INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS APPENDIX REFERENCES

In photosynthesis, solar energy is captured by light-harvesting antennae and transferred to the reaction center (RC) where it is used for transmembrane charge separation.

Light harvesting proceeds with a quantum efficiency of almost one. Recent emergence of the structures of a growing number of pigment-protein complexes (Deisenhofer et al., 1985; Kühlbrandt et al., 1994; McDermott et al., 1995; Koepke et al., 1996; Zouni et al., 2001; Jordan et al., 2001) stimulated numerous spectroscopic investigations as well as theoretical studies (van Grondelle et al., 1994; Arnett et al., 1999; Sundström et al., 1999; Schulten, 1999; van Amerongen et al., 2000; and references therein) analyzing relationships between structural features and functional properties.

With photosystem I (PS I) from Synechococcus elongatus, for the first time a high resolution x-ray structure becomes available for a photosynthetic pigment-protein complex that binds both light harvesting antenna pigments and the reaction center cofactors inseparably on the same protein subunits. It is a unique property of the PS I complex that it carries out all the primary photosynthetic processes within one and the same membrane protein complex. Thus, it offers the possibility to study the interplay of light absorption, excited state energy transfer, and trapping by electron transfer within one functional unit.

PS I is a membrane-bound pigment-protein complex that mediates the light-driven electron transfer from reduced plastocyanine or cytochrome c₆ to ferredoxin or flavodoxin (for review, see Golbeck, 1994; Brettel, 1997). The PS I complex of cyanobacteria is composed of 12 subunits. The two largest subunits (PsaA and PsaB) bind most of the antenna pigments and the following redox cofactors involved in the electron-transfer process: the primary electron donor P700 (a heterodimer of chlorophyll a(P_B) and a'(P_A), the primary acceptor A₀ (a Chl a monomer), the secondary acceptor A₁ (a phylloquinone), and F_X (a [4Fe-4S] iron sulfur cluster). The terminal electron acceptors F_A and F_B (two [4Fe-4S] iron sulfur clusters) are both coordinated by subunit PsaC, one of the three extrinsic subunits located on the stromal side.

The recently published x-ray structure of PS I from S. elongatus at 2.5-Å resolution (Jordan et al., 2001) identifies twelve protein subunits, 96 Chls, 22 carotenoids (Car), two phylloquinones, three iron-sulfur clusters, four lipids, ~200 water molecules, and a metal ion (presumably Ca²⁺). Knowledge of the detailed arrangement of the Chls makes it possible to calculate the mutual orientation of the Q_Y optical transition moments and allows to abandon the use of just averaged orientation values for theoretically predicted excitation energy transfer rates between individual pigments. This is a big step forward towards a more realistic description of the excitation energy transfer in PS I. Although, there is still left some uncertainty as long as the Förster overlap integrals (FOI) are not exactly known due to lack of knowledge on the exact spectral line positions and shapes of all Chls embedded in the protein environment and on the local dielectric constant of the protein environment.

Most of the Chls are rather densely packed within two layers at the membrane surfaces. The antenna system has been divided spatially into a central and two peripheral domains (Jordan et al., 2001). The central domain surrounds the five C-terminal -helices of PsaA and PsaB, which coordinate the cofactors involved in electron transfer. It also contains 10 Chls that are located between the stromal and lumenal layers, thereby facilitating excitation energy transfer between them.

A comparison of the mean distances between nearest neighbor Chls shows that the PS I complex is well within the values for other structurally known pigment-protein complexes (Table 1). In all cases the mean nearest neighbor distance is only slightly above the ~9-Å diameter of the chlorin ring of the Chls.

Up to now, various attempts have been made to explain spectral properties of pigment-protein complexes in terms of excitonically interacting chromophores not only for the systems with smaller center-to-center distances (rows 1-3 in Table 1; Sundström et al., 1999; Gradinaru et al., 1998; Knapp et al., 1985), but also for the Femma-Mathews-Olson (FMO)-complex (last row in Table 1) with considerable larger center-to-center distances (Pearlstein, 1992; van Amerongen et al., 2000; Owen and Hoff, 2001).

From the mean nearest neighbor distance of 9.9 Å in the PS I complex, an average interaction energy between neighboring Chls of ~70 cm¹ is calculated, which corresponds to ~6 nm splitting between the upper and lower excitonic bands. Due to the magnitude of the exciton coupling between each pigment and its nearest neighbor the spectral properties may be described by exciton states partially delocalized over the pigments. Therefore, it is essential to take into account excitonic interaction for the characterization of the electronic excited state of antenna pigments. Various spectroscopic observations have been interpreted as indication for excitonic interactions between Chls in PS I from Synechocystis (Gobets et al., 1994; Savikhin et al., 1999, 2000; Melkozernov et al., 2000; Rätsep et al., 2000) and Spirulina platensis (Engelmann et al., 2001).

In addition, the short center-to-center distances between the Chls suggest very fast energy transfer and charge separation kinetics. Indeed, a single step rate constant of energy transfer from Chl to Chl has been estimated to be ~200 fs¹, (Du et al., 1993). The fluorescence lifetime of PS I from S. elongatus has been found to be between 33 and 38 ps in good agreement between various groups (Holzwarth et al., 1993; Dorra et al., 1998; Byrdin et al., 2000; Gobets et al., 2001; Kennis et al., 2001). It has been attributed to the trapping of excitation energy via charge separation. The spectrum associated with the trapping is positive at all wavelengths and was found to be independent of the excitation wavelength (Hastings et al., 1995; Melkozernov et al., 2000). As the Chls in PS I absorb at different wavelengths, excited state energy redistribution processes directed towards thermal equilibration take place, the slowest of which have been detected with transfer lifetimes in the 2- to 15-ps range (Holzwarth et al., 1993; Dorra et al., 1998; Byrdin et al., 2000; Gobets et al., 2001; Kennis et al., 2001). The spectra associated with energy redistribution processes exhibit positive and negative amplitudes related to energy transfer between antenna Chls absorbing at different wavelengths. The shape of these spectra are strongly excitation wavelength dependent (Hastings et al., 1995; Melkozernov et al., 2000).

Regarding the spectral heterogeneity, the most intriguing feature of the PS I complex is that it contains so called "red" chlorophylls that absorb at wavelengths above 700 nm, i.e., at energies below that of the primary donor P700. Time-resolved fluorescence measurements with samples of varying red Chl content (Gobets et al., 2001) as well as numerous previous studies (e.g., Trissl, 1993; Trinkunas and Holzwarth, 1996; Pålsson et al., 1998; Byrdin et al., 2000; Gobets et al., 2001) indicate a crucial role of these red Chls in the kinetics of energy transfer and trapping. At present, it is not clear which Chls are responsible for the long wavelength absorption. Unfortunately, even a high-resolution structure does not allow a reliable assignment of spectral properties to individual Chls. Attempts to use site-directed mutagenesis or single molecule spectroscopy have not been very successful for spectral/structural assignments so far, although promising first steps have been taken in both directions (Soukoulis et al., 1999; Jelezko et al., 2000). Thus, for the time being, simulations, in which the assignment of site energies to each of the Chls is treated as an adjustable parameter, are the method of choice. Nevertheless, with the new structure available a number of useful restrictions can be applied concerning the assignment of site energies (see below).

In this paper, we present the experimentally determined steady-state optical spectra (linear dichroism (LD), circular dichroism (CD), and absorption as function of temperature) and analyze them on the basis of the 2.5-Å structure by application of exciton coupling theory (Pearlstein, 1992). An approach introduced by Fetisova et al. (1996) allows one to study the excited state dynamics of this excitonically coupled system on the basis of Förster theory. We are thus able to simultaneously simulate both absorption and emission properties of PS I with a common set of site energies. This in turn opens the opportunity to use the set of optical spectra as a fitting criterion for the transition energies of the individual (noninteracting) Chls. Recently, this approach yielded convincing results for the FMO-complex (Owen and Hoff, 2001). Besides that, the limits of the excitonic coupling theory and its applicability for the description of the excitation transfer based on Förster theory are discussed and the consequences of the structural organization on the light harvesting process are studied in a quantitative way.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS APPENDIX REFERENCES

Measurement of steady-state optical spectra

Trimeric PS I complexes were prepared from the thermophilic cyanobacterium S. elongatus as described by Fromme and Witt (1998). For absorption and CD measurements the concentrated samples were diluted to a final Chl concentration of ~10 µM with a buffer containing 20 mM Tricine (N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine, pH 7.5), 25 mM MgCl₂, 100 mM KCl, 5 mM sodium ascorbate, and 0.02% n-dodecyl--D-maltoside (-DM). For measurements at cryogenic temperatures, glycerol was added to the samples to a final concentration of 65% (w/v). Absorption spectra were recorded with a spectral resolution of 1 nm on a Cary-1E-UV/VIS spectrophotometer, Varian, Darmstadt, Germany. CD spectra were measured with spectral width of 2 nm on a JASCO spectropolarimeter model J-720, Jasco, Gross-Umstadt, Germany.

For LD measurements, the samples were oriented by squeezing a 1.25 cm × 1.25 cm × 2 cm polyacrylamide gel along two perpendicular axes allowing it to expand in the third direction resulting in dimensions of 1.0 cm × 1.0 cm × 3.1 cm. The LD of the sample is defined as the difference in the absorption of light polarized parallel and perpendicular to the stretching direction of the gel. The light is propagating perpendicular to the stretching direction of the gel. The PS I complexes are disc shaped with two long axis parallel to the membrane plane and a short axis being the crystallographic C₃ axis (z axis in the 1JB0.pdb file), which is perpendicular to the membrane plane. Considering the shape of the PSI complexes they will align with their C₃ axis perpendicular to the stretching direction. The final composition of the gel was 14.5% acrylamide + 0.5% bis-acrylamide in 20 mM Bis-Tris, 20 mM NaCl, and 0.02% ß-DM, at pH 6.5, polymerized with 0.06% N,N,N',N'-tetra-methylethylenediamene, and 0.01% ammonium peroxodisulphate solution. LD spectra were recorded with a spectral resolution of 2 nm on a Beckmann DU 62 spectrophotometer, Bechman Coulter, Unterschleissheim-Lohof, Germany, using a film polarizer (Spindler and Hoyer model 10K, Göttingen, Germany).

For low temperature measurements, the cuvette was placed in a continuous flow helium cryostat (Oxford CF 1204, Wiesbaden, Germany) or liquid nitrogen bath cryostat (Oxford DN1704).

Structure-based calculation of distances, excitonic interactions, and transfer rates between Chls

All calculations were based on the published dataset of the x-ray structure of PS I from S. elongatus with 2.5-Å resolution (1JB0.pdb file; Jordan et al., 2001) containing the coordinates of 96 Chls and 22 Cars. For calculation of center-to-center distances R between Chl molecules, the arithmetic average of the coordinates of the four pyrrol nitrogen atoms have been used as the center of the Chl molecules, instead of the central Mg atoms, as those are located out of the ring plane of the Chls (by 0.35 Å on average). For calculation of closest contacts of the  systems R between different Chls, and between Chls and Cars, first the distances from any atom of the -system of one molecule to any atom of the -system of the other molecule (Fig. 1) were calculated and afterwards, the minimum of these values was selected, indicating the actual closest contact of the respective -systems.

View larger version (26K): [in this window] [in a new window]   FIGURE 1   Molecule structure of Chl and Car. -systems are shaded. In the Chl scheme, additionally the IUPAC ring nomenclature is indicated. The QY transition moment lies approximately along the axis connecting the nitrogen atoms of rings A and C, respectively (van Amerongen et al., 2000).

A structural factor s_ij was calculated using the center-to-center distances R_ij and the orientation factor  as follows: s_ij = _ij/R, in which  is defined by  = (â · )  3(â · )( · ), with â and being unit vectors in the directions of the acceptor and donor transition dipoles and in the direction of the line joining the dipole centers.

Förster transfer rates k_ij and excitonic coupling strengths J_ij between Chls i and j are calculated based on the structural factor s_ij as follows:

(1a)

and

(1b)

The following values for the parameters are chosen: natural fluorescence lifetime  = 15 ns (Colbow, 1973), and oscillator strength of the Chl Q_Y (0-0) transition µ² = 21.5 D² (Shipman, 1977). C takes into account the influence of the medium and is given by C = (n² + 2)²/9n² (van Amerongen et al., 2000). Assuming n = 1.5 for the protein matrix, this gives C = 0.89. The Förster radius R₀ defines the distance for which the rate of energy transfer becomes equal to the intrinsic rate of radiative decay. For Chls with identical Q_Y transition energies ("isoenergetic"), a value of R₀ = 74 Å (Byrdin et al., 2000) has been used. The factor 0.667 in the denominator of Eq. 1a) corrects for the average ² contained in R.

Simulation of excitonic absorption, linear, and circular dichroism spectra

We analyze the experimental spectra with an excitonic model, which has been described previously, e.g., by Pearlstein (1992). The Hamiltonian of the system H is given by the coupling matrix containing all the 96 × 96 two-pigment coupling energies J_ij, and, on the main diagonal, the monomer transition energy of Chl i, v_i. We diagonalize H to find its eigenvalues and eigenvectors. The thus defined transitions may differ in energy, direction, and oscillator strength from those of the individual noninteracting Chls. The energies v (in cm¹) of the 96 all-complex exciton states are given by the eigenvalues of H. The components c_iK of the corresponding eigenvectors c_K = (c_1K, ... , c_iK, ... , c_96K) describe the contribution of pigment i to the exciton state K. Following Durrant et al. (1995) and Fidder et al. (1991), a delocalization parameter N_K = 1/_i(c_iK)⁴ is calculated, which equals one for monomers, two for dimers, etc.

Steady-state optical spectra arising from excitonic coupling of all individual transition moments µ_i are calculated as follows (Pearlstein, 1992):

(2a)

(2b)

(2c)

(2d)

(2e)

Here µ is the dipole strength of the transition moment on the ith molecule, _i is the unit vector in the direction of that transition, c_iK is the ith element of the eigenvector for the Kth exciton state, v is the transition energy (in cm¹), R_ij is the distance (in nm) from the center of the transition charge distribution of the ith molecule to that of the jth, _ij is the unit vector in the direction of that vector, is the unit vector in the direction of the membrane normal (z axis). µ is the dipole strength of the component of the Kth exciton transition moment perpendicular to the membrane plane, i.e., parallel to the membrane normal . The membrane normal coincides with the crystallographic C₃ axis (see the trimerization axis in Fig. 4). µ is the dipole strength of the component of the Kth exciton transition moment parallel to the membrane plane. _K is the rotational strength of the Kth exciton transition moment. Note, that the indices parallel and perpendicular refer to the membrane plane, which is oriented parallel to the stretching direction and not to the membrane normal entering Eqs. 2c and 2d.

For presentation, absorption, and linear dichroism stick spectra have been dressed by Gaussians with 360 cm¹ (~18 nm) full width at half maximum (FWHM), CD stick spectra, due to their narrowed nature, with 300 cm¹ (~15 nm) FWHM, and emission spectra with 400 cm¹ (~20 nm) FWHM. The amplitudes of these simulated spectra contain information on oscillator and rotational strength of the involved transitions.

Simulation of excitation energy transfer and trapping dynamics

The excited state dynamics of the PS I antenna-RC complex can be described by a set of coupled differential equations, the solution of which is found from the eigensystem of the system's 96 × 96 transfer matrix T (Byrdin et al., 2000). Consideration of excitonic coupling within the complex requires some modifications of T that are described below.

T is composed of transfer rates between pigments i and j, and on the main diagonal, depopulation rates. The transfer rate is the product of a structural factor, a spectral factor, and a scaling parameter defining the time axis. The structural factor k_ij is given by Eq. 1a and describes the transfer rate between Chls with identical transition energies, whereas the spectral factor is a normalized FOI between donor emission and acceptor absorption (for isoenergetic Chls FOI  1). For the calculation of the FOI, the Gaussian decomposition of the absorption spectrum of chlorophyll in diethyl ether by Shipman et al., 1976 has been slightly modified. The short-wavelength Gaussian corresponding to the 630-nm band has been slightly increased in amplitude and width with respect to the Q_Y (0-0) band to reproduce the spectral shape of Chl in micelles of ß-DM, which are expected to better mimic the biological membrane environment.

Chl emission spectra were obtained as mirror images of these absorption spectra. The differences to Chl emission spectra resulting from application of the Stepanov relation are negligible (compare Laible, 1995; Laible et al., 1998). A uniform Stokes shift of 130 cm¹ (~6.5 nm) has been used to ensure detailed balance (Laible, 1995; Laible et al., 1998). The depopulation rate k_i of the excited state i is calculated as _j k_ij with j = 1, ... , 96, in which k_ii is the rate for the intrinsic decay by fluorescence (0.5 ns¹ fixed) and, for both sites constituting P700, by charge separation (parameter k_CS). The inverse mean of all k_i is called "single step hopping time" and provides a measure of the connectivity between Chls within the PS I complex.

Lifetimes of excitation energy transfer and trapping processes _i have been determined from the eigenvalues of the transfer matrix T. The corresponding decay associated amplitudes result from the respective eigenvectors after weighting with the initial excitation distribution. Excitation was simulated as nonselective and stoichiometric, i.e., all pigments are excited with equal probability (except where otherwise indicated). A plot of the Gaussian dressed amplitudes weighted with the respective normalized oscillator strength versus the Stokes shifted transition energies of the states results in decay associated spectra (DAS). The steady-state emission spectrum is given by the lifetime weighted integral over all decay associated spectra and was calculated as _i DAS_i × _i. Of the 96 lifetime components obtained from the solution of the transfer equations, all with the exception of the very slowest few are characterized by lifetimes <2 ps and vanishing amplitudes. They mainly describe the exchange of excitation between spatially and/or spectrally neighboring states. Such processes are experimentally not resolvable, even with the high time resolution available at present. The spectrum resulting from summing up the spectra of all these processes is characteristic for very fast energy transfer processes in the bulk antenna directed towards thermal equilibration.

Alternatively, based on the transfer matrix T, Monte Carlo simulations of a random walk of the excitation through the complex have been performed. They resulted in essentially the same lifetimes but allowed in addition to follow the individual trajectory of the excitation.

The effects of excitonic coupling have been taken into account in the following way. The delocalized exciton states of strongly coupled (interaction energy >95 cm¹) aggregates are regarded as donors and acceptors, respectively. All weaker couplings are neglected as being ruled out by the site inhomogeneities. Transfer rates from/to exciton sites have been calculated as described by Fetisova et al., (1996). In short, the structural factor s² is averaged over the s²-factors of the exciton bands at the locations of the monomers and the FOI for the transfer from/to the exciton states are weighted with the normalized oscillator strength of the respective exciton band. To ensure fast thermal equilibration between exciton states, the relaxation rate from the higher exciton state has been set to 5 ps¹ and the reverse rate has been calculated from this according to Boltzmann. This results in an equilibration time between upper and lower exciton band of ~100 fs as found in the RC from purple bacteria (Vos et al., 1997) and PSII (Klug et al., 1998).

Thus, it becomes possible to simultaneously simulate all the spectral properties along with the excitation energy transfer kinetics of PS I based on a minimal set of common parameters consisting besides the two time scaling parameters (Förster radius R₀ for excitation transfer and primary charge separation rate k_CS for electron transfer) only of the assignment of transition energies to all the structurally defined Chls. As there are 96 of these, the parameter space appeared too multidimensional for a conventional fit routine. We therefore implemented a simulation program (in Mathematica 3.0) that uses the individual Chl transition energies as input values for the calculation of spectral properties and excitation transfer kinetics. Comparison of simulated and experimental optical spectra provided a criterion for the quality of the structural/spectral assignment. The assignment strategy was basically guided by symmetry relations, considerations regarding the axial ligands and H bonding to the Chls, functional arguments concerning the location of the red pigments (see below), and fine tuning of transition energies of the bulk pigments to improve the simulations of the optical spectra. The structural/spectral assignment thus determined (listed in Table 2) represents the distribution of transition energies that was found to agree best with the experimentally determined optical properties of the PS I complex.

View this table: [in this window] [in a new window]   TABLE 2   Position (A1, for example, is located in the stromal layer at 5 o'clock relative to the trimerization axis indicated by the black triangle (= 12 o'clock), see also Figs. 4 and 5), orientation ( is the angle between the transition moment and the membrane normal), QY transition energy (= color in nm), number (#) of and distance to the next carotenoid (Car), probability of being the last antenna Chl, which is visited before the excitation enters for the first time the RC calculated for homogeneous and heterogeneous site energies for the Chls in PS I (the asterisk indicates that these Chls belong to coupled aggregates if excitonic interactions >95 cm1 are taken into account)

We are aware of certain limitations of the applied simulation methods. For the sake of simplicity, the following assumptions have been made. 1) Dynamic disorder effects caused, e.g., by thermal fluctuations of Chl positions and orientations have been neglected. 2) Only Q_Y transitions have been taken into account, and possible influence of charge transfer interactions has been neglected. 3) The point dipole approach was used, although this approach may not be satisfactory where the distances between donor and acceptor are comparable with the size of the molecules (Table 3). Deviations between the interaction energies calculated by more accurate methods such as the transition monopole approximation (Chang, 1977) or the density cube method (Krueger et al., 1998) and the point dipole approach can be significant depending on the distance between the chromophores and the mutual orientation (see Appendix). However, given the lack of experimental data for the actual Coulombic interaction energies together with insufficient knowledge on the exact transition charge density distributions underlying these calculations, it appears hard to decide which of the various values is most appropriate. We therefore stick to the point dipole approach, which requires incomparably less computational efforts. 4) Above that, at distances between chromophores comparable with the van-der-Waals distance energy transfer based on the Dexter mechanism (Dexter, 1953) may become relevant, which was also excluded from consideration here.

View this table: [in this window] [in a new window]   TABLE 3   Characteristics of the strongest coupled dimers in PS I: excitonic interaction energy calculated in the point dipole approximation (J < 0  sum band () lies energetically lower), cosine of the angle  between the transition dipoles, ratio of the oscillator strength (cos  < 0  less oscillatory strength in the sum band), center-to-center and edge-to-edge distance and excitonic interaction energy calculated in the extended dipole approximation

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS APPENDIX REFERENCES

Steady-state absorption spectroscopy of the PS I complex

Previously reported spectroscopic data for PS I from S. elongatus (e.g., Pålsson et al., 1998) have been complemented by steady-state absorption measurements as a function of polarization and of temperature to obtain additional information to what extent excitonic interactions contribute to the spectral properties. Fig. 2 shows the absorption spectra of the PS I complex from S. elongatus at 5, 80, 160, 220 K, and at room temperature (RT).

View larger version (23K): [in this window] [in a new window]   FIGURE 2   Steady-state absorption of the trimeric PS I core complex from S. elongatus at various temperatures. Note the "isosbestic" points at 657, 688, and 705 nm, and the redistribution of oscillator strength into the red with decreasing temperature.

Besides the broad main peak centered at ~680 nm additional spectral features can be distinguished in the long wavelength region. A shoulder observed at ~710 nm at higher temperatures is resolved as a separate peak at lower temperatures. Generally, the low temperature spectra show distinctly more structured features than the spectra above 100 K. Both on the short- and on the long-wavelength side of the main peak separate features begin to emerge with decreasing temperature. Three "isosbestic" points can be distinguished at 657, 688, and 705 nm. Whereas the total area enclosed by the spectra stays essentially constant, a considerable redistribution of absorption can be noticed. Especially pronounced is the absorption increase with decreasing temperature in the red region above 705 nm at the expense of the 688- to 705-nm region. A weaker effect is observed on the blue side of the main peak: the area in the region between 657 and 688 nm increases at the expense of the region <657 nm upon lowering the temperature. Fig. 2 shows unambiguously that the amount of long wavelength absorption in the complex is temperature dependent. Such redistribution of absorption as a function of temperature has been found also for PS I from Synechocystis sp. (Rätsep et al., 2000) and Spirulina pl. (Cometta et al., 2000). It may be explained by an appropriate temperature dependence of the dielectric constant  of the protein environment. Charge transfer interactions could be involved as well (Beekman et al., 1997; Rätsep et al., 2000).

Fig. 3 shows polarized steady-state spectra of the PS I complex from S. elongatus. In Fig. 3 a the LD spectrum at RT is presented. For comparison, the absorption spectrum is also plotted as a dashed line. Note the different ordinate scales for the two curves. Linear dichroism spectra display the difference in absorption of light polarized parallel and orthogonal to the stretching direction of the gel. Due to the shape of the PS I complexes, they will align with their C₃ axis perpendicular to the stretching direction (see Materials and Methods). The C₃ axis is parallel to the membrane normal and has been chosen as z axis in the 1JB0.pdb file. Chlorophylls oriented in the PS I complex with their transition moments parallel to will absorb incident light polarized orthogonal to the stretching direction of the gel. On average, one-half of the chlorophylls with transition moments lying in the membrane plane will absorb incident light polarized parallel to the stretching direction.

View larger version (30K): [in this window] [in a new window]   FIGURE 3   Steady-state polarized absorption of the trimeric PS I complex from S. elongatus. (a) Linear dichroism (solid) at RT, for comparison the nonpolarized absorption from Fig. 2 (divided by 10) is shown as well (dashed). (b) Circular dichroism at RT (solid) and at 77 K (dashed). Note that only at the red side a separate peak appears at low temperature.

The maximal amplitude of the observed LD (Fig. 3 a) is approximately one-tenth of the total absorption, small negative amplitudes are observed at wavelengths <665 nm. Hence, for the latter spectral region, absorption due to transition moments with preferential orientation parallel to the membrane plane are quite well compensated by absorption due to transition moments oriented perpendicular to the membrane plane. In contrast, for longer wavelengths there is an excess of parallel absorption. The maximum of this positive LD is reached at ~685 nm, i.e., 5 nm to the red of the absorption maximum. This shows that transitions at longer wavelengths lie, in tendency, parallel to the membrane plane. The contribution of the 96 Chls to the LD signal is given by the squared cosine of the angle  between the transition moment and the membrane normal (see column 3 of Table 2). Fig. 3 b shows the CD spectra of the PS I complex from S. elongatus at 77 K and at RT. The CD = A_L  A_R at low temperature is larger than that at RT. Both spectra exhibit the positive lobe on the high energy side. The areas enclosed by the two lobes of each spectrum and the x axis are not equal, a phenomenon referred to as nonconservativity. In our case, the area of the negative lobe is about twice that of the positive one. Interestingly, the redistribution of absorption to the red with decreasing temperature found in the absorption spectrum (Fig. 2) is observed in the CD spectrum as well (compare the position of the side band in the 77-K spectrum and that of the shoulder in the RT spectrum in Fig. 3 b.) Note that the emergence of a separate peak with decreasing temperature is observed only on the low-energy side of the spectrum. This feature can be attributed to the low exciton band of coupled Chls giving rise to long wavelengths absorption. The corresponding CD from the high energy exciton band seems obscured by the main positive peak. Finally, the blue shift of the zero-crossing point with decreasing temperature is noteworthy. CD spectra for other cyanobacteria have been presented by van der Lee et al. (1993) at 77 K for Synechocystis and at RT by Cometta et al. (2000) for Spirulina pl. Both of these spectra are similar to the ones presented here: smaller positive lobe on the high-energy side and larger negative lobe on the low-energy side.

Structural organization of the PS I complex from a functional point of view

Structural elements of the light harvesting antenna

View larger version (46K): [in this window] [in a new window]   FIGURE 4   Arrangement of the pigments within the PS I complex from S. elongatus (a) top view, (b) side view; carotenoids are shown completely and marked with a capital C, Chls are represented by three atoms: the central magnesium and two nitrogens. The connecting line between the nitrogens defines the direction of the QY transition dipole. The coding by different shades of gray means: light gray-stromal, dark gray-membrane buried, black-lumenal. Chls are marked by the letter of the binding subunit and a running number. The trimerization axis is on top (black triangle). The six Chls in the center belong to the RC, the rest to the core antenna. The two vertical lines in b indicate separation of antenna Chls coordinated by PsaA/B into a central part and peripheral parts as suggested by Jordan et al. (2001).

1

1

1

View larger version (43K): [in this window] [in a new window]   FIGURE 5   Top view of separate Chl layers (a) stromal, (b) lumenal; coding as in Fig. 4. Chls indicated in Italics have no symmetric counterpart on the respective other protein subunit.

Network of energy transfer pathways

1

1

1

1

View larger version (77K): [in this window] [in a new window]   FIGURE 6   Scheme of energy transfer pathways within the PS I complex from S. elongatus. For clarity, the lumenal ring and the stromal ring are displayed as an outer ring in dark gray and an inner ring in light gray. Chls in the middle of the membrane are shown between the rings in medium gray, with the exception of the Chls in the RC, which are presented in the center. The remaining peripheral Chls have been placed outside the outer ring. All transfer rates >0.3 ps1 are shown; line thickness encodes transfer speed: thin, 0.3 to 3 ps1; medium, 3 to 11 ps1; thick, >11 ps1; wavy lines indicate enhancement of transfer rates via modified orientation by excitonic interaction, dashed lines show reduction of transfer rates by the same mechanism. Chls in bold type are visited last with a probability of >1.0% before the excitation enters the RC for the first time. Squares instead of circles indicate major symmetry breakings between PsaA and PsaB.

1

Excitonic interaction in the PS I complex

Modeling optical spectra of PS I at 295 K taking into account excitonic interactions

1

1

1

1

View larger version (24K): [in this window] [in a new window]   FIGURE 7   Two attempts to reproduce the absorption spectrum of the trimeric PS I complex from S. elongatus in the QY region. (a) Measured absorption spectrum of trimeric PS I (dash-dot) and absorption spectrum of Chl a in 80% acetone/water (v/v). (b) Gaussian band representing the QY absorption band of 96 spectrally identical Chls in PS I without excitonic coupling (solid) and with excitonic coupling (dashed). (c) Absorption of 96 Chls in PS I with heterogeneous site energies without excitonic coupling (solid) and with excitonic coupling (dashed). Taking into account spectral heterogeneity and excitonic coupling allows us to reproduce the experimental absorption spectrum of PS I (compare dash-dotted line in a and dashed line in c).

View larger version (18K): [in this window] [in a new window]   FIGURE 8   Comparison of simulated (solid lines) and measured (dotted lines) steady-state spectra (a) absorption, (b) linear dichroism, (c) circular dichroism, and (d) fluorescence emission. Whereas a and b show rather good agreement, c and d cannot reproduce the width of the experimental spectra, possibly due to the simple Gaussian dressing method used. The simulations takes into account the site energies of the Chls as in Table 2 and all excitonic interactions as calculated from the 2.5-Å structure.