9728
J. Phys. Chem. B 2000, 104, 9728-9739
•A Structural Model for the Charge Separated State P•+ 700A1 in Photosystem I from the Orientation of the Magnetic Interaction Tensors
Stephan G. Zech,†,‡ Wulf Hofbauer,† Andreas Kamlowski,‡,§ Petra Fromme,† Dietmar Stehlik,‡ Wolfgang Lubitz,† and Robert Bittl*,† Max-Volmer-Institut fu¨ r Biophysikalische Chemie und Biochemie, Technische UniVersita¨ t Berlin, Strasse des 17, Juni 135, 10623 Berlin, Germany, and Fachbereich Physik, Freie UniVersita¨ t Berlin, Arnimallee 14, 14195 Berlin, Germany ReceiVed: June 12, 2000; In Final Form: August 8, 2000
•The charge separated state P•+ 700A1 (P700 ) primary electron donor, A1 ) phylloquinone electron acceptor) in photosystem I of oxygenic photosynthesis has been investigated by EPR spectroscopy in frozen solution and •single crystals. The transient EPR spectra of P•+ 700A1 recorded in frozen solution of fully deuterated samples at X-, Q-, and W-band frequencies are shown to contain sufficient information to yield the orientation of the •+ g-tensors of both A•1 and P700 with respect to the axis connecting both spins. So far incomplete information •on the orientation of A1 relative to the membrane plane has been complemented by data from time-resolved EPR on single crystals measured at Q-band. The phylloquinone headgroup orientation evaluated from the •EPR data in the charge-separated state P•+ 700A1 is compared with the presently available X-ray structural •+ model. The g-tensor of P700 has also been determined from cw-EPR experiments at W-band on single crystals, independent of the orientational data of the P•+ 700 g-tensor from the time-resolved EPR experiments. •+ The direction of the principal axes of g(P700) differ from the molecular axes system of the chlorophylls comprising P700 as found previously in the case of P•+ 865 in bacterial reaction centers. The implications of the •+ complete structural model from the A•and P molecular magnetic interaction tensors in the active charge 1 700 •separated state P•+ A in PS I are discussed. 700 1
1. Introduction Photosystem I (PS I) is the complex in the thylakoid membrane that mediates light-induced electron transfer from plastocyanin, a mobile electron carrier located in the lumen, to ferredoxin or flavodoxin at the stromal membrane side (see e.g., refs 1-3 for overviews). PS I isolated from cyanobacteria consists of 11 protein subunits and several cofactors, acting as electron donors or acceptors. Photochemistry starts at the primary donor, P700, which is made up of a dimer of chlorophyll a (Chla) molecules (see Scheme 1 for molecular structure). After photoexcitation of P700, an electron is transferred via a chlorophyll species, A0, to a phylloquinone (vitamin K1, VK1) molecule, Al (see Scheme 1). Subsequently, the iron sulfur clusters, FX, FA, and FB are reduced. Recently, experimental evidence has emerged indicating that FB is the likely terminal acceptor in PS I and the electron donor to ferredoxin.4,5 The present model of the PS I structure is based on X-ray crystallography at 4 Å resolution.5-8 The electron density map shows the overall architecture of the complex and the functional cofactors of the energy and electron transfer (ET) system. The ET chain contains six Chla molecules which are positioned in three pairs with pseudo-C2-symmetric arrangement, defined by the symmetrically related helices of the large protein subunits PsaA and PsaB. In the most recent PS I structure model,8 the * Corresponding author, Fax: +{49|0}30-314 21122; E-mail:
[email protected]. † Technische Universita ¨ t. ‡ Freie Universita ¨ t. § Present adress: Bruker Analytik GmbH, 76287 Rheinstetten, Germany.
SCHEME 1: Molecular Structures of Phylloquinone (VK1)a and Chlorophyll a (Chla)
a For VK , the axes of the electronic g-tensor of the respective radical 1 anion (indicated by arrows) are parallel to the molecular axes system.
phylloquinones QK and QK′ have been assigned with respect to the position of the center of mass of the naphthoquinone (NQ) headgroups. The positions of the two quinones are related to each other by the same approximate C2 symmetry axis as the large protein subunits. The assignment was largely guided by •results from pulsed EPR experiments on the RP state P•+ 700A1 performed on frozen solutions9-11 and on single crystals.12 The
10.1021/jp002125w CCC: $19.00 © 2000 American Chemical Society Published on Web 09/27/2000
Structural Model for P700•+A1•- in PS I elongated axes of the electron density blobs were interpreted to represent the orientation of the long molecular axes of the phylloquinone headgroups.8 This axis is, to a good approximation, collinear with the y-axis of the electronic g-tensor of the radical anion, A•1 (see Scheme 1). The axis through the two carbonyl oxygens (x-axis) could not yet be assigned from the X-ray data8 but has been modeled according to EPR results.13,14 Information on the protein environment of the quinones, i.e., on the amino acids constituting the A1 binding pocket, has not been reported so far from X-ray crystallography. However, the environment of A1 must be designed to fulfill the various functional features, e.g., the redox potential with the extremely negative value of approximately -0.8 V.3,15 To obtain as much information as possible on the quinone binding in PS I, further experiments have been performed on 16 and on the functional RP the photoaccumulated radical A•1 •- 14 A . From the orientation dependence of the cwstate, P•+ 700 1 EPR signal of photoaccumulated A•measured in oriented PS 1 I particles at X-band, it was possible to deduce two of the three angles necessary to fully describe the orientation of the A1 headgroup with respect to the protein complex.16 Supplementary information on orientation and distance had to be used from •- 13,17,18 for a full studies on the functional RP state P•+ 700A1 •- 16 structural model of A1 . In the other study,14 the relative orientation as well as the orientation with respect to the crystal axes of the accessible magnetic interaction tensors of the •P•+ 700A1 state was determined from transient EPR at X-band in PS I single crystals. However, while the orientation of the inplane axis of the carbonyl bond of the VK1 headgroup is well defined, the second in-plane axis could only be determined with a large error margin.14 The goal of this paper is not only to improve the accuracy but also to present complete structural parameters for both cofactor radical ions. First, we present transient EPR spectra of •the P•+ 700A1 state of fully deuterated PS I particles in frozen solution measured at different frequency bands (X-, Q-, and W-band). From this the relative orientation of the magnetic •+ g-tensors of A•1 and P700 is deduced with improved accuracy compared with the previous results on protonated samples.13 Second, the orientation of all principal axes of the magnetic •+ tensors g(A•1 ), g(P700) is determined with respect to the crystallographic c-axis (equivalent to the membrane normal) •from TR-EPR spectra of P•+ 700A1 at Q-band in PS I single crystals. From these data, complete information is obtained on the orientation of the functional VK1 headgroup and the g(P•+ 700) tensor with respect to the PS I reaction center complex. To compare the EPR-deduced orientation of the g-tensor axes with the molecular structure obtained from the X-ray structure analysis, independent knowledge is required on the orientation of the g-tensor axes with respect to the molecular axes. In that respect the state of knowledge is quite different for the two g-tensors under study in this paper. For quinones, the collinearity of g-tensor and molecular axes as shown in Scheme 1 is based on the theory of Stone.19,20 Convincing experimental confirmation is available from orientation-selective studies of model quinones in vitro as well as from stable radical ion states (see ref 21 for a review) and from the transient radical pair state in bRC.22 For the primary donor P•+ 865 in Rb. sphaeroides R-26, a substantial deviation of the g-tensor axes from the molecular axes of the bacteriochlorophyll (BChl) a constituents of the dimer has been observed by high-field EPR performed on RC single crystals.23 A rationale for this result may be found in the composition of P865 as a dimer of BChl a molecules with
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9729 unequal charge and spin density distribution over the two halves (see, e.g., ref 24). This, however, cannot explain the observation that nearly the same g-tensor axis orientations are found for the heterodimer mutants,25 because for these heterodimers the unpaired spin density is located exclusively on one dimer half. The considerable uncertainty in the assignment and interpretation of the g(P) tensor axes underlines the necessity for further experimental data for primary electron donors P in different reaction centers. In this study we provide the orientation for the g-tensor axes of P•+ 700 in PS I single crystals. 2. Materials and Methods 2.1. Sample Preparation. Preparation of Deuterated PS I. For the deuterated preparation, cells of the thermophillic cyanobacterium S. elongatus were grown as described elsewhere26 and adapted to D2O by subsequent increase of the D2O/ H2O ratio from 0 to over 95% in steps of 10% over a period of three months. All preparation steps were performed in H2Obased buffers as described earlier.27 Samples (2 mL) of the PS I preparations were concentrated to 50 µL, and 1 mL of D2Obased buffer was added, followed by an additional concentration step. This washing procedure was performed in order to replace possible exchanged deuterons and was repeated twice more. The probe in buffer at pD 6.4 had a Chla concentration of 2.7 mM for X- and Q-band experiments and 11 mM for W-band experiments. The samples were frozen in the dark without an external magnetic field applied. Therefore, magnetic field induced partial orientation of the reaction centers, as found in a recent study,28 is irrelevant in our case. Single Crystals of PS I. Time-resolved EPR measurements at Q-band have been performed on the same crystals of PS I from S. elongatus as previously used for X-band experiments.12,14 For W-band experiments on P•+ 700, smaller PS I crystals were mounted in synthetic quartz capillaries (o.d. 0.9 mm, i.d. 0.7 mm) with the morphological axis either parallel or perpendicular to the sample tube axis and frozen in liquid nitrogen under continuous illumination in order to photochemically generate P•+ 700. Orientation-dependent spectra were obtained by rotation of the sample holder rod in steps of 5°. 2.2. EPR measurements. Transient EPR at X- and Q-band. The transient EPR spectra have been recorded in the direct detection mode as described in ref 29. Complete 2D-data sets (time vs magnetic field) were acquired, and the spectra were obtained by digital boxcar integration of the time signal within a specified integration time gate. The magnetic field has been calibrated using a DPPH sample with giso ) 2.00351(1), ∆BPP ) 0.1 mT calibrated against Li:LiF (g ) 2.002293).31 All spectra have been obtained at low microwave powers, where the signal decay is limited by T1. Transient EPR at W-band. To illuminate the samples in the W-band resonator, a Spectra Physics Quanta Ray DCR-2(10) Q-switched, frequency-doubled (532 nm) Nd:YAG laser was used as a repetitive (10 Hz, 8 ns pulse width) light source. The light path consisted of a silica optical fiber (SpecTran HCGM0365T, 365 µm core diameter, 430 µm outer diameter) with one end inserted into the sample tube. The attenuated laser beam was focused on the other end of the silica fiber using an f ) 50 mm lens as described in ref 30. The laser pulse energy at the sample location was generally limited by the damage threshold of the fiber at the insertion point; here, about 0.5 mJ were used. The transient signal was recorded and averaged using a LeCroy 9410 digital storage oscilloscope. The experiments were con-
9730 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Zech et al.
trolled by the BRUKER Xepr software, customized by homewritten PulseSpeL routines yielding 2D data sets as for X- and Q-band. Other specifications were as for cw-EPR at W-band. cw-EPR at W-band. The cw-EPR measurements at 94 GHz were performed using a BRUKER Elexsys E680 spectrometer with an OXFORD CF935 helium cryostat. The magnetic field axis was calibrated using a Li:LiF sample (g ) 2.002293),31 measured at two different microwave frequencies (νl ≈ 94.0 GHz, ν2 ≈ 94.3 GHz). Reproducibility of the calibration has been found to be better than 0.03 mT. To compensate for slight changes (|∆ν| < 0.1 GHz) of the microwave frequency between individual spectra due to varying resonance frequencies of the resonator for different sample orientations, all spectra have been normalized to a frequency of 94.0 GHz. Analysis of the Crystal Spectra. The spin Hamiltonian for the six PS I sites per unit cell can be written as 6
H ˆ )
∑ i)1
µBB0
ˆi RigR-1 i S
p
where Ri are the rotation matrices representing the elements of the underlying P63 symmetry group. Accordingly, for each paramagnetic species up to six EPR lines are observable for an arbitrary orientation of the crystal relative to the magnetic field B0. For B0 || c (where c denotes the orientation of the crystallographic c-axis) the six lines collapse into one; for B0 ⊥ c, the six lines degenerate pairwise, resulting in three lines with equal intensities. 3. Results •3.1. Transient EPR Spectra of P•+ 700A1 in Frozen Solution at X-, Q-, and W-Band Frequencies. To derive information about the orientation of the g-tensors with respect to the dipolar coupling axis, transient EPR spectra of the radical pair •P•+ have been measured and analyzed by numerical 700A1 simulations according to the “correlated coupled radical pair” (CCRP) model (see ref 32 for a review). In Figure 1 the transient •in fully deuterated PS I from S. EPR spectra of P•+ 700A1 elongatus measured in frozen solution at three different microwave frequencies (X-, Q-, and W-band) are shown. Compared to the protonated samples published previously13 the deuterated spectra show a better spectral resolution due to the reduction of the hyperfine coupling constants (hfcs). •RelatiVe Orientation of g(P•+ 700), g(A1 ), and zD Obtained from Spectral Simulations. For simulations according to the CCRP model, a large set of magnetic interaction parameters is required. These are the g-tensor principal values for both radicals involved, the orientation of the tensor axes with respect to each other and to the axis of dipolar coupling, zD (axis connecting •P•+ 700 and A1 ), the magnitude of the dipolar and isotropic spin-spin coupling, D and J, and the respective line widths, •∆B(P•+ 700) and ∆B(A1 ). Most of these parameters have been determined by independent W-band experiments on the individual radical ions (see Table 1). The g-tensor principal values given in Table 1 agree well with earlier results.16,33-35 Adjustable parameters are therefore only the line widths and the relative •+ orientations of g(A•1 ), g(P700), and zD, which are of particular interest in this study. However, for some of these parameters restrictions have been provided by previous results (see refs 13 and 17 and Table 2 of ref 14) and are used as starting points for our simulation. According to the CCRP model, the spin-polarized spectrum for a coupled two-spin system without hfcs consists of four lines
•Figure 1. Transient EPR spectra of P•+ 700A1 in fully deuterated PS I from Synechococcus elongatus measured at three different microwave frequencies. Experimental conditions: temperature, 80 K. X-band: microwave (mw) frequency νmw ) 9.728 GHz; microwave power Pmw ) 63 µW; integration time gate, 0.5-1.5 µs after the laser flash. Q-band: νmw ) 34.00 GHz, Pmw ) 134 µW, 1.0-3.5 µs. W-band: νmw ) 94.11 GHz, Pmw ) 20 µW, 2.6-4.6 µs. Parameters used for the simulations (dashed lines) are given in Table 1. The field positions corresponding to the principal values of g(A•1 ) are indicated on top of each spectrum.
with equal intensity but with alternating signs of polarization. At W-band frequency, the differences in g-values between P•+ 700 and A•1 are much larger than the spin-spin coupling for most orientations of the complex with respect to B0. For this case, the four transitions can be assigned in good approximation to flips of an individual electron spin. The superposition of one absorptive and one emissive line represents the subspectrum of A•1 (solid line in Figure 2A), whereas the superposition of the other two represents the subspectrum of P•+ 700 (dashed line in Figure 2A). This simplifies the understanding of the results obtained by simulation discussed below. Orientation of g(A•1 ) with Respect to zD. The improved spectral resolution obtained at W-band for the deuterated sample allows the different residual line width parameters, ∆B(A•1 ) ), to be determined with high precision (absolute and ∆B(P•+ 700 error about (0.02 mT). The more reliable the line width parameters are, the smaller is the range of orientations that are consistent with the experimental data. Spectral simulations discussed in previous publications yielded a nearly parallel 13,14,17,36,37 As orientation of the gx-axis of A•1 (xQ) and zD.
Structural Model for P700•+A1•- in PS I
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9731
TABLE 1: Parameters Obtained from Simulations of the ••+ EPR Specta of P•+ 700A1 or the Individual Radicals, P700 and •A1 , in Deuterated PS I from S. elongatus g-Tensorsa (this work) b P•+ 700 (deut. PS I) c P•+ (PS I X-tals) 700 d A•1 (deut. PS I)
gx
gy
gz
2.00308 2.00309 2.00622
2.00264 2.00260 2.00507
2.00226 2.00223 2.00218
Isotropic Linewidthse (this work) ) ) 0.31 mT ∆B(P•+ 700 ∆B(A•) ) 0.25 mT 1 D ) -170 µT J ) 1 µT
Spin-Spin Coupling9,11
Geometrical Parametersf (this work) g(P•+ 700)
in
g(A•1 )
zD in g(A•1 )
RP
βP
γP
81°
126° θD
182° φD
90°
0°
a The absolute error for the g-values is (0.00005. The error for the g-value differences, i. e. |gi - gj|, is about (0.00002. b Obtained from W-band spectra of photooxidized P•+ 700 in deuterated PS I (data not shown). c Obtained from fit of the W-band spectra of photo-oxidized d P•+ 700 in PS I single crystals (Figures 7 and 8). Obtained from W-band spectra of photochemically reduced A•1 in deuterated PS I (data not shown). e ∆B corresponds to the peak-to-peak linewidth in firstderivative spectra. f The Euler angles (defined in ref 13) describe the •relative orientation of g(P•+ 700) to g(A1 ). The polar and azimuthal angles, θ and φ, respectively, describe the orientation of zD relative to g(A•1 ): zD ) (-sin θ cos φ, sin θ sin φ, cos θ).
•Figure 2. Simulation of the TR-EPR spectrum of P•+ 700A1 at W-band ••+ frequency. A: Subspectra of A1 (solid line) and P700 (dashed line) after powder averaging. B: Experimental spectrum (solid line, identical to Figure 1, top) and simulation (dashed line) which is the sum of solid and dotted contribution in A.
evident from the excellent agreement between experiment and simulation shown in Figure 2B, an orientation with xQ || zD is also corroborated by the W-band spectrum of the deuterated sample. Due to the weak angular dependence of the effective dipolar coupling near zD || B0, the accuracy is limited to about
Figure 3. Representation of the orientational data obtained from multifrequency time-resolved and cw-EPR data in this study. The structural parameters are collected in Tables 2 and 4. The orientation of the dipolar axis zD has been derived by pulsed EPR experiments on ••+ 12 the P•+ 700A1 state in single crystals of PS I. The g-tensors for P700 and •A1 are represented as ellipsoids. zD depicts the angular dependence of the dipolar coupling. The relative orientations of these tensors have been deduced by TR-EPR experiments on frozen solution (Figures 1 •+ and 2). The orientation of g(A•1 ) and g(P700) with respect to the •crystallographic c-axis have been determined by TR-EPR, on P•+ 700A1 in PS I single crystals at Q-band (Table 4, Figures 4-6). The g(P•+ 700) tensor has been obtained independently by W-band cw-EPR on P•+ 700 in PS I single crystals (Table 4, Figures 7-8). The plane of the figure is the (c, zD) plane.
(10°. Figure 3 visualizes the xQ || zD orientation for A•1 , as well as the relative orientation of the other magnetic interaction •tensors of the RP state P•+ 700A1 as obtained from single-crystal studies which are discussed below. Orientation of g(P•+ 700) with respect to zD. The orientation ) with respect to zD can now be determined with for g(P•+ 700 much higher accuracy than previously done for the protonated samples.13 This orientation has been obtained in three steps: First, the angle between the gz-axis of g(P•+ 700) (zP) and zD is obtained from the high-field part of the spectrum. Afterward the orientation of the xP and yP-axes are derived by rotation about the zP-axis (by Euler angle RP). In a third step, the •orientation of g(P•+ 700) relative to g(A1 ) is adjusted by rotation about zD (Figure 3). The values used for the simulations shown in Figures 1 and 2 are included in Table 1. The highest sensitivity of the polarization pattern on the g(P•+ 700) orientation is observed for the angle ∠(zP, zD). The perpendicular orientation of the zQaxis of g(A•1 ) to zD gives rise to an emission at the high-field edge around gz(A•1 ) (Figure 2A, solid line). However, in •22 Q contrast to P•+ 865 A in Zn-bRCs, an emission at the high field
9732 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Zech et al.
TABLE 2: Orientation fo the Dipolar Axis, zD, with Respect •to the Principal Axes of g(Q•-) and g(P•+) for P•+ 700A1 in PS I •+ •and P865QA in bRCs ∠(zD, g(Q•-)) •P•+ 700A1 in PS I this work ref 13 ref 28 •a P•+ 865QA in bRCs
∠(zD, g(P•+))
xQ
yQ
zQ
xP
yP
zP
0° 0° 20° 68°
90° 90° 75° 25°
90° 90° 78° 79°
83° 76° 71° 78°
55° 54° 45° 24°
36° 39° 51° 69°
a Angles obtained from X-ray data41 and using option II for the 23 g(P•+ 865) tensor.
TABLE 3: Parameters Used for Simulations of the •Transient EPR Spectra of P•+ 700A1 in Single Crystals (in addition to those given in Table 1) HFC Principal Values 33,38
2-CH3
Ax
Ay
Az
9.0 MHz
12.8 MHz
9.0 MHz
Residual Linewidths (this work) ∆B(P•+ ) ) 0.70 mT ∆B(A•700 1 ) ) 0.40 mT Geometrical Parametersa RA A(2-CH3) to n to a
g(A•1 )
33 g(A•1 )
25
(this work)
βA
γA
0°
0°
θn
φn
-65.3°
-12.4°
The angles are defined in Table 1.
•edge of the spectrum is not observed for P•+ 700A1 (Figure 1, top). Because the rather well-defined emissive contribution due to the A•1 subspectrum (Figure 2A, solid line) has to be compensated by an E/A pattern in the P•+ 700 subspectrum (Figure 2A, dotted line), an angle ∠(zP, zD) much smaller than the magic angle (54.7°) is required. Due to the precise line •+ widths ∆B(A•1 ) and ∆B(P700), only a small range of 34° < ∠(zP, zD) < 39° is in agreement with the experimental spectrum. The best fit is achieved for ∠(zP, zD) ) 36° (Figure 3), which is significantly smaller than the angle determined from partially oriented whole cells28 (see Table 2). Simulations of X-Band and Q-Band Spectra. Because we are primarily interested in the orientational parameters, we have concentrated so far on the W-band spectrum, which is most appropriate for this purpose. In Figure 1, the simulations using the same parameter set (Table 1) are shown as dotted lines also for X- and Q-band spectra. For all frequency bands, the simulations are very good for the A•1 part of the spectra (low field). For the P•+ 700-part of the spectra, the agreement is still very good at Q-band, whereas differences in the high-field part are visible for the X-band spectrum, which are the result of anisotropic contributions to the line width, either by hfcs or relaxation processes. These anisotropic contributions, however, are not considered in this study because of the large number of parameters involved for which, at present, no independent experimental information is available. However, using the orientational parameters given in Table 1 and including one hfc tensor for the 2-methyl group33,38 (Table 3), an excellent fit (not •shown) is also achieved for the W-band spectrum of P•+ 700A1 in 13 protonated PS I. •3.2. Transient EPR Spectra of P•+ 700A1 in PS I Single Crystals at Q-Band. The combination of xQ || zD (Table 2) and ∠(zD, n) ) 27° ( 5°12 directly yields the angle between
Figure 4. Angular dependence of the transient Q-band EPR spectra •of the radical pair state P•+ 700A1 in single crystals of PS I. The same crystals have been measured before at X-band.12,14 The orientation of the crystallographic c-axis (parallel to the membrane normal, n) is nearly parallel to the rotation axis, r.12,14 Since r is perpendicular to B0, c is almost perpendicular to B0 for all spectra. For comparison, the spectrum recorded in frozen solution is shown on top. All spectra have been normalized to a mw frequency of 34.0 GHz. Experimental conditions: T ) 80 K, mw power 840 µW, digital boxcar time gate (0.4-2.4) µs. At each of the 128 field points, 128 time traces have been averaged.
the xQ-axis and n: ∠(xQ; n) ) 27° (Figure 3). However, information on the angles of the other axes, yQ and zQ, requires additional experiments on oriented samples, e.g., single crystals. Here we use the spectral separation of the contributions from •+ •+ •A•1 and P700 in the RP state P700A1 achieved in single crystals at Q-band. This allows to determine the orientation of the yQand zQ-axes of g(A•1 ) from the orientation dependent effective g-value of A•. 1 •The top trace of Figure 4 shows the spectrum for P•+ 700A1 recorded in frozen solution at Q-band. The spectrum shows an E/A/A/E/A polarization pattern. Predominantly around gy(A•1 ), a partially resolved hf structure arising from the 2-CH3 group of the phylloquinone molecule is visible, which has also been observed in cw-spectra of photoaccumulated 16,33 A•1 . Rotation with Axis Parallel to the Rotation Axis. The lower •part of Figure 4 contains the transient EPR spectra of P•+ 700A1 in a PS I single crystal with the rotation angle R varied in steps of 10° through a 120° range. The crystallographic c-axis is
Structural Model for P700•+A1•- in PS I oriented nearly parallel to the rotation axis, r. For all spectra shown, the magnetic field B0 is almost perpendicular to c. All spectra show an (A)/(E)/A/E polarization pattern from which an angle between c and zD consistent with the 27° ( 5° determined from ESEEM experiments12 can be derived (see discussion of the X-band spectra14 for the same crystals). Due to the increased spectral resolution at Q-band, the •gx(A•1 ) and gy(A1 ) principal values are separated and can also clearly be distinguished from g(P•+ 700). This allows the orientation of the xQ-axis with respect to the c-axis to be estimated from the width of the transient spectra. As seen in Figure 4, the single-crystal spectra do not extend as far downfield as the powder spectrum. Numerical simulations (not shown) require that the angle between the xQ-axis and c are in the range 15° < ∠(xQ, c) < 40° to account for the experimentally observed g-anisotropy in Figure 4. The single-crystal spectra are therefore consistent with ∠(xQ, c) ) 27° deduced above and are also in agreement with the orientation dependent X-band cw-EPR 16 which yielded an angle of spectra of photoaccumulated A•1 about 31° (see also Table 4). In a next step, information on the so far poorly defined orientation of the yQ- and/or zQ-axes, and hence on the orientation of the plane of the quinone headgroup, is deduced. With the known information on the orientation of the xQ-axis, two limiting orientations of the quinone plane with respect to the effective g-value of A•1 can be distinguished: (i) The quinone plane lies parallel to the plane defined by c and zD, (c, zD)-plane. By 90° rotation about the zD || xQ-axis (Figure 3), the second limiting orientation is derived: (ii) The quinone plane lies perpendicular to the (c, zD)-plane. Both extreme orientations show different effective g-values of both ions upon rotation about r || c. For orientation (i), the geff value for A•1 falls in the range 2.00218 e geff e 2.00531 (gmin ) gz; gmax ) (g2x cos2 63° + g2y cos2 27°)1/2). For orientation (ii) the variation is smallest: 2.00301 e geff e 2.00507 (gmin ) (g2x cos2 63° + g2y cos2 27°)1/2; gmax ) gy). The difference in the upper limits of geff is quite small, whereas the lower limits differ significantly. Unfortunately, the lower limit is hard to observe because in the •+ high field area the subspectra of A•1 and P700 are superimcan be posed. Nevertheless, the effective g-value of A•1 evaluated quite well from the center of the hf splitting of the 2-methyl group. As seen especially for rotation angles R ) 20° and 30°, the center of the quartet splitting observed in the high field part is located at geff ) 2.0031 ( 0.0001. This value is similar to the lower limit of case (ii) and indicates qualitatively that the orientation of the quinone plane is closer to possibility (ii) than to (i). Quantitative information is obtained more effectively from the rotation with c perpendicular to r as described next. Rotation with the c-Axis Perpendicular to the Rotation Axis. In Figure 5, the single crystal spectra for the rotation with c perpendicular to the rotation axis, c ⊥ r, are depicted. Again, the polarization pattern shows the same qualitative dependence on the rotation angle R as previously found for the X-band spectra,14 but a significantly better resolution is obtained at Q-band. The most interesting orientation is c || B0, where all six PS I monomers are magnetically equivalent. The observed E/A/E/A pattern is expected for ∠(zD, B0) < 54.7°, i.e., for angles between zD and B0 smaller than the magic angle. However, considering the line-broadening due to hf interactions, this polarization pattern can only be observed if the g-difference of both radicals is sufficiently large. At Q-band this separation is achieved while this was not the case at X-band where only an E/A pattern14 is found at this orientation.
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9733
Figure 5. Angular dependence of the transient Q-band EPR spectra •of P•+ 700A1 recorded in a PS I single crystal which has been oriented with the c-axis perpendicular to the rotation axis r. The spectrum recorded in frozen solution is shown on top. For experimental conditions see Figure 4. The rotation angle is set to R ) 0° for B0 parallel to the crystallographic c-axis. Due to the crystal symmetry, at this orientation all six radical pairs in the unit cell become magnetically equivalent. At this orientation the polarization pattern exhibits a clear E/A/E/A •+ structure showing that the signals due to A•1 and P700 are spectrally separated.
Orientation of g(A•1 ) Obtained from Simulation. For analysis of the transient EPR spectra of single crystals, the same program as previously used for X-band experiments14 has been utilized. For protonated samples, the anisotropic hfc of the 14,16,33,38 has to be included explicitly to 2-methyl group of A•1 improve the quality of the simulations.14 The principal values of this hfc tensor and the direction of the tensor axes relative to 33,38 or g(A•1 ) have been measured by ENDOR spectroscopy •- 33 W-band EPR, on photoaccumulated A1 , respectively, and are given in Table 3. In Figure 6A the experimental spectrum for c || B0 (solid line) is shown along with spectral simulations referring to the cases (i), with dotted lines and (ii) (dash-dotted line). The parameters used for simulation are collected in Tables 1 and 3. The two cases discussed above can now be distinguished according to their different geff (A•1 ) values. The effective are g ) 2.00599 in case (i) and geff ) g-values for A•eff 1 2.00539 in case (ii). The dotted line in Figure 6B shows the simulation with best agreement with the experimentally observed spectrum (solid line). For this simulation, the angle between the quinone plane and the (c, zD)-plane is 65°, corresponding
9734 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Zech et al. TABLE 4: Angles between the Principal Axes of g(Q•-) and g(P•+) with respect to the Membrane Normal, n, Determined for the RP State of PS I and the Individual Radicals ∠(n, g(Q•-)) •a P•+ 700A1 (this work) •+ b P700 (this work) •- 28c P•+ 700A1 •- 16d A1 41e Q•A in bRCs •+ P865 in bRCs23f a
∠(n, g(P•+))
xQ
yQ
zQ
xP
yP
zP
27°
79°
65°
36° 31° 97°
85° 64° 7°
55° 76° 87°
69° 68° 48°
36° 38° 44°
62° 61° 81°
12°
78°
90° b
Obtained from time-resolved EPR experiments. Obtained from W-band cw-EPR on PS I single crystals. c Data derived from partially oriented whole cells of S. liVidus. Angles calculated from the Euler angles given in ref 28. d Data derived from oriented membranes. Angles calculated from the Euler angles given in ref 16. e Angles between the pseudo-C2-axis (membrane normal) and the molecular axes of QA. f Measured at W-band in single crystals of Rb. sphaeroides R-26. Option II of ref 23 has been used.
Figure 6. A: Simulations of the Q-band transient EPR spectrum for •P•+ 700A1 with c || B0 (R ) 0° in Figure 5) for different assumed orientations of the quinone plane with respect to the (c, zD) plane. The orientation of the xQ axis and the complete g-tensor of P•+ 700 have been held constant. Solid line: Experimental spectrum. Dotted line: Quinone plane parallel to (c, zD). Dashed line: Quinone plane perpendicular to (c, zD). B: Comparison of the experimental spectrum (solid line) with the simulation yielding the best fit (dotted). The corresponding simulation parameters are summarized in Tables 1 and 3. The four small vertical lines at the bottom represent the hyperfine splitting due to the 2-CH3 group of A•1 with Aeff ) (10.3 ( 0.6) MHz for this particular orientation.
to an effective g-value of geff(A•1 ) ) 2.00547. Nearly the same value (geff(A•1 ) ) 2.0055 ( 0.0001) is derived from the center of the partially resolved quartet splitting of the 2-methyl group, indicated as vertical lines (see bottom of Figure 6). Taking into account the error for the orientation of the xQ-axis (about (10°), the error of the angle between the quinone plane and the (c, zD)-plane is estimated to be (20°. For the simulation in Figure 6B, the orientation of the membrane normal, n, within the g-tensor axis frame of A•(used as reference system) is 1 expressed as polar and azimuthal angles in Table 3 (bottom) and visualized in Figure 3 as a vertical line. Orientation of g(P•+ 700) Deduced from Time-ResolVed EPR Data. The combined orientational data from frozen solution experiments and the orientation of g(A•1 ) with respect to the membrane normal from the single-crystal experiments can be used to derive information on the orientation of g(P•+ 700) in the charge-separated RP state, which then can provide a prediction for the g-tensor orientation of chemically stabilized P•+ 700 in single crystals. The angles of the principal axes of g(P•+ 700) with respect to the membrane normal as shown in Figure 3 are included in Table 4. In the single-crystal spectrum for c || B0 (Figures 5 and 6B), the evaluated g(P•+ 700) orientation yields an effective g-value of geff ) 2.00261. This agrees with the experimentally observed zero-crossing point (geff ≈ 2.0026) for the P•+ 700 part in Figures 5 and 6B and confirms the g-tensor orientation of P•+ 700 obtained from the frozen solution spectra. However, the orientation of g(P•+ 700) has been obtained so far
Figure 7. Experimental (left) and simulated (right, with central line positions connected) rotation pattern of the EPR spectra of photochemically generated P•+ 700 in a PS I single crystal at W-band. All spectra are normalized to 94.0 GHz. The rotation axis is approximately parallel (3° deviation) to the crystallographic c-axis, leading almost to a pairwise degeneracy of the six lines and resulting effectively in a 2:2:2 line structure with 60° periodicity.
only indirectly by a combination of various structural data gathered from time-resolved EPR. Although the transient EPR •spectra of P•+ 700A1 are very sensitive to the orientation of •+ g(P700) with respect to zD, they are much less sensitive to rotations of g(P•+ 700) about the zD-axis. Therefore, an independent direct determination of the orientation of g(P•+ 700) with respect to the crystallographic c-axis is desirable. This has been accomplished by cw-EPR at W-band or photochemically generated P•+ 700 in PS I single crystals. 3.3. cw-EPR at W-Band on P•+ 700 in PS I Single Crystals. Orientation-dependent spectra of P•+ 700 have been taken for two different rotation axes of the PS I single crystals. In Figure 7 (left side), the crystal is oriented with its c-axis nearly parallel to the rotation axis. A 60° periodicity as seen in Figure 7 is expected for an orientation of B0 ⊥ c and thus r || c. In Figure 8, the crystal is oriented nearly perpendicular to the rotation
Structural Model for P700•+A1•- in PS I
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9735 the crystal axes system agrees well with that obtained less directly by transient EPR on the RP state (Table 4). However, significant deviations (up to 20°) are found with respect to the orientation determined from partially oriented whole cells of S. liVidus.28 The orientation of g(P•+ 700) in ref 28 is in agreement neither with the time-resolved spectra in Figure 1 nor with the single-crystal experiments on P•+ 700 (Figures 7 and 8). 4. Discussion
Figure 8. Experimental (left) and simulated (right) rotation pattern of the EPR spectra of P•+ 700 at W-band with the rotation axis approximately perpendicular (88°) to the c-axis of the PS I single crystal. The spectra vary between almost complete degeneracy of all six lines at R ) 0° and 180° (B0 || c almost realized) and a 4:2 pattern at R ) 90°.
axis. The observation of only one degenerate line for 0° and 180° with B0 || c confirms the crystal orientation with r ⊥ c. Numerical Simulations of the P•+ 700 W-Band Spectra. The usual way of analyzing orientation dependent cw-EPR lines in single crystals is to extract the individual line position for each site by deconvolution of the spectra. In the present case this approach, however, is problematic due to several reasons. Even at W-band frequencies, the g-anisotropy of P•+ 700 is too small to resolve the single crystal spectra into individual lines. The necessary deconvolution of six inhomogeneously broadened lines suffers from numerical instabilities, especially for orientations where lines are nearly degenerate. The obtained line positions are also sensitive to the chosen line width. Consequently, an unambiguous assignment of line positions to individual sites is not always possible. A complementary strategy to analyze the spectra is a simulation of all spectra of the rotation pattern. Here, the knowledge of the crystal symmetry group can be utilized from the beginning, by which most of the problems associated with degeneracy are discriminated. The simulation can be refined using least-squares fitting of the simulation parameters. A C++ program has been written to perform these simulations. The input parameters are (i) the symmetry group of the crystal (P63, fixed), (ii) the principal values of g, (iii) the orientation of g relative to the crystal axes for an unspecified site, (iv) the inhomogeneous line width, and (v) the rotation axis of the crystal. These parameters were fitted to all spectra simultaneously. The simulated sets of spectra are shown next to the experimental data in Figures 7 and 8. Line positions have been indicated as well. The resulting fit parameters include a slight misalignment of the crystallographic c axis, which turns out to be inclined relative to the rotation axis r by 3° (Figure 7) and 88° (Figure 8), respectively. The fitted principal values (Table 1) are in good agreement with frozen solution studies using submillimeter cw-EPR35 on protonated samples and the values derived in this study from P•+ 700 W-band spectra of deuterated PS I in frozen solution (Table 1, spectrum not shown). Furthermore, the orientation of g(P•+ 700) determined directly in
4.1. Summary of Information on the Orientation of •+ g(A•1 ) and g(P700) in the Protein Framework. As mentioned above, Figure 3 includes all information for the tensor axis •orientations obtained from the RP state P•+ 700A1 in PS I. The basic features are summarized as follows: The orientation of •+ g(A•1 ), g(P700), and zD relative to each other is obtained from TR-EPR experiments on deuterated samples measured in frozen solution at high-field (Figures 1 and 2). Table 2 also includes the angles between the other g-tensor principal axes and the dipolar axis, which are compared to previously published results. For the quinone orientation, the angles turn out to be identical to those published earlier for protonated PS I at K-band17 and W-band.13 A comparison with the respective orientation for •22 (see Table 2) P•+ 865QA in Zn-bRCs of Rb. sphaeroides R-26 shows that the orientation of the quinones with respect to zD differs drastically (up to about 70°) between PS I and bRCs.13,39 For the primary donors, the yP and zP axes differ less but still by about 30°, whereas ∠(xP, zD) is roughly the same in both systems. The dipolar axis zD has been located within the protein framework by pulsed EPR experiments on PS I single crystals12 and corroborated by transient X-band EPR data14 and this study. The complete orientation of the g(A•1 ) with respect to the •membrane normal, n, has been derived by TR-EPR on P•+ 700A1 in single crystals (Figures 4 to 6). From the effective g value of A•1 observed at the specific orientation c || B0, the angle between the quinone (x, y) plane and the (c, zD) plane could be determined to 65° ( 20°. The angles between the principal axes of g(A•1 ) and the membrane normal (equivalent to c) are collected in Table 4. Within the respective experimental errors, the orientation of g(A•1 ), derived here for the charge separated •state P•+ 700A1 , is in reasonable agreement (deviations < 15°) with other experimental results.16,28 However, large orientational changes between the quinones in bRCs and PS I occur with respect to both the dipolar axis (see Table 2 and 13) and the membrane normal n (see Table 4). The orientation of the g-tensor of P•+ 700 has been obtained in two ways: directly from simulation of the cw-EPR spectra of P•+ 700 measured at W-band in PS I single crystals, and independently from a combination of various time-resolved EPR •experiments on the charge separated state P•+ 700A1 . Finally, •+ Table 4 compares the orientation of g(P700) with the g-tensor 23 Despite the similar molecular orientation for P•+ 865 in bRCs. framework, the angles in Table 4 reveal differences in orienta•+ tion between g(P•+ 700) and g(P865) of up to about 20°. In particular, the deviation of the yP-axis from the local C2 symmetry axis (parallel to the crystallographic c-axis and membrane normal n) is considerably larger for P•+ 700 than for . Because the Chla and BChla constituents of P P•+ 700 and P865 865 are perpendicular to the membrane plane6,7,40,41 the zP-axis is expected to be parallel to the membrane plane. However, as evident from the angles in Table 4, the zP-axis is not parallel to the membrane plane, i.e., it is not perpendicular to the membrane
9736 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Figure 9. Location and orientation of the electron acceptor A1 (red) as determined in this study within the protein framework of PS I as deduced by X-ray diffraction8 (PDB entry 1C51, see text for detailed discussion). A: The view direction is perpendicular to the plane defined by the crystallographic c and a-axes. Note that the projection plane for •P•+ 700A1 is different from Figure 3 (obtained by 120° rotation about the c-axis). B: The view direction is along the crystallographic c-axis fiom the stromal side (projection onto the (a, b) plane). The chlorophyll molecules labeled A and A0 have been omitted in this projection.
normal. Interestingly, the zP-axis of g(P•+ 700) deviates even more from the average molecular z-axis (normal to the molecular planes of the two chlorophylls) than that of g(P+ 865) in bRCs. In summary, complete structural information is available on the •paramagnetic species P•+ 700 and A1 from EPR data and will now be included into the current PS I model available from X-ray crystallography at 4 Å resolution.5,8 4.2. Location and Orientation of A•1 within the Protein Frame of PS I. The expanded experimental data available from the present work allows to position and orient the phylloquinone headgroup in the PS I reaction center protein framework (Figure 9). The figure includes only the backbone of the pairs of helices (primed and unprimed) termed m, n and o and the connecting loops as taken from the recent X-ray structure.8 Convincing evidence has accumulated that this protein part represents the essential binding structure for the inner membrane cofactors of the charge-transfer chain. The two branches of cofactors that are related by an approximate C2-symmetry axis raise the question whether ET proceeds unidirectionally along one branch (as found in bRCs), or whether both branches are active. A recent study based on
Zech et al. the observation of multiexponential reoxidation kinetics of A•1 and photovoltaic measurements suggests that both branches are active.42 This has to be compared with the observation that only one phylloquinone Al site appears to be photoaccumulated and therefore to be preferentially redox active.43 A simple argument comes from the large number of resolved magnetic interaction properties collected for A•1 , which can only result from a single phylloquinone. Moreover, the photoaccumulated A•1 state16 was shown to have identical properties to the A•1 in the •functional charge-separated state P•+ 700A1 observed by time44 resolved EPR measurements. With respect to the latter, one has to keep in mind that time-resolved EPR is only sensitive to a time window longer than a few 10-8 s. Thus, a shorter lived charge separated state including the second branch cannot be excluded. This putative second phylloquinone may be reduced and reoxidized too fast to be EPR detectable, which also requires that the interacting successive acceptors are reoxidized on the same fast time scale. Moreover, this phylloquinone would not be photoaccumulated, which is difficult to rationalize. Recent experiments43 on photoaccumulated A•1 in PS I mutants with specific deletions of asymmetrically positioned stromal protein subunits confirm that only one quinone site is photoaccumulated. In addition, indications are provided that this site corresponds to the one termed QK in ref 8, i.e., the one distal to the trimer axis that is associated with the unprimed helices.43 This has been chosen here for representation in Figure 9. Position of A1. Obviously, EPR cannot decide between symmetry equivalent positions, e.g., between the two cofactor branches. In that respect the choice made in Figure 9 is arbitrary. If the spin carrying Chla half of the P700 dimer is chosen to be eC1′, the EPR active A•1 is positioned on a torus of distance (25.4 ( 0.3) Å9,11 from eC1′ and a diameter determined by the angle ∠(c, zD) of 27° ( 5°.12 Independent information on the orientation of the crystal axis a or b (as deduced for instance from the morphological crystal shape in ref 12) selects six symmetry equivalent positions for A•1 on this torus. In contrast to position 1 chosen in ref 12, in this work the alternative position 3 was selected (Figure 9). Using the refined structural model,8 the average center-to-center distances of the Al position 3 (Figure 9) to the other members of the electron transfer chain are eC2′-A1 ) (17 ( 2) Å, eC3′ -A1 ) (9 ( 2) Å, and FX-A1 ) (13 ( 2) Å. As previously found for position 1,12 also position 3 agrees well within error with the distances estimated from electron-transfer rates.3 A significant aspect of this choice is the analogy to the situation in bRC. There the predominant spin density is located on the BChl half of the P•+ 865 dimer ligated to the L-side of the RC protein, whereas the primary quinone QA is bound to the M-side. The corresponding arrangement is realized for PS I with eC1′ ligated to the m′ helix and Al chosen in Figure 9 associated with the unprimed m, n, and o stromal helices. Note that the equivalent C2 symmetric situation would be reached when the whole sequence of arguments is repeated but with the eC1 half of the P700 dimer chosen as the spin carrying Chla (not shown in Figure 9). However, recent ADMR, data on 3P700 and ENDOR experiments on P•+ 700 in PS I mutants where the ligating histidines have been replaced show that the spin density is located on eC1′.45 Orientation of A1. Two additional in-plane axes of the phylloquinone headgroup were determined in this work. First, the 2-methyl hyperfine coupling (Table 3) yields an angle of the C-CH3 bond relative to the c-axis of 52° ( 10° (Table 5). Second, the orientation of the VK1 molecular axes to the membrane normal follows from that of the g-tensor axes given in Table 4. With these axis orientations two A1-orientations are
Structural Model for P700•+A1•- in PS I
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9737
TABLE 5: Comparison of Structural Parameters for A•1 with the Quinone QK from the X-ray Structural Model8 parameter ∠(yQ, n) b ∠(ya,b Q , a-axis) distancec ∠(zD, n)d e ∠(za,b D , a-axis) ∠(xQ, zD) ∠(xQ, n) ∠(zQ, n)f ∠(C2-CH3, n)g a,b h ∠(ya,b Q , zD )
EPRa A1
X-ray8 QK
79° ( 10° 68° ( 20° •P•+ 700A1 25.4 ( 0.3 Å 27° ( 5° 0° ( 10° 0° ( 10° 27° ( 10° 65° ( 7° 52° ( 10° 68° ( 20°
77° ( 5° 78° ( 5° eC1′-QK 25.5 ( 1 Å 28° ( 3° 10° ( 5° 14° ( 10° 17° ( 10° 59° ( 10° 45° ( 10° 68° ( 5°
•Parameters for the charge-separated state P•+ 700A1 as obtained in b this work (except where indicated). Angle between the projection of yQ onto the a, b plane and the a-axis. Note that the angles are measured to the same a-axis. c Distance between the electron spins within the RP.9,11 d Angle between the dipolar axis and the crystallographic c-axis.12 e Angle between the projection of zD onto the a, b plane and the a-axis.12 f This angle is geometrically equivalent to the angle between the quinone plane and the membrane plane. However, it is different from the angle between the quinone plane and the (c, zD) plane (65° ( 20°) and the identical values are accidental. g Angle between the bond-direction of the 2-methyl group of A1 and n. h Angle between yQ and zD, both projected onto the a, b plane. a
consistent. They are related by a 180° rotation around the outof-plane zQ-axis of VK1. In one orientation the phytyl chain points toward the central C2-symmetry axis and the lumenal site, whereas in the other orientation the chain points away from the C2-symmetry axis and toward the stromal side. The choice made for Figure 9 is again in analogy to the bRC case where the tail points away from the central C2-symmetry axis. Accordingly, the aromatic benzene ring part of VK1 points toward FX. In Table 5, selected structural parameters deduced from the X-ray model8 for the QK site are compared to those •derived here for the charge-separated RP state P•+ 700A1 (Figure 9). Within the given errors (Table 5), the EPR data and the X-ray structural parameters for the pair eC1′-QK are in acceptable agreement.8 In summary, a complete model for the position and orientation of the VK1 headgroup within the PS I reaction center is obtained. While symmetry equivalent positions cannot be excluded on the basis of the EPR data alone, reliable independent information (closeness to the average “acceptor-plane” defined by the centers •of all its Chla constituents and analogy to the active P•+ 865QA state in bRC) made it possible to arrive at the choice shown in Figure 9. Consequences for the A1 Site Model and Comparison with the Current X-ray Structure. The improved structural knowledge of the VK1 orientation immediately raises the question about the consequences for the model of the A1 site proposed before.46 Although the present X-ray structural model of PS I5,8 does not specify the amino acid side chains, the helices m and n could be oriented reliably by (i) the His ligation to P700 and (ii) by pointing the polar side of the n-helix toward the stromal side.46 With this alignment of the m and n helices and the correspondingly assigned amino acid locations, the model showed a remarkable stability for the particular structural feature of a π-stacking between the VK1 headgroup and the first tryptophan residue in the n-helix (W697 for PsaA or W677 for PsaB in S. elongatus).46 Additional modeling on the basis of the VK1 coordinates used for Figure 9 indeed confirmed this π-stacking
in a configuration with favorable π-bond overlap (B. AltenbergGreulich, private communication). It should be noted that by ESEEM spectroscopy, a 14N nuclear quadrupolar coupling has been detected and assigned to a tryptophan residue in the 47 A π-π interaction with an immediate surrounding of A•1 . aromatic amino acid residue has also been proposed earlier on the basis of the binding affinities of various artificial quinones substituted into the A1 site.48 Another remarkable feature of the A1 site model46 (in particular in comparison with the QA site in bRC) is the lack of an H-bond between VK1 and a histidine residue in the m-helix because no such histidine exists at the stromal end of the m-helix. This is in contrast to the situation in the bRC where a His is found at the appropriate position in the d-helix. Nevertheless, an asymmetric H-bond situation is expected with respect to the two quinone CO groups in A•1 which is based on the experimentally observed asymmetric spin-density distribution of this radical. Such a situation was also proposed on theoretical grounds by O’Malley.49 Experimental evidence for the asymmetry of the spin-density distribution relies mainly on the significantly increased methyl hfc in the A•1 site as compared 33,38 With plastoquinone-9 substito VK•1 in organic solvents. tuted for VK1 in the A•site of PS I, a corresponding 1 asymmetric spin density distribution has recently been observed for more ring positions.50 Interestingly, an asymmetric spin density distribution, related to an asymmetric H-bond situation, is also a characteristic feature of the first acceptor QA in bRC (for an overview see, e.g., ref 21). With respect to the position of the hydrocarbon tail at the quinone ring, the asymmetric and alternating spin density is, however, inverted between QA in bRC and A•1 in PS I. This is consistent with a preferential H-bond to the carbonyl group next to the phytyl chain of A1, whereas the stronger H-bond for QA is found for the CO group in a position distal to the tail.21 On the basis of this result we gained an independent argument for the choice of the specific A1 orientation chosen in Figure 9. Only for this case, the phytyl chain is next to the CO group with the dominant H-bond pointing toward the stromal side. The EPR derived position and orientation of Al incorporated into the X-ray strucural model8 indicate a proximity to the stromal n-o loop with possible H-bonding to the backbone, whereas an H-bond to the lumenal carbonyl group is less likely. With a dominant H-bond to the stromal carbonyl group, the phytyl chain also has to point toward the stromal side. Finally, we note that for the C2 symmetric QK′ site, the equivalent A1 orientation (not shown in Figure 9) remains as an alternative consistent with all data presented in this study. Details of the A1 binding site on a molecular level are expected soon from an improved X-ray structural model which should also show individual amino acids. From improved structural and spectroscopic data, more detailed information on the specific protein-cofactor interactions in the A1 site are expected that will provide a better understanding of the functional properties of the cofactors in PS I. 4.3. Orientation of g(P•+ 700) with Respect to the Molecular Axis System of P700. In this work, we determined the orientation of the P•+ 700 g-tensor axes, with respect to the crystallographic c-axis. Even when the orientation of the crystals with respect to their a- and b-axes are known, a 6-fold ambiguity will remain, due to the 6-fold crystal symmetry. In addition, the orientation •+ •of g(P•+ 700) with respect to the axis connecting P700 and A1 (zD, see Figure 3) is known from the transient EPR data (see above). If the position of the active quinone is known, then the
9738 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Zech et al. TABLE 6: Orientation of the g-Tensor of P•+ with Respect to the Molecular Out-of-Plane z-Axis of the Chla Molecule, zChl ∠(zChl, g(P•+)) P•+ 700
(this work)a
b23 P•+ 865
xP
yP
zP
pos. 3 (eC1′-QK) pos. 1 (eC1′-QK′)
50° ( 9° 72° ( 10°
78° ( 5° 121° ( 3°
137° ( 6° 37° ( 8°
option II option I
70° 128°
101° 96°
23° 38°
The A•1 position is either 3 (for eC1′-QK) or 1 (eC1′-QK′) given in ref 12. The error is mainly determined by the remaining uncertainty in a,b ref position of A•1 (∠(zD , a) ) 0° ( 10°, see Table 5 and ref 12). b Time-resolved EPR on P•+ Q•- at W-band22 supported option II of 865 A 23. a
Figure 10. Principal axis system of the g-tensor of P•+ 700 with respect to the molecular framework of the spin carrying Chla molecule of 12 P•+ 700. It is assumed that the active quinone A1 is located at position 3 as used for Figure 9, with the radical pair state associated with eC′1-QK. Refer to Table 6 for the angles between the g-tensor axes (xP, yP, zP) and the molecular z-axes of the Chla, zChl. The orientation of the Chla molecule with respect to the c-axis has been chosen according 51 A: view direction to ENDOR data on P•+ 700 in PS I single crystals. perpendicular to c and parallel to the plane normal; zChl, of the Chla molecule. B: view direction parallel to the c-axis.
orientation of g(P•+ 700) with respect to the molecular axes system can be derived unambiguously. Figure 10 shows the orientation of the g-tensor axes (xP, yP, zP) within the molecular structure of the spin-carrying Chla molecule of P•+ 700. For the picture it is assumed that the active quinone is represented by QK (as shown in Figure 9) and the spin-carrying chlorophyll is eC1′.45 The exact orientation of the Chla molecule has been obtained previously by ENDOR 51 Here, we focus experiments on P•+ 700 in PS I single crystals. on the discussion of the angles between the molecular z-axis of the eC1′ chlorophyll, zChl, and the principal axes of g(P•+ 700). Table 6 summarizes the relevant data, i.e., the angles between the axes of g(P•+ 700) and zChl. As seen from Table 6, the g-tensor of P•+ 700 is rotated with respect to the molecular axis system of P•+ 700. This deviation is seen for both of the two possible ET chains, represented either by the pair eC1′-QK′ or eC1′-QK. For comparison, the respective angles of two orientations of P•+ 865 in bRCs23 are also included. According to the theory of Stone for the g-tensor of planar π-systems19,20 a collinearity of the zP-axis and the molecular z-axis would be expected. A deviation from this coplanarity in the primary donors might arise from their dimeric structure. However, experimental evidence against this assumption is given
by the study of the orientation of the g-tensor in bRC heterodimer mutant HL(M202)25 where the unpaired electron on P•+ 865 resides on only one BChl molecule. In this study a similar tilt between the zP-axis and the molecular z-axis as in the dimeric system has been found.25 In a theoretical study52 the experimentally observed deviation from collinearity has been interpreted in terms of a out-of-plane rotation of the acetyl group of BChl. However, this argument for the deviation of zP from the molecular z-axis does not hold for P700 since the acetyl group of BChla is replaced by an unpolar vinyl group in Chla see Scheme 1). It has been shown that the spin density distribution in the •+ 51 P•+ 700 is even more asymmetric than for P865. A localization of the electron on one dimer half should result in a g-tensor with zP perpendicular to the π-plane of the macrocycle and the xP and yP axes lying in the molecular plane. A deviation from this symmetry is expected when the π-system deviates from planarity or when strong interactions with the environment are present. Consequently, we have to anticipate that the significant difference in the g-tensor axes from the ideal geometry is caused by interactions with the surrounding that are not yet fully understood. Possible reasons can be found in the influence of the second dimer half, the effect of neighboring pigments and/or the ligating amino acids. Clearly, thorough high-field EPR studies of model systems in oriented matrices or single crystals hand in hand with advanced theoretical calculations of both the magnitude and orientation of the g-tensor are required to finally solve this interesting question. Irrespective of the unidentified source of the deviation of molecular and g-tensor axes for a Chla species as found in this study for P700, this deviation seems to be a common feature of the chlorophyll ions in photosynthetic protein complexes. Acknowledgment. We are grateful to B. Altenberg-Greulich (EMBL Heidelberg, Germany) for refinement of the molecular modeling of the A1 site and valuable discussions. N. Krauss, O. Klukas, P. Jordan (FU Berlin, Germany) are kindly acknowledged for stimulating discussions about the PS I structure. This work has been supported by Deutsche Forschungsgemeinschaft (SFB 498, Teilprojekte Al, A3 and C5; Schwerpunktprogramm Hochfeld-EPR) and Fonds der Chemischen Industrie (to W.L. and R.B.). References and Notes (1) Golbeck, J. H. In The Molecular Biology of Cyanobacteria; Bryant, D. A., Ed.; Vol. 1 of AdVances in Photosynthesis; Kluwer Academic Publishers: Dordrecht, 1994; pp 319-360. (2) Fromme, P. Curr. Opin. Struct. Biol. 1996, 6, 473. (3) Brettel, K. Biochim. Biophys. Acta 1997, 1318, 322. (4) Golbeck, J. H. Photosynth. Res. 1999, 61, 107.
Structural Model for P700•+A1•- in PS I (5) Klukas, O.; Schubert, W.-D.; Jordan, P.; Krauss, N.; Fromme, P.; Witt, H. T.; Saenger, W. J. Biol. Chem. 1999, 274, 7351. (6) Krauss, N.; Schubert, W.-D.; Klukas, O.; Fromme, P.: Witt, H. T.: Saenger, W. Nature Struct. Biol. 1996, 3, 965. (7) Schubert, W.-D.; Klukas, O.; Krauss, N.; Saenger, W.; Fromme, P.; Witt, H. T. J. Mol. Biol. 1997, 272, 741. (8) Klukas, O.; Schubert, W.-D.; Jordan, P.; Krauss, N.; Fromme, P.; Witt, H. T.; Saenger, W. J. Biol. Chem. 1999, 274, 7361, Brookhaven Protein Data Bank entry 1C51. (9) Zech, S. G.; Lubitz, W.; Bittl: R. Ber. Bunsen-Ges. Phys. Chem. 1996, 100, 2041. (10) Hara, H.; Dzuba, S. A.; Kawamori, A.; Akabori, K.; Tomo, T.; Satoh, K.; Iwaki, M.; Itoh, S. Biochim. Biophys. Acta 1997, 1322, 77. (11) Bittl, R.; Zech, S. G. J. Phys. Chem. B 1997, 101, 1429. (12) Bittl, R.; Zech, S. G.; Fromme, P.; Witt, H. T.; Lubitz, W. Biochemistry 1997, 36, 12001. (13) van der Est, A.; Prisner, T.; Bittl, R.; Fromme, P.; Lubitz, W.; Mo¨bius, K.; Stehlik, D. J. Phys. Chem. B 1997, 101, 1437. (14) Kamlowski, A.; Zech, S. G.; Fromme, P.; Bittl, R.; Lubitz, W.; Witt, H. T.; Stehlik, D. J. Phys. Chem. B 1998, 102, 8266. (15) Golbeck, J. H. Curr. Opin. Struct. Biol. 1993, 3, 508. (16) MacMillan, F.; Hanley, J.; van der Weerd, L.; Knu¨pling, M.; Un, S.; Rutherford, A. W. Biochemistry 1997, 36, 9297-9393. (17) Stehlik, D.; Bock, C. H.; Petersen, J. J. Phys. Chem. 1989, 93, 1612. (18) Kothe, G.; Weber, S.; Ohmes, E.; Thurnauer, M. C.; Norris, J. R. J. Phys. Chem. 1994, 98, 2706. (19) Stone, A. J. Proc. R. Soc. A 1963, 271, 424. (20) Stone, A. J. Mol. Phys. 1963, 6, 509. (21) Lubitz, W.; Feher, G. Appl. Magn. Reson. 1999, 17, 1. (22) Prisner, T. F.; van der Est, A.; Bittl, R.; Lubitz, W.; Stehlik, D.; Mo¨bius, K. Chem. Phys. 1995, 194, 361. (23) Klette, R.; To¨rring, J. T.; Plato, M.; Mo¨bius, K.; Bo¨nigk, B.; Lubitz, W. J. Phys. Chem. 1993, 97, 2015. (24) Lendzian, F.; Huber, M.; Isaacson, R. A.; Endeward, B.; Plato, M.; Bo¨nigk, B.; Mo¨bius, K.; Lubitz, W.; Feher, G. Biochim. Biophys. Acta 1993, 1183, 139. (25) Huber, M.; To¨rring, J. T. Chem. Phys. 1995, 194, 379. (26) Ro¨gner, M.; Nixon, P. J.; Diner, B. A. J. Biol. Chem. 1990, 256, 6189. (27) Fromme, P.; Witt, H. T. Biochim. Biophys. Acta 1998, 1365, 175. (28) Berthold, T.; Bechthold, M.; Heinen, U.; Link, G.; Poluektov, O.; Utschig, L.; Tang, J.; Thurnauer, M.; Kothe, G. J. Phys. Chem. B 1999, 103, 10733. (29) Zech, S. G.; Bittl, R.; Gardiner, A. T.; Lubitz, W. Appl. Magn. Reson. 1997, 13, 517. (30) Hofbauer, W.; Bittl, R. BRUKER Report 1998, 145, 38.
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9739 (31) Stesmans, A.; van Gorp, G. ReV. Sci. Instrum. 1989, 60, 2949. (32) Stehlik, D.; Mo¨bius, K. Annu. ReV. Phys. Chem. 1997, 48, 745. (33) Teutloff, C.; MacMillan, F.; Bittl, R.; Lendzian, F.; Lubitz, W. In Magnetic Resonance and Related Phenomena; Ziessow, D., Lubitz, W., Lendzian, F., Eds.; Vol. 2; Technische Universitat: Berlin, 1998; pp 806807. (34) Prisner, T. F.; McDermott, A. E.; Un, S.; Norris, J. R.; Thurnauer, M. C.; Griffin, R. G. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 9485. (35) Bratt, P. J.; Rohrer, M.; Krzystek, J.; Evans, M. C. W.; Brunel, L.-C.; Angerhofer, A. J. Phys. Chem. B 1997, 101, 9686. (36) van der Est. A.; Sieckmann, I.; Lubitz, W.; Stehlik, D. Chem. Phys. 1995, 194, 349. (37) Weber, S.; Ohmes, E.; Thurnauer, M. C.; Norris, J. R.; Kotlle, G. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 7789. (38) Rigby, S. E. J.; Evans; M. C. W.; Heathcote, P. Biochemistry 1996, 35, 6651. (39) Fu¨chsle, G.; Bittl, R.; van der Est, A.; Lubitz, W.; Stehlik, D. Biochim. Biophys. Acta 1993, 1142, 23. (40) Ermler, U.; Fritzsch, G.; Buchanan, S. K.; Michel, H. Structure 1994, 2, 925. (41) Stowell, M. H. B.; McPhillips, T. M.; Rees, D. C.; Soltis, S. M.; Abresch, E.; Feher, G. Science 1997, 276, 812. (42) Joliot, P.; Joliot, A. Biochemistry 1999, 38, 11130. (43) Yang, F.; Shen, G.; Schluchter, W. M.; Zybailow, B. L.; Ganago, A. O.; Vassiliev, I. R.; Bryant, D. A.; Golbeck. J. H. J. Phys. Chem. B 1998, 102, 8288. (44) Bittl, R.; Zech. S. G.; Teutloff, C.; Krabben, L.; Lubitz, W. In Photosynthesis: Mechanisms and Effects; Garab, G., Ed.; Kluwer Academic Publishers: Dordrecht, 1998; pp 509-514. (45) Krabben, L.; Schlodder, E.; Jordan, R.; Carbonara, D.; Giacometti, G.; Lee, H.; Webber, A. N.; Lubitz, W. Biochemistry 2000, in press. (46) Kamlowski, A.; Altenberg-Greulich, B.; van der Est, A.; Zech, S. G.; Bittl, R.; Fromme, P.; Lubitz, W.; Stehlik, D. J. Phys. Chem. B 1998, 102, 8278. (47) Hanley, J.; Deligiannakis, Y.; MacMillan, F.; Bottin, H.; Rutherford, A. W. Biochemistry 1997, 36, 11543. (48) Iwaki, M.; Itoh, S. Biochemistry 1991, 30, 5347. (49) O’Malley, P. Biochim. Biophys. Acta 1999, 1411, 101. (50) Zybailov, B.; van der Est, A.; Zech, S. G.; Teutloff, C.; Johnson, W.; Shen, G.; Bittl, R.; Stehlik, D.; Chitnis, P. R.; Golbeck, J. H. J. Biol. Chem. 2000, 275, 8531. (51) Ka¨ss, H. Die Struktur des prima¨ ren Donators P700 in Photosystem I - Untersuchungen mit Methoden der stationa¨ ren und gepulsten Elektronenspinresonanz; Ph.D. Thesis, Technische Universitat Berlin, 1995. (52) Mo¨bius, K.; Plato, M. In The Reaction Center of Photosynthetic Bacteria - Structure and Dynamics; Michel-Beyerle, M.-E., Ed.; SpringerVerlag: Berlin, 1996; pp 63-80.