Localization of Two Phylloquinones, QK and QK′, in an Improved Electron Density Map of Photosystem I at 4-Å Resolution*

An improved electron density map of photosystem I from Synechococcus elongatus calculated at 4-Å resolution for the first time reveals a second phylloquinone molecule and thereby completes the set of cofactors constituting the electron transfer system of this iron-sulfur type photosynthetic reaction center: six chlorophyll a, two phylloquinones, and three Fe4S4 clusters. The location of the newly identified phylloquinone pair, the individual plane orientations of these molecules, and the resulting distances to other cofactors of the electron transfer system are discussed and compared with those determined by magnetic resonance techniques.

An improved electron density map of photosystem I from Synechococcus elongatus calculated at 4-Å resolution for the first time reveals a second phylloquinone molecule and thereby completes the set of cofactors constituting the electron transfer system of this iron-sulfur type photosynthetic reaction center: six chlorophyll a, two phylloquinones, and three Fe 4 S 4 clusters. The location of the newly identified phylloquinone pair, the individual plane orientations of these molecules, and the resulting distances to other cofactors of the electron transfer system are discussed and compared with those determined by magnetic resonance techniques.
The electron transfer processes of oxygenic photosynthesis, as observed in cyanobacteria, eukaryotic algae, and higher plants, involve two distinct types of photosynthetic reaction centers located in the thylakoid membrane. Photosystem II catalyzes the light-driven luminal oxidation of water and the reduction of plastoquinone near the stromal side of the photosynthetic membrane. Photosystem I (PSI) 1 luminally oxidizes the soluble electron donor plastocyanin (alternatively cytochrome c 6 ) and stromally reduces the extrinsic electron acceptor ferredoxin or flavodoxin. The reduced ferredoxin induces the reduction of NADP ϩ , a reaction catalyzed by ferredoxin: NADP ϩ reductase. Photosystem I receives electrons from photosystem II via an intermediate plastoquinone pool, the cytochrome b 6 /f complex, and water soluble electron carriers. The difference in proton concentration across the thylakoid membrane, which results from the proton pumping of the plastoquinone pool and the cytochrome b 6 /f complex, the stromal consumption of protons by NADP ϩ reduction, and the luminal release of protons following water oxidation, is used by the ATP-synthase for phosphorylation of ADP to ATP (1,2).
Cyanobacterial PSI consists of 11 subunits referred to as PsaA to PsaF and PsaI to PsaM. An x-ray structural model of a cyanobacterial PSI complex from the thermophile Synechococcus elongatus has been postulated on the basis of an electron density map calculated at 4-Å resolution (3,4). Despite the comparatively low resolution, it was possible to suggest an assignment of 43 ␣-helices to the individual subunits of PSI by correlating the information provided by the electron density map with available biochemical and biophysical data. Furthermore, the electron density map allowed the positions of 89 Chl a molecules, constituents of both the core antenna system and electron transfer system, one phylloquinone, and three ironsulfur clusters to be modeled.
The electron transfer reactions of PSI are initiated through excitation of the primary electron donor P700 positioned near the luminal side of the membrane-integral complex. Structurally, P700 consists of a chlorophyll a dimer (eC 1 /eC 1 ), whose mutually parallel dihydroporphyrin ring planes are aligned with the membrane normal. Upon excitation, P700* passes an electron to the primary electron acceptor A (probably eC 2 or eC 2 ; see below). Spectroscopically, the first electron acceptor has been identified as A 0 , in all probability one (though possibly either) of the pair of Chl a monomers denoted eC 3 and eC 3 in the structural model of PSI (3,4). This process occurs with a rate constant of about 5⅐10 11 s Ϫ1 (5). The charge separation P700 . ϩ A 0 . is spatially extended across the membrane by electron transfer from the radical A 0 . to the next electron acceptor spectroscopically referred to as A 1 ; the rate constant is estimated to be 2-5⅐10 10 s Ϫ1 (for a review, see Ref. 5). A 1 is now generally agreed to be a phylloquinone (6,7). Due to the difficulty of locating the small phylloquinone molecules in low resolution electron density maps and because of the stability of the radical state P700 . ϩ A 1 . , the position and orientation of A 1 relative to the PSI holocomplex has recently received increased attention, especially by improved EPR techniques. These have, inter alia, determined the distance between A 1 . and P700 . ϩ to be ϳ25.4 Å (8,9,10). A relative position for A 1 was derived through orientation-dependent pulsed EPR measurements on PSI single crystals (10). Geometrically, this position was found to correspond to Q K , a single phylloquinone assigned to a well defined pocket in the earlier electron density map (4). The assignment * This work was supported by the Deutsche Forschungsgemeinschaft (Sonderforschungsbereich 312), the Bundesministerium fü r Bildung und Forschung, the Fonds der Chemischen Industrie, and the Deutsche Agentur fü r Raumfahrt-Angelegenheiten. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Present address: Gesellschaft fü r Biotechnologische Forschung, Mascheroder Weg 1, D-38124 Braunschweig, Germany.
¶ To whom correspondence should be addressed. Tel.: 49 30 838 6326; Fax: 49 30 838 6702; E-mail: nkrauss@chemie.fu-berlin.de. 1 The abbreviations used are: PSI, photosystem I; PsaA and PsaB, large, central subunits of PSI, encoded by genes psaA and psaB; Є (a; b), angle between vectors a and b; a,b-plane, crystallographic plane parallel to the membrane plane; C 2 (AB), axis of pseudo-2-fold symmetry relating subunits PsaA and PsaB and also respective branches of the electron transfer system; Chl a, chlorophyll a; c-axis, crystallographic c-axis parallel to the membrane normal; eC 1 and eC 1 , luminal Chl a cofactors of the electron transfer system and its pseudosymmetric counterpart; eC 1 , pertaining to both eC 1 and eC 1 ; eC 2 and eC 3 , second and third pair of Chl a cofactors of the electron transfer system; eC X , eC Y , distances between named cofactor pairs (averaged value of pseudosymmetric branches); Q K and Q K , phylloquinone cofactors of the electron transfer system; F 1 and F 2 , preliminary x-ray structural model names for F A and F B (F B and F A ); P700, A 0 , A 1 , F X , F A , and F B , spectroscopically identified cofactors of the electron transfer system of PSI as follows: primary electron donor (dimer of Chl a molecules), primary (single Chl a), secondary (phylloquinone), intermediate, and two terminal (Fe 4 S 4 clusters) electron acceptors; m, n, and o and m, n, and o, ␣-helix nomenclature. of this position, however, remained internally uncorroborated, since an expected pseudosymmetrically positioned second phylloquinone could not be identified at the time.
The three terminal cofactors of the electron transfer system are iron-sulfur centers, F X being closest to P700, followed by F 1 and F 2 (we retain this nomenclature, for the present, to emphasize the remaining structural ambiguity in their assignment to the known cofactors F A and F B , although see Refs. 11-13 as well as Ref. 14 for recent results correlating F A with F 1 and F B with F 2 ).
In the following, we describe an improved model of the electron transfer system of PSI based on the present electron density map at 4-Å resolution (14). This map reveals the position of the second phylloquinone molecule and allows the spatial positioning of all 11 cofactors of the electron transfer system of PSI. The positions and orientations of individual cofactors are discussed and compared with structural information derived from spectroscopic data.

EXPERIMENTAL PROCEDURES
Calculation of an Improved Electron Density Map-The phases for the electron density map presented here were derived using essentially the same data described previously (4), although a new native data set with a resolution of 3.5 Å and an additional mercury derivative data set have been included (14). Using the program SHARP (15) instead of the earlier combination VECREF/MLPHARE (16,17) and including a total of five heavy atom derivative data sets, it was possible, by incorporating new minor sites, to derive a significantly improved heavy atom model. The program SOLOMON (16) has been employed in the solvent flattening procedure. Due to the low diffraction quality of heavy atom derivative crystals, experimentally obtained phase information is still limited to a resolution of 4 Å. Since no additional phase information at higher resolution could be achieved by phase extension using density modification techniques, the electron density map was calculated at a resolution of 4 Å. It reveals more detailed information on the polypeptide chain folding than previous maps as well as the complete cofactor set of PSI. For the detailed procedure and statistics for the determination of this electron density map, see Ref. 14.
Model Building-The previously reported model of the electron transfer system (3,4) has been used as a basis for the present cofactor model. The Chl a head groups are visible as almost quadratically flat density pockets. The positions and orientations of the Chl a molecules are modeled by 4-fold symmetrical porphyrin moieties, since the present resolution does not permit their asymmetric features to be defined unambiguously. Similarly, the phylloquinone molecules are represented by their naphthalene moieties to interpret the corresponding elongated ellipsoidal electron density. Neither the phylloquinone side chains nor the oxygen atoms have been included in the model.
Chl a cofactors of the electron transfer system were placed into the electron density using the program O (18), and their positions were optimized using the real space refinement procedure as provided by this program.

Distances between Cofactors and Associated Errors-For
Chl a molecules center-to-center distances were calculated between the central Mg 2ϩ ions, while for iron-sulfur clusters and phylloquinones the centroid of the cluster and naphthalene model, respectively, have been used. Edge-to-edge distances of cofactors important for the kinetics of electron transfer were determined between the outer atoms of the porphyrin, naphthalene, and iron-sulfur cluster models, respectively. For iron-sulfur clusters, edge-to-edge distances have been determined between the iron and sulfur atoms of the clusters, as modeled.
The estimated errors for center-to-center and edge-to-edge distances are on the order of Ϯ1 and Ϯ 2 Å, respectively, the latter reflecting the larger uncertainties in the orientations of the planar cofactors within their molecular planes.

Two Symmetrically Arranged Density Pockets Assigned to the
Positions of the Phylloquinones-In our previous x-ray structural model of PSI (4), only a tentative positional description of a single phylloquinone was included, assigned to an electron density pocket located between F X and eC 3 . The lack of a second, pseudosymmetrically positioned phylloquinone, however, prevented an internal corroboration of this identification.
The new electron density map now reveals two such electron density structures symmetrically positioned on either side of the pseudo-2-fold rotation axis C 2 (AB) and located between eC 3 and F X . These have been assigned to the phylloquinone electron acceptors Q K and Q K (Fig. 1). The latter is equivalent to the position Q K identified previously (4). Note that following our earlier convention of priming cofactors coordinated by primed ␣-helices, the position previously denoted Q K will be renamed Q K (coordinated by ␣-helices m-n), while the new second phylloquinone will be referred to as Q K (coordinated by m-n). Q K (Q K ) is situated slightly luminally of and close to the N terminus of ␣-helix n (nЈ) and immediately adjacent to the loop connecting ␣-helices m and n (mЈ and nЈ). The corresponding electron density is clearly separated from that of the neighboring ␣-helices (Fig. 1). Facing away from the loop m-n (mЈ-nЈ), each phylloquinone is additionally delimited by the long loop n-o (nЈ-oЈ) connecting the C-terminal end of n (nЈ) to the stromal end of o (oЈ).
In addition to Q K and Q K , a significantly more symmetrical arrangement of ␣-helices and connecting loops on either side of the pseudo-2-fold axis C 2 (AB) is now apparent in it vicinity as compared with the previously published electron density map. Whereas the earlier model of the ␣-helix m almost passed through the position now assigned to Q K , the stromal end of m now has a comparable inclination relative to the membrane normal as its pseudosymmetric partner mЈ. The loops m-n (mЈ-nЈ) connecting ␣-helix m (mЈ) to the "surface" ␣-helix n (nЈ) are similar in both shape and length (Fig. 2).
The electron densities of both Q K and Q K are elongatedly ellipsoidal (Fig. 1). As a result, the long molecular axis of the naphthoquinone moiety may be identified with some confidence. However, the plane orientation as well as the quinone oxygen atoms remain indeterminate. As a result, the phylloquinone molecules have been modeled by their naphthalene backbones only. These naphthalene models were placed into the electron density optimizing their positions and the orientations of their long molecular axes. The molecular plane of Q K was then rotated around the long axis to align the molecular plane with the vector eC 1 -Q K , to account for the observation that the carbonyl O-O-axis is approximately aligned with the FIG. 1. Stereo view of the electron density structures interpreted as Q K and Q K . They are largely separate from the surrounding electron density corresponding to ␣-helices, inter-helical loops and eC 3 , eC 3 . This figure was produced using O (18). vector P700 . ϩ -A 1 . (19). Since the electron spin density is primarily located on either eC 1 or eC 1 (20) (although which one remains to be clarified), the procedure was repeated to align the molecular plane of Q K with the vector eC 1 -Q K . Similarly, two plane orientations were obtained for Q K .
The long molecular axes of both Q K and Q K are observed to be inclined by 13 Ϯ5°relative to the membrane plane (equivalent to the crystallographic a,b-plane). Projected onto the a,b-plane, the long molecular axis of Q K (Q K ) describes an angle of 18°(60°) to the crystallographic a-axis. The axes of Q K and Q K form an angle of 42°with each other.
The Iron-Sulfur Clusters-The positions of the iron-sulfur clusters correspond to the highest electron density observed (21). F X was tentatively modeled by fitting a Fe 4 S 4 cluster into the electron density. Contouring the electron density map at 11 S.D. above the mean density reveals a tetrahedrally distorted electron density structure associated with F X (Fig. 3). The most likely explanation is, that this tetrahedron is equivalent to the arrangement of the four iron atoms of the Fe 4 S 4 cluster. Modeling a Fe 4 S 4 cluster into this tetrahedral shape results in a good structural match, while the four sulfur atoms lie outside the contour, in agreement with their lower density of electrons. Interestingly, the derived orientation of F X upholds the 2-fold symmetry of C 2 (AB), a fact that had been assumed on grounds of symmetry yet had remained unsubstantiated. The observation that the g XX principal axis of the g tensor of reduced F X is oriented perpendicular to the thylakoid membrane (22) now favors one of two alternative assignments of g tensor axes to the distorted cubane structure of Fe 4 S 4 clusters. According to EPR studies on Fe 4 S 4 model compounds (23,24), our structural model and the EPR results are in agreement with the assignment, where each of the three principal magnetic axes is normal to one of the mutually orthogonal faces of the distorted Fe 4 S 4 cube (25).
The orientations of F 1 and F 2 have been inferred from the 2Fe 4 S 4 ferredoxin structure from Peptostreptococcus asaccharolyticus (26) used as a model for PsaC (27). They are in agreement with those derived independently by EPR experiments on PSI single crystals (28). For F 1 and F 2 , such tetragonally shaped density structures as observed in the case of F X are not evident as the electron density "outside" the membrane-integral region is less well defined.

DISCUSSION
Overall Cofactor Arrangement-The electron transfer system of PSI constitutes the innermost cylindrical core of the larger, membrane-integral photosynthetic reaction center complex. A set of 10 ␣-helices, five from each of the two central subunits, PsaA and PsaB, tightly encloses the electron transfer FIG. 2. The core of the reaction center of PSI. a, view direction perpendicular to the C 2 (AB)-axis onto median plane of all cofactors of the electron transfer system. The palisade of ␣-helices surrounding the electron transfer system and the loops connecting these ␣-helices are rendered in black and gray. The luminal loop i-j could not unambiguously be located in the electron density map. The naphthalene backbone of the modeled phylloquinone orientation as well as all remaining cofactors of the electron transfer system are depicted in white. b, view direction from the stromal side onto the membrane plane. The pseudo-C 2 (AB)-axis passes through the center of the Fe 4 S 4 cluster F X . For clarity, only the stromal loop region j-k coordinating the iron sulfur cluster F X is shown. This figure was produced using Setor (37) .   FIG. 3. Stereoscopic depiction of the electron density of F X contoured at 1.2 and at 11 above mean density. The tetragonal shape of F X in the 11 contour reveals the absolute orientation of F X observed to comply with the pseudosymmetry of C 2 (AB) (arbitrarily oriented axis C 2 (AB) not shown). For F 1 and F 2 , such tetrahedrally shaped density structures are not evident. This figure was produced using BobScript (38). system, separating it from the surrounding antenna system (4). The electron transfer system itself consists of two symmetrically arranged cofactor branches (Fig. 2). Now that all cofactors of the electron transfer system have been identified, this pseudosymmetry is seen to encompass the whole of the membrane-integral region, extending from the pair eC 1 /eC 1 near the luminal side to Q K /Q K near the stromal side (Fig. 2). The iron-sulfur cluster F X is located on the pseudo-2-fold axis C 2 (AB), completing the symmetrical arrangement at the stromal edge of the membrane-integral subunits. Merely the two stromal iron-sulfur clusters F 1 and F 2 (F A and F B ), coordinated by the extrinsic subunit PsaC, do not adhere to this 2-fold symmetry.
In the direction parallel to the membrane normal, the membrane-integral cofactors divide the membrane into four sections of roughly comparable width, here denoted eC 1 -eC 2 , eC 2 -eC 3 , eC 3 -Q K , and Q K -F X . The height difference for eC 1 -eC 2 , eC 2 -eC 3 , eC 3 -Q K , and Q K -F X amount to 5.9, 8.6, 7.8, and 8.8 Å, respectively, while the total distance eC 1 -F X is 31.1 Å. This corresponds to fractional distances of 0.19, 0.28, 0.25, and 0.28, respectively (Fig. 4a).
Photovoltage measurements on oriented PSI thylakoid membranes estimated values of fractional dielectrically weighted transmembrane distances of 0.62 for P700-A 0 (compare with eC 1 -eC 3 , 0.47), 0.16 for A 0 -A 1 (compare with eC 3 -Q K , 0.25), and 0.22 for A 1 -F X (compare with Q K -F X , 0.28) (29, 30). These relative distances, especially for the pair P700-A 0 , do not correspond to the x-ray structural model distances eC 1 -eC 3 as well as one might have expected. Because the distances A 0 -A 1 (eC 3 -Q K ) and A 1 -F X (Q K -F X ) are comparable (29), matching our observations, the distance P700-A 0 (eC 1 -eC 3 ) has clearly been overestimated by the photovoltage measurements relative to the other distances. Possibly, the fast rate of charge separation results in a significant error for the distance eC 1 -eC 3 (alternatively, the dielectric constant around P700 may differ substantially from that nearer the middle of the membrane), giving rise to the observed distortion.
Intercofactor Distances from X-ray Structure and Spectroscopic Studies-Comparisons of structural and spectroscopic data have recently been published based on models of PSI derived at 4.5-and 4-Å resolution (4,5). These studies, however, included none or, in the latter case, a single phylloquinone position designated Q K (now renamed Q K ). Here we will include the latest structural results and compare these to the available spectroscopic data.
The Moser-Dutton "ruler" (31) (an empirical first order relationship between electron transfer rates and shortest edge-toedge distances of the cofactors involved) provides a simple tool to estimate the "optimal" electron transfer rates from struc- FIG. 4. Schematic representations of the electron transfer system. a, cofactor distribution along the membrane normal showing distances in Å and fractional distances. The observed center-to-center distances (Ϯ1 Å) (b) and the edge-to-edge distances (Ϯ2 Å) (c) are indicated. Except for the interplane distance between eC 1 and eC 1 (3.6 Å) the values of the other edge-to-edge distances have been determined with an accuracy of 0.5 Å. d, the individual distances and angles between the cofactors eC 1 , eC 1 , Q K , and Q K are shown. They correspond to the center of the phylloquinones and are independent of the phylloquinone plane orientations. Present distances and angles are largely in agreement with those published previously (4). Note the schematic nature of these diagrams; true distances are supplied but may not be measured directly. tural data, the optimal electron transfer rate being achieved when the sum of the standard reaction free energy and reorganization energy is essentially zero (32). In Table I, the edgeto-edge distances and the optimal (i.e. fastest theoretically possible) electron transfer rates derived, using the above relationship, are listed.
Chlorophyll a Cofactors-The Chl a molecules, eC 1 , eC 1 , eC 2 , eC 2 , eC 3 , and eC 3 constitute the luminal half of the electron transfer system. eC 1 and eC 1 have been identified as structural components of the spectroscopically identified primary electron donor P700; eC 2 and eC 2 are referred to as the accessory chlorophylls; while either or both of eC 3 and eC 3 have been assigned to the spectroscopically identified primary electron acceptor A 0 .
As noted (4,5), the edge-to-edge distance between eC 1 and eC 3 of 13.3 Å is too long to be compatible with the electron transfer rate of 5⅐10 11 s Ϫ1 (for a review, see Ref. 5) determined for the primary electron transfer step (see Table I, Fig. 4c). It therefore seems likely that neither eC 3 nor eC 3 (the spectroscopically identified electron acceptor A 0 ), but one or both of eC 2 and eC 2 , represent an additional intermediate (i.e. true primary electron acceptor) with a calculated optimal electron transfer rate from eC 1 to eC 2 of 4⅐10 12 s Ϫ1 and eC 2 to eC 3 of 1⅐10 12 s Ϫ1 (see Table I).
The Phylloquinone Electron Acceptors-The average centerto-center distance between Q K and the neighboring cofactor eC 3 amounts to 8.7 Å. Combining data from pulsed EPR experiments on the position of the quinone cofactor (relative to P700 and the crystallographic a-axis) with the x-ray structural model of PSI yielded a comparable center-to-center distance estimate of 7.5 Ϯ 2 Å for eC 3 -Q K (10).
The center-to-center distance between Q K and F X has similarly been estimated to be 14 Ϯ 2 Å by EPR measurements (10), matching that of the x-ray structural model (Q K /Q K )-F X ϭ 14.3 Ϯ 1/14.1 Ϯ 1 Å (Fig. 4b). Although the orientation of the molecular plane of Q K about the naphthalene long axis could not be determined from the electron density map, we determined the edge-to-edge distances to the surrounding cofactors. Compared with the overall errors estimated for edge-to-edge distances (Ϯ2 Å), the distances observed for the two Q K plane orientations modeled prove insignificant.
The averaged edge-to-edge distance eC 3 -Q K of 4.8 Ϯ 2 Å is somewhat shorter than the value of A 0 -A 1 Յ 7.8 Å estimated from pico-nanosecond laser spectroscopy (33) ( Table I). The difference between these values is possibly due to the inequivalence of the reaction free energy and reorganization energy, causing the distance to be overestimated (33). The averaged edge-to-edge distance between Q K and F X is 11.3 Å Ϯ 2 Å, which agrees well with the 10.7 Å suggested previously (34) ( Table I).
The discrepancy between the electron transfer rate derived from the edge-to-edge distance eC 1 -Q K and those rates reported for the charge recombination reaction P700 . ϩ A 1 . 3 P700A 1 proves to be slightly more problematic; the latter is roughly 4⅐10 3 s Ϫ1 (5) ( Table I). According to the Moser-Dutton approximation (31), the optimal (i.e. fastest possible) transfer rate for a direct recombination through the distance eC 1 -Q K (20.5 Å) would be in the range of 5.0⅐10 2 s Ϫ1 . The reason for the observed rates being faster than that calculated from the corresponding edge-to-edge distance for this pair of cofactors is unclear, although an intermediate step in recombination could provide an explanation of this difference.
Correlation of A 1 with Q K or Q K Ј-The identification of two phylloquinones and their introduction to the x-ray structural model gives new impetus to the question of which one of Q K or Q K corresponds to the spectroscopically identified cofactor A 1 .
To analyze the current possibilities, we derived two orientational models for each phylloquinone (see "Results"). The orientation and position of each of these models (two for each Q K and Q K ) are quantified by the parameters shown in Fig. 5. They are successively compared with the corresponding values derived mainly from EPR experiments on oriented PSI particles or PSI single crystals (Table II).
The distance between P700 . ϩ and A 1 . has been placed at 25.4 Ϯ 0.3 Å (8,9). The center-to-center distances of all four combinations (eC 1 -Q K , eC 1 -Q K , eC 1 -Q K , and eC 1 -Q K ) are compatible with the EPR data (Table II, Fig. 4d). ␣-The inclination of P700 . ϩ -A 1 . relative to the membrane normal (c-axis) is 27 Ϯ 5° (10,35). The corresponding values for eC 1 -Q K , eC 1 -Q K , eC 1 -Q K , and eC 1 -Q K are 23, 28, 31, and 29 Ϯ 2°, respectively (Table II, Fig. 4d; see also ␣ in Fig. 5). A unique assignment of P700 . ϩ A 1 . is thus not obtained, although eC 1 -Q K and eC 1 -Q K give the closest agreement. -The angle between the projection of P700 . ϩ -A 1 . onto the a,b-plane and the a-axis ( in Fig. 5c) is put at 0 Ϯ 10°by EPR simulations (10). It should be noted that because of the inherent D 3 symmetry of these EPR techniques, the above values do not only hold for the crystallographic a-axis but also for the equivalent directions described by the vectors b and (a ϩ b). In the following, the term "a-axis" therefore includes the true crystallographic a-axis, as well as its repetitions at 60°intervals in the a,b-plane. The corresponding values for derived from the x-ray structural model are 12°(eC 1 -Q K ), 10°(eC 1 -Q K ), 21°(eC 1 -Q K ), and 5°(eC 1 -Q K ). Here eC 1 -Q K yields the best agreement, although eC 1 -Q K and to a lesser degree eC 1 -Q K cannot (entirely) be excluded.
-The axis defined by the phylloquinone oxygen atoms has been reported to be essentially parallel (Ϯ10°) to the vector P700 .
ϩ -A 1 . (19). Although the x-ray models of Q K (Q K ) were derived by aligning the phylloquinone plane with the vector eC 1 -Q K or eC 1 -Q K (eC 1 -Q K , eC 1 -Q K ), corresponding to P700 . ϩ -A 1 . , this leaves a possible deviation between the O-Oaxis and eC 1 -Q K within the phylloquinone plane ( in Fig. 5c). This deviation is observed to be 15 and 14 Ϯ 10°(15 and 1°), respectively, for the vectors listed above, if the center of the phylloquinone is taken to represent the center of spin density of A 1 . . These values change by as much as 3°if the electron spin density is assumed to be centered on either one of the naphthoquinone model ring systems (Table II). Here eC 1 -Q K clearly shows the best correspondence to the EPR data. ␤-Because EPR results indicate the phylloquinone O-Oaxis to be aligned with P700 .  Table II) is assumed to be equivalent in size to the angle ␣ ( 208 (P700 . ϩ -A 1 . ; c-axis), Fig. 5a Table II).
For all x-ray structurally derived models, the inclinations are in the range 50 -60°; the match between EPR-and x-ray structural values, therefore, is less than satisfactory. ␥-In the EPR model of Kamlowski et al. (36), the methyl group of the phylloquinones either faces the lumen or the stroma, while the C 2 -C methyl bonds subtend an angle of 35 Ϯ 20°with respect to the c-axis (36). The equivalent value for the x-ray structural model phylloquinone plane orientations (␥ in Fig. 5b, Table II) are in the range of 45-53°. Since the error of this value may be as large as 10°, the values do not necessarily contradict the EPR estimate of 35 Ϯ 20° (36). This criterion, however, does not aid in distinguishing the four cofactor pairs of the x-ray structural model.
-In the present x-ray structural model, the angles between the longest molecular axis of Q K (Q K ) and the plane defined by the c-axis and the vectors equivalent to P700 . ϩ -A 1 . (eC 1 -Q K , eC 1 -Q K , eC 1 -Q K , and eC 1 -Q K ) are 30, 68, 81, or 68°, respectively ( in Fig. 5, Table II). At present, no estimate for the angle is available from spectroscopic investigations. Combining the conclusions from the geometric comparison of the spectroscopic data and x-ray structural model indicates that eC 1 -Q K shows the closest correspondence with the pair P700 .
ϩ -A 1 . . Nevertheless, the pairs eC 1 -Q K and eC 1 -Q K similarly mostly remain within the error limits. Overall, eC 1 -Q K is most poorly correlated with the EPR data. CONCLUSION A complete and internally consistent set of cofactors of the electron transfer system of PSI has been modeled into an improved electron density map at 4-Å resolution. The corrected model of the ␣-helices in the vicinity of the phylloquinone molecules leads to a better 2-fold symmetrical arrangement in this region of the PSI core. The distances for Q K and Q K to other cofactors compare favorably with those suggested from spectroscopic measurements. Although the orientation of the phylloquinones remains undetermined about the long axis of the molecular plane, the pair P700 . ϩ -A 1 . is best correlated with eC 1 -Q K followed by eC 1 -Q K and eC 1 -Q K . "Long axis" is equivalent to the phylloquinone long molecular axis. b The term "a-axis" not only describes the true crystallographic a-axis but also the b-axis, as well as the (a ؉ b)-axis (60°from either a-or b-axis) including their negative directions, since these are indistinguishable in the EPR data.
c Values in parentheses indicate the two extreme cases where the spin density of A 1 . is located on either one of the two naphthoquinone rings.
Especially in projection onto the a,b-plane these may differ substantially from those obtained using the molecular centroids (see dotted lines in Fig. 5). d "Molecular plane" implies molecular plane of the phylloquinone molecules.