The Primary Electron Donor of Photosystem II of the Cyanobacterium

7351

2008, 112, 7351–7354 Published on Web 05/30/2008

The Primary Electron Donor of Photosystem II of the Cyanobacterium Acaryochloris marina Is a Chlorophyll d and the Water Oxidation Is Driven by a Chlorophyll a/Chlorophyll d Heterodimer T. Renger*,† and E. Schlodder‡ Institute of Chemistry and Biochemistry, Free UniVersity Berlin, Fabeckstrasse 36a, D-14195 Berlin, Germany, and Max Volmer Laboratory for Biophysical Chemistry, Technical UniVersity Berlin, Strasse des 17. Juni 135, D-10623 Berlin, Germany ReceiVed: March 04, 2008; ReVised Manuscript ReceiVed: April 30, 2008

We present a theoretical analysis of the flash-induced absorbance difference spectrum assigned to the formation of the secondary radical pair P+QA- in photosystem II of the chlorophyll d-containing cyanobacterium Acaryochloris marina. An exciton Hamiltonian determined previously for chlorophyll a-containing photosystem II complexes is modified to take into account the occupancy of certain binding sites by chlorophyll d instead of chlorophyll a. Different assignments of the reaction center pigments to chlorophyll a or d from the literature are investigated in the calculation of the absorbance difference spectrum. A quantitative explanation of the experimental data requires one chlorophyll a molecule per reaction center, located at the site of PD1. The remaining sites are occupied by chlorophyll d and pheophytin a. By far, the lowest site energy is found for the accessory chlorophyll of the D1 branch, ChlD1, which represents the sink of excitation energy and therefore the primary electron donor. The cationic state is stabilized at the chlorophyll a, which drives the oxidation of water. I. Introduction The chlorophyll d-containing cyanobacterium Acaryochloris marina possesses the unique capability to use light with wavelengths larger than 700 nm for the oxidation of water. It lives in an ecological niche where light with higher energies does not arrive because it is absorbed by other organisms that contain chlorophyll a (Chla) as the major light-harvesting pigment.1 Acaryochloris marina has adapted to these light conditions by utilizing chlorophyll d (Chld), instead of Chla, as the major pigment.2 Chld absorbs about 35 nm to the red of Chla1 and has a somewhat higher oxidation potential than Chla in vitro.3 Interestingly, besides the large majority of Chld, there is a small amount of Chla present in A. marina, where the exact amount was found to depend on the light intensity.4 The ratio between Chla and pheophytin a (Pheoa) in whole cells of A. marina increases from 1 to 2.3 when going from low to high light intensities.4 Taking into account that at least one5,6 and most likely two7 (as in Chla-containing photosystem II (PSII))8 Pheoa are bound in the reaction center (RC) of PS-II of A. marina, the above low-light-intensity ratio gives an upper limit of 2 Chla molecules bound per PS-II RC. The functional importance and the location of this small Chla content in A. marina are controversially discussed in the literature.3,4,6,9–11 From light-induced cation-minus-neutral FTIR difference spectra, Tomo et al.9 concluded that the positive charge is stabilized on a Chld, which is located at the PD1 site in the special pair of two Chld molecules. In contrast to this conclusion, Schlodder et al.6 showed that the main bleaching * To whom correspondence should be addressed. † Free University Berlin. ‡ Technical University Berlin.

10.1021/jp801900e CCC: $40.75

in the Soret region at 435 nm and the spectrum of the absorbance increase in the NIR upon photooxidation are identical with flashinduced absorbance changes measured for Chla-containing PSII, implying that the PD1 site is occupied by a Chla. In the Qy region, the absorbance difference spectrum measured by Schlodder et al.6 shows a main bleaching at around 727 nm at 77 K, which was confirmed by Itoh et al.,10 whereas the one measured by Tomo et al.9 is at 713 nm. At present, the most likely reason for the difference is the use of different excitation conditions. Whereas Schlodder et al.6 and Itoh et al.10 measured absorbance changes induced by single turnover flashes, Tomo et al.9 measured the difference between the absorbance during illumination for 1 s and in the dark. They assume that oxidized P is photoaccumulated during the illumination. However, the measured bleaching at around 713 nm, which has been observed neither by Schlodder et al.6 nor by Itoh et al.,10 most likely indicates the accumulation of unwanted products (e.g., ChlZ), which cannot be excluded, especially in the presence of silicomolybdate. Although the P+QA--PQA absorbance difference spectra in the Qy region measured by Schlodder et al.6 and by Itoh et al.10 are very similar, the interpretations are different. Itoh et al.10 assume the main bleaching in their spectra to arise from the bleaching of the low-energy exciton state of a Chld homodimer special pair, designated by them as P727, and an electrochromic shift due to the accessory chlorophyll ChlD1, which is assigned to be also a Chld. Schlodder et al.6 infer that the main bleaching is solely caused by the electrochromic shift of a Chld in the site of ChlD1 and that the special pair is either a Chla homo- or a Chla/Chld heterodimer. The main purpose of the present letter is to provide a structure-based calculation of the above-discussed absorbance  2008 American Chemical Society

7352 J. Phys. Chem. B, Vol. 112, No. 25, 2008

Letters

difference spectrum in the Qy region and, on the basis of a comparison with experimental data,6 to decide which site in the RC of PS-II of A. marina is occupied by Chld and which by Chla. II. Theoretical Methods An exciton Hamiltonian derived previously on the basis of the crystal structure of Chla-containing PS-II of T. elongatus8 and calculation of optical spectra of D1-D2-cytb559 complexes12 and of PS-II core complexes13,14 is used as a starting point. The Hamiltonian is modified to take into account a replacement of Chla by Chld at certain sites in the following way. The local transition energies (the so-called site energies) at those sites which are occupied by Chld are treated as parameters in the fit of optical spectra of A. marina. The about 1.3 times higher dipole strength15 of the S0 f S1 transition of Chld, as compared to that of Chla, is taken into account by scaling the transition dipole moments and the transition charges used in the calculation of the excitonic couplings with the TrEsp (transition-charges-from-electrostatic-potential)method,16 accordingly. Concerning the two Pheoa molecules in the RC of PS-II of T. elongatus, we assume that they are both present in A. marina and that their spectroscopic properties are unchanged, that is, we use the same site energies as those determined previously.12–14 The theory of optical spectra, developed17 and applied12–14 previously, takes into account the excitonic couplings between pigments and the dynamic and static modulation of pigment transition energies by the protein. The dynamic modulation is described by a spectral density that was extracted from fluorescence line-narrowing spectra and the temperature dependence of the absorption spectrum.13,17 The static modulation is taken into account by a Monte Carlo method, in which an average is performed over 30000 randomly generated homogeneous spectra, which are each obtained for a different realization of site energies. Each particular site energy is taken from a Gaussian distribution function centered around a (mean) site energy. A width of 200 cm-1 as determined previously13 is assumed for the distribution function, except for the accessory chlorophyll, ChlD1, for which a smaller value is assumed, as previously.13,14 In the calculation of the absorbance spectrum of the complex in the state P+QA-, it is assumed that the cation resides on the special pair chlorophyll PD1 of the D1 branch, as in Chlacontaining PS-II RCs.13,18 The absorbance of the complex in the state P+QA- is obtained by setting the transition dipole strength and the excitonic couplings involving PD1 to zero and by taking into account the electrochromic shift of the site energies of the remaining pigments by the positive charge on PD1 and the negative charge on QA. As described in detail before,12,13 the electrochromic shift of the site energy of a pigment is obtained from the Coulomb interaction of the difference vector ∆µb between excited- and ground-state permanent dipole moments of that pigment and the charge density of the oxidized PD1 and the reduced QA. The latter are approximated by positive and negative elementary charges, respectively, that are evenly distributed over the atoms contributing to the conjugated π systems. Local field and screening effects of the Coulomb coupling by the polarizability of the protein are considered in an effective way by using a factor 1/
eff in the calculation of the Coulomb coupling. The magnitude and direction of the ∆µb vectors for Chla were estimated from Stark spectra, as discussed previously.13 Unfortunately, no Stark data on Chld are available in the literature yet. We took the same ∆µb vectors as those for Chla as a first approximation

Figure 1. Different model calculations of the P+QA--PQA absorbance difference spectrum (upper and middle panels) are compared with experimental data6 (bottom panel); T ) 77 K. In the upper panel, a model is considered with three Chld and one Chla, assuming the following site energies for the Chl’s in units of nm (in the order of PD1, PD2, ChlD1, ChlD2): 666, 700, 724, 700. In the middle panel, a model with four Chld is considered with site energies (same order and units as before) of 724, 700, 700, 700 (black curve), 700, 724, 700, 700 (red curve), 724, 724, 700, 700 (blue curve). The site energies of the two Pheoa correspond to wavelengths of 672 and 675 nm for the D1 and the D2 branch, respectively.12,13 The same width of 200 cm-1 was assumed for the inhomogeneous distribution function of all six site energies, and an effective dielectric constant of
eff ) 1.520 was used in the calculation of electrochromic shifts.

since both molecules have a very similar structure. Preliminary experimental Stark spectra on Chld support this approximation.19 The values for the electrochromic shifts and the excitonic couplings, used in the calculations of the spectra, are given in the Supporting Information (SI). III. Results We first aim at a qualitative explanation of the experimental P+QA--PQA absorbance difference spectrum of A. marina, measured6 at 77 K that is shown in the lower panel of Figure 1. For simplicity, a minimal model is considered first, where only two different values are assumed for the site energies of the Chld molecules, which are determined from the calculation of the spectrum and comparison with the experimental data. The same inhomogeneous width is assumed for the distribution function of all site energies. To investigate the origin of the low-energy bleaching at around 727 nm, two different models are considered. In model (a) PD1 is a Chla, whereas PD2, ChlD1, and ChlD2 are Chld molecules, and the 727 nm bleaching is assumed to arise from an electrochromic band shift of ChlD1.6,10 In model (b), PD1, PD2, ChlD1, and ChlD2 are Chld molecules, and the bleaching around 727 nm is assumed to arise due to photooxidation of PD1.9,10 The two possibilities (a) and (b) are considered in the upper and middle panels of Figure 1, respectively. The experimental spectrum can be qualitatively described in model (a) by assuming a strongly red-shifted transition energy

Letters

Figure 2. Comparison of calculations of the P+QA--PQA spectrum, assuming a Chla homo- or a Chla/Chld heterodimer in the special pair, with experimental data,6 T ) 77 K. In the heterodimer calculations, the site energies of the Chl’s in units of nm (in the order of PD1, PD2, ChlD1, ChlD2) are 680, 693, 724, 700. The respective values in the homodimer calculations are 666, 666, 724, 695. The inhomogeneous width of ChlD1 was chosen to be smaller than that of the remaining pigments (see text).

of ChlD1. In model (b), irrespective of whether the two Chld of the special pair are assumed to have the same (blue curve) or different site energies (black and red curves), the absorbance increase observed in the experiment at around 720 nm (bottom panel in Figure 1) cannot be explained. The calculated positive band is either absent or on the low-energy side of the bleaching and smaller in amplitude than the experimental one. The qualitative deviations in model (b) cannot be overcome by varying the site energies further, as long as the special pair has a contribution to the low-energy bleaching, as shown in the SI. In particular, red shifting the site energy of ChlD1 such that it and the special pair both contribute to the long-wavelength bleaching10 does not describe the experimental data. The reason for the discrepancy lies in the optical properties of the special pair: (i) There is a large redistribution of the oscillator strength to the low-energy exciton state, which carries 75% of the dipole strength of the PD1 and PD2 pigments in the case of a homodimer. (ii) There is no strong positive band in its absorbance difference spectrum. In the following, it is investigated whether it is possible to obtain quantitative agreement of model (a) with the experimental data. For this purpose, all three Chld molecules are allowed to have different site energies, and a smaller width of the distribution function of the site energy of ChlD1 is assumed.13,14 In addition, we consider the possibility that the site energy of the Chla at the PD1 site in the special pair heterodimer is shifted with respect to the value in the Chla homodimer (666 nm)12,13 because of the close proximity of the neighboring Chld molecule at the PD2 site. The latter might have an influence due to charge density and dispersive interaction and electron exchange. Alternative to a Chla/Chld heterodimer, we consider also a Chla homodimer. In the latter case, we apply the site energies determined earlier12,13 for T. elongatus since the D1-D2 polypeptides in A. marina and T. elongatus exhibit a high sequence homology (66% identity for D1 and 88% identity for D2).21 In the heterodimer model, almost perfect quantitative agreement with the experimental data is obtained, whereas in the Chla homodimer case, the bleaching at short wavelengths is considerably stronger in the calculations than that in the experiment, as seen in Figure 2. A detailed analysis of the pigment contributions to the calculated spectrum is given in the SI. We note that up-shifting of the site energy of the Chla

J. Phys. Chem. B, Vol. 112, No. 25, 2008 7353

Figure 3. Identities of Chl’s and wavelengths corresponding to the site energies inferred from the present calculations. The values for Pheo’s were assumed to be the same as those in Chla-containing PS-II RCs.12,13

Figure 4. Disorder-averaged equilibrated population of local excited states (eq 1) at T ) 300 K.

homodimer, keeping both site energies equal, does not improve the fit of the experimental data. From this result, we conclude that A. marina most likely contains a Chla/Chld heterodimer special pair. The optimal site energies and identities of Chl’s in the RC of PS-II of A. marina inferred from the present calculations are summarized in Figure 3. The optimal inhomogeneous width for the distribution function of the site energy of ChlD1 is 60 cm-1, and that of the remaining pigments is 200 cm-1. IV. Discussion The calculations suggest that the special pair in the PS-II RC of A. marina is a Chla/Chld heterodimer and the accessory Chl sites, ChlD1 and ChlD2, are both occupied by Chld. Although the site energies of the PD1 and the PD2 sites just differ by 13 nm (see Figure 3), we are quite confident that PD1 is a Chla and PD2 a Chld for the following reasons: (i) there is independent evidence for at least one Chla in the special pair from the P+QA--PQA spectrum in the Soret and NIR regions,6 as discussed in the Introduction, and (ii) in the case of a Chla homodimer, we would expect very similar site energies as those in Chla-containing PS-II RCs (666 nm12,13). The present assignment of a Chla in the PD1 binding site seems to be in contrast with the FTIR data of Tomo et al.9 As noted in the Introduction, the long illumination in their experiment could have led to oxidation of other pigments. In addition, Hastings et al.11 pointed out that the FTIR bands of Chld

7354 J. Phys. Chem. B, Vol. 112, No. 25, 2008

Letters

identified by Tomo et al.9 in the P+-P spectrum might not belong to the oxidized pigment but could as well be due to electrochromically shifted vibrational bands of pigments in the neighborhood of the oxidized pigment. The present calculations reveal a remarkable difference in site energies between ChlD1 and ChlD2 that corresponds to a 24 nm red shift of the local transition of ChlD1 (see Figure 3). It is interesting to note that in Chla-containing PS-II RCs, also a red shift of the site energy of ChlD1 occurs, by 15 nm with respect to that of ChlD2.13 As a consequence, in both RCs, the lowest excited state is localized strongly on ChlD1. To identify the primary electron donor in A. marina, we calculate the disorder-averaged population of local excited states, assuming a fast equilibration of excitons

Pm(eq) )

〈

∑ |cm(M)|2 M

e-EM/kT

∑e

-EN/kT

N

〉

(1) dis

Here, M counts the exciton states with energies EM, and |cm(M)|2 describes the quantum mechanic probability of finding pigment m excited in exciton state M. If one assumes fast equilibration of excitons prior to electron transfer, the decay of excitation energy by electron transfer is described by the rate constant k ) Pm(eq)kintr,12,13 where m is the primary electron donor and kintr is the rate constant for electron transfer between the primary donor and its neighbor. As seen in Figure 4, the thermal occupation probability Pm(eq) is by far the largest for m ) ChlD1, even at physiological temperature (T ) 300 K). Hence, excitation energy is funneled from the Chld-containing antenna to this site, and primary electron transfer starts from there. The excited electron at ChlD1 is transferred to PheoD1. A transfer to PD1 would be uphill in energy since PD1 has a smaller oxidation potential (i.e., a higher HOMO) and a larger transition energy and, therefore, a higher LUMO. PheoD1 also has a larger transition energy, but this difference is overcompensated for by the higher oxidation potential (The experimental oxidation potentials of Pheoa, Chld, and Chla in vitro (acetonitrile) were determined as3 +1.14, +0.88, and +0.81 V, respectively.). However, the driving force for primary electron transfer of about 0.15 eV is only half of the 0.3 eV that one estimates in this simple picture for the same reaction in Chla-containing PS-II. In the case of the latter, despite decades of research, there is still no conclusive picture about the mechanistic and kinetic details of the reaction. Recently, experimental evidence was reported that the primary electron donor is ChlD1 and that the primary electron acceptor is PheoD1 from pump-probe experiments in the visible22 and visible/mid-infrared23 spectral range. However, the reported timescales differ by a factor of 5. Whereas Holzwarth et al.22 inferred a time constant of 3 ps for reduction of PheoD1, the one of Groot et al.23 is 600-800 fs. From structure-based simulations of light harvesting and trapping of excitation energy in PS-II core complexes, we have inferred an even faster time constant of 300 fs for reduction of PheoD1 in the RC.24 A. marina offers the possibility to study the driving force dependence of the primary electron-transfer reaction in oxygenic photosynthesis and to draw conclusions about time scales and mechanistic details. Interpretation of the spectra of A. marina will be much simpler than those for Chlacontaining PS-II because of the better separation of optical bands. Since the oxidation potential of Chld is larger than that of Chla,3 the electron hole is transferred from ChlD1 to PD1, where it drives the oxidation of water. Obviously, a high enough

oxidation potential for the water splitting was not the reason to place a Chla at the PD1 site in A. marina. An interesting question in that respect is: Is Chla required to localize the cationic state at PD1? In Chla-containing photosystem II complexes, the cationic state also localizes at PD1,13,18 although all Chl’s are of the same type. One explanation could be that a delicate protein environment is needed to achieve the latter localization, whereas in the evolution of A. marina, it was simpler to keep one Chla at PD1 to reach the same goal. Acknowledgment. Financial support by the German Research Foundation through collaborative research centers Sfb 429 (TP A9 to T.R.) and Sfb 498 (TP A6 to E.S.) is gratefully acknowledged. Supporting Information Available: Excitonic couplings, electrochromic shifts, best fit of model (b), and detailed analysis of the calculations in Figure 2. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Larkum, A. W. D.; Ku¨hl, M. Trends Plant Sci. 2005, 10, 355. (2) Miyashita, H.; Ikemoto, H.; Kurano, N.; Adachi, K.; Chihara, M.; Miyashi, S. Nature 1996, 383, 402. (3) Kobayashi, M.; Ohashi, S.; Iwamoto, K.; Shiraiwa, Y.; Kato, Y.; Watanabe, T. Biochim. Biophys. Acta 2007, 1767, 596. (4) Mimuro, M.; Akimoto, S.; Gotoh, T.; Yokono, M.; Akiyama, M.; Tsuchiya, T.; Miyshita, H.; Kobayashi, M.; Yamazaki, I. FEBS Lett. 2004, 556, 95. (5) Razeghifard, M. R.; Chen, M.; Hughes, J. L.; Freeman, J.; Krausz, E.; Wydrzynski, T. Biochemistry 2005, 44, 11178. (6) Schlodder, E.; Cetin, M.; Eckert, H.-J.; Schmitt, F.-J.; Barber, J.; Telfer, A. Biochim. Biophys. Acta 2007, 1767, 589. (7) Chen, M.; Telfer, A.; Lin, S.; Pascal, A.; Larkum, A. W. D.; Barber, J.; Blankenship, R. E. Photochem. Photobiol. Sci. 2005, 4, 1060. (8) Loll, B.; Kern, J.; Saenger, W.; Zouni, A.; Biesiadka, J. Nature 2005, 438, 1040. (9) Tomo, T.; Okubo, T.; Akimoto, S.; Yokono, M.; Miyashita, H.; Tsuchiya, T.; Noguchi, T.; Mimuro, M. Proc. Natl. Sci. U.S.A. 2007, 104, 7283. (10) Itoh, S.; Mino, H.; Itoh, K.; Shigenaga, T.; Uzumaki, T.; Iwaki, M. Biochemistry 2007, 46, 12473. (11) Hastings, G.; Wang, R. Photosynth. Res. 2008, 95, 55. (12) Raszewski, G.; Saenger, W,; Renger, T. Biophys. J. 2005, 88, 986. (13) Raszewski, G.; Diner, B.; Schlodder, E.; Renger, T. Biophys. J. 2008, doi:10.1529/biophysj.107.123935, in press, available online. (14) Schlodder, E.; Renger, T.; Raszewski, G.; Coleman, W. J.; Nixon, P. J.; Cohen, R. O.; Diner, B. A. Biochemistry 2008, 47, 3143. (15) Hughes, J. L.; Razeghifard, R.; Logue, M.; Oakley, A.; Wydrzynski, T.; Krausz, E. J. Am. Chem. Soc. 2006, 128, 3649. (16) Madjet, M. E.; Abdurahman, A.; Renger, T. J. Phys. Chem. B 2006, 110, 17268. (17) Renger, T.; Marcus, R. A. J. Chem. Phys. 2002, 116, 9997. (18) Diner, B. A.; Schlodder, E.; Nixon, P. J.; Coleman, W. J.; Rappaport, F.; Lavergne, J.; Vermaas, W. F. J.; Chisholm, D. A. Biochemistry 2001, 40, 9265. (19) Stanislaw Krawczyk (private communication). (20) The present
eff ) 1.5 is slightly smaller than the
eff ) 2.0 used in calculations on Chl-containing PS-II core complexes.13 This difference, inferred from the calculations of the spectra, might not be real and could just compensate for a slightly larger change in permanent dipole moment of Chld than Chla, which was assumed to be equal in the present calculations. (21) Swinley, W. D.; Chen, M.; Cheung, P. C.; Conrad, A. L.; Dejesa, L. C.; Hao, J.; Honchak, B. M.; Karbach, L. E.; Kurdoglu, A.; Lahiri, S.; Mastrian, S. D.; Miyachita, H.; Page, L.; Ramakrishna, P.; Satoh, S.; Sattley, W. M.; Shimada, Y.; Taylor, H. L.; Tomo, T.; Tsuchiya, T.; Wang, Z.; Raymond, J.; Mimuro, M.; Blankenship, R. E.; Touchman, J. W. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 2005. (22) Holzwarth, A. R.; Mu¨ller, M. G.; Reus, M.; Nowaczyk, M.; Sander, J.; Ro¨gner, M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 6895. (23) Groot, M.-L.; Pawlowicz, N. P.; Van Wilderen, L. J. G. W.; Breton, J.; Van Stokkum, I. H. M.; van Grondelle, R. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 13087. (24) Raszewski, G.; Renger, T. J. Am. Chem. Soc. 2008, 130, 4431.

JP801900E