Phys. Status Solidi B 248, No. 4, 833–838 (2011) / DOI 10.1002/pssb.201000856 b solidi p status s s www.pss-b.com physica basic solid state physics Quantum coherence in photosynthetic complexes

Tessa R. Calhoun1,2 and Graham R. Fleming*,1,2

1 Department of Chemistry, University of California, Berkeley, CA 94720, USA 2 Physical Biosciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA

Received 14 August 2010, revised 5 October 2010, accepted 6 October 2010 Published online 12 January 2011

Keywords dynamics, energy transfer, excitons, structures, two-dimensional electronic spectroscopy

* Corresponding author: e-mail grfl[email protected], Phone: 510-643-2735, Fax: 510-642-6340

The initial steps of photosynthesis require the absorption and artificial photosynthesis, the mechanism by which nature subsequent transfer of energy through an intricate network of achieves this efficiency is unknown. Recent experiments have pigment–protein complexes. Held within the protein scaffold of revealed the presence of long-lived quantum coherences in these complexes, molecules are densely packed photosynthetic pigment–protein complexes spanning bacterial and fixed in specific geometries relative to one another resulting and plant species with a variety of functions and compositions. in Coulombic coupling. Excitation energy transfer through Its ubiquitous presence and wavelike energy transfer implicate these systems can be accomplished with near unity quantum quantum coherence as key to the high efficiency achieved by efficiency [Wraight and Clayton, Biochim. Biophys. Acta 333, photosynthesis. 246 (1974)]. While replication of this feat is desirable for

ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction The ability to harvest and utilize between different organisms, those of light-harvesting energy from the sun has been perfected by nature to sustain antenna vary greatly. Figure 1 shows the structures for a life for organisms ranging from towering redwood trees to variety of photosynthetic complexes including the reaction microscopic bacteria. In all of these systems, success hinges center for purple bacteria [4], the Fenna–Matthews–Olson on efficient energy transfer through highly engineered Complex (FMO) from green sulfur bacteria [5], light networks of pigment–protein complexes. These networks harvesting complex II (LHCII) from higher plants [6], and are extensive requiring the transfer of excitation energy from phycoerythin 545 (PE545) from marine algae [7]. With the antennae complexes, where light energy is absorbed, to exception of the bacterial reaction center in Fig. 1a, these reaction centers, where the excitation is converted into seemingly different complexes all perform the same general chemical energy. Such performance is not currently possible function of absorbing and transferring energy. In addition, all in artificial photosynthetic systems. A direct comparison can of these complexes have experimentally exhibited long- be made to the internal quantum efficiency in organic lived quantum coherence [8–11] as will be discussed in the semiconductors [1] which is limited by the relatively small following sections. exciton diffusion lengths of 5 nm [2]. Natural photosyn- Despite their many differences, all of the structures in thesis, however, maintains nearly 100% quantum efficiency Fig. 1 contain a very dense packing of held in over lengths scales of 20–100 nm [3]. As such, the a precise configuration by the surrounding protein. As mechanism of energy transfer in natural photosynthesis is opposed to the common synthetic approach of using different an active area of research. molecules for donor and acceptor, nature relies on a small subset of similar molecules (Fig. 2) and carefully tunes their 2 Photosynthetic systems electronic properties to perform a variety of different 2.1 Structure The structures of pigment–protein com- functions. This results in broad absorption bands, which is plexes revealed by X-ray crystallography has significantly beneficial for maximizing light absorption, but also causes enhanced our understanding of photosynthetic systems. complex energy landscapes making analysis of these While structures of reaction centers are generally similar systems difficult. The majority of chromophores in all

ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim b solidi

p status s s physica

834 T. R. Calhoun and G. R. Fleming: Quantum coherence in photosynthetic complexes

Figure 1 (online colour at: www.pss-b.com) Crystal structures photosynthetic pigment–protein complexes that have exhibited excitonic coherence: (a) bacterial reaction center [4, 9], (b) FMO [5, 8], (c) LHCII [6, 10], and (d) PE545 [7, 11]. photosynthetic complexes are -based and can be 2.1.1 Chromophores The first, and most obvious, described by a ‘‘four orbital’’ model generating the two method to manipulate the properties of a chromophore for a lowest energy singlet excitations aligned along orthogonal specific function is to change the molecule itself. Figure 2 gives axes of the ring, Qy and Qx [12]. Focusing on the ground and the chemical structures for the chromophore molecules present HOMO–LUMO (Qy) transition can reduce the description to in the complexes in Fig. 1. Slight alterations of the substituent a simple two-level system. Even with this simple picture, groups on the main ring gives rise to the differences between nature creates an array of variability through the use of three (BChl), in which the Qy excitation is in the key mechanisms. near-IR, and (Chl) which absorbs in the visible. Smaller variations result in different types of BChl and Chl (i.e., Chla and Chlb) shifting the electronic excitations by O O 10 nm. Removing the metal center from BChl and Chl gives H H H H rise to bacteriopheophytins and pheophytins, respectively, N N N N which are often found in reaction centers where they play a key H Mg H role in electron transfer. Also shown is a pigment which H N N H N N appears as an opened ring but is often found in a linear conformation and is the only type of photosynthetic H H H H chromophore covalently bound to the protein [12]. All of these O O O O H3COOC H3COOC molecules absorb and transfer energy in the initial steps of OR OR photosynthesis. bacteriopheophytin a bacteriochlorophyll a λQy = 748 nm λQy = 771 nm H 2.1.2 Environment From Fig. 1 it is clear that the O chromophores are in close proximity to the surrounding protein. The protein matrix is an inhomogenous environment N N N N in which each chromophore experiences unique pigment– Mg Mg protein interactions dependent on its specific position or H N N H N N ‘‘site.’’ Shifting of the chromophores’ electronic transitions to lower energies can arise through hydrogen bonding H H H H O O interactions or electrochromic effects from nearby amino O O H3COOC H3COOC acids [14]. As an example, the FMO complex (Fig. 1b) OR OR contains seven identical BChl molecules, but their site λQy = 660 nm λQy = 642 nm energies range from 790 to 820 nm due to different pigment– protein interactions [15]. The dynamic effects of the protein O O will be discussed in Section 4.

N N H 2.1.3 Electronic coupling The average distance H H N N between neighboring chromophore molecules in most photo- synthetic complexes is 10 A˚ [16]. Here electrostatic interactions give rise to Coulombic coupling. As a result, excitations are delocalized and can be described by the Frenkel HOOC COOH exciton model represented by the Hamiltonian [Eq. (1)] phycoerythrobilin Figure 2 Molecular structures of common photosynthetic pig- XN X ments. The R group is a long hydrocarbon, phytyl chain. lQy denotes He ¼ enjin hjn þ JnmðÞjin hjm þ jim hjn (1) the absorption maximum of the Qy transition in diethyl ether [13]. n¼1 n

ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com Original Paper

Phys. Status Solidi B 248, No. 4 (2011) 835

for N chromophores where en are transition energies to the the first two terms are stationary, describing populations, excited electronic states jni. In our discussion, en is the Qy while the last two terms, which evolve with phase factors transition of the chromophore in the n-th site of the protein, a related to the energy difference between excitons, are so-called ‘‘site energy.’’ The excitonic coupling between the quantum coherences. If energy transfer through the complex n-th and m-th chromophores in different sites of the protein were to occur on the same timescale of these superpositions, is given by Jnm. The eigenvalues of Eq. (1) are the it would behave quantum-mechanically instead of in a spectroscopically observable exciton energies. Given that classical, hopping manner. This wavelike motion of energy the Coulombic coupling term, J, is dominated by the is reversible, avoiding local minima, as the excitation distance between molecules and the relative angles of their searches for the energy sink to exit the complex. transition dipoles, it is apparent how the exact architecture of these structures gives rise to unique and complex 3 Experiments electronic energy landscapes. 3.1 Two-dimensional electronic spectroscopy Two-dimensional (2D) electronic spectroscopy has recently 2.2 Energy transfer The dynamics of excitation emerged as a powerful technique for the study of energy transfer in photosynthetic complexes have recently photodynamics in a variety of condensed phase systems been reviewed [17, 18]. Historically, energy transfer has been including molecular aggregates [26], semiconductor nanos- described by the classical Fo¨rster theory [19]. In this picture, tructures [27], and conjugated polymers [11]. Being after a donor is excited, the bath equilibrates quickly to localize sensitive to the phase evolution in Eq. (2), it is an ideal the excitonic wavefunction on the donor. The excitation is experimental tool to study energy transfer in photosynthetic subsequently transferred nonradiatively from the donor to an systems. Details of the theory [28] and experimental acceptor exciton. The fast bath equilibration destroys phase apparatus [29] have been described elsewhere. Briefly, an information leading to an irreversible, hopping motion of the ultrafast laser source is split into three successive pulses energy. Fo¨rster theory,however, is limited to a regime in which which are focused onto the sample in a box geometry (Fig. 3; the coupling between chromophores is weaker than their left). The resulting signal emerges in phase-matched respective interactions with the bath [20]. In the opposite direction collinear with a fourth, local oscillator pulse regime, where system–bath interactions are very weak, generating an interferogram as they are spectrally-resolved coherent dynamics are expected. In the limit of very weak on a CCD camera. The first interaction with the sample system–bath interactions, Redfield theory [21] or its modified creates a coherence between the ground and an excited state form [22] is expected to be accurate. However, photosynthetic that is allowed to propagate for a time t before the second light harvesting systems operate in the intermediate regime pulse arrives. This second interaction can create a population where there are no large or small parameters. This makes or electronic coherence that evolves for a waiting time (T)as perturbative approaches inappropriate and, in addition, the described by Eq. (2). The third pulse acts as the probe finite relaxation timescale of the bath cannot be neglected. resulting in the generation of the signal field. If pulse 1 These considerations led Ishizaki and Fleming [23] to develop precedes pulse 2 (þt) in the box geometry shown in Fig. 3, a formally exact approach—the reduced hierarchy method— that incorporates a finite bath relaxation time scale via a spectral density. More recently, Nalbach and Thorwart [24] have developed a path integral method that agrees quantitat- ively with the reduced hierarchy method in all parameter ranges. Both methods are computationally intensive and what remains to be done is to develop an approximate method that is quantitatively accurate for coherent dynamics in the parameter range relevant to photosynthetic light harvesting. Finally, Tao and Miller [25] described a semi-classical approach which captures the essential features of the exact results for the FMO complex. Such an approach enables the connection of quantum dynamical calculations with classical molecular dynamics simulations. All the methods predict longer lived coherence than the Redfield approach [23]. If we examine the dynamic evolution with the time propagation of the density matrix [Eq. (2)], Figure 3 (online colour at: www.pss-b.com) Left: Pulse geometry jiCðtÞ hjCðtÞ ¼ jja 2jie hje þ jjb 2jie hje and timing used in 2D electronic spectroscopy experiments. Right: 1 1 2 2 Model 2D relaxation spectrum of a dimer system exhibiting a iðv1v2Þt þ ab e jie1 hje2 (2) positive cross peak below the diagonal from population relaxation and a negative cross peak above the diagonal from excited state iðv1v2Þt þ a be jie2 hje1 absorption.

www.pss-b.com ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim b solidi

p status s s physica

836 T. R. Calhoun and G. R. Fleming: Quantum coherence in photosynthetic complexes

the signal produced in the phase-matched direction collinear with the local oscillator will be a photon echo arising from ‘‘rephasing’’ pathways. For t, when pulse 2 arrives at the sample before pulse 1, a ‘‘nonrephasing,’’ free-induction decay signal will be measured. At each T, t is scanned producing a series of interferograms, and Fourier-transform- ation along the t-axis produces the final 2D relaxation spectrum (Fig. 3; right). The 2D spectrum can be thought of as a map that correlates absorption and emission events with excitation on the x-, vt-axis, and emission on the y-, vt-axis. Signals along the diagonal correspond to features in linear absorption spectra while off-diagonal features arise from electronic coupling, population transfer/relaxation, and excited state absorption. Electronic coherences during T Figure 4 (online colour at: www.pss-b.com) Beating of the non- generate features both on and off of the diagonal that rephasing, diagonal signal as a function of T in LHCII arising from oscillate in amplitude with a frequency equal to the energy excitonic coherence at 77 K [10]. difference between the excitons participating.

3.2 FMO 2D electronic spectroscopy was first used to underlying pattern essential to the design of these natural investigate excitonic coherence in the FMO complex. systems that gives rise to this phenomenon. LHCII of higher Present in green sulfur bacteria, FMO acts as an energetic plants contains >50% of all Chl on Earth making it wire connecting the chlorosome antenna complex to the particularily well suited to answer this question. The reaction center. The structure of FMO in Fig. 1b shows seven structure of LHCII (Fig. 1c) is trimeric with each monomer BChl molecules held in each monomer of the trimer. While containing eight Chla and six Chlb molecules. It is previous 2D experiments on FMO elucidated the energy immediately obvious that this is a much more complex transfer pathways [30], a new 2D experiment was conducted system than FMO and, therefore, presents several exper- to search for the beating signal arising from quantum imental challenges. First, the broad absorption in the visible coherences. Not only was signal consistent with excitonic afforded by the two distinct types of chromophores requires coherence detected, it was observed to be surprisingly long- broadband excitation. In addition, the sheer number of lived, lasting beyond the 660 fs duration of the experiment. chromophores generates a complex energy landscape and This timescale is also longer than many of the exciton congested spectra. The same methodology applied to FMO lifetimes in FMO supporting the notion that energy transfer would result in spectra of multiple overlapping features all through the system is largely wavelike as opposed to beating with different frequencies making it impossible to hopping. While these initial experiments were conducted extract information. Previous theoretical work in our group at 77 K, new work has revealed prolonged beating to 300 fs at had shown that diagonal and antidiagonal coherence signals 277 K [31]. can be separated through analysis of the rephasing and These experimental results have spawned a wide breadth nonrephasing 2D contributions, respectively [36]. of theoretical work. A variety of theoretical approaches have Nonrephasing diagonal cuts of the 2D data for LHCII at been used to explore quantum coherence and its effects in 77 K are shown in Fig. 4 as a function of waiting time. photosynthetic complexes as recently reviewed [32]. Beating due to excitonic coherence is clearly visible for the Simulation of the dynamics of excitonic coherence in FMO 500 fs duration of the experiment in the the more intense first predicted significant quantum coherence in a photo- Chla region on the left and in the weaker Chlb signal on the synthetic complex at ambient temperatures [33]. This study right. Fourier transformation along the T-axis produced a went on to illustrate how quantum coherence in FMO coherence power spectrum in which the exciton energy promotes unidirectional energy flow to the reaction center. levels could be experimentally extracted for the first time Quantum coherence in FMO has also been shown through despite the overlapping signals [10]. theoretical calculations to increase the efficiency of energy transfer when aided by noise from the protein environment 3.4 Other photosynthetic complexes Recently, [34, 35]. excitonic coherence has been observed with 2D spec- troscopy in antennae complexes from photosynthetic marine 3.3 LHCII While the FMO studies have provided algae at 294 K [11]. The structure for one of these complexes several exciting results, its biological role is relatively small, is given in Fig. 1d. The large separation (20 A˚ ) of the bilin and the question remained as to whether the quantum chromophores makes the presence of long-lived coherence mechanical motion of energy in photosynthetic complexes especially remarkable. The fact that the chromophores are was a universal phenomenon. Given the structural differ- covalently bound to the protein is unique to these systems ences apparent in Fig. 1, the ubiquitous presence of long- and may compensate for the relatively spacious organization lived excitonic coherence would provide evidence of an [11].

ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com Original Paper

Phys. Status Solidi B 248, No. 4 (2011) 837

Experiments other than 2D spectroscopy have also been 5 Conclusions The existence of long-lived coherence used to investigate quantum coherence in photosynthetic in a wide range of photosynthetic systems from bacteria complexes. A two color photon echo technique was used to to plants suggests an inherent design principle that has isolate excitonic coherence signals between BChl and survived evolution. Current experiments and theory imply bacteriopheophytin in the reaction center of purple bacteria that this phenomenon arises not only from the chromophores at 77 and 180 K [9]. More recently, an angle-resolved themselves, but from an active protein environment. A full transient grating experiment separated excitonic coherence understanding of the role of quantum mechanical energy from population transfer in the major antennae complex in transfer in natural photosynthesis could provide insight purple bacteria (LH2) at ambient temperatures [37]. The role into fields ranging from alternative energy to quantum of excitonic coherence in LH2 had previously been discussed information. in regard to the exciton delocalization in this system [38, 39], however no techniques were available at the time to fully Acknowledgements This work was supported by the understand the relevance of quantum coherence effects to the Director, Office of Science, Office of Basic Energy Sciences, of overall dynamics. the US Department of Energy under Contract DE-AC02- 05CH11231 and by the Chemical Sciences, Geosciences, and 4 Protein environment Unlike most artificial sys- Biosciences Division, Office of Basic Energy Sciences, US tems, nature has housed its active media in a very Department of Energy Sciences, of the US Department of Energy inhomogeneous and carefully designed environment. The under contract DE-AC03-76SF000098. results of initial experiments suggested the protein played an integral role in preserving the electronic coherence [8] while References theoretical modeling of long-lived coherence in the bacterial [1] P. Peumans, V. Bulovic, and S. R. Forrest, Appl. Phys. Lett. reaction center required strong correlation between fluctu- 76(19), 2650–2652 (2000). ations arising from the protein environment [9]. Theoretical [2] S. R. Scully and M. D. McGehee, J. Appl. Phys. 100(3), 5 calculations have specifically implicated the role of the (2006). a-helices as a source of this correlation [40], and recent [3] G. D. Scholes, J. Phys. Chem. Lett. 1(1), 2–8 (2010). advancements in the theoretical framework used to treat [4] A. Camara-Artigas, D. Brune, and J. P. Allen, Proc. Natl. electronic energy transfer in photosynthetic complexes have Acad. Sci. USA 99(17), 11055–11060 (2002). allowed the role of the protein environment to be explored in [5] A. Camara-Artigas, R. E. Blankenship, and J. P. Allen, more depth [32]. Figure 5 illustrates the effects of positively Photosynth. Res. 75(1), 49–55 (2003). and negatively correlated fluctuations on the electronic states [6] J. Standfuss, A. C. T. van Scheltinga, M. Lamborghini, and W. Ku¨hlbrandt, EMBO J. 24(5), 919–928 (2005). of a dimer system. In the case of positive correlation, the [7] K. E. Wilk, S. J. Harrop, L. Jankova, D. Edler, G. Keenan, F. energy barrier between the states is reduced resulting in more Sharples, R. G. Hiller, and P. M. G. Curmi, Proc. Natl. Acad. delocalization, promoting coherence. For negative corre- Sci. USA 96(16), 8901–8906 (1999). lation, the opposite is true and coherence is prevented. As the [8] G. S. Engel, T. R. Calhoun, E. L. Read, T. K. Ahn, T. Mancˇal, determination of correlation as being positive or negative Y. C. Cheng, R. E. Blankenship, and G. R. Fleming, Nature is based on the position and orientation of the molecules 446(7137), 782–786 (2007). relative to common protein phonon modes to which they are [9] H. Lee, Y. C. Cheng, and G. R. Fleming, Science 316(5830), coupled, the importance of the chromophores’ organization 1462–1465 (2007). within the protein becomes apparent. [10] T. R. Calhoun, N. S. Ginsberg, G. S. Schlau-Cohen, Y. C. Cheng, M. Ballottari, R. Bassi, and G. R. Fleming, J. Phys. Chem. B 113(51), 16291–16295 (2009). positive-correlation negative-correlation [11] E. Collini, C. Y. Wong, K. E. Wilk, P. M. G. Curmi, P. Brumer, 8 8 and G. D. Scholes, Nature 463(7281), 644–647 (2010). [12] R. E. Blankenship, Molecular Mechanisms of Photosynthesis (Blackwell Science, Oxford, 2002). 4 4 [13] M. Kobayashi, M. Akiyama, H. Kano, and H. Kise, in: q 2 and , edited by B. Grimm, R. J. Porra, W. Ru¨diger and H. Sheer (Springer, New York, 0 0 2006), pp. 79–94. [14] H. van Amerongen, L. Valkunas, and R. van Grondelle, -4 -4 Photosynthetic Excitons (World Scientific, Singapore, 2000). -4 048 -4 048 q q [15] J. Adolphs, F. Mu¨h, M. Madjet, and T. Renger, Photosynth. 1 1 Res. 95(2), 197–209 (2008). Figure 5 Schematic of the adiabatic potential energy surfaces for [16] D. Noy, C. C. Moser, and P. L. Dutton, Biochim. Biophys. the electronic states of a dimer in the presence of two phonon Acta 1757(2), 90–105 (2006). coordinates (q1 and q2) fluctuating with positive and negative [17] T. Renger, V. May, and O. Ku¨hn, Phys. Rep. 343, 137–254 correlation [32]. (2001).

www.pss-b.com ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim b solidi

p status s s physica

838 T. R. Calhoun and G. R. Fleming: Quantum coherence in photosynthetic complexes

[18] Y. C. Cheng and G. R. Fleming, Annu. Rev. Phys. Chem. 60, [31] G. Panitchayangkoon, D. Hayes, K. A. Fransted, J. R. Caram, 241–262 (2009). E. Harel, J. Wen, R. E. Blankenship, and G. S. Engel, Proc. [19] T. Fo¨rster, Ann. Phys. 437, 55–75 (1948). Natl. Acad. Sci. USA 107(29), 12766–12770 (2010). [20] G. D. Scholes, Annu. Rev. Phys. Chem. 54(1), 57–87 (2003). [32] A. Ishizaki, T. R. Calhoun, G. S. Schlau-Cohen, and G. R. [21] A. G. Redfield, IBM J. Res. Dev. 1(1), 19–31 (1957). Fleming, Phys. Chem. Chem. Phys. 12(27), 7319–7337 (2010). [22] M. Yang and G. R. Fleming, Chem. Phys. 275, 355–372 [33] A. Ishizaki and G. R. Fleming, Proc. Natl. Acad. Sci. USA (2002). 106(41), 17255–17260 (2009). [23] A. Ishizaki and G. R. Fleming, J. Chem. Phys. 130, 234111 [34] M. Mohseni, P. Rebentrost, S. Lloyd, and A. Aspuru-Guzik, (2009). J. Chem. Phys. 129(17), 174106 (2008). [24] P. Nalbach and M. Thorwart, Personal Communication. [35] A. W. Chin, A. Datta, F. Caruso, S. F. Huelga, and M. B. [25] G. Tao and W. H. Miller, J. Chem. Phys. Lett. 1(6), 891–894 Plenio, New J. Phys. 12(6), 065002 (2010). (2010). [36] Y. C. Cheng and G. R. Fleming, J. Phys. Chem. A 112(18), [26] N. S. Ginsberg, Y. C. Cheng, and G. R. Fleming, Acc. Chem. 4254–4260 (2008). Res. 42(9), 1352–1363 (2009). [37] I. P. Mercer, Y. C. El-Taha, N. Kajumba, J. P. Marangos, [27] T. Zhang, I. Kuznetsova, T. Meier, X. Li, R. P. Mirin, P. J. W. G. Tisch, M. Gabrielsen, R. J. Cogdell, E. Springate, Thomas, and S. T. Cundiff, Proc. Natl. Acad. Sci. USA and E. Turcu, Phys. Rev. Lett. 102(5), 057402 (2009). 104(36), 14227–14232 (2007). [38] T. Meier, Y. Zhao, V. Chernyak, and S. Mukamel, J. Chem. [28] D. M. Jonas, Annu. Rev. Phys. Chem. 54, 425–463 (2003). Phys. 10(8), 3876–3893 (1997). [29] T. Brixner, T. Mancˇal, I. V. Stiopkin, and G. R. Fleming, J. [39] O. Ku¨hn and V. Sundstro¨m, J. Chem. Phys. 107(11), 4154– Chem. Phys. 121(9), 4221–4236 (2004). 4164 (1997). [30] T. Brixner, J. Stenger, H. M. Vaswani, M. Cho, R. E. [40] F. Mu¨h, M. E. A. Madjet, J. Adolphs, A. Abdurahman, B. Blankenship, and G. R. Fleming, Nature 434(7033), 625– Rabenstein, H. Ishikita, E. W. Knapp, and T. Renger, Proc. 628 (2005). Natl. Acad. Sci. USA 104(43), 16862–16867 (2007).

ß 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.pss-b.com