules are symmetrically located with respect to the C 2 symmetry axis of the protein heterodimer.Whether the PsaB-side quinone as well as the PsaA-side quinone are active for electron transfer has been the subject of considerable recent debate.Optical studies have shown that the reoxidation of A 1 Ϫ is biphasic (t1 ⁄2 of 5-30 ns and 150 -300 ns) (8,9), but the assignments and interpretations of these two kinetic phases are still controversial (10).To explain the biphasic kinetics, Brettel and co-workers (8,10,11) originally proposed a model where rapid equilibration between A 1 and F X would give rise to the faster phase of A 1 Ϫ reoxidation, whereas the depletion of this quasi-equilibrium by electron transfer to F A /F B would give rise to the slower phase.The same phylloquinone was proposed to participate in both phases.Joliot and Joliot (12) showed, however, that the relative amplitudes of the fast and slow phases were insensitive to the membrane potential, inconsistent with the attribution of the fast phase to the formation of equilibrated state, A 1 Ϫ F X ^A1 F X Ϫ , an electrogenic electron transfer step (13).They suggested that the biphasic kinetics could come from either of the following two models: (i) a single phylloquinone molecule in two different conformational states having different A 1 Ϫ reoxidation rates, or (ii) two different phylloquinones, one on each side of the reaction center and oxidized by F X at different rates.The former would constitute a unidirectional pathway, and the latter would be bidirectional.The x-ray structure reveals no obvious differences between the binding pockets of the two phylloquinones nor a difference in the phylloquinone-F X edge-to-edge distances (6.8 Å).Recently, Guergova-Kuras et al. (14), using site-directed mutants of Chlamydomonas reinhardtii, provided the first experimental support for the participation of both the PsaA-side and PsaBside quinones in electron transfer to F X .Site-directed replacement of the tryptophan directly -stacked to the PsaA-side phylloquinone with phenylalanine (PsaA-W693F) selectively slowed the slower phase of A 1 Ϫ reoxidation, whereas the same mutation of the corresponding tryptophan on the PsaB side (PsaB-W673F) selectively slowed the faster phase.Xu and coworkers (10), using the cyanobacterium Synechocystis sp.PCC 6803, confirmed the room temperature optical results on the PsaA-W693F and the PsaB-W673F mutations, but although transient EPR studies provided experimental evidence for PsaA-side electron transfer, there was no indication of PsaBside electron transfer.Similarly, Cohen et al. (15) constructed site-directed mutants in which the axial methionine ligands to chlorophylls eC-A 3 and eC-B 3 were changed to leucine, and they found an effect on the transient EPR signal only on the PsaA-side mutant.They concluded that electron transfer occurs primarily along the PsaA branch in cyanobacteria, although a minor fraction of electron transfer along the PsaB branch could not be excluded.Fairclough and co-workers (16), however, constructed a site-directed replacement with histidine of a methionine ligand to chlorophyll eC-A 3 in C. reinhardtii, PsaA-M684H, that still allows photoautotrophic growth despite having blocked electron transfer to the PsaA branch phylloquinone.This observation implies that the PsaB branch is active for electron transfer and that its activity is sufficient to support photoautotrophic growth.This mutation and the corresponding PsaB-side ligand replacement, PsaB-M664H, were constructed by Fairclough et al. (16) and by Ramesh et al. (17).Both mutations block or slow electron transfer from A 0 Ϫ to A 1 on their respective branches, implying that in C. reinhardtii both PsaA and PsaB branches are active in electron transport.
All of the carotenoids in the electron density map derived from the x-ray crystal structure of PS I were modeled as ␤-carotene (2).Seventeen were in an all-trans configuration.Two were modeled as 9-cis; another showed a 13-cis configuration, and two were shown with 2-cis double bonds: 9,9Ј-di-cis and 9,13Ј-di-cis (for numbering, see Fig. 1 and Ref. 18).The carotenoids are arranged in six clusters with each cluster containing 2, 3, or 6 carotenoids.(see Fig. 9 of Ref. 3).In cyanobacteria and in other carotenoid-producing organisms, carotenoids are synthesized from isoprenoid precursors.The first step specific for carotenoid biosynthesis is the head-to-head-condensation of geranylgeranyl pyrophosphate (see Fig. 1).This reaction is catalyzed by phytoene synthase and forms phytoene, a C 40 carotenoid with three conjugated double bonds.Phytoene desaturase (encoded by crtP (pds)) catalyzes the double dehydro-genation of phytoene to form -carotene, which possesses seven conjugated double bonds.Another enzyme, -carotene desaturase (encoded by crtQ (zds)), is responsible for the dehydrogenation of -carotene to form lycopene, a carotenoid with 11 conjugated double bonds.Lycopene cyclase catalyzes the formation of ␤-ionylidene rings at both ends of lycopene to form ␤-carotene, still with 11 conjugated double bonds but with the terminal double bonds being part of the rings.␤-Carotene can then be hydroxylated in the rings by specific ␤-carotene hydroxylases to form ␤-cryptoxanthin, isocryptoxanthin, and/or zeaxanthin (Fig. 1).
Brettel (11) and Joliot and Joliot (12) proposed that the reduction of the PS I phylloquinone electron acceptor, A 1 , produces electrochromic band shifts centered at 450 and 500 -510 nm of one or more nearby carotenoid molecules.According to the x-ray crystal structure, several carotenoids with different orientations and configurations and modeled as ␤-carotene are indeed present in the immediate vicinity of each of the two phylloquinone molecules (Fig. 2).If both branches of the electron transport chain were active, then each of the phylloquinones should show electrochromic shifts of two different sets of carotenoids, the absorption spectra of which might be slightly different.To examine this possibility, we have altered the carotenoid biosynthetic pathway to replace ␤-carotene with carotenoids having shorter -conjugation, the absorbance bands of which are blue-shifted compared with ␤-carotene.This replacement should shift the carotenoid electrochromism to shorter wavelengths, allowing the WT carotenoid band shifts associated with A 1 Ϫ formation to be clearly identified.The results presented here show that associated with each of the two phases of the A 1 Ϫ reoxidation is the loss of electrochromic band shifts on two different sets of carotenoid molecules with different absorbance spectra.These spectral differences are consistent with the differences in configuration of the carotenoids close to phylloquinones on the PsaA and PsaB branches as seen in the x-ray crystal structure.These results therefore support a model for electron transport in PS I in which both the PsaA-and PsaB-side phylloquinones participate in electron transfer to F X .
FIG. 1.The biosynthetic pathway leading to ringed carotenoids in cyanobacteria starting from geranylgeranyl pyrophosphate (GGPP).The molecules are depicted in the all-trans configuration.The standard numbering is given in the last structure.

Construction of Mutants-
The genes for phytoene desaturase (pds) and -carotene desaturase (zds) were deleted separately in the genome of WT Synechocystis sp.PCC 6803 and replaced by a kanamycin resistance cassette.A PCR copy of the gene to be deleted (pds or zds) that included 500 bases upstream and downstream of the gene was ligated onto a pGEM-T plasmid vector (Promega A1360) using T4 DNA ligase and amplified in XL1-Blue Escherichia coli competent cells (Stratagene 200249).In this resulting plasmid, the first and last 50 bases of the gene together with the ligated pGEM-T plasmid vector between them were copied by PCR using primers designed to give unique restriction enzyme sites to the termini of the PCR product.The PCR product was digested with the appropriate restriction enzymes and then ligated using T4 DNA ligase to a kanamycin resistance cassette with the same sticky ends as the PCR product.This final plasmid construct was then amplified in XL1-Blue E. coli competent cells and used to transform the WT Synechocystis sp.PCC 6803 strain.Transformation of Synechocystis sp.PCC 6803 WT strain was as described in Nixon et al. (19).Successful transformants were selected on Petri plates containing BG-11 medium, 5 mM glucose, 50 g/ml kanamycin and grown in very dim light (ϳ0.1 mol photons m Ϫ2 s Ϫ1 ) at 30 °C.Kanamycin-resistant colonies representing successful transformants were replated seven times in BG-11 plates containing 5 mM glucose and 200 g/ml kanamycin to obtain full segregation.Deletion mutations were confirmed by extracting total pigments from the cells and analyzing for the complete absence of ␤-carotene by HPLC (see below).The cells were also unable to produce ␤-carotene under nonselective conditions (without kanamycin).
PS I Preparation-WT Synechocystis sp.PCC 6803 cells were grown in liquid BG-11 containing 5 mM glucose medium as described (20 -22).The mutants were grown in BG-11 containing 5 mM glucose in very dim light (ϳ0.1 mol photons m Ϫ2 s Ϫ1 ). Isolation of the thylakoid mem- branes from these cells was performed as described previously (20,21).The thylakoid membranes were suspended at a concentration of 1 mg/ml chlorophyll in Buffer A (25% glycerol, 20 mM CaCl 2 , 5 mM MgCl 2 , 50 mM MES, pH 6.0).After addition of a one-ninth volume of 10% (w/v) n-dodecyl ␤-D-maltoside (␤-DM) (final concentration of 1% (w/v)), the suspension was stirred for 20 min and then centrifuged at 55,000 rpm in a 70Ti rotor (Beckman-Coulter) for 30 min.The supernatant was then loaded onto an ϳ500-ml weak anion exchange Toyopearl DEAE-650S (Supelco) column pre-equilibrated with Buffer B (Buffer A containing 0.03% ␤-DM) and washed with Buffer B containing 19 mM MgSO 4 .Elution of PS I complex from the column was performed by using 500 ml of Buffer B, forming an increasing salt gradient from 19 to 30 mM MgSO 4 .At the end of the gradient, the salt concentration was increased to 50 mM MgSO 4 .The eluent fractions containing PS I were pooled and concentrated in a Millipore concentrator utilizing a YM 100 Diaflo ultrafiltration membrane.The concentrated samples were then desalted by passing them through a gel filtration column (Econo-Pac 10 DG, Bio-Rad) pre-equilibrated with Buffer B. The samples were then frozen in liquid N 2 and stored at Ϫ80 °C until use.
Pigment Analysis-The chlorophyll content of the cells was determined by extraction with 100% methanol and using the 79.24 ml/mg extinction coefficient of Lichtenthaler ( 23) at 665 nm.For total pigment analysis, PS I complexes were extracted with a 50:50 acetone/methanol mixture under dim light.The mixture was centrifuged for 2 min (10,000 ϫ g), and a 50-l aliquot of the supernatant was immediately filtered (0.45 m, Millex-LH PTFE) and injected onto a reverse phase HPLC column (Zorbax-ODS, 4.6 mm x 15 cm, Rockland Technologies) equilibrated with methanol/ethyl acetate (68:32, v/v).The separation was performed in isocratic mode using the same solvent system (24) at a flow rate of 0.25 ml/min for 60 min at room temperature.The relative chlorophyll and carotenoid concentrations were determined by measuring the area of their respective peaks in the HPLC and dividing by the extinction coefficients according to Giorgi et al. (24).
EPR Spectroscopy-X-band (9.4 GHz) EPR studies were performed using a Bruker ECS-106 spectrometer and a standard mode resonator (ST 8615) equipped with a slotted port for light entry.Cryogenic temperatures were maintained with a liquid helium cryostat and an ITC-4 temperature controller (Oxford, UK).The microwave frequency was measured with a Hewlett-Packard 5340A frequency counter, and the magnetic field was calibrated using ␣,␣Ј-diphenyl-␤-picrylhydrazyl as the standard.Sample temperatures were monitored by a calibrated thermocouple located 3 mm beneath the bottom of the quartz sample tube and referenced to liquid N 2 .The spectra of F A Ϫ /F B Ϫ in WT and mutant PS I complexes were obtained as described previously (25).Glycine buffer (1 M) was added to bring the pH to 10.0 (100 mM final concentration), and sodium dithionite was added to a final concentration of 50 mM.The sample was incubated in darkness on ice for 30 min, after which the sample was frozen in liquid N 2 and the EPR spectrum was recorded at 15 K. Q-band (34 GHz) EPR studies were performed using a Bruker ER 5106 QT-W1 resonator equipped with a port for sample illumination.Cryogenic temperatures were maintained with a Bruker ER4118CV liquid nitrogen cryostat and a Bruker ER4121 temperature control unit.The microwave frequency was measured with a Bruker ER035M NMR gaussmeter.The magnetic field was calibrated with ␥,␥-bisdiphenylene-␤-phenylallyl complexed 1:1 with benzene, which has no detectable g anisotropy at 34 GHz.Photoaccumulation of A 1 Ϫ and F X Ϫ in all PS I samples was carried out as described in Ref. 26.The pH of the sample was adjusted to 10.0 with glycine buffer (1 M) (final concentration of 100 mM), and sodium dithionite was added to a final concentration of 50 mM.After incubation for 20 min in the dark on ice, the sample was frozen in liquid nitrogen and placed in the microwave cavity.A pre-illumination spectrum was then recorded at 205 K.The sample was then illuminated with a 20-milliwatt He-Ne laser for 40 min at 205 K.The laser was then turned off, and the light-induced spectrum was recorded at the same temperature.The light-minus-dark difference spectrum at 205 K showed primarily A 1 Ϫ , as signals from F X Ϫ and the other iron-sulfur clusters were only detectable at very low temperatures (less than 20 K).
Optical Spectroscopy-Charge recombination kinetics from F A Ϫ /F B Ϫ to P700 ϩ nm were measured at 435 nm in a laboratory-built flash-detection spectrophotometer similar to that originally described by Joliot et al. (27).PS I complexes at a Chl a concentration of 14 g/ml were suspended in 25 mM Tris-HCl (pH 8.3), 10 M sodium ascorbate, plus 4 M DCPIP (or 250 M sodium ascorbate without DCPIP), and 0.03% ␤-DM.Under these conditions, P700 is fully reduced, and all electron acceptors are oxidized prior to excitation (26).Charge separation was induced by a saturating flash (30 mJ) provided by a linear flash lamp pulsed dye laser (model

FIG. 2. The electron transfer cofactors from the x-ray crystallographic structure of PS I from T. elongatus at 2.5 Å resolution (2).
The figure shows the clusters of carotenoids near the phylloquinones PhyQ A and PhyQ B .The ␤-carotene fam carotenoids are color-coded as follows: red, cis; yellow, all-trans.The phytyl tails of the chlorophylls have been omitted for simplicity.Note the greater number of carotenoid cis double bonds on the PsaA side as compared with the PsaB side.The coordinates of the structure were taken from the Brookhaven Protein Data Bank (entry 1JBO).
LFDL-3, Cynosure Inc.) using rhodamine 590 perchlorate (Exciton).The reduction of P700 ϩ by recombination was measured at 435 nm using detecting flashes provided by an EG & G FX199U flash lamp and passed through a Jobin-Yvon HL300 monochromator.
Measurements of the kinetics of A 1 Ϫ oxidation in the nanosecond to microsecond time scale were performed on whole cells and isolated PS I complexes in a flash-detection spectrophotometer as described previously (14,28).Whole cells were suspended in 20 mM HEPES (pH 7.2), 20% (w/v) Ficoll, and 5 M carbonyl cyanide (4-(trifluoromethoxy)phenyl)hydrazone.The PS I complexes were suspended at a concentration of 40 g/ml under the same conditions as for charge recombination, except that the sodium ascorbate and DCPIP concentrations were 10 mM and 40 M, respectively.Decay-associated spectra of the kinetic phases were derived from a global multiexponential fit of the kinetic components obtained at each wavelength, using the program MEXFIT (29).Charge separation was induced by a 5-ns (full width at halfmaximal) light pulse at 700 nm using a Nd:YAG pumped LDS 698 dye exciting ϳ70% of the reaction centers.Absorbance changes were followed from 5 ns to 20 s using detecting flashes provided by an OPO Continuum (OPO Panther, type II) from 300 to 540 nm, frequency doubled for wavelengths less than 410 nm.For detection in the UV, a fluorescent glass (Sumita Optical Glass, Lumilass-B) was used to convert the UV photons to the blue where the blocking filters perform better.
Measurements of the variable fluorescence yield in whole cells were used to quantify functional PS II relative to the WT using a Joliot-type flash detection spectrophotometer as described previously (30).Cells were suspended in 50 mM HEPES/KOH (pH 7.5) at a concentration equivalent to an OD 730 ϭ 0.9 and were incubated for 10 min in the dark in 0.3 mM p-benzoquinone and 0.3 mM potassium ferricyanide prior to the start of the experiment.The cells were then incubated in the dark at room temperature in the presence of 20 M DCMU and 20 mM NH 2 OH, the latter added ϳ30 s before the start of the measurement.The variable fluorescence yield of PS II was then measured by detecting flashes following each of 20 saturating light flashes (18 Hz).

RESULTS
Gene Knock-out Mutations-The crtP (pds) and crtQ (zds) genes were deleted independently in Synechocystis sp.PCC 6803 as described under "Materials and Methods."The pds Ϫ and zds Ϫ mutants were difficult to select as the gene deletions caused the mutants to be acutely light-sensitive.This photosensitivity was most probably due to the inability of the carotenoids with shorter conjugated chains to quench chlorophyll triplet states, which in the presence of light and oxygen lead to the formation of toxic singlet excited O 2 ( 1 ⌬ g ) (31).Consequently, to drive genomic segregation to completion, the pds Ϫ and zds Ϫ mutants were selected on BG-11 containing 5 mM glucose plates prepared with fresh 200 g/ml kanamycin and grown under very dim light (ϳ0.1 mol photons m Ϫ2 s Ϫ1 ).The reduced light intensity resulted in a very low rate of cell growth.These knock-out mutations also resulted in the complete loss of the PS II reaction centers in both mutant strains, as indicated by the absence of fluorescence induction in the presence of DCMU and hydroxylamine (see "Materials and Methods").Several failed attempts at the biochemical isolation of PS II complexes from thylakoid membranes (21) also indicated the absence of PS II.The absence of PS II is likely a consequence of the inability to assemble PS II complexes, where only phytoene or -carotene is available as a replacement for ␤-carotene.
The PS I concentration of the mutants, on a per chlorophyll basis, was close to that of the WT strain, as equivalent loading of detergent-solubilized thylakoid membranes (based on chlorophyll content) onto the DEAE column leads to similar recoveries of PS I complexes.However, the amount of chlorophyll isolated from the pds Ϫ and zds Ϫ strains (chlorophyll/cell) was approximately one-half that of the WT strain from cultures of the same volume (18 liters) and optical density (OD 730 ϳ 1.5).In Fig. 3a, the absorption spectrum of the WT PS I complex is compared with that of the pds Ϫ and zds Ϫ strains following normalization to the chlorophyll Q y absorption band.Particu-larly noticeable in the absorption spectra of the mutants is the loss of absorption in the 450 -525 nm region, indicating the loss of ␤-carotene biosynthesis in the mutants.Consistent with this spectrum and the loss of ␤-carotene and its derivatives, the PS I complexes isolated from the pds Ϫ and zds Ϫ strains are bluer in color than their WT homologues.In the zds Ϫ strain spectrum (Fig. 3a), however, there is an enhanced absorption in the 350 -425 nm region, which is attributed to the presence of the 7-conjugated double bond carotenoid -carotene and its derivatives (Fig. 3b).-Carotene absorbs maximally at ϳ400 nm in solvents with low polarizability (32).The pds Ϫ cores, on the other hand, show an absorbance spectrum that is essentially identical to that of pure Chl a, indicating the lack of synthesis of both ␤-carotene and -carotene.HPLC analysis (see below) confirmed these observations.Instead of ␤-carotene or -carotene, the pds Ϫ strain produces the 3-conjugated double bond carotenoid phytoene.Phytoene absorbs maximally at ϳ280 nm in solvents with low polarizability (32).In the red region, the maximum absorption corresponding to the Q y transitions of the chlorophylls in the WT PS I complexes peaks at 680.6 nm.For the -carotene desaturase and phytoene desaturase mutants, the maximum absorption of the Q y transition was shifted to 678.8 and 679.0, respectively.The possible reasons for this will be discussed below.Fig. 3b shows the spectral subtractions of the Q y -normalized absorbance spectra, WT minus pds Ϫ and zds Ϫ minus pds Ϫ .The former yields the absorbance spectra of ␤-carotene, together with its isomers and hydroxy derivatives, that are bound to the WT PS I complex, whereas the latter, that of -carotene and its isomers, that are bound to the zds Ϫ PS I complex.The maximum height of the zds Ϫ carotenoid absorbance spectrum is ϳ35% that of the WT carotenoid absorbance spectrum.As the extinction coefficients of ␤and -carotene are practically equivalent, the loss of absorption indicates that some of the carotenoid-binding sites in the PS I complexes of the zds Ϫ strain are devoid of carotenoids.This finding is also confirmed by the HPLC analysis below.Also noticeable in these spectral subtractions is the difference in shape of the cumulative absorbance of the carotenoids in the WT (mainly ␤-carotene) versus that in the zds Ϫ strain (all -carotene).The absorbance peaks of -carotene (385, 406, and 432 nm) in the spectral subtractions are sharper and better defined, whereas those of ␤-carotene (434, 468, 496 nm) are less so.This is mainly because linear carotenoids tend to have sharper absorption peaks than carotenoids with cyclic ends (32).
Pigment Analysis-The PS I complex in the x-ray crystal structure is a trimer with each monomer binding 96 Chls a and 22 carotenoid molecules.Fig. 4a shows an HPLC of the total pigment extract of WT PS I complex run on a reverse phase column.Three peaks for carotenoids are observed together with a peak for Chl a.The absorbance maxima of all of the carotenoid peaks are at 450 nm (not shown), suggesting that the different carotenoids found in PS I all have the same extent of carbon-carbon -electron conjugation.Because the method used for elution was reverse phase HPLC, we expected that the most nonpolar molecules will elute last.We assigned the peak eluting at ϳ30 min to ␤-carotene not only because of its position in the elution profile but also because of its relative abundance compared with the other carotenoid peaks.This assignment was also confirmed by the observation of the same retention time and absorption spectrum for a ␤-carotene standard (not shown).Noticeable in this peak is a shoulder at ϳ32 min that we assigned to several forms of cis-␤-carotene.The absorbance spectrum of the sample eluting at that time showed a "cisband," a feature present at ϳ140 nm to the blue of the red-most absorption band of carotenoids.Because of the C 2h symmetry present in all-trans-carotenoids, this transition was absent or less prominent in all-trans-carotenoids but became more visible in cis-carotenoids because trans-to-cis isomerization breaks the C 2h symmetry (33,34).We assigned the peak eluting at ϳ10 min to all-trans-zeaxanthin, and this was confirmed by an identical retention time and absorption spectrum of a standard zeaxanthin sample (not shown).Zeaxanthin (Fig. 1, bottom) is isoelectronic with ␤-carotene and contains two hydroxyl groups that are present in each of the terminal ␤-ionylidene rings, which makes it relatively more polar causing it to elute much earlier than ␤-carotene.The peak eluting at ϳ17 min was assigned to a mixture of cis-␤-cryptoxanthin and/or isocryptoxanthin, which were also found in substantial amounts in PS I preparations of Synechococcus elongatus (35).These two carotenoids only differ from each other by the position of the hydroxyl group on the ring (Fig. 1, bottom) and were expected to have identical absorption spectra (18,36).For simplicity, the mention of ␤-carotene fam (␤-carotene family) will henceforth represent all the carotenoids (zeaxanthin, ␤-cryptoxanthin, and isocryptoxanthin and their cis-trans isomers) "isoelectronic" (i.e. the same number of -conjugated double bonds) with ␤-carotene.
Fig. 4b shows the HPLC chromatogram of the total pigment extract for the zds Ϫ PS I complex.The absorption spectra of all the carotenoid peaks have their absorption maxima at ϳ400 nm (not shown), which indicates that the carotenoids in this complex are isoelectronic.From the biosynthetic pathway of carotenoids for cyanobacteria shown in Fig. 1, it is most likely that these carotenoids are isomeric -carotenes, which would accumulate in the absence of the zds gene.However, the possibility of a derivative form of -carotene with cyclohexenyl or ␤-ionylidene rings cannot be completely discounted as lycopene cyclase is still present in this mutant and could use -carotene as a substrate.These derivatives of -carotene would also show absorption spectra similar to acyclic or linear -carotene because the extent of their -electron conjugation would not be affected by the cyclization of the ends of the molecule.For simplicity, the mention of -carotene fam (-carotene family) in this paper represents all the isoelectronic carotenoids with 7 -conjugated double bonds produced in this mutant.The main peak eluting at ϳ29 min is all-trans--carotene, and the small shoulder before that eluting at ϳ27 min is a cis--carotene.A minor peak eluting at ϳ33 min is assigned to another form of -carotene, perhaps another cis, where the cis configuration is at or near the end of the -electron conjugation.The cis-band in this form has a lower intensity than in central cis-carotenoids (34).
For the pds Ϫ mutant complex (Fig. 4c), only one distinct pigment peak is present in addition to Chl a.The peak eluting at ϳ30 min was assigned to phytoene (maximum absorbance at 288 nm) (not shown).The earlier peak at ϳ5 min is due to the acetone/methanol solvent front in which the pigment extracts were dissolved before injection into the HPLC.
Table I summarizes the approximate stoichiometry of the chlorophyll and carotenoids extracted from the WT and mutant PS I complexes.The number of carotenoids present here per WT complex (ϳ26, normalized to 96 chlorophylls) is slightly greater than the 22 reported in the Thermosynechococcus elon-  gatus crystal structure (2), possibly reflecting differences in the preparations.Noticeable in Table I is the reduced amount of carotenoid in the mutant PS I complexes.Because the native protein evolved binding ␤-carotene fam , -carotene fam and phytoene were not likely to produce a tight fit in the carotenoidbinding sites as the more saturated carotenoids were nonplanar at the ends because of more sp 3 -hybridized carbon centers with tetrahedral geometry and greater rotational degrees of freedom.The carotenoids are therefore more loosely bound and may either not have bound in vivo or may have been lost during the purification process.X-and Q-band EPR-According to the x-ray crystallographic structure of the PS I reaction center of T. elongatus (2), several carotenoids are located in the immediate vicinity of each of the phylloquinones of PS I (Fig. 2).To check if a change in the number and structure of carotenoids in PS I has had an effect on the EPR spectra of the electron acceptors, X-band and Qband EPR spectra were recorded.X-band (9.4 GHz) EPR spectroscopy was performed on the WT and mutant (pds Ϫ and zds Ϫ ) PS I complexes to examine iron-sulfur clusters F A Ϫ /F B Ϫ .The pH of the samples was adjusted to 10.0, and sodium dithionite was added to a final concentration of 50 mM (see "Materials and Methods"), and the EPR spectrum was recorded at 15 K.The g values and the relative spin concentrations of reduced F A (g ϭ 2.05, 1.94, and 1.85) and F B (g ϭ 2.07, 1.92, and 1.88) in the pds Ϫ and zds Ϫ PS I mutants were identical to those of the WT (Fig. 5).Similar spectra were recorded when the samples were adjusted to pH 8.3 in the presence of sodium ascorbate and illuminated during freezing to 15 K (data not shown).Thus, the absence of and/or replacement of ␤-carotene fam with carote- noids of shorter conjugated length in PS I did not affect either the ability to form F A Ϫ /F B Ϫ or the electronic structures of the F A Ϫ /F B Ϫ iron-sulfur clusters at low temperature.Fig. 6 shows the Q-band (34 GHz) EPR spectra of the WT and the mutant PS I complexes after photoaccumulation at 205 K, a technique that generates A 1 Ϫ and F X Ϫ in a background of chemically prereduced F A Ϫ /F B Ϫ .Note that at 205 K, signals from the iron-sulfur clusters are not detectable, hence the spectra shown are mainly due to A 1 Ϫ and perhaps a small amount of A 0 Ϫ .The WT and mutant spectra very closely resembled each other.At 34 GHz, the field-dependent g anisotropy dominated the spectrum of A 1 Ϫ , which allowed for the partial resolution of the turning points at g xx ϭ 2.0062 and g zz ϭ 2.0021, values that agree with previous studies (26,(37)(38)(39).Because the g yy component of the tensor is hidden by the methyl proton hyperfine lines that arise from the high spin density at carbon 2 of the phylloquinone anion radical (A 1 Ϫ ), it is hard to determine its exact value.To better resolve the value of the g yy tensor, PS I from Synechocystis sp.PCC 6803 cells grown in 92% D 2 O were used in a similar experiment (37) to suppress the intensity of the proton hyperfine couplings.Deuteration resulted in the narrowing of the line widths of the g components and revealed a value of 2.0051 for the g yy component in WT PS I (37).The Q-band EPR spectra indicated that the g tensor values and the hyperfine couplings for phyllosemiquinone were virtually identical in the WT and mutant ؊ in WT, zds ؊ , and pds ؊ PS I complexes at 205 K.The samples were suspended to a concentration of 1 mg of Chl a ml Ϫ1 in Buffer A (see "Materials and Methods") to which were added 100 mM glycine buffer (pH 10) and 50 mM sodium dithionite.The preillumination and post-illumination spectra were recorded as described under "Materials and Methods."The instrument settings were as follows: microwave power, 1 milliwatt; microwave frequency, 34.056 GHz; modulation frequency, 100 kHz; modulation amplitude, 1 G; time constant, 10 ms; conversion time, 10 ms; 50 scans were averaged.
strains.We concluded that the ability to photoreduce phyllosemiquinone and the electronic structure of phyllosemiquinone in the WT and mutant PS I reaction centers was insensitive to the absence and/or the replacement of ␤-carotene fam with -carotene fam or phytoene.
Electron Transfer from A 1 Ϫ to F X -Fig.7 shows kinetic traces of the absorbance changes at 380 and 480 nm after the flash-induced formation of the phyllosemiquinone anion A 1 Ϫ for the WT, pds Ϫ , and zds Ϫ PS I complexes.The transients were best fit with double exponential functions.Similar kinetic profiles were also determined at different wavelengths from 300 -540 nm in whole cells and PS I complexes, and the kinetic data were subjected to global fit analysis.The results are summarized in Table II.For the WT PS I complexes, the faster decay component has a t1 ⁄2 ϭ 8 -9 ns (20 -25% of total decay amplitude estimated by comparing the relative amplitudes of the UV signals at 380 -390 nm of Fig. 8, see below), whereas the slower component has a t1 ⁄2 ϭ 152-180 ns (ϳ75-80% of total decay).These time constants and relative amplitudes were similar to previous determinations on different cyanobacterial PS I preparations (9), although considerable variability in the relative amplitudes of the fast and slow phases has been found with PS I complexes from the eukaryote, spinach, C. reinhardtii, and Chlorella.In spinach, depending on preparation methods, the ratio of the amplitudes of the fast to the slow component varied considerably from 2:1 to 1:2, and in whole cells of C. reinhardtii and Chlorella, the difference spectra showed 1:2 and 1:1 relative amplitudes, respectively.For the pds Ϫ and zds Ϫ strains, the relative amplitudes of the fast and slow phases were similar to the WT.However, the time constants in the PS I mutant complexes were somewhat longer, particularly for the slow phase (1.5 times) compared with WT (see Table II).
To verify that the fast and slow phases indeed arise from A 1 Ϫ oxidation, the global exponential fits for these phases were compared in the UV and in the near-UV where the phyllosemiquinone minus phylloquinone difference spectrum is expected to contribute most strongly.Fig. 8 shows the A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ difference spectra from 300 to 540 nm obtained by global exponential fit analysis of the kinetic data for the WT, pds Ϫ , and zds Ϫ PS I complexes.The slow phase in the UV and near-UV light difference spectra of the WT shows peaks at 375, 405, and 430 nm, with a shoulder at 340 nm and a minimum at 315 nm.In the fast phase difference spectrum, there is a peak at 400 and 425 nm, but the shorter wavelength peak, the shoulder, and the minimum are all shifted by 10 -15 nm to the red with respect to the slow phase.In the case of the mutants, however, the fast and slow phase difference spectra looked more similar to each other except for the presence of a 340 nm shoulder in the slow phase difference spectra and a small 385 nm feature present in the fast phase difference spectra.The contribution of A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ is expected to be rela-tively small in this region.With the possible exception of a contribution from FeS Ϫ to the spectrum of the slow phase (e.g. at 340 nm), the similarity of the fast and slow phase difference spectra to each other and to the Chromatium menasemiquinone-menaquinone (Vit K 2 Ϫ ⅐ Ϫ Vit K 2 ) difference spectra (40) supports the attribution of the fast and slow phases largely to the oxidation of phyllosemiquinone.The PS I phyllosemiquinone-phylloquinone (Vit K 1 Ϫ ⅐ Ϫ Vit K 1 ) difference spectra, however, appear to be shifted to shorter wavelength by about 30 nm, compared with the difference spectra reported by Romijn and Amesz (40).
Fig. 8 shows the absorption difference spectra of the fast and slow phases of the A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ difference spectra of the WT, zds Ϫ , and pds Ϫ PS I cores to 540 nm.Some of the most marked features of the visible difference spectrum were located at Ն440 nm, where the contribution of the phyllosemiquinone minus phylloquinone difference spectrum was expected to be small and relatively flat (40).The A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ for the WT showed a marked second derivative-shaped signal centered at ϳ450 nm for both the fast and slow phases.Another derivative-shaped signal was centered at ϳ506 nm for the fast phase and ϳ500 nm for the slow phase.The spectral differences between the fast and slow phases in this region have not been reported previously, as the earlier work did not extend beyond 500 nm (44).All of these features have been attributed to electrochromic band shifts of carotenoids coupled to A 1 Ϫ reduction (11,12).The construction of the pds Ϫ and zds Ϫ strains provided a test of these assignments.As shown in Fig. 8, the loss of ␤-carotene fam results in the complete disappearance of the 500 and 506 nm features but only a small change in the 450 nm-centered feature.Clearly the former was due entirely to ␤-carotene fam , whereas the latter was not.The absence of the 500 and 506 nm features in the pds Ϫ and zds Ϫ spectra was caused either by the replacement of ␤-carotene fam by more saturated carotenoids, phytoene or -carotene, respectively, which are at least 50 nm blue-shifted compared with ␤-carotene fam , or the absence of carotenoid in the carotenoid-binding sites that were nearest to A 1 .The second derivative-shaped feature centered at 450 nm was likely due to an electrochromic band shift of a chlorophyll or chlorophylls located close to A 1 Ϫ .The minimum of this signal appears at 448 and 452 nm for the fast and slow phases, respectively.Also part of the fit for the global decomposition of the transient kinetics was a component that remained following the relaxations in the submicrosecond range.As the only species remaining are P700 ϩ and FeS [A/B] Ϫ , this component corresponds to the difference spectrum P700 ϩ FeS

Because the contribution of the FeS [A/B]
Ϫ Ϫ FeS [A/B] was so much smaller than that of P700 ϩ Ϫ P700, this spectrum corresponded primarily to the latter.This difference spectrum, shown in Fig. 9, shows a large bleaching at 435 nm and a derivative-shaped spectrum centered at about 518 nm.The latter feature disappeared in the P700 ϩ FeS [A/B] Ϫ Ϫ P700 FeS [A/B] difference spectrum in the pds Ϫ and zds Ϫ strains, indicating that this feature was most likely due to a ␤-carotene fam that is located close to P700.
To better resolve the spectra arising from carotenoid electrochromism coupled to A 1 ] mutant for the fast and slow phases of A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ and for P700 ϩ FeS [A/B] Ϫ Ϫ P700 FeS [A/B] (Fig. 10).The spectra were normalized to the 430 nm signal of P700 ϩ FeS [A/B] Ϫ Ϫ P700 FeS [A/B] measured at 20 s.In principle, all absorbance changes other than those arising from the carotenoids alone should disappear.It should be kept in mind, however, that spectral subtractions in the region of strong chlorophyll absorption corresponded to differences be-tween two large signals, thus creating a greater uncertainty in the difference spectra in this region.The A 1 double difference for the WT Ϫ zds Ϫ and for the WT Ϫ pds Ϫ both show very similar three peaked differences, consistent with their arising from electrochromic red shifts of carotenoids.Both the positions of the inflections (corresponding to the carotenoid absorbance maxima) and their displacement from each other (30 -35 nm) were consistent with their identification with the ␤-carotene fam .A comparison of the fast and slow phases in both cases indicated that the slow phase was displaced to the blue relative to the fast phase by ϳ6 nm.The difference between the two phases implied that the carotenoids that detected the fast and slow phases were located in different environments.This difference in the locations of the carotenoid band shifts was also observed in WT cells (Fig. 11), indicating that it is an intrinsic property of the reaction center and not an artifact of detergent isolation.The reasons for this displacement will be discussed below.
Fig. 10 also shows the P700 double difference spectrum for the WT and each of the mutants.Although there are some differences in the 400 -450 nm range, these double difference spectra also indicate the clear involvement of ␤-carotene fam in electrochromism tied to P700 oxidation.By taking the P700 double difference of zds Ϫ minus pds Ϫ , band shifts are apparent that likely arise from the two longer wavelength peaks of -carotene fam .These are responsible for the differences ob- served in the WT minus zds Ϫ and the WT minus pds Ϫ difference spectrum between 400 and 450 nm.Thus, the carotenoid or carotenoids that sense the oxidation of P700 are ␤carotene fam in the WT and are replaced by -carotene fam in the zds Ϫ strain.The double difference zds Ϫ minus pds Ϫ in the A 1 case leaves smaller residuals than those observed for P700, possibly indicating a lower probability of the -carotene fam replacing ␤-carotene fam in the neighborhood of the phylloquinones.
Charge Recombination from F A /F B Ϫ to P700 ϩ -The slowing of the slow phase of oxidation of A 1 Ϫ in the mutants compared with WT (Table II) could reflect a change in the reduction potential of A 1 /A 1 Ϫ (Equations 1 and 2).Any mutation-induced changes in the reduction potential of A 1 /A 1 Ϫ could be quantified by measuring the rate of charge recombination (V obs, Equation 3) between P700 ϩ and F A /F B Ϫ , assuming that the rate of acceptor side equilibration was rapid relative to the rate of charge recombination and that the intrinsic rate of charge recombination (k in ϭ 95 s (41)) of P700 ϩ A 1 Ϫ 3 P700A 1 did not change in the mutants.The rate of charge recombination should then be proportional to the equilibrated concentration of A 1 Ϫ (Equations 1 and 2) determined by equilibrium constant K AF (Equation 4).An increase in the reduction potential of A 1 /A 1 Ϫ would consequently accelerate charge recombination.The rate of charge recombination between P700 ϩ and F A /F B Ϫ was measured by saturating flash excitation followed by detection at 435 nm, a minimum of the P700 ϩ Ϫ P700 difference spectrum (Fig. 9).Fig. 12 shows an increase in the rate of charge recombination from a t1 ⁄2 of 35 ms ( ϭ 51 ms) in the WT to a t1 ⁄2 of 19 ms ( ϭ 27 ms) in the zds Ϫ strain (Fig. 12).A similar experiment performed on PS I complexes of WT and the pds Ϫ strain gave a t1 ⁄2 of 33 and 24 ms, respectively (not shown).The change in recombination rate corresponded to an increase in reduction potential of A 1 /A 1 Ϫ of 16 and 8 mV, respectively, for the zds Ϫ and pds Ϫ complexes. where

DISCUSSION
␤-Carotene fam Is Needed for PS II Assembly-Carotenoids play multiple roles in photosynthesis.They are accessory pigments used for light-harvesting and for photoprotection in chlorophyll-protein complexes by quenching of chlorophyll triplet states (42).We show here that the replacement of ␤-carotene fam with shorter conjugated chain carotenoids in Synechocystis sp.PCC 6803 results in the loss of PS II complexes in the thylakoid membranes.This observation is consistent with the observations of Trebst and Depka (43) in which inhibition of phytoene desaturase by norflurazon and fluridone results in the loss of the D1 polypeptide.Either the inability to quench chlorophyll triplet states accelerates the rate of D1 photodamage and degradation or the lack of an appropriately shaped carotenoid results in an inability to correctly assemble PS II reaction centers.In either case, accumulation of PS II in the thylakoid membranes is prevented.
Spectroscopic Characterization-The replacement or absence of ␤-carotene and its hydroxylated derivatives results in a 2-nm blue shift of the Q y absorption maximum of the Chls in the PS I complexes From the 2.5 Å x-ray crystallographic structure of PS I in T. elongatus, close associations are observed between the carotenoids and the chlorophylls of the reaction center (2,3).Because the ringed carotenoids (␤-carotene, zeaxanthin, ␤-cryptoxanthin, and isocryptoxanthin) are replaced by ones with shorter -conjugation (-carotene or phytoene) or no carotenoid at all, structural adjustments are to be expected.Such structural changes could result in alterations in local dielectric, in excitonic interaction between pigments, or in a loss of carotenoid-associated chlorophyll-binding sites.That the last two of these contribute to the above-mentioned absorbance changes will be documented in a forthcoming publication.
Forward Electron Transfer from A 1 Ϫ to F X -Brettel et al. (44) showed in P700 ϩ A 1 Ϫ charge recombination experiments with prereduced F A and F B at 10 K (t1 ⁄2 Ϸ 50 s) that the A 1 Ϫ Ϫ A 1 difference spectrum bears a close resemblance to the absorbance difference spectrum for Vit K 1 . Ϫ Vit K 1 (40).This resemblance is consistent with the presence of phylloquinone in PS I complexes (45) and provided strong evidence for the identification of A 1 Ϫ with phyllosemiquinone.Brettel et al. (44) were the first to measure the UV and visible region absorbance changes associated with the forward oxidation of A 1 Ϫ .The A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ absorbance difference spectra were measured in PS I complexes from the cyanobacteria Synechocystis sp.PCC 6803 (8), Synechococcus sp.(11), and spinach (46).These spectra have now been examined with improved signal/noise and over a more extended wavelength range in PS I complexes from Synechocystis sp.PCC 6803 (the present work) and in whole cells of C. reinhardtii (47).These spectra bear a close resemblance in the UV to the difference spectra generated by P700 ϩ A 1 Ϫ charge recombination (44).The forward oxidation of A 1 Ϫ has been shown to be biphasic with half-times (relative amplitudes) of 25 (65%) and 150 ns (35%) in spinach PS I complexes (46) and 7 (33%) and 190 ns (66%) in Synechocystis sp.PCC 6803 (8).Similar biphasic kinetics have been observed in Chlorella sorokiniana (12) and in C. reinhardtii (14).The A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ difference spectra for the two phases are shown in Fig. 8 for WT Synechocystis sp.PCC 6803 and the two knock-out strains pds Ϫ and zds Ϫ .They both bear an overall resemblance in the UV to Chromatium Vit K 2 . Ϫ Vit K 2 and to the in vitro Vit K .Ϫ Vit K 1 difference spectrum (11,40).The spectra observed here peak at 375-380 nm instead of the 390 nm observed in the in vitro spectrum and are shifted by about 30 nm to shorter wavelength compared with the Chromatium difference spectrum.There is also additional fine structure in the biological spectra, as observed by Brettel (8,11).In the case of the WT, the slow and fast phases of A 1 Ϫ oxidation show similar spectral components, the relative amplitudes of which are different.A component at 385 nm appears to dominate in the fast phase spectrum, and one at 378 nm dominates in that of the slow phase.Also a shoulder at 340 nm in the slow phase appears to be missing in the fast phase.The fast and slow phase UV spectra are more similar in the case of the pds Ϫ and zds Ϫ strains.In the two mutants, the 385 nm component is smaller than the 378 and 405 nm components resulting in a closer resemblance in the spectra of the two phases.These observations argue in favor of the fast and slow phases arising from the same chemical species, possibly in somewhat different environments, the differences for which are more marked in the WT difference spectrum.It is also possible that there is an additional small contribution from F X Ϫ to F A /F B electron transfer that contributes to the slow phase in all three strains.The shoulder at 340 nm in the slow phase might arise from such a contribution.
The Vit K 1 .Ϫ Vit K 1 in vitro difference spectrum shows only small and flat absorbance changes above 440 nm (40).In this range there are marked features (450 -470 nm and 500 -520 nm) in the A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ difference spectra of the fast and slow phases that most likely come from electrochromic band shifts of pigments located in close proximity to A 1 , reflecting the anionic charge present in the A 1 Ϫ state.The pigments responsible for these electrochromic band shifts were suggested to be carotenoids (11,12).A comparison of the A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ from the WT and from the two carotenoid mutants indicates slight modifications in the 450 -470 nm range and large differences in the 500 -520 nm range.In both the pds Ϫ and the zds Ϫ strains, the band shifts in the 500 -520 nm range have completely disappeared in the difference spectra of both phases.However, the large second derivative-shaped signal in the 430 -470 nm range has remained largely intact, including the displacement of the fast phase minimum at 448 to 452 nm in the slow phase.A derivative-shaped shoulder present at 460 nm in the WT fast phase has disappeared in the mutants.By taking the double difference [A 1 mutant , one obtains the spectra shown in Fig. 10.Here one can see that there are three electrochromic bands that undergo a red shift in both the fast and slow phases.The 30 -35 nm spacing between these shifts is consistent with their arising from ␤-carotene fam .It is clear that the band shifts are not located in the same positions for the fast and slow phases, with the band shifts displaced by ϳ6 nm to higher wavelength for the fast phase as compared with the slow phase.This observation supports the idea that the A 1 species in each of the two phases is present in a different environment, giving rise to electrochromic band shifts of a different set of carotenoid pigments.That the second derivative-shaped feature in the 430 -470 nm range remains prominent in the mutant difference spectra, despite the absence of carotenoids absorbing in this range, argues that the remaining band shift most likely arises from one or more chlorophyll molecules.
There has been considerable controversy concerning whether one or both of the phylloquinones are active for electron transfer in PS I to explain the biphasic kinetics in the electron transfer step between A 1 Ϫ and the FeS clusters.Brettel (8) and Setif and Brettel (46) had initially attributed the presence of the biphasic kinetics of A 1 Ϫ oxidation to the equilibrium between A 1 and F X , the reduction potentials of which were proposed to be close.The fast phase would then arise from the establishment of the equilibrium A 1 Ϫ F X i A 1 F X Ϫ , whereas the slow phase would be limited by the electron transfer from F X to the F A /F B centers (8,46).Joliot and Joliot (12) have argued that as the electron transfer from A 1 Ϫ to F X is electrogenic (13), then if the fast phase were indeed the establishment of the equilibrium concentrations of A 1 Ϫ and F X Ϫ , its relative amplitude should be sensitive to the transmembrane electric field.They showed that, on the contrary, the relative amplitudes of the fast and slow phases were independent of the membrane potential.Leibl and co-workers (13,48) have also concluded from photovoltage measurements that the rate of electron transfer from F X Ϫ to F A /F B ( Յ50 ns) is faster than that of the slow phase of A 1 Ϫ oxidation.Both of these experimental observations thus argue against the equilibrium model and in favor of one where either the two quinones operate independently in a two-pathway model or where only one of the quinones is functional, but where there is a heterogeneity in the centers giving rise to two different rate constants for A 1 Ϫ to F X electron transfer (12).
Site-directed mutations have been constructed in C. reinhardtii and in Synechocystis sp.PCC 6803 in the region of each of the two phylloquinones.Mutations PsaA-W693F and PsaB-W673F (10,14) resulted in a substantial slowing, respectively, of the slow and fast phases of A 1 Ϫ forward oxidation.In each case, the rates of the complementary fast and slow phases, respectively, were unchanged as were the relative amplitudes of the fast and slow phases.Similar results were obtained with the PsaA-and PsaB-side mutations, PsaA-S692C and PsaB-S672C in Synechocystis sp.PCC 6803, although the effect on the fast kinetic phase of the PsaB-side mutant was not as pronounced as in the PsaB-W673F (10,47).These results argue in favor of the two-pathway model in which the fast phase would arise from the PsaB side and the slow phase the PsaA side.Also supporting the two-pathway model are EPR-detected kinetics of P700 ϩ A 1 Ϫ charge recombination at 100 K following illumination at 205 K in the presence of sodium dithionite.Two phases are observed for charge recombination, the slower of which is lost upon illumination in the PsaA-W693H and PsaA-W693L mutants (49).Xu et al. (10), while confirming the room temperature optical results with both sets of mutants, were unable to observe a slowing of the fast phase using transient EPR.They concluded that the PsaA branch is active but that there was no EPR-based evidence for electron transfer up the PsaB branch.Thus, according to their data, a unidirectional scheme using the A pathway of electron transfer with center heterogeneity could not be ruled out.The inability of Xu et al. (10) to observe the same relative amplitude of fast phase by transient EPR as they observe in optical experiments, however, may be due to the inability of transient EPR to detect, even indirectly, a kinetic phase of Ͻ10 ns, especially when the fast kinetic phase represents a minority of the electron flux to F X .
The present work adds support for the two-pathway model and provides additional spectroscopic markers with which to distinguish the two electron transfer pathways.We have shown here that the carotenoid pigments that act as electrochromic detectors for the fast and slow phases are different and are shifted 6 nm with respect to each other.That these same difference spectra are observed in PS I complexes and in PS I in whole cells, with similar rates for the fast and slow phases and with similar relative amplitudes, would argue against the two phases originating from center heterogeneity.In the ␤-carotene-deficient pds Ϫ and zds Ϫ strains, that the A 1 Ϫ (FeS) Ϫ A 1 (FeS) Ϫ difference spectra of the fast and slow phases become practically identical with little change in relative amplitude compared with WT also argues that the spectral differences between the fast and slow phases and the biphasic electron transfer do not arise from a heterogeneity between PS I reaction centers.The spectral differences in the two phases in WT most likely arise then from carotenoids with different conformations and/or configurations or carotenoids in different environments associated with each of the two phylloquinones.These spectral differences associated with each phase thus argue for the participation in electron transfer of both the PsaA-side and PsaB-side quinones.
Examination of the T. elongatus PS I x-ray crystal structure (2) indeed shows that there are structural differences in the carotenoids associated with the PsaA and PsaB sides of the reaction center in the immediate vicinity of the phylloquinones.Two ␤-carotene fam carotenoids with cis double bonds are found near the phylloquinone on the PsaA side, whereas only one is found on the PsaB side.On the PsaA side, there is a 9,13-di-cis (see numbering in Fig. 1) located at a distance of 19 Å from PhQ A (edge-to-edge distance).There is also a 13-cis ␤carotene fam carotenoid 25 Å away.On the PsaB side, the only cis-␤-carotene fam carotenoid found near phylloquinone bound by PsaB is 9-cis, located 22 Å away.Hashimoto and Koyama (50) have shown in n-hexane that the absorption spectrum of 9-cis-␤-carotene is 6 nm shifted to the blue compared with the corresponding all-trans isomer.The 13-cis isomer, on the other hand, is shifted by 11 nm to the blue, again compared with the corresponding all-trans isomer.In general, the nearer the cis double bond is to the center of the carotenoid, the greater the extent of the blue shift (50).A 9,13-di-cis-␤-carotene is expected to be even more blue-shifted than its corresponding mono-cis isomers.It is most likely that the carotenoids nearest to the phylloquinones all experience electrochromic effects but that the cumulative effect of having two cis-carotenoids (having a total of three cis double bonds) on the PsaA side shifts the overall absorption spectrum of the carotenoids on the PsaA side 6 nm to the blue compared with the carotenoids on the PsaB side with only one cis double bond between them.
The differences in the minima of the second derivativeshaped feature in the 430 -470 nm range between the fast and slow phases also likely reflect a difference in the electrochromic effects of the PsaA-and PsaB-side phyllosemiquinone anions on nearby chlorophylls.Consistent with this observation, Dobek and Brettel (51) have recently reported differences in the spectra of the fast and slow phases of A 1 Ϫ oxidation in the 660 -694 nm range that likely reflect electrochromic effects on different reaction center chlorophylls.The absorbance spectra of most of the PS I component chlorophylls are not known, however, making it difficult to make structural assignments to the detecting chlorophylls.
The zds Ϫ and pds Ϫ strains both show a marked slowing of the slow phase of A 1 Ϫ oxidation (t1 ⁄2 ϳ 250 Ϯ 18 ns) compared with WT (t1 ⁄2 ϳ170 Ϯ 15 ns) (Table II).There may also be a slight slowing of the fast phase, but the alteration in rate, if it exists, is within the noise level of the measurement.The mutant strains also show an acceleration of the rate of charge recombination relative to WT.These observations are understandable in terms of a model proposed by Agalarov and Brettel (9) in which A 1 Ϫ 3 F X electron transfer is exergonic on one side of the reaction center and endergonic on the other side, most likely the PsaB and PsaA sides, respectively.We pro-pose that the replacement of ␤-carotene fam by shorter chain carotenoids or the loss of ␤-carotene fam in the mutant strains results in the raising of the reduction potential of the phylloquinones.As the Marcus curve is parabolic for the A 1 Ϫ 3 F X electron transfer, a given change in the free energy of the reaction will have a larger effect on the reaction rate of the endergonic than on the exergonic reaction (Fig. 13).This is indeed what we observe (Table II and Fig. 7).
The rate of charge recombination is governed by the product of the intrinsic rate (k in ) of charge recombination between P700 ϩ and A 1 Ϫ times the concentration of A 1 Ϫ .The latter is determined by the relative reduction potentials of the A 1 /A 1 Ϫ and (F A /F B )/(F A /F B ) Ϫ redox couples.The more positive the reduction potential of A 1 /A 1 Ϫ the higher the concentration of A 1 Ϫ following charge separation and the higher the rate of charge recombination.Assuming the intrinsic rate of P700 ϩ A 1 Ϫ recombination to be the same in the mutants and WT, then the higher rate of charge recombination in the mutants reflects an increase in the reduction potential of A 1 /A 1 Ϫ .Recombination should occur predominantly through the PsaA-side phylloquinone, the more positive of the two resident quinones.The more rapid recombination in the mutants is thus consistent with the slowed rate of the slow phase forward reaction, with both arising from an increased reduction potential of the PsaA-side phylloquinone (Fig. 13).The 1.4 -1.8-fold increase in rate of charge recombination corresponds to an increase in reduction potential of A 1 /A 1 Ϫ of 8 -16 mV.Although less visible, it is likely that the PsaB-side phylloquinone has also undergone a similar increase in reduction potential.The existence of endergonic and exergonic components of the kinetics of oxidation of A 1 Ϫ is most consistent with the bidirectional model.Fig. 2 shows the cluster of carotenoids located close to each of the phylloquinones in the x-ray crystal structure of PS I.The increase in reduction potential of the phylloquinones in the mutants suggests that the loss/replacement of the carotenoids close to the phylloquinone in the WT results in a stabilization of the phyllosemiquinone relative to phylloquinone.One of the striking characteristics of the phylloquino- nes of PS I is their very low reduction potential.These have been estimated to be Ϫ531 and Ϫ686 mV for the PsaA-and PsaB-side PhQ/PhQ Ϫ redox couples, respectively (52).The carotenoids are apparently one contributing factor to this low potential, and a role for these molecules in photosynthesis was not appreciated previously.
The small increase in the A 1 /A 1 Ϫ reduction potential associated with the loss/replacement of the carotenoid could result from a small conformational shift of the peptide backbone in the vicinity of the phylloquinone, a factor that Ishikita and Knapp (52) have signaled as being a determining factor for quinone redox behavior in PS I. We have also found that the loss/replacement of carotenoid results in the depletion of the PsaF and PsaL subunits in the isolated complex along with some long wavelength chlorophylls. 2van der Est and co-workers (53) have argued that the loss of PsaF in a psaF deletion mutant (accompanied by the loss of chlorophylls 1138 and 1139) could influence the redox properties of PhQ A by increasing access to bulk water.van der Est and co-workers (53) report, relative to WT PS I complexes, an increased rate of PhQ A Ϫ oxidation by F X in the presence of Triton X-100 but a slowing of the rate by 10% in the presence of ␤-DM, the detergent used here.The slowing of the slow phase of the A 1 Ϫ oxidation kinetics that we observe here in the pds Ϫ and zds Ϫ mutants is larger (50%) relative to WT (Table II and Fig. 7) than what was reported by van der Est and co-workers in Synechococcus sp.PCC 7002 (53) and observed here both in vivo and in isolated PS I complexes from Synechocystis sp.PCC 6803.Although the relative amplitudes of the effect on rate may be due to species differences, it is likely that the loss/ replacement of carotenoid, with consequences for both PsaF and chlorophyll binding, enhances the impact on the reduction potential of PhQ A /PhQ A Ϫ , beyond what is observed by deletion of PsaF alone.Although we interpreted above the difference in kinetic effects of carotenoid loss/replacement on the relative energetics of the PsaA and PsaB branches, it is also possible that the structural perturbations associated with the carotenoid loss/replacement have a larger impact on A 1 /A 1 Ϫ reduction potential on the PsaA side.

FIG. 3 .
FIG.3.a, absorption spectra of the WT, zds Ϫ , and pds Ϫ PS I complexes after normalization at the Q y absorbance maximum.The inset shows an expansion of the region of maximum absorption of the Q y optical transitions.b, subtractions of the Q y -normalized spectra of the PS I complexes of WT minus pds Ϫ and of zds Ϫ minus pds Ϫ , revealing the absorbance spectra of the carotenoids present in the WT and zds Ϫ strains.

FIG. 4 .
FIG. 4. HPLCs of total pigment extracts from PS I complexes of the WT (a), zds ؊ (b), and pds ؊ strains (c).The pigments were extracted, and the HPLC was run as described under "Materials and Methods."The chromatograms were detected at 450, 400, and 288 nm, respectively, for the WT, zds Ϫ , and pds Ϫ strains.

FIG. 7 .
FIG. 7. Kinetics of A 1؊ oxidation in the WT (f), zds ؊ (‚), and pds ؊ (E) PS I complexes monitored at 380 (left) and 480 nm (right) following an actinic flash exciting 70% of the reaction centers.The PS I complexes were suspended as described under "Materials and Methods."

FIG. 9 . 1 ؊(
FIG. 9. Comparison of the zds ؊ and the pds ؊ decay-associated spectra with that of WT for P700 ؉ FeS [A/B] ؊ ؊ P700 FeS [A/B] in PS I complexes.This spectrum was detected at 20 us following the actinic flash.

FIG. 11 .
FIG. 11.Decay-associated spectra in Synechocystis sp.6803 whole cells of the 8 ns (f, fast phase) and 180 ns ( , slow phase) components from a global analysis of kinetic data in WT.The cells were suspended in 20 mM HEPES (pH 7.2), 20% (w/v) Ficoll, and 5 M carbonyl cyanide (4-(trifluoromethoxy)phenyl)hydrazone.FIG.12. Charge recombination kinetics of electron transfer from FeS [A/B]؊ to P700 ؉ monitored at 435 nm (minimum of P700 ؉ ؊ P700) after a saturating actinic flash in WT and zds ؊ PS I complexes.The complexes were suspended in 25 mM Tris-HCl (pH 8.3), 10 M sodium ascorbate, 4 M DCPIP, and 0.03% ␤-DM.The chlorophyll a concentration for both samples was 14 g/ml.

FIG. 13 .
FIG. 13.Relative reduction potentials of photosystem I cofactors in the WT and mutant PS I complexes.Shown in the inset is a portion of the inverted region of Marcus curve.

TABLE I
Carotenoid composition of WT and mutant PSI complexes