Removal of the PsaF Polypeptide Biases Electron Transfer in Favor of the PsaB Branch of Cofactors in Triton X-100 Photosystem I Complexes from Synechococcus sp. PCC 7002

Authors

  • Rufat Agalarov,

    1. Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA
    Search for more papers by this author
  • Martin Byrdin,

    1. Institut de Biologie Physico-Chimique, UMR 7141 CNRS/Paris 6, Paris, France
    Search for more papers by this author
    • §

      Current address: Martin Byrdin, CEA, iBiTecS, Service de Bioénergétique Biologie Structurale et Mécanismes, Gif-sur-Yvette, France.

  • Fabrice Rappaport,

    1. Institut de Biologie Physico-Chimique, UMR 7141 CNRS/Paris 6, Paris, France
    Search for more papers by this author
  • Gaozhong Shen,

    1. Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA
    Search for more papers by this author
  • Donald A. Bryant,

    1. Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA
    Search for more papers by this author
  • Art Van Der Est,

    1. Department of Chemistry, Brock University, St. Catharines, ON, Canada
    Search for more papers by this author
  • John H. Golbeck

    Corresponding author
    1. Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA
    2. Department of Chemistry, The Pennsylvania State University, University Park, PA
      *Corresponding author email: jhg5@psu.edu (John H. Golbeck)
    Search for more papers by this author

  • This invited paper is part of the Symposium-in-Print: Photosynthesis.

*Corresponding author email: jhg5@psu.edu (John H. Golbeck)

Abstract

Continuous wave (CW) and transient electron paramagnetic resonance studies have implied that when PsaF is removed genetically, the double reduction of A1A is facile, the lifetime of A1A is shorter and the ratio of fast to slow kinetic phases increases in PS I complexes isolated with Triton X-100 (Van der Est, A., A. I. Valieva, Y. E. Kandrashkin, G. Shen, D. A. Bryant and J. H. Golbeck [2004] Biochemistry43, 1264–1275). Changes in the lifetimes of A1A and A1B are characteristic of mutants involving the quinone binding sites, but changes in the relative amplitudes of A1A and A1B are characteristic of mutants involving the primary electron acceptors, A0A and A0B. Here, we measured the fast and slow phases of electron transfer from A1B and A1A to FX in psaF and psaE psaF null mutants using time-resolved CW and pump-probe optical absorption spectroscopy. The lifetime of the fast kinetic phase was found to be unaltered, but the lifetime of the slow kinetic phase was shorter in the psaF null mutant and even more so in the psaE psaF null mutant. Concomitantly, the amplitude of the fast kinetic phase increased by a factor of 1.8 and 2.0 in the psaF and psaE psaF null mutants, respectively, at the expense of the slow kinetic phase. The change in ratio of the fast to slow kinetic phases is explained as either a redirection of electron transfer through A1B at the expense of A1A, or a shortening of the lifetime of A1A to become identical to that of A1B. The constant lifetime and the characteristics of the near-UV spectrum of the fast kinetic phase favor the former explanation. A unified hypothesis is presented of a displacement of the A-jk(1) α-helix and switchback loop, which would weaken the H-bond from Leu722 to A1A, accounting for the acceleration of the slow kinetic phase, as well as weaken the H-bond from Tyr696 to A0A, accounting for the bias of electron transfer in favor of the PsaB branch of cofactors.

Introduction

The X-ray crystal structure of trimeric Photosystem I (PS I) from Synechococcus elongatus (since renamed Thermosynechococcus elongatus) has been solved to a resolution of 2.5 Å, which allows the architecture of the pigments, electron transfer cofactors and protein subunits to be accurately depicted (1). Each PS I monomer is composed of 12 subunits (PsaA to PsaF, PsaI to PsaM and PsaX), 96 chlorophyll (95 Chl a plus 1 Chl a′) molecules, 22 β-carotenes, two phylloquinones (PhQ), three [4Fe-4S] clusters, four lipids and a relatively large number of bound water molecules. The PsaA and PsaB heterodimers provide ligands for the electron transfer cofactors from P700 through FX, which form two branches on either side of a pseudo-C2 axis of symmetry (Fig. 1). From a functional as well as structural standpoint, the electron transfer chain begins at the Chl a/Chl a′ dimer (Chl a′ is a 132 epimer of Chl a [2]) (P700), and bifurcates into two branches, each containing two Chl a monomers (A-1 and A0) and a phylloquinone (A1). Although it has not yet been settled whether P700 (eC-A1/eC-B1)** or the A-1A (eC-A2) and A-1B (eC-B2) chlorophylls act as the primary electron donor (3), there is general agreement that A0A (eC-A3) and A0B (eC-B3) are the primary electron acceptors, and that A1A (QK-A) and A1B (QK-B) are the secondary electron acceptors. The two branches converge at FX, an interpolypeptide [4Fe-4S] cluster ligated by two cysteine residues from PsaA and two cysteine residues from PsaB. The terminal [4Fe-4S] clusters FA and FB are ligated by eight cysteine residues on PsaC, which is a ferredoxin-like protein located on the stromal side of the PS I complex (4). Subunits PsaD and PsaE flank the PsaC subunit on either side, stabilizing its association with the PS I core and providing binding sites for soluble electron carriers such as ferredoxin and flavodoxin (reviewed in Fromme et al. [5]). Subunits PsaL, PsaI and PsaM are believed to be involved in the trimerization of PS I, while other subunits, including PsaF and PsaJ, either indirectly or directly, bind Chl a molecules that increase the optical cross section of the antenna system (1).

Figure 1.

 Depiction of the cofactors according to the 2.5 X-ray crystal structure of cyanobacterial Photosystem I. The six excitonically coupled chlorophyll molecules that comprise the heart of the reaction center are depicted as P700 (Chl a), A-1A, A0A on the A-branch and P700 (Chl a′), A-1B, A0B on the B-branch. These spectroscopic names correspond to the crystallographic names (in parenthesis): P700 (eC-A1/eC-B1), A-1A (eC-A2), A-1B (eC-B2), A0A (eC-A3) and A0B (eC-B3). The electron transfer step from A1A to FX is responsible for the slow kinetic phase of phyllosemiquinone oxidation and the electron transfer step from A1B to FX is considered responsible for the fast kinetic phase of phyllosemiquinone oxidation. Note the high degree of C2 symmetry, with an axis that passes through the center of P700 and FX.

The electron transfer kinetics among the cofactors have been well studied. In particular, the oxidation of phyllosemi-quinone is biphasic, with half-times of 25 and 150 ns in PS I from spinach (6) and half-times of 7 and 200 ns in PS I from Synechocystis sp. PCC 6803 (7). The assignment of the fast and slow kinetic phases of phyllosemiquinone oxidation to A1B and A1A was first proposed by Joliot and Joliot (8) and is supported by time-resolved optical data from mutants surrounding the quinone binding sites in PS I of algae and cyanobacteria (9–11). The salient feature of these mutants is that the lifetimes, but not the ratios, of the fast and slow kinetic phases are altered as the environment of the quinone is changed. This is expected because the dynamics of light-induced charge separation is thought to involve only the six excitonically coupled Chl molecules that comprise the primary donor, accessory Chls and acceptor Chls, and not the quinone secondary acceptors. Accordingly, alterations around the quinones should change the kinetics, but not the relative amplitudes of the fast and slow kinetic phases. Conversely, alterations in and around the six Chl molecules that form the heart of photochemical charge separation process would be expected to change the relative amplitudes, but not the kinetics of the fast and slow kinetic phases. Indeed, when the H-bond from a nearby Tyr to the primary acceptor chlorophylls A0A and A0B is removed by conversion to a Phe, the relative amplitudes, but not the lifetimes, of the two kinetic components of phylloquinone oxidation are affected (12). The mutation near A0A increases the fraction of the faster component at the expense of the slower component, with the opposite effect seen in the A0B mutant (12). This result was interpreted as a decrease in the relative use of the targeted branch.

Depending on the detergent used in the isolation procedure, a number of studies have noted differences in the kinetics or spectroscopic behavior of the phylloquinone cofactors in PS I. Early biochemical studies showed that the relatively strong detergent Triton X-100 was able to remove PsaF (then known as subunit III) from Swiss Chard PS I complexes (13) and that it could extract up to 60 bound Chls from barley (14) and spinach (15) PS I complexes. The photoaccumulated, X-band electron paramagnetic resonance (EPR) spectrum of A1 in PS I complexes prepared from spinach using the relatively mild detergent digitonin showed the presence of partially resolved hyperfine couplings from the methyl protons of the ring –CH3 group (16), whereas the spectrum of A1 in PS I complexes prepared using Triton X-100 had lost much of the hyperfine structure so that it was not as broad and therefore not as easily distinguishable from the spectrum of A0 (17). These differences were attributed to an unspecified perturbation of the PS I complex induced by Triton X-100 (18). The relative amplitudes of the fast and slow kinetic phases of A1 oxidation were reported to be dependent on the method used to isolate the PS I complexes (6). When Triton X-100 was used to prepare TSF1 particles, the fast phase was found to dominate; when digitonin or a mixture of digitonin and deoxycholate were used to prepare D144 or PSI-110 particles, roughly equal amounts of fast and slow kinetic phases were found; and when no detergent was used to prepare enriched PS I fragments (a Yeda press was used to fractionate the membranes), the fast kinetic phase accounted for only 30% of the electron transfer (6).

Because these spectroscopic changes appeared to correlate with the loss of the PsaF polypeptide, psaF null mutants were generated and studied in Synechococcus sp. PCC 7002 (19,20). The first of two studies (19) showed that the A1A quinone was susceptible to double reduction in a low-temperature photoaccumulation protocol when PS I from the psaF and psaE psaF null mutants was isolated with Triton X-100, but not when PS I was prepared with n-dodecyl-β-d-maltopyranoside. Double reduction occurs only when the singly reduced phylloquinone or menaquinone is protonated during or prior to the second reduction. PsaF indirectly binds a number of Chl molecules, and its loss could open a water channel to the A1A quinone binding site. The loss of PsaF could additionally render Chls bound near the hydrophobic tail of the quinone susceptible to removal by Triton X-100, which could further increase the hydrophilicity of the A1A quinone site. A second study (20) showed that the changes in the transient EPR kinetics and spectra for the psaF and psaE psaF null mutants were consistent with a faster electron transfer from A1A to FX, a slower forward electron transfer from A0 to A1 and slightly smaller hyperfine couplings. In principle, the decrease in the hyperfine coupling can come about as a result of a decrease in the strength of the hydrogen bond from the Leu722 to C4 of the A1A quinone. A weaker H-bond would drive the redox potential of A1A more reducing, which would decrease the free energy difference between A0 and A1, and increase the free energy difference between A1 and FX. Given that this electron transfer step is usually described by the normal region of the Marcus curve, this would lead to a faster electron transfer from A1A to FX and a slower forward electron transfer from A0 to A1. Although only the slow kinetic phase is directly measured by transient EPR, it could be inferred from spectral modeling studies that the relative amplitude of the fast kinetic phase had increased at the expense of the slow kinetic phase.

Given this chain of logic, it was important to confirm the rate and relative contribution of electron transfer from A1A and A1B to FX in the mutants lacking PsaF or PsaE and PsaF. Here, we use time-resolved optical absorption spectroscopy to measure the fast and slow kinetic phases attributed to the oxidation of A1B and A1A, respectively. We find that the slow kinetic phase attributed to A1A to FX electron transfer indeed becomes faster and the amplitude of the fast kinetic phase increases up to a factor of 2 or more in the PsaF and PsaE PsaF subunit deletion mutants. The latter effect is best explained as a redirection of electron transfer through A1B at the expense of A1A.

Materials and Methods

Preparation of PS I complexes from the psaF, psaJ and psaE psaF null mutants.  The psaF, psaJ and psaE psaF null mutants of Synechococcus sp. PCC 7002 were constructed as described previously (19). The psaF and psaJ genes were insertionally inactivated with DNA fragments encoding the aphII gene, which encodes aminoglycoside phosphotransferase II and confers resistance to kanamycin. To produce a psaE psaF mutant, the psaF::aphII strain was transformed with a construction in which the aadA gene, which confers resistance to streptomycin and spectinomycin, was inserted into the coding sequence of psaE. Trimeric PS I complexes were prepared from these strains using the detergents n-dodecyl-β-d-maltopyranoside or Triton X-100 as previously described (19).

Low-temperature fluorescence spectroscopy.  Fluorescence emission spectra were measured with an SLM 8000C spectrofluorometer, which had been upgraded for computerized data acquisition, as described previously (21). To measure fluorescence emission at 77K, PS I complexes were diluted in 25 mm HEPES-NaOH, pH 7.0 containing 60% (vol/vol) glycerol to a concentration of 2 μg Chl mL−1 prior to freezing in liquid nitrogen. The excitation wavelength was 440 nm.

Continuous wave (CW) time-resolved optical spectroscopy at 380 nm.  The fast and slow kinetic phases of menasemiquinone oxidation were measured optically on the nanosecond time scale by monitoring the flash-induced transient absorption change at 380 nm using a time-resolved spectrophotometer similar to that described in Gerken et al. (22). The excitation beam was provided by a frequency-doubled (λ = 532 nm), Q-switched Nd:YAG laser (DCR-11; Spectra Physics, Mountain View, CA) operated in the short pulse mode (∼3 ns). The measuring light was provided by a xenon flash that was tailored with a bank of inductors and capacitors to produce a relatively flat top for 5 μs. The 380 nm measuring beam was selected using a combination of narrow-band (8 nm) interference and colored glass filters (to block the scattered laser flash) prior to the sample and in front of the Si photodiode detector (FND 100Q; EG&G). The photocurrent was changed to a voltage across a 50 Ω resistor, and the signal was amplified using a laboratory-built gain block (30 dB, 500 Hz–1.7 GHz) and recorded using a digitizing oscilloscope (DSA 602A with amplifier plug-in 11A52; Tektronix, Beaverton, OR). A photodiode connected to a 11A72 plug-in detected the laser flash and initiated the data acquisition. The rise-time of the detection system was measured using Ru(bipy)3Cl2 luminescence to be 3 ns (23). The baseline was recorded after every flash by mechanically blocking the excitation flash, and the no-flash signal was subtracted from that recorded with flash excitation on every cycle. Typically, 1024 or 2048 cycles of flash minus no-flash transients were averaged at a repetition rate of 1 Hz. Software written in LabView controlled the timing sequence, acquisition and manipulation of the data. The sample was contained in a 10 × 10mm standard quartz cuvette. All spectroscopic measurements were performed at room temperature. Kinetic traces were analyzed by fitting with a multiexponential function plus a constant using the Marquardt least-squares algorithm programmed in IGOR Pro v. 5.2 (Wavemetrics, Portland, OR).

Pump-probe optical spectroscopy in the near-UV and visible region.  Measurements of the kinetics of menasemiquinone oxidation in the nanosecond to microsecond time scale were performed on isolated PS I complexes in a flash-detection spectrophotometer as described previously (24). The PS I complexes were suspended at a concentration of 40 μg mL−1 in buffer with sodium ascorbate and 2,6-dichlorophenolindophenol concentrations of 10 mm and 40 μm, 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 adapted from Grzybek et al. (25). Charge separation was induced by a 5 ns (full width at half maxima) light pulse at 700 nm using a Nd:YAG-pumped LDS 698 dye exciting ∼70% of the PS I complexes. Absorbance changes were followed from 5 ns to 20 μs 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 blue photons, which can be more effectively attenuated by filters.

Results

Characterization of PS I complexes from the psaF, psaJ and psaE psaF mutants

Three small membrane-intrinsic polypeptides, PsaF, PsaJ, and PsaX, are bound near the PsaA/PsaB interface that is distal to the three-fold symmetry axis of PS I trimers (1,5). The X-ray structure of T. elongatus PS I shows that PsaX has a single transmembrane α-helix and binds a single Chl a molecule; however, no psaX gene occurs in the genome of Synechococcus sp. PCC 7002, and no evidence for the presence of a subunit homologous to PsaX has been obtained for PS I complexes from either this cyanobacterium or from Synechocystis sp. PCC 6803. The PsaJ polypeptide also has a single transmembrane α-helix; this polypeptide ligates three Chl a molecules and is involved in binding a cluster of five β-carotene molecules. PsaE is a stromal polypeptide composed of a five-stranded β-sheet and a single 3(10) helix that has contacts with PsaF as well as with PsaA. The C-terminal domain of PsaF is mainly membrane-intrinsic and possesses one transmembrane α-helix and two partial membrane helices. Although this subunit does not directly ligate any Chl a molecules, it is involved in binding a cluster of five β-carotene molecules and appears to shield a cluster of several Chl a molecules from the bulk lipids of the thylakoid membrane (1,5).

The psaF and psaJ genes form a dicistronic psaFJ operon in Synechococcus sp. PCC 7002 (26). The psaF and psaJ genes were insertionally inactivated individually with a DNA fragment encoding the aphII gene, which encodes aminoglycoside phosphotransferase and confers resistance to kanamycin (26). The psaJ mutant had no detectable growth defect, but the psaF mutant grew about 20% slower than the wild-type strain under all conditions tested. Gel electrophoretic analyses of trimeric PS I complexes isolated from the psaJ mutant previously showed that only the PsaJ subunit was missing, while analyses of PS I complexes isolated from the psaF mutant showed that both the PsaJ and PsaF polypeptides were missing (19). Trimeric PS I complexes of the psaE psaF mutant strain were completely devoid of the PsaE, PsaF and PsaJ polypeptides (19).

Figure 2 shows low-temperature fluorescence emission spectra for trimeric PS I complexes isolated from the wild type and from the three mutants strains by using the detergents n-dodecyl-β-d-maltopyranoside and Triton X-100. Wild-type Synechococcus sp. PCC 7002 PS I complexes prepared with these detergents were identical and had a fluorescence emission maximum at 716 nm. Although removal of PsaJ eliminates the ligands for three Chl a molecules, the 77 K fluorescence emission spectra of PS I trimers from the psaJ mutant that had been isolated with either detergent were indistinguishable from the spectra for the wild-type PS I complexes (Fig. 2C). However, the fluorescence emission maxima of PS I complexes isolated from the psaF mutant were blueshifted by 3 nm to 713 nm, and these complexes clearly had reduced fluorescence emission at longer wavelengths (Fig. 2A,B). No further changes in fluorescence emission were found when PsaE was additionally eliminated.

Figure 2.

 Fluorescence emission spectra of PS I complexes isolated from wild type and the psaF, psaJ and psaE psaF null mutant strains. (A) 77 K emission spectra of the PS I complexes isolated with n-dodecyl-β-d-maltopyranoside; (B) 77 K emission spectra of the PS I complexes isolated with Triton X-100; (C) comparison of 77K emission spectra of the wild-type and psaJ mutant PS I complexes isolated with Triton X-100. Spectra shown are the average of four measurements and have been normalized to facilitate comparison. The excitation wavelength is 440 nm.

Kinetics of menasemiquinone oxidation measured at 380 nm

Figure 3 shows the time course of quinone oxidation measured at 380 nm using a CW spectrometer for PS I trimers isolated with Triton X-100 from Synechococcus sp. PCC 7002. Unlike most other species, the quinone in the A1A and A1B sites of PS I in Synechococcus sp. PCC 7002 is menaquinone-4 rather than phylloquinone (27). However, this difference has little or no impact on the spectroscopic properties because the two quinones differ only in the degree of saturation in their side chains. Menaquinone-4 has an unsaturated geranylgeranyl side chain compared to the mostly saturated phytol side chain found in phylloquinone. The onset of the laser flash results in an absorbance increase in the near-UV due to the reduction of menaquinone-4 by the primary electron acceptor chlorophyll(s). The kinetics of electron transfer to the quinone occur in 13–35 ps (28–30), and are therefore too rapid to be observed, but the kinetics of electron transfer from the quinone are biphasic and occur in the tens and hundreds of nanosecond time range and are readily resolved. The decay of the menasemiquinone anion in wild-type PS I complexes (Fig. 3, top) can be fitted with biphasic kinetics, with lifetimes (and percentage of the total amplitude) of 20.5 ns (32%) and 278 ns (68%). In Synechocystis sp. PCC 6803 PS I, a similar flash-induced transient decays with biphasic kinetics, with lifetimes (and percentage of the total amplitude) of 15 ns (15%) and 256 ns (85%). The lifetimes are similar in the two cyanobacterial PS I complexes, but the relative contribution of the fast kinetic phase is greater in PS I from Synechococcus sp. PCC 7002 than in PS I from Synechocystis sp. PCC 6803. The decay of the menasemiquinone anion in the psaF null mutant (Fig. 3, middle) can be fitted with biphasic kinetics, with lifetimes (and percentage of the total amplitude) of 18.9 ns (59%) and 162 ns (41%). The total amplitude of the absorbance change is slightly lower than for the wild-type PS I complexes after the difference in the Chl concentration is taken into account, which could indicate either loss of photoactive centers or the presence of an unresolved faster kinetic component. The kinetic trace of the PS I complexes from the psaE psaF null mutant (Fig. 3, bottom) similarly shows a biphasic decay of the menasemiquinone radical, with lifetimes (and percentage of the total amplitude) of 19.9 ns (64%) and 125 ns (36%). In this mutant, the optical transient decays to a stable (on the nanosecond timescale) absorbance below the baseline, thus representing a net bleaching, and it returns back to the baseline with a lifetime of 30 μs (data not shown). The decay-associated spectrum of the bleaching from 360 to 520 nm is characteristic of a Chl triplet (data not shown). Because this triplet is most likely generated by intersystem crossing in the antenna Chls, and has a rise time on the order of subnanoseconds, its rise and subsequent decay on the microsecond timescale is not expected to compromise the analysis of the kinetics of the fast and slow kinetic phases.

Figure 3.

 Time-resolved optical kinetics at 380 nm in PS I complexes from the wild type (top), from the psaF mutant (middle) and the psaE psaF mutant (bottom) isolated with Triton X-100. The ΔA380 kinetics are presented on a linear time scale so that the fast and slow kinetic phases attributed to A1A and A1B oxidation can be easily visualized. The wild-type sample was measured at 5 μg mL−1 and the psaF and psaE psaF mutant samples were measured at 10 μg mL−1 Chl in 25 mm Tris, pH 8.3, 10 mm sodium ascorbate and 500 μm 2,6-dichlorophenolindophenol. The computer-generated exponential fit is shown as a solid line. Results of the exponential fits are displayed as fit curves, where each individual component is plotted with a vertical offset relative to the next component (with a longer lifetime) or the baseline, the offset being equal to the amplitude of the latter component. The relative contributions of each kinetic phase can be judged by the intersection of the fit line with the abscissa. The residual of the fit is depicted above the main trace.

The residuals shown above the decay curves show no indication for the presence of any additional kinetic components in either wild-type PS I or in PS I from the psaF and psaE psaF null mutants. The time course of menasemiquinone oxidation in PS I trimers isolated with n-dodecyl-β-d-maltopyranoside from the psaF and psaE psaF null mutants was similar to that of the wild type, indicating that the presence of a stronger detergent is required to observe the changes in relative amplitude and kinetics.

Flash-induced difference spectra of the fast and slow kinetic phases

Figure 4 shows a global fit of the time course of menasemiquinone oxidation measured in the near-UV and blue using a pump-probe spectrometer for PS I trimers isolated from Synechococcus sp. PCC 7002. Transient absorption signals were recorded in the nanosecond to microsecond timescale at wavelengths from 330 to 435 nm and from 435 to 540 nm after selective excitation at 700 nm with a 3 ns laser flash. Global analysis of the data for the wild-type PS I complexes (Fig. 4, top) reveals four components—two decay in the nanosecond time range, one decays in the microsecond time range and one component that did not decay on the timescale of the experiment. The spectrum of the nondecaying component was similar to the flash-induced P700+ [FA/FB]minus P700 [FA/FB] difference spectrum, and represents forward electron transfer of the electron past A1A and A1B to the terminal electron acceptors. The amplitude of the component that decays with a lifetime of 5.3 μs was so small that the spectrum could not be identified; however, its lifetime is similar to that of the decay of a Chl triplet. The fast and slow nanosecond kinetic phases (lifetimes of 14 and 264 ns, respectively) show an initial positive absorption change from 330 to 420 nm in the near-UV that is characteristic of a semiquinone anion radical. Both kinetic phases also show a positive absorption change in the blue, particularly at 480 nm, characteristic of an electrochromic bandshift induced on a nearby carotenoid by the presence of the negative charge on A1A and A1B (31). The amplitude of the spectrum representing the fast kinetic phase is smaller than that of the spectrum representing the slow kinetic phase (ratio of 38:62 at 380 nm), and is roughly in line with the direct measurement by the CW flash kinetic spectrometer. Global analyses of the data for the PS I complexes isolated from the psaF and psaE psaF null mutants similarly reveal four components, the slowest two of which are similar to those observed for wild-type PS I complexes. In the PS I complexes of the psaE psaF null mutant (Fig. 4, bottom), the lifetime of the fast kinetic phase remained unaffected, but that of the slow kinetic phase became faster (lifetime of 190 ns). In addition, the amplitude of the spectrum represented by the fast kinetic phase was greater than that of the slow kinetic phase (ratio of 58:42 at 380 nm) in the psaE psaF null mutant relative to that observed for wild-type PS I.

Figure 4.

 Spectra of the fast and slow kinetic phases derived from global decomposition of pump-probe kinetic experiments carried out in the UV and blue regions. Top: PS I complexes isolated from the wild type using Triton X-100. Bottom: PS I complexes isolated from the psaE psaF mutant using Triton X-100.

Comparison of transient EPR and optical data

Table 1 shows lifetimes and relative amplitudes of the fast and slow kinetic phases compared with those determined previously by transient EPR (20). The latter detects only the slow kinetic phase from the oxidation of A1A, but the relative amplitudes of the slow and fast kinetic phases can, in principle, be inferred by admixing simulated spectra of the P700+A1A and P700+FX radical pairs. The lifetime of the optically determined fast kinetic phase from A1B remains unchanged, but the lifetime of the slow kinetic phase from A1A oxidation becomes faster as PsaF and then PsaE are removed. The same trend is noted in the kinetics of A1A oxidation by transient EPR. Equally important is that the ratio of the fast to the slow kinetic phases changes from 32:68 in the wild type to 59:41 in the psaF null mutant to 64:36 in the psaE psaF null mutants. Hence, the amplitude of the fast kinetic phase in the latter is twice that of the wild-type PS I complexes, accounting for two-thirds rather than one-third of the total absorbance change. These data support the inference from the transient EPR data that the relative amplitude of the fast kinetic phase increases in PS I complexes lacking PsaF or PsaF and PsaE (20).

Table 1.   Comparison of lifetimes and relative amplitudes of the fast and slow kinetic phases by time-resolved optical spectroscopy and transient electron paramagnetic resonance (EPR) spectroscopy.
 OpticalEPR
Fast*Slow*Ratio†Fast*Slow*Ratio
  1. nd = not determined. *ns. †This value corresponds to the ratio of PsaB-side to PsaA-side electron transfer only if the differential extinction coefficients of A1A/A1A and A1B/A1B would be identical.

Wild type20.527832:68nd34017:83
PsaF18.916259:41nd25035:65
PsaE PsaF20.012564:36nd21055:45

Closer inspection of the data in Table 1 shows that the lifetimes obtained from the transient EPR data are consistently longer than those from the optical data and although the ratio of the fast and slow phases shows the same trends there are significant differences in the values obtained from the two methods. The likely reason for these differences is the uncertainty in the magnetic and geometric parameters needed for the analysis of the EPR data. In particular, the differences in the lifetimes suggest that there may be a systematic error in the treatment of lifetime broadening in the EPR analysis. To confirm that the discrepancies in Table 1 are indeed a result of the data analysis and not a systematic difference between the two methods, we used the kinetics and relative amplitudes of the fast and slow phases obtained from the optical data in Table 1 to calculate selected EPR transients. A comparison of the predicted and experimental EPR transients is shown in Fig. 5. At the chosen field position (20) the signal from P700+A1 is absorptive (positive) while that from P700+(FeS) is emissive (negative) and the increase in the amount of the fast phase in the psaF and psaE psaF null mutants results in a decrease in the absorptive signal from P700+A1 and a corresponding increase in the emissive P700+(FeS) signal at early time. As can be seen, the EPR transients are reproduced well using the lifetimes and amplitude ratios of the two phases from the optical data. Thus, all three methods (transient EPR, time-resolved CW optical spectroscopy and pump-probe optical spectroscopy) show that in the absence of the PsaF polypeptide, the use of Triton X-100 to isolate PS I complexes leads to changes in the lifetime of the slow kinetic phase and in the ratio of fast and slow kinetic phases associated with A1B and A1A oxidation.

Figure 5.

 Room temperature EPR transients of PS I complexes from the wild-type (top), the psaF null mutant (middle) and the psaE psaF null mutant (bottom), all isolated with Triton X-100. The transients are the same as those presented in fig. 3 (bottom left) of Van der Est et al. (20). The solid lines are transients calculated using the lifetimes and relative amplitudes of the electron transfer phases determined from the optical data in Table 1. Labels A and E refer to absorption and emission, respectively.

Discussion

The time-resolved optical kinetic experiments described here confirm previous transient EPR studies (20) in that the slow kinetic phase of menasemiquinone oxidation attributed to A1A is accelerated and that the fast kinetic phase had increased at the expense of the slow kinetic phase in PS I complexes isolated with Triton X-100 from psaF and psaE psaF null mutants. One additional result from the EPR studies was that the shapes of the P700+A1 and P700+FX spectra were altered in the latter such that they showed a greater amount of absorbance on the high field end of the P700+ region. This change could be reproduced by postulating that a small amount of singlet-triplet mixing occurs in the P700+A0 radical pair state as a result of an increase in its lifetime. Such an increase would be expected if the shorter lifetime of P700+A1 were due to a change in the midpoint potential of the quinone to a more negative value. This proposed change in the midpoint potential would be consistent with the slightly smaller hyperfine coupling from the 2-methyl group of menaquinone-4 in the mutant PS I complexes (20). In wild-type PS I complexes this coupling is large because of alternating spin density around the quinone ring as a consequence of an asymmetrical H-bonding scheme to the protein. In principle, a decrease in the hyperfine coupling can come about as a result of either a more symmetrical H-bonding arrangement, i.e. a second H-bond, in this instance to the C1 carbonyl group, or a weaker H-bond, in this instance from PsaA-Leu722 to the C4 carbonyl group. We favor the latter because in the absence of PsaF it is likely that the detergent would disorder the region of the protein containing PsaA-Leu722. Moreover, a weaker H-bond should result in a more negative redox potential for the A1A/A1A redox couple whereas the second H-bond would have the opposite effect. Provided that any effect to the FX cluster by the loss of PsaF or PsaE and PsaF is sufficiently small, a more negative midpoint potential for the A1A/A1A redox couple should result in a larger free energy change between A1A and FX, and hence, a faster rate of electron transfer. It should also result in a smaller free energy difference between A0A and A1A and hence a slower electron transfer rate, provided this reaction lies in the noninverted region of the Marcus curve. This assessment is supported by the increased lifetime for A0A observed by transient EPR as well as the significantly shorter lifetime for A1A observed by both transient EPR and time-resolved optical spectroscopy.

Although the hypothesis of a change in the H-bonding network to A1A accounts for the changes in the lifetime of both A0A and A1A, it does not explain the changes in the ratio of the fast and slow kinetic phases. As discussed in Van der Est et al. (20), the EPR data do not show whether the increased fraction of fast electron transfer in the mutants occurs in the PsaA or PsaB branch but the location of PsaF as a transmembrane polypeptide with significant contacts with the stromal A-jg(1) loop region of PsaA, suggests that it is in the former. In principle, there are three possible explanations for the differences in the PS I complexes of the subunit deletion mutants and the wild type.

The first possibility is that electron transfer to A1A is less efficient in the mutant complexes so that the relative reduction yield of A1A is decreased with no compensatory increase in the relative reduction yield of A1B. This hypothesis, however, can be ruled out because none of the previously characterized signatures of charge recombination prior to formation of P700+A1 were observed, i.e. an increased yield of 3P700 formation as well as an additional kinetic component developing in the tens of nanosecond time range (32,33).

The second possibility is that the relative yield of electron transfer in the two branches is unchanged but the combined action of removing PsaF and isolating PS I with Triton X-100 decreases the lifetime of A1A as suggested in Van der Est et al. (20). This hypothesis is supported by the structural arguments given above and the indications in the transient EPR that the A0 to A1 electron transfer may be slowed. However, the optical data presented here show that the electron transfer lifetime of the fast phase is unaltered in the mutant PS I complexes. Thus, if the generally accepted assignment of the fast phase to PsaB branch electron transfer in the wild type is correct, then in the mutants the PsaA branch electron transfer rate would have to be increased to match precisely the same rate as in the PsaB branch. Moreover, this increase would only occur in a fraction of the PS I particles. In the remaining fraction the rate would also be increased relative to that in wild-type PS I but by a smaller amount that depends on whether both PsaE and PsaF are removed or only PsaF. In addition, the relative size of the two fractions would also depend on which of the subunits were removed. Although this hypothesis was initially put forward (20), from the optical data presented here and from the above arguments, it now seems unlikely that all of these conditions could be met.

The third possibility is that the relative yield of electron transfer through A1B is increased. This hypothesis provides a satisfying explanation for the observed lifetimes and intensity ratios. In this scenario, the fast phase kinetics would be unaffected by the mutations because it occurs only in the PsaB branch, which has few contacts with the stromal region of PsaF, while the lifetime of the slow phase would become faster as a result of changes in the midpoint potential of A1A. However, this explanation suffers from lack of an obvious reason as to why changes on the stromal side of the PS I complex should alter the directionality of electron transfer. Indeed, point mutation studies have shown consistently that changes near the A1A and A1B quinones change the lifetimes of the slow and fast kinetic phases but not their relative amplitudes while changes near the A0A and A0B Chls have been shown to alter the amplitude ratio of the two kinetic phases.

An earlier finding that the absorbance difference spectra of the slow and fast kinetic phases assigned to the respective A1A and A1B radicals in Chlamydomonas reinhardtii differ slightly in the near-UV can be used to discriminate between the second and third hypotheses (33). It was proposed that part of the differences between the decay-associated spectra of the fast and slow phases in the near-UV components reflect a contribution from different electron acceptors, i.e. from an inter-Fe/S cluster electron transfer event. The spectrum of PS I isolated from wild-type Synechocystis sp. PCC 6803 also shows maxima at 380 and 400 nm while the spectrum of the slow kinetic phase shows a shoulder from 330 to 350 nm that is largely lacking in the fast kinetic phase (31). The same difference is noted in PS I isolated from wild-type Synechococcus sp. PCC 7002. The flash-induced difference spectrum of the enhanced fast kinetic phase in the psaE psaF null mutant similarly lacks this shoulder (Fig. 4, bottom), thereby lending support to the proposal that electrons are being redirected through A1B at the expense of A1A.

The speeding up of the slow kinetic phase attributed to A1A oxidation can be rationalized as an increase in the hydrophilicity of the site (19). The contacts between PsaF and the A-jk(1) surface loop of PsaA that contains the H-bonded Leu722 residue are shown in Fig. 6. These extensive interactions have led to the proposal that in the absence of PsaF, Triton X-100 removes Chl molecules and opens a water channel to the A1A quinone. Although PsaF does not directly participate in ligating Chl a molecules, the combined loss of PsaF and PsaJ apparently resulted in the removal of Chl a molecules beyond those lost when PsaJ was eliminated. The loss of these additional Chl a molecules (and possibly β-carotenes as well) is reflected in the blueshifted fluorescence emission maximum and the loss of long wavelength fluorescence emission at 77 K (Fig. 2). Their removal would allow menaquinone-4 to undergo protonation and facile double reduction in a low-temperature photoaccumulation protocol. The loss of these contacts may lead to a change in the repositioning of the A-jk(1) α-helix, switchback and random coil to attain a new minimum energy configuration, thereby altering the distance between Leu722 and the carbonyl oxygen of the quinone. A longer distance would result in a weaker H-bond, which would lead to the weaker hyperfine coupling observed by transient EPR spectroscopy (20).

Figure 6.

 Depiction of the region surrounding the A1A quinone, including the A-jk(1) helix that has numerous contacts with PsaF. Top: View from the stromal side in the plane of the membrane. Bottom: View normal to the membrane. The H-bond from Tyr676 to A0A is depicted; the N-terminus of PsaF was deleted in both views for clarity.

In contrast, any explanation for a redirection in electron transfer would need to take into account changes in the ligands or environment of the six excitonically coupled Chls that participate in the initial act of charge separation. It is interesting that when PsaA-Tyr696 is changed to Phe in C. reinhardtii, the amplitude of the fast component attributed to A1B oxidation increases at the expense of the slow kinetic component attributed to A1A oxidation. Conversely, when PsaB-Tyr676 is changed to Phe, the amplitude of the slow kinetic component increases at the expense of the fast kinetic component. These results were interpreted as a decrease in the relative use of the targeted branch (12). PsaA-Tyr696 is located at the proximal end of the α-helical region of the A-jk(1) loop, and is H-bonded to the 131 keto oxygen of A0A, the primary acceptor Chl (Fig. 6). One attractive hypothesis is that in the absence of PsaF, the loss of contact with the A-jk(1) loop causes a wholesale movement of the α-helix, switchback and random coil away from the electron transfer cofactors. This would not only serve to weaken the H-bond from Leu722 to the A1A quinone, which is likely the cause of the acceleration in the slow kinetic phase of A1A oxidation, but it might simultaneously weaken the H-bond from Tyr696 to A0A Chl, which would result in a redirection of electron transfer in favor of the PsaB branch cofactors. Thus, the two observables appear to be inextricably related, and may derive from a common cause, which is a new equilibrium position of the A-jk(1) loop attained in the absence of PsaF. This mechanism would explain the partial loss of the hyperfine couplings (18) and the dominance of the fast kinetic phases (6) observed earlier in spinach PS I particles isolated with Triton X-100: in both instances, the removal of chlorophylls (14,15) and PsaF (13) would be expected in these eukaryotic PS I complexes.

If this hypothesis is correct it is important to also consider how the redirection of electron transfer from one branch to the other could occur. Recently, an analysis of ultrafast transient absorbance data has led to a new model for the initial charge separation (3). In this model, the charge separation is postulated to occur between Chls A-1 and A0 from the excited state of the six excitonically coupled core Chls. Further, it was suggested that the charge separation should be reversible to explain the observation of fluorescence components that are longer lived than the modeled trapping lifetime. Bidirectional electron transfer was not discussed in any detail in Muller et al. (3) but is easily incorporated into this model by postulating that charge separation in either branch can occur from the excited states of the six coupled Chls. If the charge separation is reversible then a change in the forward and reverse rates in one branch quite naturally leads to a relative increase in charge separation in the other branch. Thus, alteration to A0A by weakening the H-bond donated by PsaA-Tyr696 would be expected to shift the charge separation toward the PsaB branch of electron transfer cofactors.

Footnotes

  • **

    The spectroscopic names for the electron transfer cofactors are used throughout this article; the crystallographic names are shown in parenthesis.

Acknowledgments

Acknowledgements— This work was supported by the US National Science Foundation grant MCB-0519743 to D.A.B. and J.H.G., by ACI Jeune chercheur to F.R. from the French Ministry for Research and by NSERC to A.v.d.E.

Ancillary