- Email: [email protected]
High-field (94-GHz) EPR spectroscopy on the S2 multiline signal of photosystem II
FEBS Letters 580 (2006) 3605–3609
High-field (94-GHz) EPR spectroscopy on the S2 multiline signal of photosystem II Christian Teutloffa, Sven Keßena, Jan Kernb, Athina Zounib, Robert Bittla,* b
a Freie Universita¨t Berlin, Institut fu¨r Experimentalphysik, Arnimallee 14, 14195 Berlin, Germany Technische Universita¨t Berlin, Max-Volmer-Laboratorium (PC 14), Straße des 17. Juni 135, 10623 Berlin, Germany
Received 4 April 2006; revised 15 May 2006; accepted 16 May 2006 Available online 24 May 2006 Edited by Peter Brzezinski
Abstract The multiline signal of the S2 state in Photosystem II was measured both in frozen-solution and single-crystal preparations from the cyanobacterium Thermosynechococcus elongatus. The frozen-solution EPR spectrum shows a gaussian-like line shape without any resolution of Mn hyperfine couplings. This line shape can be understood on the basis of the single-crystal spectra, where a strong orientation dependence of partially resolved hyperfine structures appears. Simulation of the frozensolution spectrum on the basis of Mn hyperfine couplings taken from published pulse-ENDOR data yields a fully rhombic g-matrix for the multiline signal with principal components 1.997, 1.970, and 1.965. The resulting isotropic g-value giso = 1.977 is surprisingly small compared to other manganese complexes containing manganese ions in the formal oxidation states three and four. Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Keywords: Photosystem II; Oxygen evolving complex; Manganese cluster; EPR; Multiline signal
1. Introduction Photosystem II (PSII) is the protein complex in green plants and cyanobacteria responsible for light-driven water oxidation and concomitant oxygen evolution (see e.g. [1]). X-ray crystallography [2,3] shows that the metal center, supposed to be the catalytic site where the water-splitting reaction takes place, is composed of four manganese ions and a calcium ion (Mn4Ca-cluster). In close vicinity to the metal cluster a redox-active tyrosine is placed which couples the light-induced electron transfer chain of organic cofactors within PSII to the metal cluster. The manganese–calcium cluster cycles during the enzymatic reaction through five states (S0–S4) [4] and accumulates the four redox equivalents necessary for water oxidation. The dark stable ground state is the S1-state. The S2 state can either be populated by a saturating laser flash at room temperature or trapped by continuous illumination at low temperatures and is paramagnetic with a half-integer spin state. Therefore, it is accessible to conventional EPR spectroscopy. Two EPR signals were detected from the S2-state: the multiline signal
* Corresponding author. Fax: +49 30 838 56046. E-mail address: [email protected] (R. Bittl).
(MLS) composed of 18–20 lines centered at g 2 [5], and the g 4.1 signal [6,7]. Here we focus on the MLS. This signal is the result of antiferromagnetic exchange coupling between the manganese ions which leads to a S = 1/2 ground state. In the past, a controversy regarding the number of manganese ˚ hrling and ions responsible for the S2-MLS existed. While A Pace [8] favor a model including only two Mn ions, more common are models invoking four Mn ions [9–13]. The model using only two Mn ions contradicts the published 55Mn-ENDOR data [11,13], and based on the current X-ray crystallographic model of the Mn4Ca cluster with all five ions within a very tight pocket [2,3], the model assuming four interacting Mn ions seems more realistic. Since the discovery of the MLS [5] a number of different parameter sets for the hyperfine couplings (hfcs) of the manganese nuclei and the g-matrix of the S = 1/2 ground state have been derived from EPR spectroscopy at S-, X-, and Q-band frequencies [8–10,12]. Additional information on the manganese hfcs has been gained from X- and Q-band pulse ENDOR spectroscopy [11,13]. However, the complicated structure of the S2-MLS arising from the exchange coupling between the manganese ions, the high nuclear spin (I = 5/2) of Mn with anisotropic hyperfine coupling, and g-anisotropy make a distinction of the quality of the different parameter sets extremely difficult. In the case of binuclear Mn model complexes a multifrequency approach including W-band (94-GHz) EPR has proven successful for resolving such ambiguities [14,15]. The first pioneering step towards high-field/high-frequency EPR of the Mn4Ca cluster of PSII are experiments at W-band presented by Kawamori et al. [16] at the 2004 Photosynthesis Congress. There, using the cw EPR technique the S2-MLS signal was observed in PSII single crystals but no MLS from frozen-solution samples has been reported so far. Here we report pulse EPR measurements at 94 GHz on both frozen solution and single crystals of PSII from Thermosynechococcus (T.) elongatus. Based on the hyperfine data from Q-band 55Mn-ENDOR [13] we deduce from the 94-GHz frozen-solution spectrum the principal components of the rhombic g-matrix for the MLS spectrum of the S2-state of PSII.
2. Materials and methods Preparation and crystallization of PSII core complexes (PSIIcc) from T. elongatus are described elsewhere [17]. PSIIcc in 100 mM PIPES buffer at pH 7.0, including 5 mM CaCl2 and 0.03% b-DM, with a chlorophyll concentration of 13 mM were filled into quartz tubes (0.7 mm i.d., 0.87 mm o.d.) and frozen in liquid nitrogen. The frozen
0014-5793/$32.00 Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2006.05.039
3606 solution was then illuminated about 2 min at 200 K with a 180 W halogen lamp immediately before the EPR experiments. The sample tube was transferred after illumination into the cryostat pre-cooled to 80 K. Single crystals of PSIIcc soaked in 100 mM PIPES buffer at pH 7.0, including 5 mM CaCl2, 10% w/w PEG 2000 and 25% w/w glycerol, were sucked with a drop of soaking buffer into quartz capillaries and rapidly frozen in liquid nitrogen in an arbitrary orientation. PulseEPR experiments were done at 5 K as the MLS could be detected only at temperatures below 10 K. Field-swept echo (FSE) spectra were acquired using a two pulse sequence p/2–s–p–s-echo with pulse lengths of 28 ns and 56 ns for the p/2- and p-pulses, respectively and s = 300 ns on an Elexsys E680 (Bruker, Rheinstetten, Germany) including an E680-PU power upgrade and a Bruker Teraflex EN6001021H ENDOR resonator. The large spectral width of the signal exceeded the field range covered by the room temperature coils, therefore the measurements had to be performed by sweeping the main coils. For field calibration, the signal of the YD radical with well-known g-values was used as an internal standard. Derivative-type spectra were obtained from the FSE spectra by the pseudo-modulation function of the Bruker Xepr software. To obtain the g-matrix principal components from the frozen-solution spectra, simulations were done using a self-written computer programme that calculates the resonance positions by second-order perturbation theory. This programme takes into account a spin Hamiltonian including the electron Zeeman and the hyperfine interaction terms. As shown by Haddy et al. [18], already at 9 GHz third-order corrections are smaller than 0.5% and thus the second-order treatment is well-sufficient for 94 GHz. In order to account for the large quadru˚ hrling and Pace [8] the correpole couplings in the parameter set of A sponding simulation was performed using EasySpin [19].
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
a
b
c
3. Results and discussion 3200
FSE EPR spectra of frozen PSIIcc solutions are shown in Fig. 1. The spectrum after illumination (a), before illumination (b), and the light–dark difference spectrum (c) are presented. The main difference between spectra (a) and (b) and the only prominent signal in the difference spectrum (c) is a very broad line assigned to the light-induced MLS of the S2-state. In the g = 2.005 region the strong contribution of the YD radical is visible in the spectrum after illumination and the dark spectrum. Due to the optimization of the acquisition parameters for the MLS and changes in the YD intensity upon sample illumination, this signal is also present in the difference spectrum. Furthermore six troughs (negative intensities) can be seen in the spectrum after illumination and the dark spectrum. These belong to Mn2+ traces within the sample. Such Mn2+ signals sharpen and show up with high intensity at high-field/high-frequency EPR (see e.g. [20]). The negative signal sign of the Mn2+ contribution relative to that of YD and the MLS is due to the different transition moment of the S = 5/2 state of Mn2+ compared to the two S = 1/2 states. The flip angle of the microwave pulses was optimized for the S = 1/2 systems and, consequently, not for Mn2+. This effect can be used to separate the echo signals arising from different spin states [21] by appropriate choice of the integration window. However, since the Mn2+ signal contribution is almost absent in the difference spectrum, no such separation was performed here. Additionally, a much broader signal contribution is detected centered at about 3265 mT in the spectrum after illumination and the dark spectrum. Like the Mn2+ signal, this intensity vanishes in the light–dark spectrum and remains unassigned here. The main light-induced signal is centered at gcenter = 1.976 with an approximate width (full-width at halfmaximum) of 85 mT and a full-width of about 202 mT which is roughly 25 mT broader compared to X-band (data not
3300
3400 B / mT
3500
3600
Fig. 1. 94-GHz FSE spectra of frozen solution PSIIcc samples from T. elongatus. (a) Spectrum after sample illumination, (b) prior to illumination, and (c) light–dark difference spectrum. T = 5 K; 4096 accumulations per field point; 204 ls shot repetition time.
shown). The signal shape can be well-described as gaussian and no resolution of 55Mn hyperfine couplings is visible in contrast to spectra recorded at X-band frequencies where the S2MLS shows prominent hyperfine features (about 18 resolved lines in cw mode). Rather drastic changes in the appearance of EPR spectra at different microwave frequencies are common for manganese clusters [14,15] and the S2-MLS shows already at Q-band frequencies [13] reduced hyperfine resolution compared to X-band spectra. Strongly orientation-dependent hyperfine structure of the S2-MLS has been reported by Kawamori et al. [16] for cw EPR experiments at 94 GHz on PSII single crystals. At all orientations of the PSII single crystal with respect to the magnetic field [16] the intensity of the hyperfine lines seems smaller compared to the broad underlying signal than in experiments on frozen solution at X-band frequencies. To correlate the broad featureless light-induced signal in our 94-GHz pulse EPR experiments on frozen solution to the earlier observation of the S2-MLS at 94 GHz by cw EPR on PSII single crystals [16] we have also recorded corresponding spectra for PSIIcc single crystals. Two single-crystal FSE spectra are shown in Fig. 2. These are spectra recorded after illumination and no light–dark correction has been performed. Both spectra have a slightly smaller width than the frozen-solution spectrum and are centered at different effective g-values as expected for an anisotropic g-matrix but correspond in general to the broad signal observed for the frozen-solution sample. The
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
3607
a
0˚
crystal orientation
b
likely, because we observe, as Kawamori et al. [16], strongly orientation-dependent hyperfine resolution of the S2-MLS when recording FSE spectra for single crystals. This strongly orientation-dependent hyperfine resolution in the single-crystal spectra is difficult to reconcile with g-strain effects but can easily be understood as an orientation-dependent superposition of g- and hyperfine structure in the spectra leading to a destructive interference of spectral features at certain crystal orientations and in particular in frozen-solution samples. The same effect seems to mask the hyperfine resolution of the S0-MLS in frozen solution already at Q-band frequencies, where a FSE spectrum without visible hyperfine structure has been reported [13]. Fig. 3 shows the light–dark spectrum (a, solid black line) from Fig. 1 together with simulations based on different parameter sets for S2-MLS simulations [9–13] (see Table 1). From these parameter sets that of Peloquin et al. [11] (d) and Kulik et al. [13] (f) satisfactorily reproduce the position of the spectrum center, i.e. employ isotropic g-values giso = 1.977 consistent with the experimental spectrum (a, solid black line). The giso-values of the parameter sets [9,10,12] are in the range of 1.983–1.997 and are significantly too large.
c
g-value 2.000 1.976 1.953
70˚ d
3300
3350
3400
3450
3500
3550
B / mT a Fig. 2. 94-GHz EPR spectra of a PSIIcc single crystal from T. elongatus after illumination. Experimental FSE spectra (a, c) and derivative-type spectra obtained by numerical pseudo-modulation (b, d) are shown for two crystal orientations with respect to the magnetic field. T = 5 K; 4096 accumulations per field point; 204 ls shot repetition time.
two presented spectra are extreme examples with respect to the hyperfine resolution. Pronounced hyperfine structure is visible in the upper trace (a) recorded at an angle of 0° between the magnetic field and an arbitrary reference axis. In spectrum (c), recorded after a rotation of 70° about an axis perpendicular to the magnetic field, no hyperfine structure is resolved. This effect is better visualized by the 1st derivatives (b) and (d) of the original spectra (a) and (c), respectively. In both derivative spectra the complete range of the Mn2+ signal has been removed for clarity. The derivative spectrum (b, 0°) from the PSIIcc single crystal at 94 GHz is similar to typical S2-MLS spectra recorded by cw EPR on frozen PSII solution at X-band and to some of the 94-GHz spectra presented by Kawamori et al. [16]. In contrast, the derivative spectrum (d, 70°) shows almost no hyperfine structure. Indications of typical Mn hyperfine splittings of the order of 7–8 mT are visible only in the narrow spectral range around 3380–3400 mT. Two likely origins for the weak to missing hyperfine resolution in our 94-GHz spectra are obvious. First, g-strain effects become increasingly important at higher fields. Second, g-anisotropy at 94 GHz might be of similar magnitude than the hyperfine splitting. Both effects would lead to a loss of hyperfine resolution. We consider the second possibility more
b
c
d
e
f
3200
3300
3400 B / mT
3500
3600
Fig. 3. Experimental 94-GHz S2-MLS (solid black line (a) and dashed black lines (b–f)) and spectrum simulations (colored lines (a–f)) using different parameter sets. The experimental spectrum has been shifted on the magnetic field axis in the dashed black lines (b–e) for better comparison. Simulation parameters are given in Table 1: (a) this work; (b) Zheng and Dismukes [9]; (c) Hasegawa et al. [10]; (d) Peloquin et al. [11]; (e) Charlot et al. [12]; (f) Kulik et al. [13].
3608
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
Table 1 Parameter sets used for simulation of the S2-MLS Reference
MnA (MHz) ax, ay, az
MnB (MHz) ax, ay, az
MnC (MHz) ax, ay, az
MnD (MHz) ax, ay, az
gx, gy, gz
[9] [10] [11] [12] [13] This work
237, 237, 248, 232, 232, 232, 324, 351, 235, 235, n.d.a
237, 237, 291, 284, 200, 200, 240, 306, 185, 185, n.d.a
257, 257, 110, 106, 311, 311, 248, 263, 310, 310, n.d.a
280, 280, 295, 304, 180, 180, 181, 184, 175, 175, n.d.a
2.00, 2.00, 1.98 1.993, 1.990, 1.976 1.97, 1.97, 1.99 1.988, 1.985, 1.975 1.97, 1.97, 1.99 1.997, 1.970, 1.965
a
237 245 270 312 265
237 194 250 225 245
337 117 270 217 265
300 294 240 193 230
We have employed the hyperfine parameters from [13] for the simulation shown in Fig. 3(a) (solid red line).
In order to judge the quality of the different parameter sets with respect to the hfcs and the g-anisotropy, we have shifted the experimental spectrum in traces (b, c, e) of Fig. 3 on the field axis to fit the center of the simulated spectra and plotted it (dashed black lines) together with each simulated spectrum. The spectra calculated on the basis of the parameter sets of Hasegawa et al. [10] (c), Peloquin et al. [11] (d) and Kulik et al. [13] (f) are all narrower than the experimental spectrum. The parameter sets of Zheng and Dismukes [9] for an assumed MnIII MnIV composition of the S2-state (b) and of Charlot 3 et al. [12] (e) yield simulated spectra with a width roughly in agreement with the experimental spectrum. The spectra calculated with the parameter set of Zheng and Dismukes [9] assumIV ing a MnIII composition of the S2-state and with those 3 Mn ˚ from Ahrling and Pace [8] are significantly too broad (spectra not shown) and these parameter sets have been excluded before, e.g. by Charlot et al. [12]. The similar widths of the calculated spectra for the parameter sets of Hasegawa et al. [10] (c), Peloquin et al. [11] (d), and Kulik et al. [13] (f) on one hand and of Zheng and Dismukes [9] (b) and Charlot et al. [12] (e) on the other can easily be understood. The first three parameter sets all assume a g-anisotropy of about 0.02 and employ hfcs with very similar second moments corresponding to four effective hfcs of 237, 242, and 237, respectively, leading to spectra of comparable width. The parameters of Zheng and Dismukes [9] (b) include the same g-anisotropy but the second moment of the hfcs is substantially larger corresponding to four effective hfcs of 262 MHz. Charlot et al. [12] (e) assume a slightly smaller g-anisotropy of only 0.013 but also use hfcs with a large second moment corresponding to four hfcs of 258 MHz. Drastic differences arise in the visible hyperfine resolution of the calculated spectra based on these parameter sets. The parameters of Hasegawa et al. [10] result in a simulated spectrum (c) lacking any hyperfine resolution like the experimental data. Similarly, the parameters of Charlot et al. [12] give only very little hyperfine resolution (e). The data of Peloquin et al. [11] and Kulik et al. [13] show weak but clearly visible hyperfine structure (d) and (f), respectively. In contrast, the parameters of Zheng and Dismukes [9] yield a well-resolved hyperfine structure (b) which is not observed experimentally. These differences can again immediately be explained. Hasegawa et al. [10] and Charlot et al. [12] invoke fully rhombic gand hyperfine coupling matrices while all other parameter sets assume only axial g- and hyperfine coupling matrices. Furthermore, the dispersion between the smallest and the largest assumed hyperfine coupling and the anisotropies of the individual hfcs is largest in the parameter set of Hasegawa et al. [10] leading to a complete loss of any hyperfine resolution
in the calculated spectrum while the most pronounced hyperfine resolution occurs for the parameters of Zheng and Dismukes [9] assuming two isotropic hfcs. We consider this strong dependence of the hyperfine resolution in the simulated spectra on minor differences between the employed hfcs as further evidence, next to the strong orientation dependence in the single-crystal data and the findings for the S0-MLS at Q-band, to dismiss g-strain effects as the major cause for the absent hyperfine resolution in our frozen-solution spectrum. Even though the spectra calculated on the basis of the parameter sets of Zheng and Dismukes [9] (b) and Charlot et al. [12] (e) show a width roughly in agreement with our experimental spectrum we exclude these from the analysis of our data because they contain hfcs significantly larger than those deduced from 55Mn-ENDOR experiments [11,13]. Instead, we use for the simulation of our 94-GHz S2-MLS the most recent hfcs from the pulsed Q-band 55Mn-ENDOR study of Kulik et al. [13]. Based on these hfcs we obtain the g-matrix principal values 1.997, 1.970, and 1.965 for the S2-MLS. The corresponding spectrum is shown in Fig. 3 (a, red line). The anisotropy of this g-matrix is within the range of anisotropies given previously (see e.g. Tables 1 and 2 in [12]). The resulting giso = 1.977 is identical to the giso of Peloquin et al. [11], however the two g-matrices show clearly different anisotropies. Simulations using the hyperfine parameters of Peloquin et al. [11] require identical g-matrix principal components as given above for a best fit (data not shown). Since the second moment of the hfcs of Hasegawa et al. [10] is very similar to the latter two, it is not surprising that we obtain under the assumption of these hfcs again identical g-matrix principle values. It is interesting to note that a scaling of the hfcs of Charlot et al. [12] by 0.93 to fit the second moment of the hfcs from Kulik et al. [13] yields, together with our g-matrix, a calculated spectrum without any hyperfine structure and in very good agreement with the experiment (data not shown). From our simulations, we estimate an error margin of ±0.002 for the individual g-components. The g-matrix principal values deduced from the simulation of the MLS might show a small additional dependence on the used hyperfine coupling parameters. However, as expected for a gaussian line shape, the isotropic g-value giso = 1.977 derived from the spectral simulation is very close to the g-value of the spectrum center at gcenter = 1.976. Therefore, the giso-value can be considered to be experimentally determined and is independent of the actual hfcs used for the simulations. The giso = 1.975 found in the W-band single-crystal study of Kawamori et al. [16] is slightly smaller than the value given here. The giso = 1.977 is much smaller than the corresponding values observed for dimeric manganese complexes containing
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
MnIII and MnIV ions (see e.g. [22] for model complexes and [14] for Mn catalase). Similar small giso = 1.977 values are, however, frequently found for mixed-valence complexes with MnII and MnIII ions [15]. The small giso = 1.977 is an additional interesting discrepancy between the EPR properties of the S2-state MLS compared to binuclear model complexes. Previously, e.g. the relaxation properties of the MLS [23] have been found to differ strongly from those found for dimeric model complexes composed of MnIII and MnIV ions [24]. These particular properties of the water-oxidizing complex likely arise from the coupling of more than two manganese ions to the S = 1/2 spin ground state of the Mn4Ca cluster in the S2-state of PSII. Reconciling the small giso = 1.977 of the S2-MLS reported here with coupled MnIII and MnIV ions on the basis of DFT calculations will yield valuable insight into the electronic structure of the water-oxidizing complex in PSII. In summary, we have presented a light-induced spectrum in PSIIcc measured at 94 GHz by pulse EPR on frozen solution and representative spectra from a single crystal. Based on the similarity of the frozen-solution spectrum with those from the single crystal which have previously been assigned to the S2-MLS [16] we also attribute this signal to the S2-MLS. The frozen-solution spectrum lacks the characteristic hyperfine resolution of the S2-MLS at lower frequencies and the hyperfine resolution is strongly orientation-dependent in the single crystal. This leads to the conclusion of destructive interference between hyperfine and g-resolution at W-band as the primary cause for the particular spectral shape. Simulation of the frozen-solution spectrum using hfcs from 55Mn-ENDOR yields a fully rhombic g-matrix with substantial anisotropy. The observed giso-value of the S2-MLS is unexpectedly small for a cluster composed of MnIII and MnIV ions. Further experiments to obtain complete rotation patterns for the S2-MLS in PSIIcc single crystals by pulsed 94-GHz EPR are ongoing in our laboratory. These rotation patterns will yield additional constraints for the magnetic resonance parameters characterizing the S2-state of the water-oxidizing complex in PSII and, thereby, provide details of its electronic structure. Acknowledgements: We are grateful to D. DiFiore and C. Lu¨neberg for technical assistance. This work has been supported by Deutsche Forschungsgemeinschaft (Sfb 498; TP C5, C7).
References [1] Wydrzynski, T.J. and Satoh, K., Eds., (2005). Photosystem II The Light-Driven Water: Plastoquinone Oxidoreductase. Advances in Photosynthesis and Respiration, vol. 22, Springer, Dordrecht, The Netherlands. [2] Ferreira, K.N., Iverson, T.M., Maghlaoui, K., Barber, J. and Iwata, S. (2004) Architecture of the photosynthetic oxygenevolving center. Science 303, 1831. [3] Loll, B., Kern, J., Saenger, W., Zouni, A. and Biesiadka, J. (2005) ˚ resolution Towards complete cofactor arrangement in the 3.0 A structure of photosystem II. Nature 438, 1040. [4] Kok, B., Forbush, B. and McGloin, M. (1970) Cooperation of charges in phototsynthetic O2 evolution: 1. A linear 4 step mechanism. Photochem. Photobiol. 11, 457. [5] Dismukes, G.C. and Siderer, Y. (1981) Intermediates of a polynuclear manganese center involved in photosynthetic oxidation of water. Proc. Natl. Acad. Sci. USA 78, 274. [6] Casey, J.L. and Sauer, K. (1984) EPR detection of a cryogenically photogenerated intermediate in photosynthetic oxygen evolution. Biochim. Biophys. Acta 767, 21.
3609 [7] Zimmermann, J.L. and Rutherford, A.W. (1984) EPR studies of the oxygen-evolving enzyme of photosystem II. Biochim. Biophys. Acta 767, 160. ˚ hrling, K.A. and Pace, R. (1995) Simulation of the S2 state [8] A multiline electron paramagnetic resonance signal of photosystem II: a multifrequency approach. Biophys. J. 68, 2081. [9] Zheng, M. and Dismukes, G.C. (1996) Orbital configuration of the valence electrons, ligand field symmetry, and manganese oxidation states of the photosynthetic water oxidizing complex: analysis of the S2-state multiline EPR signals. Inorg. Chem. 35, 3307. [10] Hasegawa, K., Kusunoki, M., Inoue, Y. and Ono, T. (1998) Simulation of the S2-state multiline EPR signal in oriented photosystem II membranes: structural implications for the manganese cluster in an oxygen-evolving complex. Biochemistry 37, 9457. [11] Peloquin, J.M., Campbell, K.A., Randall, D.W., Evanchik, M.A., Pec oraro, V.L., Armstrong, W.H. and Britt, R.D. (2000) 55Mn ENDOR of the S2-state multiline EPR signal of photosystem II: implications on the structure of the tetranuclear Mn cluster. J. Am. Chem. Soc. 122, 10926. [12] Charlot, M.-F., Boussac, A. and Blondin, G. (2005) Towards a spin coupling model for the Mn4 cluster in photosystem II. Biochim. Biophys. Acta 1708, 120. [13] Kulik, L.V., Epel, B., Lubitz, W. and Messinger, J. (2005) 55Mn Pulse ENDOR at 34 GHz of the S0 and S2 states of the oxygenevolving complex in photosystem II. J. Am. Chem. Soc. 127, 2392. [14] Scha¨fer, K.-O., Bittl, R., Lendzian, F., Barynin, V., Weyhermu¨ller, T., Wieghardt, K. and Lubitz, W. (2003) Multifrequency EPR investigation of Dimanganese catalase and related Mn(III)Mn(IV) complexes. J. Phys. Chem. B 107, 1242. [15] Teutloff, C., Scha¨fer, K.O., Sinnecker, S., Barynin, V., Bittl, R., Wieghardt, K., Lendzian, F. and Lubitz, W. (2005) High-field EPR investigations of MnIIIMnIV and MnIIMnIII states of dimanganese catalase and related model systems. Magn. Reson. Chem. 43, S51. [16] Kawamori, A., Shen, J.-R., Mino, H., Furukawa, K., Matsuoka, H. and Kato, T. (2005) High frequency EPR of the Mn-cluster in single crystals of cyanobacterial photosystem II, in: Photosynthesis: Fundamental Aspects to Global Perspectives; Proceedings of the 13th International Congress of Photosynthesis (van der Est, A. and Bruce, D., Eds.), pp. 406–408, International Society of Photosynthesis; Allen Press Inc. [17] Kern, J., Loll, B., Lu¨neberg, C., DiFiore, D., Biesiadka, J., Irrgang, K.D. and Zouni, A. (2005) Purification, characterisation and crystallisation of photosystem II from Thermosynechococcus elongatus cultivated in a new type of photobioreactor. Biochim. Biophys. Acta 1706, 147. [18] Haddy, A., Waldo, G.S., Sands, R.H. and Penner-Hahn, J.E. (1994) Simulation of multifrequency EPR spectra from Mn(III)/ Mn(IV) catalase of Lactobacillus plantarum using a new approach based on perturbation theory. Inorg. Chem. 33, 2677. [19] Stoll, S. and Schweiger, A. (2006) EasySpin, a comprehensive software package for spectral simulation and analysis in EPR. J. Magn. Reson. 178, 42. [20] Rohrer, M., Bru¨gmann, O., Kinzer, B. and Prisner, T.F. (2001) High-field/high-frequency EPR spectrometer operating in pulsed and continuous-wave mode at 180 GHz. Appl. Magn. Reson. 21, 257. [21] Hofbauer, W. and Bittl, R. (2000) A novel approach to separating EPR lines arising from species with different transition moments. J. Magn. Reson. 147, 226. [22] Policar, C., Knu¨pling, M., Frapart, Y.-M. and Un, S. (1998) Multifrequency high-field EPR study of binuclear Mn(III)Mn(IV) complexes. J. Phys. Chem. B 102, 10391. [23] Lorigan, G.A. and Britt, R.D. (1994) Temperature-dependent pulsed electron paramagnetic resonance studies of the S2 state multiline signal of the photo-synthetic oxygen-evolving complex. Biochemistry 33, 12072. [24] Kulik, L.V., Lubitz, W. and Messinger, J. (2005) Electron spin– lattice relaxation of the S0 state of the oxygen-evolving complex of photosystem II and of dinuclear manganese model complexes. Biochemistry 44, 9368.
High-field (94-GHz) EPR spectroscopy on the S2 multiline signal of photosystem II Christian Teutloffa, Sven Keßena, Jan Kernb, Athina Zounib, Robert Bittla,* b
a Freie Universita¨t Berlin, Institut fu¨r Experimentalphysik, Arnimallee 14, 14195 Berlin, Germany Technische Universita¨t Berlin, Max-Volmer-Laboratorium (PC 14), Straße des 17. Juni 135, 10623 Berlin, Germany
Received 4 April 2006; revised 15 May 2006; accepted 16 May 2006 Available online 24 May 2006 Edited by Peter Brzezinski
Abstract The multiline signal of the S2 state in Photosystem II was measured both in frozen-solution and single-crystal preparations from the cyanobacterium Thermosynechococcus elongatus. The frozen-solution EPR spectrum shows a gaussian-like line shape without any resolution of Mn hyperfine couplings. This line shape can be understood on the basis of the single-crystal spectra, where a strong orientation dependence of partially resolved hyperfine structures appears. Simulation of the frozensolution spectrum on the basis of Mn hyperfine couplings taken from published pulse-ENDOR data yields a fully rhombic g-matrix for the multiline signal with principal components 1.997, 1.970, and 1.965. The resulting isotropic g-value giso = 1.977 is surprisingly small compared to other manganese complexes containing manganese ions in the formal oxidation states three and four. Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Keywords: Photosystem II; Oxygen evolving complex; Manganese cluster; EPR; Multiline signal
1. Introduction Photosystem II (PSII) is the protein complex in green plants and cyanobacteria responsible for light-driven water oxidation and concomitant oxygen evolution (see e.g. [1]). X-ray crystallography [2,3] shows that the metal center, supposed to be the catalytic site where the water-splitting reaction takes place, is composed of four manganese ions and a calcium ion (Mn4Ca-cluster). In close vicinity to the metal cluster a redox-active tyrosine is placed which couples the light-induced electron transfer chain of organic cofactors within PSII to the metal cluster. The manganese–calcium cluster cycles during the enzymatic reaction through five states (S0–S4) [4] and accumulates the four redox equivalents necessary for water oxidation. The dark stable ground state is the S1-state. The S2 state can either be populated by a saturating laser flash at room temperature or trapped by continuous illumination at low temperatures and is paramagnetic with a half-integer spin state. Therefore, it is accessible to conventional EPR spectroscopy. Two EPR signals were detected from the S2-state: the multiline signal
* Corresponding author. Fax: +49 30 838 56046. E-mail address: [email protected] (R. Bittl).
(MLS) composed of 18–20 lines centered at g 2 [5], and the g 4.1 signal [6,7]. Here we focus on the MLS. This signal is the result of antiferromagnetic exchange coupling between the manganese ions which leads to a S = 1/2 ground state. In the past, a controversy regarding the number of manganese ˚ hrling and ions responsible for the S2-MLS existed. While A Pace [8] favor a model including only two Mn ions, more common are models invoking four Mn ions [9–13]. The model using only two Mn ions contradicts the published 55Mn-ENDOR data [11,13], and based on the current X-ray crystallographic model of the Mn4Ca cluster with all five ions within a very tight pocket [2,3], the model assuming four interacting Mn ions seems more realistic. Since the discovery of the MLS [5] a number of different parameter sets for the hyperfine couplings (hfcs) of the manganese nuclei and the g-matrix of the S = 1/2 ground state have been derived from EPR spectroscopy at S-, X-, and Q-band frequencies [8–10,12]. Additional information on the manganese hfcs has been gained from X- and Q-band pulse ENDOR spectroscopy [11,13]. However, the complicated structure of the S2-MLS arising from the exchange coupling between the manganese ions, the high nuclear spin (I = 5/2) of Mn with anisotropic hyperfine coupling, and g-anisotropy make a distinction of the quality of the different parameter sets extremely difficult. In the case of binuclear Mn model complexes a multifrequency approach including W-band (94-GHz) EPR has proven successful for resolving such ambiguities [14,15]. The first pioneering step towards high-field/high-frequency EPR of the Mn4Ca cluster of PSII are experiments at W-band presented by Kawamori et al. [16] at the 2004 Photosynthesis Congress. There, using the cw EPR technique the S2-MLS signal was observed in PSII single crystals but no MLS from frozen-solution samples has been reported so far. Here we report pulse EPR measurements at 94 GHz on both frozen solution and single crystals of PSII from Thermosynechococcus (T.) elongatus. Based on the hyperfine data from Q-band 55Mn-ENDOR [13] we deduce from the 94-GHz frozen-solution spectrum the principal components of the rhombic g-matrix for the MLS spectrum of the S2-state of PSII.
2. Materials and methods Preparation and crystallization of PSII core complexes (PSIIcc) from T. elongatus are described elsewhere [17]. PSIIcc in 100 mM PIPES buffer at pH 7.0, including 5 mM CaCl2 and 0.03% b-DM, with a chlorophyll concentration of 13 mM were filled into quartz tubes (0.7 mm i.d., 0.87 mm o.d.) and frozen in liquid nitrogen. The frozen
0014-5793/$32.00 Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2006.05.039
3606 solution was then illuminated about 2 min at 200 K with a 180 W halogen lamp immediately before the EPR experiments. The sample tube was transferred after illumination into the cryostat pre-cooled to 80 K. Single crystals of PSIIcc soaked in 100 mM PIPES buffer at pH 7.0, including 5 mM CaCl2, 10% w/w PEG 2000 and 25% w/w glycerol, were sucked with a drop of soaking buffer into quartz capillaries and rapidly frozen in liquid nitrogen in an arbitrary orientation. PulseEPR experiments were done at 5 K as the MLS could be detected only at temperatures below 10 K. Field-swept echo (FSE) spectra were acquired using a two pulse sequence p/2–s–p–s-echo with pulse lengths of 28 ns and 56 ns for the p/2- and p-pulses, respectively and s = 300 ns on an Elexsys E680 (Bruker, Rheinstetten, Germany) including an E680-PU power upgrade and a Bruker Teraflex EN6001021H ENDOR resonator. The large spectral width of the signal exceeded the field range covered by the room temperature coils, therefore the measurements had to be performed by sweeping the main coils. For field calibration, the signal of the YD radical with well-known g-values was used as an internal standard. Derivative-type spectra were obtained from the FSE spectra by the pseudo-modulation function of the Bruker Xepr software. To obtain the g-matrix principal components from the frozen-solution spectra, simulations were done using a self-written computer programme that calculates the resonance positions by second-order perturbation theory. This programme takes into account a spin Hamiltonian including the electron Zeeman and the hyperfine interaction terms. As shown by Haddy et al. [18], already at 9 GHz third-order corrections are smaller than 0.5% and thus the second-order treatment is well-sufficient for 94 GHz. In order to account for the large quadru˚ hrling and Pace [8] the correpole couplings in the parameter set of A sponding simulation was performed using EasySpin [19].
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
a
b
c
3. Results and discussion 3200
FSE EPR spectra of frozen PSIIcc solutions are shown in Fig. 1. The spectrum after illumination (a), before illumination (b), and the light–dark difference spectrum (c) are presented. The main difference between spectra (a) and (b) and the only prominent signal in the difference spectrum (c) is a very broad line assigned to the light-induced MLS of the S2-state. In the g = 2.005 region the strong contribution of the YD radical is visible in the spectrum after illumination and the dark spectrum. Due to the optimization of the acquisition parameters for the MLS and changes in the YD intensity upon sample illumination, this signal is also present in the difference spectrum. Furthermore six troughs (negative intensities) can be seen in the spectrum after illumination and the dark spectrum. These belong to Mn2+ traces within the sample. Such Mn2+ signals sharpen and show up with high intensity at high-field/high-frequency EPR (see e.g. [20]). The negative signal sign of the Mn2+ contribution relative to that of YD and the MLS is due to the different transition moment of the S = 5/2 state of Mn2+ compared to the two S = 1/2 states. The flip angle of the microwave pulses was optimized for the S = 1/2 systems and, consequently, not for Mn2+. This effect can be used to separate the echo signals arising from different spin states [21] by appropriate choice of the integration window. However, since the Mn2+ signal contribution is almost absent in the difference spectrum, no such separation was performed here. Additionally, a much broader signal contribution is detected centered at about 3265 mT in the spectrum after illumination and the dark spectrum. Like the Mn2+ signal, this intensity vanishes in the light–dark spectrum and remains unassigned here. The main light-induced signal is centered at gcenter = 1.976 with an approximate width (full-width at halfmaximum) of 85 mT and a full-width of about 202 mT which is roughly 25 mT broader compared to X-band (data not
3300
3400 B / mT
3500
3600
Fig. 1. 94-GHz FSE spectra of frozen solution PSIIcc samples from T. elongatus. (a) Spectrum after sample illumination, (b) prior to illumination, and (c) light–dark difference spectrum. T = 5 K; 4096 accumulations per field point; 204 ls shot repetition time.
shown). The signal shape can be well-described as gaussian and no resolution of 55Mn hyperfine couplings is visible in contrast to spectra recorded at X-band frequencies where the S2MLS shows prominent hyperfine features (about 18 resolved lines in cw mode). Rather drastic changes in the appearance of EPR spectra at different microwave frequencies are common for manganese clusters [14,15] and the S2-MLS shows already at Q-band frequencies [13] reduced hyperfine resolution compared to X-band spectra. Strongly orientation-dependent hyperfine structure of the S2-MLS has been reported by Kawamori et al. [16] for cw EPR experiments at 94 GHz on PSII single crystals. At all orientations of the PSII single crystal with respect to the magnetic field [16] the intensity of the hyperfine lines seems smaller compared to the broad underlying signal than in experiments on frozen solution at X-band frequencies. To correlate the broad featureless light-induced signal in our 94-GHz pulse EPR experiments on frozen solution to the earlier observation of the S2-MLS at 94 GHz by cw EPR on PSII single crystals [16] we have also recorded corresponding spectra for PSIIcc single crystals. Two single-crystal FSE spectra are shown in Fig. 2. These are spectra recorded after illumination and no light–dark correction has been performed. Both spectra have a slightly smaller width than the frozen-solution spectrum and are centered at different effective g-values as expected for an anisotropic g-matrix but correspond in general to the broad signal observed for the frozen-solution sample. The
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
3607
a
0˚
crystal orientation
b
likely, because we observe, as Kawamori et al. [16], strongly orientation-dependent hyperfine resolution of the S2-MLS when recording FSE spectra for single crystals. This strongly orientation-dependent hyperfine resolution in the single-crystal spectra is difficult to reconcile with g-strain effects but can easily be understood as an orientation-dependent superposition of g- and hyperfine structure in the spectra leading to a destructive interference of spectral features at certain crystal orientations and in particular in frozen-solution samples. The same effect seems to mask the hyperfine resolution of the S0-MLS in frozen solution already at Q-band frequencies, where a FSE spectrum without visible hyperfine structure has been reported [13]. Fig. 3 shows the light–dark spectrum (a, solid black line) from Fig. 1 together with simulations based on different parameter sets for S2-MLS simulations [9–13] (see Table 1). From these parameter sets that of Peloquin et al. [11] (d) and Kulik et al. [13] (f) satisfactorily reproduce the position of the spectrum center, i.e. employ isotropic g-values giso = 1.977 consistent with the experimental spectrum (a, solid black line). The giso-values of the parameter sets [9,10,12] are in the range of 1.983–1.997 and are significantly too large.
c
g-value 2.000 1.976 1.953
70˚ d
3300
3350
3400
3450
3500
3550
B / mT a Fig. 2. 94-GHz EPR spectra of a PSIIcc single crystal from T. elongatus after illumination. Experimental FSE spectra (a, c) and derivative-type spectra obtained by numerical pseudo-modulation (b, d) are shown for two crystal orientations with respect to the magnetic field. T = 5 K; 4096 accumulations per field point; 204 ls shot repetition time.
two presented spectra are extreme examples with respect to the hyperfine resolution. Pronounced hyperfine structure is visible in the upper trace (a) recorded at an angle of 0° between the magnetic field and an arbitrary reference axis. In spectrum (c), recorded after a rotation of 70° about an axis perpendicular to the magnetic field, no hyperfine structure is resolved. This effect is better visualized by the 1st derivatives (b) and (d) of the original spectra (a) and (c), respectively. In both derivative spectra the complete range of the Mn2+ signal has been removed for clarity. The derivative spectrum (b, 0°) from the PSIIcc single crystal at 94 GHz is similar to typical S2-MLS spectra recorded by cw EPR on frozen PSII solution at X-band and to some of the 94-GHz spectra presented by Kawamori et al. [16]. In contrast, the derivative spectrum (d, 70°) shows almost no hyperfine structure. Indications of typical Mn hyperfine splittings of the order of 7–8 mT are visible only in the narrow spectral range around 3380–3400 mT. Two likely origins for the weak to missing hyperfine resolution in our 94-GHz spectra are obvious. First, g-strain effects become increasingly important at higher fields. Second, g-anisotropy at 94 GHz might be of similar magnitude than the hyperfine splitting. Both effects would lead to a loss of hyperfine resolution. We consider the second possibility more
b
c
d
e
f
3200
3300
3400 B / mT
3500
3600
Fig. 3. Experimental 94-GHz S2-MLS (solid black line (a) and dashed black lines (b–f)) and spectrum simulations (colored lines (a–f)) using different parameter sets. The experimental spectrum has been shifted on the magnetic field axis in the dashed black lines (b–e) for better comparison. Simulation parameters are given in Table 1: (a) this work; (b) Zheng and Dismukes [9]; (c) Hasegawa et al. [10]; (d) Peloquin et al. [11]; (e) Charlot et al. [12]; (f) Kulik et al. [13].
3608
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
Table 1 Parameter sets used for simulation of the S2-MLS Reference
MnA (MHz) ax, ay, az
MnB (MHz) ax, ay, az
MnC (MHz) ax, ay, az
MnD (MHz) ax, ay, az
gx, gy, gz
[9] [10] [11] [12] [13] This work
237, 237, 248, 232, 232, 232, 324, 351, 235, 235, n.d.a
237, 237, 291, 284, 200, 200, 240, 306, 185, 185, n.d.a
257, 257, 110, 106, 311, 311, 248, 263, 310, 310, n.d.a
280, 280, 295, 304, 180, 180, 181, 184, 175, 175, n.d.a
2.00, 2.00, 1.98 1.993, 1.990, 1.976 1.97, 1.97, 1.99 1.988, 1.985, 1.975 1.97, 1.97, 1.99 1.997, 1.970, 1.965
a
237 245 270 312 265
237 194 250 225 245
337 117 270 217 265
300 294 240 193 230
We have employed the hyperfine parameters from [13] for the simulation shown in Fig. 3(a) (solid red line).
In order to judge the quality of the different parameter sets with respect to the hfcs and the g-anisotropy, we have shifted the experimental spectrum in traces (b, c, e) of Fig. 3 on the field axis to fit the center of the simulated spectra and plotted it (dashed black lines) together with each simulated spectrum. The spectra calculated on the basis of the parameter sets of Hasegawa et al. [10] (c), Peloquin et al. [11] (d) and Kulik et al. [13] (f) are all narrower than the experimental spectrum. The parameter sets of Zheng and Dismukes [9] for an assumed MnIII MnIV composition of the S2-state (b) and of Charlot 3 et al. [12] (e) yield simulated spectra with a width roughly in agreement with the experimental spectrum. The spectra calculated with the parameter set of Zheng and Dismukes [9] assumIV ing a MnIII composition of the S2-state and with those 3 Mn ˚ from Ahrling and Pace [8] are significantly too broad (spectra not shown) and these parameter sets have been excluded before, e.g. by Charlot et al. [12]. The similar widths of the calculated spectra for the parameter sets of Hasegawa et al. [10] (c), Peloquin et al. [11] (d), and Kulik et al. [13] (f) on one hand and of Zheng and Dismukes [9] (b) and Charlot et al. [12] (e) on the other can easily be understood. The first three parameter sets all assume a g-anisotropy of about 0.02 and employ hfcs with very similar second moments corresponding to four effective hfcs of 237, 242, and 237, respectively, leading to spectra of comparable width. The parameters of Zheng and Dismukes [9] (b) include the same g-anisotropy but the second moment of the hfcs is substantially larger corresponding to four effective hfcs of 262 MHz. Charlot et al. [12] (e) assume a slightly smaller g-anisotropy of only 0.013 but also use hfcs with a large second moment corresponding to four hfcs of 258 MHz. Drastic differences arise in the visible hyperfine resolution of the calculated spectra based on these parameter sets. The parameters of Hasegawa et al. [10] result in a simulated spectrum (c) lacking any hyperfine resolution like the experimental data. Similarly, the parameters of Charlot et al. [12] give only very little hyperfine resolution (e). The data of Peloquin et al. [11] and Kulik et al. [13] show weak but clearly visible hyperfine structure (d) and (f), respectively. In contrast, the parameters of Zheng and Dismukes [9] yield a well-resolved hyperfine structure (b) which is not observed experimentally. These differences can again immediately be explained. Hasegawa et al. [10] and Charlot et al. [12] invoke fully rhombic gand hyperfine coupling matrices while all other parameter sets assume only axial g- and hyperfine coupling matrices. Furthermore, the dispersion between the smallest and the largest assumed hyperfine coupling and the anisotropies of the individual hfcs is largest in the parameter set of Hasegawa et al. [10] leading to a complete loss of any hyperfine resolution
in the calculated spectrum while the most pronounced hyperfine resolution occurs for the parameters of Zheng and Dismukes [9] assuming two isotropic hfcs. We consider this strong dependence of the hyperfine resolution in the simulated spectra on minor differences between the employed hfcs as further evidence, next to the strong orientation dependence in the single-crystal data and the findings for the S0-MLS at Q-band, to dismiss g-strain effects as the major cause for the absent hyperfine resolution in our frozen-solution spectrum. Even though the spectra calculated on the basis of the parameter sets of Zheng and Dismukes [9] (b) and Charlot et al. [12] (e) show a width roughly in agreement with our experimental spectrum we exclude these from the analysis of our data because they contain hfcs significantly larger than those deduced from 55Mn-ENDOR experiments [11,13]. Instead, we use for the simulation of our 94-GHz S2-MLS the most recent hfcs from the pulsed Q-band 55Mn-ENDOR study of Kulik et al. [13]. Based on these hfcs we obtain the g-matrix principal values 1.997, 1.970, and 1.965 for the S2-MLS. The corresponding spectrum is shown in Fig. 3 (a, red line). The anisotropy of this g-matrix is within the range of anisotropies given previously (see e.g. Tables 1 and 2 in [12]). The resulting giso = 1.977 is identical to the giso of Peloquin et al. [11], however the two g-matrices show clearly different anisotropies. Simulations using the hyperfine parameters of Peloquin et al. [11] require identical g-matrix principal components as given above for a best fit (data not shown). Since the second moment of the hfcs of Hasegawa et al. [10] is very similar to the latter two, it is not surprising that we obtain under the assumption of these hfcs again identical g-matrix principle values. It is interesting to note that a scaling of the hfcs of Charlot et al. [12] by 0.93 to fit the second moment of the hfcs from Kulik et al. [13] yields, together with our g-matrix, a calculated spectrum without any hyperfine structure and in very good agreement with the experiment (data not shown). From our simulations, we estimate an error margin of ±0.002 for the individual g-components. The g-matrix principal values deduced from the simulation of the MLS might show a small additional dependence on the used hyperfine coupling parameters. However, as expected for a gaussian line shape, the isotropic g-value giso = 1.977 derived from the spectral simulation is very close to the g-value of the spectrum center at gcenter = 1.976. Therefore, the giso-value can be considered to be experimentally determined and is independent of the actual hfcs used for the simulations. The giso = 1.975 found in the W-band single-crystal study of Kawamori et al. [16] is slightly smaller than the value given here. The giso = 1.977 is much smaller than the corresponding values observed for dimeric manganese complexes containing
C. Teutloff et al. / FEBS Letters 580 (2006) 3605–3609
MnIII and MnIV ions (see e.g. [22] for model complexes and [14] for Mn catalase). Similar small giso = 1.977 values are, however, frequently found for mixed-valence complexes with MnII and MnIII ions [15]. The small giso = 1.977 is an additional interesting discrepancy between the EPR properties of the S2-state MLS compared to binuclear model complexes. Previously, e.g. the relaxation properties of the MLS [23] have been found to differ strongly from those found for dimeric model complexes composed of MnIII and MnIV ions [24]. These particular properties of the water-oxidizing complex likely arise from the coupling of more than two manganese ions to the S = 1/2 spin ground state of the Mn4Ca cluster in the S2-state of PSII. Reconciling the small giso = 1.977 of the S2-MLS reported here with coupled MnIII and MnIV ions on the basis of DFT calculations will yield valuable insight into the electronic structure of the water-oxidizing complex in PSII. In summary, we have presented a light-induced spectrum in PSIIcc measured at 94 GHz by pulse EPR on frozen solution and representative spectra from a single crystal. Based on the similarity of the frozen-solution spectrum with those from the single crystal which have previously been assigned to the S2-MLS [16] we also attribute this signal to the S2-MLS. The frozen-solution spectrum lacks the characteristic hyperfine resolution of the S2-MLS at lower frequencies and the hyperfine resolution is strongly orientation-dependent in the single crystal. This leads to the conclusion of destructive interference between hyperfine and g-resolution at W-band as the primary cause for the particular spectral shape. Simulation of the frozen-solution spectrum using hfcs from 55Mn-ENDOR yields a fully rhombic g-matrix with substantial anisotropy. The observed giso-value of the S2-MLS is unexpectedly small for a cluster composed of MnIII and MnIV ions. Further experiments to obtain complete rotation patterns for the S2-MLS in PSIIcc single crystals by pulsed 94-GHz EPR are ongoing in our laboratory. These rotation patterns will yield additional constraints for the magnetic resonance parameters characterizing the S2-state of the water-oxidizing complex in PSII and, thereby, provide details of its electronic structure. Acknowledgements: We are grateful to D. DiFiore and C. Lu¨neberg for technical assistance. This work has been supported by Deutsche Forschungsgemeinschaft (Sfb 498; TP C5, C7).
References [1] Wydrzynski, T.J. and Satoh, K., Eds., (2005). Photosystem II The Light-Driven Water: Plastoquinone Oxidoreductase. Advances in Photosynthesis and Respiration, vol. 22, Springer, Dordrecht, The Netherlands. [2] Ferreira, K.N., Iverson, T.M., Maghlaoui, K., Barber, J. and Iwata, S. (2004) Architecture of the photosynthetic oxygenevolving center. Science 303, 1831. [3] Loll, B., Kern, J., Saenger, W., Zouni, A. and Biesiadka, J. (2005) ˚ resolution Towards complete cofactor arrangement in the 3.0 A structure of photosystem II. Nature 438, 1040. [4] Kok, B., Forbush, B. and McGloin, M. (1970) Cooperation of charges in phototsynthetic O2 evolution: 1. A linear 4 step mechanism. Photochem. Photobiol. 11, 457. [5] Dismukes, G.C. and Siderer, Y. (1981) Intermediates of a polynuclear manganese center involved in photosynthetic oxidation of water. Proc. Natl. Acad. Sci. USA 78, 274. [6] Casey, J.L. and Sauer, K. (1984) EPR detection of a cryogenically photogenerated intermediate in photosynthetic oxygen evolution. Biochim. Biophys. Acta 767, 21.
3609 [7] Zimmermann, J.L. and Rutherford, A.W. (1984) EPR studies of the oxygen-evolving enzyme of photosystem II. Biochim. Biophys. Acta 767, 160. ˚ hrling, K.A. and Pace, R. (1995) Simulation of the S2 state [8] A multiline electron paramagnetic resonance signal of photosystem II: a multifrequency approach. Biophys. J. 68, 2081. [9] Zheng, M. and Dismukes, G.C. (1996) Orbital configuration of the valence electrons, ligand field symmetry, and manganese oxidation states of the photosynthetic water oxidizing complex: analysis of the S2-state multiline EPR signals. Inorg. Chem. 35, 3307. [10] Hasegawa, K., Kusunoki, M., Inoue, Y. and Ono, T. (1998) Simulation of the S2-state multiline EPR signal in oriented photosystem II membranes: structural implications for the manganese cluster in an oxygen-evolving complex. Biochemistry 37, 9457. [11] Peloquin, J.M., Campbell, K.A., Randall, D.W., Evanchik, M.A., Pec oraro, V.L., Armstrong, W.H. and Britt, R.D. (2000) 55Mn ENDOR of the S2-state multiline EPR signal of photosystem II: implications on the structure of the tetranuclear Mn cluster. J. Am. Chem. Soc. 122, 10926. [12] Charlot, M.-F., Boussac, A. and Blondin, G. (2005) Towards a spin coupling model for the Mn4 cluster in photosystem II. Biochim. Biophys. Acta 1708, 120. [13] Kulik, L.V., Epel, B., Lubitz, W. and Messinger, J. (2005) 55Mn Pulse ENDOR at 34 GHz of the S0 and S2 states of the oxygenevolving complex in photosystem II. J. Am. Chem. Soc. 127, 2392. [14] Scha¨fer, K.-O., Bittl, R., Lendzian, F., Barynin, V., Weyhermu¨ller, T., Wieghardt, K. and Lubitz, W. (2003) Multifrequency EPR investigation of Dimanganese catalase and related Mn(III)Mn(IV) complexes. J. Phys. Chem. B 107, 1242. [15] Teutloff, C., Scha¨fer, K.O., Sinnecker, S., Barynin, V., Bittl, R., Wieghardt, K., Lendzian, F. and Lubitz, W. (2005) High-field EPR investigations of MnIIIMnIV and MnIIMnIII states of dimanganese catalase and related model systems. Magn. Reson. Chem. 43, S51. [16] Kawamori, A., Shen, J.-R., Mino, H., Furukawa, K., Matsuoka, H. and Kato, T. (2005) High frequency EPR of the Mn-cluster in single crystals of cyanobacterial photosystem II, in: Photosynthesis: Fundamental Aspects to Global Perspectives; Proceedings of the 13th International Congress of Photosynthesis (van der Est, A. and Bruce, D., Eds.), pp. 406–408, International Society of Photosynthesis; Allen Press Inc. [17] Kern, J., Loll, B., Lu¨neberg, C., DiFiore, D., Biesiadka, J., Irrgang, K.D. and Zouni, A. (2005) Purification, characterisation and crystallisation of photosystem II from Thermosynechococcus elongatus cultivated in a new type of photobioreactor. Biochim. Biophys. Acta 1706, 147. [18] Haddy, A., Waldo, G.S., Sands, R.H. and Penner-Hahn, J.E. (1994) Simulation of multifrequency EPR spectra from Mn(III)/ Mn(IV) catalase of Lactobacillus plantarum using a new approach based on perturbation theory. Inorg. Chem. 33, 2677. [19] Stoll, S. and Schweiger, A. (2006) EasySpin, a comprehensive software package for spectral simulation and analysis in EPR. J. Magn. Reson. 178, 42. [20] Rohrer, M., Bru¨gmann, O., Kinzer, B. and Prisner, T.F. (2001) High-field/high-frequency EPR spectrometer operating in pulsed and continuous-wave mode at 180 GHz. Appl. Magn. Reson. 21, 257. [21] Hofbauer, W. and Bittl, R. (2000) A novel approach to separating EPR lines arising from species with different transition moments. J. Magn. Reson. 147, 226. [22] Policar, C., Knu¨pling, M., Frapart, Y.-M. and Un, S. (1998) Multifrequency high-field EPR study of binuclear Mn(III)Mn(IV) complexes. J. Phys. Chem. B 102, 10391. [23] Lorigan, G.A. and Britt, R.D. (1994) Temperature-dependent pulsed electron paramagnetic resonance studies of the S2 state multiline signal of the photo-synthetic oxygen-evolving complex. Biochemistry 33, 12072. [24] Kulik, L.V., Lubitz, W. and Messinger, J. (2005) Electron spin– lattice relaxation of the S0 state of the oxygen-evolving complex of photosystem II and of dinuclear manganese model complexes. Biochemistry 44, 9368.