Structural Basis of Cyanobacterial Photosystem II Inhibition by the

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 286, NO. 18, pp. 15964 –15972, May 6, 2011 © 2011 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

Structural Basis of Cyanobacterial Photosystem II Inhibition by the Herbicide Terbutryn* Received for publication, December 23, 2010, and in revised form, February 17, 2011 Published, JBC Papers in Press, March 2, 2011, DOI 10.1074/jbc.M110.215970

Matthias Broser‡1, Carina Glo¨ckner‡1, Azat Gabdulkhakov§2, Albert Guskov§3, Joachim Buchta¶, Jan Kern‡4, Frank Mu¨h‡, Holger Dau¶, Wolfram Saenger§5, and Athina Zouni‡6 From the ‡Institut fu¨r Chemie/Max-Volmer-Laboratorium fu¨r Biophysikalische Chemie, Technische Universita¨t Berlin, Strasse des 17. Juni 135, D-10623 Berlin, Germany, the §Institut fu¨r Chemie und Biochemie/Kristallographie, Freie Universita¨t Berlin, Takustrasse 6, D-14195 Berlin, Germany, and the ¶Fachbereich Physik, Freie Universita¨t Berlin, Arnimallee 14, D-14195 Berlin, Germany Herbicides that target photosystem II (PSII) compete with the native electron acceptor plastoquinone for binding at the QB site in the D1 subunit and thus block the electron transfer from QA to QB. Here, we present the first crystal structure of PSII with a ˚ . The crystallized PSII bound herbicide at a resolution of 3.2 A core complexes were isolated from the thermophilic cyanobacterium Thermosynechococcus elongatus. The used herbicide terbutryn is found to bind via at least two hydrogen bonds to the QB site similar to photosynthetic reaction centers in anoxygenic purple bacteria. Herbicide binding to PSII is also discussed regarding the influence on the redox potential of QA, which is known to affect photoinhibition. We further identified a second and novel chloride position close to the water-oxidizing complex and in the vicinity of the chloride ion reported earlier (Guskov, A., Kern, J., Gabdulkhakov, A., Broser, M., Zouni, A., and Saenger, W. (2009) Nat. Struct. Mol. Biol. 16, 334 –342). This discovery is discussed in the context of proton transfer to the lumen.

The process of photosynthesis converts solar energy into biochemically amenable energy. A distinction is made between oxygenic and anoxygenic photosynthesis. In the latter sulfur compounds, hydrogen gas or organic materials serve as electron source. In contrast, oxygenic photosynthesis in higher plants, algae, and cyanobacteria uses water as an electron source and generates molecular oxygen, thereby maintaining

* This work was supported by the Deutsche Forschungsgemeinschaft within the framework of Sfb 498 (projects A4, C6, and C7) and the center of excellence on Unifying Concepts in Catalysis (UniCat, project B1), coordinated by the Technische Universita¨t Berlin. The crystallographic part of this work was supported in part by the Program of the Russian Academy of Science on Molecular and Cellular Biology. The atomic coordinates and structure factors (codes 3PRQ and 3PRR) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/). 1 Both authors contributed equally to this work. 2 Present address: Institute of Protein Research, Russian Academy of Sciences, 142290 Pushchino, Moscow Region, Russia. 3 Present address: Biomedical Structural Biology, Nanyang Technological University, Proteos, 61 Biopolis Dr., 138673 Singapore. 4 Present address: Physical Biosciences Division, Lawrence Berkeley National Laboratory, One Cyclotron Rd., Berkeley, CA 94720. 5 To whom correspondence may be addressed. Tel.: 49-30-838-53412; Fax: 49-30-838-56702; E-mail: [email protected]. 6 To whom correspondence may be addressed. Tel.: 49-30-314-25650; Fax: 49-30-314-21122; E-mail: [email protected].

15964 JOURNAL OF BIOLOGICAL CHEMISTRY

the level of oxygen in the atmosphere. Water oxidation takes place at the large homodimeric protein-cofactor complex photosystem II (PSII),7 a light-driven water:plastoquinone oxidoreductase harbored in the thylakoid membrane (1–3). The structure of the PSII core complex (PSIIcc) from the thermophilic cyanobacterium Thermosynechococcus elongatus is known from x-ray crystallographic studies at a current resolution of 2.9 Å (4, 5). The photochemical reaction center (RC) in PSII is of type II and structurally related to the RC of purple bacteria (pbRC) (6), which perform anoxygenic photosynthesis. The PSII-RC contains four chlorophyll a (Chla) molecules (PD1, PD2, ChlD1, and ChlD2), two pheophytins (PheoD1 and PheoD2), and two plastoquinones (PQ) (QA and QB) with a nonheme iron in between. These cofactors are embedded in a heterodimeric protein matrix formed by subunits D1 and D2 (systematic names: PsbA and PsbD, respectively) and are arranged in two pseudo-C2 symmetric branches with respect to a 2-fold rotation axis, which crosses the non-heme iron and is oriented normal to the membrane plane. Photons from sunlight are collected by antenna proteins of PSII, and the excitation energy is transferred to the RC, where it gives rise to the formation of the radical pair PD1䡠⫹PheoD1䡠⫺ and subsequent electron transfer to the fixed single-electron transmitter QA. The electron hole at PD1䡠⫹ is able to abstract electrons via the redox-active tyrosine YZ (Tyr-161A) from the heteronuclear Mn4Ca cluster located at the lumenal (donor) side of PSII. Coupled to a series of oxidation states (S-states) of this metal cluster, water is oxidized to molecular oxygen. The primary quinone QA is located close to the cytoplasmic (acceptor) side in subunit D2 and reduces in turn the terminal electron acceptor QB at the symmetry-related site in subunit D1. In the next turnover, the semiquinone QB⫺ is reduced again and leaves, after double protonation, its binding site in form of a plastoquinol (PQH2). PQH2 transfers the electrons to the cytochrome b6f complex and is thereby reoxidized to PQ. The reduction equivalents are further transferred to photosystem I. The connecting membrane region between PSII and the cytochrome b6f 7

The abbreviations and trivial names used are: PSII, photosystem II; PSIIcc, photosystem II core complex; ␤-DM, n-dodecyl-␤-D-maltoside; Chl, chlorophyll; DCMU, 3-(3,4-dichlorophenyl)-1,1-dimethylurea; RC, reaction center; pbRC, purple bacteria reaction center; Pheo, pheophytin; PQ, plastoquinone(s); PQH2, quinole; PF, prompt fluorescence; PDB, Protein Data Bank; ioxynil, 4-hydroxy-3,5-diiodobenzonitrile; terbutryn, 2-tert-butylamino-4ethylamino-6-methylthio-1,3,5-triazine.

VOLUME 286 • NUMBER 18 • MAY 6, 2011

Crystal Structure of Terbutryn-bound Photosystem II complex contains a plastoquinone pool and provides PQ molecules for the exchange with PQH2 (2, 7). A third PQ molecule, QC, is located close to the QB site, but its role is not yet fully understood (4). The reaction pattern of the light-induced charge separation in PSII partially resembles that of pbRC (8) due to their structural similarity, which was already proposed in the 1980s by Michel et al. (9). The crystal structure of the pbRC of Blastochloris viridis was first reported by Deisenhofer et al. (10). The functional core of pbRC is a heterodimer of the L and M subunits that bind all active cofactors: four bacteriochlorophylls, two bacteriopheophytins, two quinones, and a non-heme iron center. Mainly based on sequence homologies, subunit D1 of PSIIcc was proposed to correspond to subunit L in pbRC, and subunit D2 was proposed to correspond to subunit M (11). In 2001, Zouni et al. (12) could verify these assumptions with the first three-dimensional crystal structure of cyanobacterial PSIIcc. It was shown that the redox active cofactors of both RCs (PSIIcc and pbRC) are coordinated in a similar manner (12). Despite high similarities, differences between both systems regarding the electron acceptor side were observed. In pbRC, several types of quinones are used as electron acceptors, and also QA and QB are not chemically identical in many species, whereas the formation of the respective quinol (QBH2) proceeds in an analog way to PQH2 formation in PSII (13–15). A common feature of the QB site in both PSII and pbRC is the binding of inhibitors with high affinity. Because these inhibitors compete with the native quinone molecules, they are able to block the electron transfer from QA to QB (16, 17). A very large number of compounds inhibit PSII electron transport, but only very few substances bind to pbRC. Some of the PSII inhibitors are commercially used as herbicides (for a review, see Ref. 18). These herbicides belong to different chemical classes (e.g. urea, triazine, and phenol derivatives (e.g. 3-(3,4-dichlorophenyl)1,1-dimethylurea (DCMU), 2-tert-butylamino-4-ethylamino6-methylthio-1,3,5-triazine (terbutryn), and 4-hydroxy-3,5-diiodobenzonitrile (ioxynil), respectively)). Triazines are also able to inhibit the electron transfer in pbRC. Herbicide binding to PSII has been studied for several decades using a variety of methods. Extensive mutational studies at the acceptor side of PSII from various species were applied to identify amino acids participating in herbicide binding (for a review, see Ref. 16). These studies indicate, inter alia, a role of Ser-264A in herbicide binding, especially for triazines. Naturally occurring herbicide resistance in higher plants and cyanobacteria, which was first found in atrazin-resistant Amaranthus hybridus (19), could also be ascribed to the amino acid substitution of Ser-264A (20). Differential scanning and isothermal titration calorimetry, circular dichroism spectroscopy, and oxygen evolution measurements were used to investigate binding of herbicides to isolated PSIIcc from T. elongatus (21). In this study, differences in the binding parameters for the herbicide classes were observed. Phenolic herbicides were found to bind endothermically and destabilize PSIIcc, whereas triazines and urea-type herbicides (e.g. terbutryn and DCMU, respectively) showed a negative binding enthalpy and tend to stabilize the complex. In other studies, an influence of herbicides on the redox properties of QA was detected (22, 23). Depending on the MAY 6, 2011 • VOLUME 286 • NUMBER 18

type of the applied herbicide, the redox potential of QA is shifted to lower (phenolic herbicides) or higher (urea-type herbicide) potentials. Study of herbicide binding to pbRC benefited from the elucidation of structural information by x-ray crystallography. The first structural information on pbRC with bound herbicides was attained at 2.9 Å resolution by Michel et al. (9). More accurate descriptions of herbicide interaction at the QB site of pbRC, especially for triazines (atrazine and terbutryn), became possible at higher resolutions. These structures showed a similar hydrogen bonding pattern for atrazine (2.35 Å resolution, PDB code 5PRC (24)) and terbutryn (2.00 Å resolution, PDB code 1DXR (25)) in the QB pocket, which is discussed in more detail under “Results.” Lacking direct structural information on herbicide binding to PSII, the pbRC-herbicide structures served so far as models for PSII inhibition (14) and were used as starting points for theoretical studies (26). Besides experimentally verifying the theoretical models of herbicide binding to PSII, an actual crystal structure of a PSIIccherbicide complex is essential to provide a structural basis for understanding the observed redox potential shift of QA as a consequence of herbicide treatment. In this study, we present the first x-ray crystal structure of a complex of PSIIcc and the triazine derivative terbutryn with a resolution of 3.2 Å. These structural data also revealed a novel and significantly shifted position of the chloride ion associated with the Mn4Ca cluster. This observation will be discussed in the context of the proposed role of chloride in proton transfer to the lumen.

EXPERIMENTAL PROCEDURES Sample Preparation—Oxygen-evolving PSIIcc was isolated from thylakoid membranes of the cyanobacterium T. elongatus, purified, and separated into monomeric and dimeric PSIIcc by anion exchange chromatography as described previously (27). Samples were stored in liquid nitrogen until use. Single Flash-induced Chla Fluorescence Measurements—For fluorescence measurements, samples of PSIIcc were diluted to 10 ␮M Chla in buffer A containing 25 mM MES (pH 6.3), 1 M betaine, 15 mM NaCl, 5 mM MgCl2, 5 mM CaCl2, and 0.03% (w/w) n-dodecyl-␤-D-maltoside (␤-DM). The influence of the herbicide terbutryn on the electron transfer to QB was studied by applying a concentration series of the herbicide. Terbutryn dissolved in DMSO was added to buffer A containing 0.143 ␮M dimeric PSIIcc (10 ␮M Chla). The final herbicide concentrations ranged from 10⫺12 to 10⫺4 M. Controls contained no herbicide but 2% DMSO. The DMSO concentrations used in all experiments had no influence on the PSIIcc oxygen evolution activity (28). Single flash-induced Chla fluorescence was measured using a laboratory-built apparatus as described (29 –31). Saturating flash excitation of samples was provided by a frequency-doubled Q-switched Nd:YAG laser (Continuum Minilite II, wavelength ⫽ 532 nm, full width at half-maximum ⫽ 5 ns). A series of 32 flashes with a flash interval of 0.7 s and laser pulse intensities of 13 mJ cm⫺2 was applied. The first data point was recorded 76 ␮s after the excitation flash. The amplitude of the fast decay phase of the prompt fluorescence (PF; as plotted in Fig. 1) was defined as the difference between the maximum JOURNAL OF BIOLOGICAL CHEMISTRY

15965

Crystal Structure of Terbutryn-bound Photosystem II fluorescence yield reached after flash excitation (FM) and the amplitude after 7.2 ms (F7.2), according to the following equation, PF ⫽

FM ⫺ F7.2 FM ⫺ F700

(Eq. 1)

where F700 represents the fluorescence level reached about 700 ms after flash excitation. In control samples, about 80% of the total QA reoxidation occurs within this time range. The obtained value for PF was plotted over the herbicide concentration and fitted by a standard binding function. Crystallization—For the co-crystallization of dimeric PSIIcc and terbutryn, PSIIcc (4 mM Chla in buffer C containing 100 mM PIPES-NaOH, pH 7.0, 5 mM CaCl2, 0.03% (w/w) ␤-DM (27)) were illuminated for 10 s using a two-armed Halolux 250 cold light source (Streppel, Wermelskirchen-Tente, Germany), followed by the addition of 2 mM terbutryn dissolved in DMSO (yielding 2% DMSO in the sample) and slight vortexing. Samples were dark-adapted for a time period of 120 min. For batch crystallization, PSIIcc-terbutryn samples were mixed with the same volume of precipitant solution, composed of 6.4 – 8.2% PEG 2000, 100 mM PIPES-NaOH (pH 7), and 5 mM CaCl2. The final terbutryn concentration in each crystal setup was 1 mM. After careful vortexing, a volume of about 5 ␮l of the obtained solution was centered for crystallization in either a glass capillary (75-mm length, 1.15-mm inner diameter) or a polytetrafluoroethylene tube (about 6-mm length, 1-mm inner diameter) and kept in the dark for 3–5 days at 20 °C. Data Collection and Processing—X-ray diffraction data from several PSIIcc-terbutryn crystals were collected using monochromatic synchrotron radiation at a wavelength of 0.93 Å at beamline ID 14-4 of ESRF (Grenoble, France). The data with the highest resolution were processed with XDS (32) to 3.2 Å resolution (see Table 1). The structure was solved by molecular replacement with the program PHASER (33) using the 2.9 Å resolution structure of PSII as a search model (PDB code 3BZ1 (4)). Model rebuilding and refinement were performed with COOT (34) and the CNS 1.2 package (35), respectively, employing non-crystallographic symmetry. The rigid body procedure was used for refining the protein part of the complex, whereas the annealing procedure was applied for refining the cofactors and side chains of the protein subunits. The final model shows R/Rfree factors of 0.269/0.299, with root mean square deviations from an ideal geometry of 0.013 Å for bond lengths and 2.2° for bond angles. Terbutryn molecules were placed in the appropriate difference electron density in Fo ⫺ Fc maps.

RESULTS Single Flash-induced Chla Fluorescence Measurements in the Presence of Terbutryn—The maximal concentration of terbutryn useable for co-crystallization with PSIIcc was limited to 1 mM due to the low solubility of the herbicide in aqueous solution. In order to ensure complete saturation of the QB binding sites with terbutryn, the minimal terbutryn concentration nec⫺ essary for full inhibition of the electron transfer from QA to QB had to be determined. Inhibition of oxygen evolution in isolated


FIGURE 1. Terbutryn titration by flash-induced fluorescence measurements. The fast phase amplitude of the PF (as defined under “Experimental Procedures”) is plotted over the terbutryn concentration (open circles). The solid line represents the fit of the data by a single-site binding function, which shows half-inhibition at a terbutryn concentration of ⬃2 ⫻ 10⫺7 M. The inset illustrates the disappearance of the fast decay phase at high terbutryn concentrations.

PSIIcc by herbicides was already investigated in former studies (21), using a Clark-type electrode (36). However, based on the occurrence of high residual activity, Zimmermann et al. (21) suggested the extent of inhibition to be maximally about 50% under these conditions (Imax ⫽ 45 ⫾ 2%). This residual activity was already observed in other studies, but its origin remains unclear. It has been proposed that damage to the QB binding site due to detergent treatment results in a reduced possibility of herbicide binding (37). However, this argument disagrees with the similar dissociation constants found for herbicide binding to detergent-solubilized PSIIcc and to thylakoid membrane fragments from spinach (38). Moreover, it was suggested that the artificial electron acceptor 2,6-dichloro-1,4-benzoquinone (used in Ref. 21) is able to capture electrons not only at the QB binding pocket but also directly from QA, thereby maintaining oxygen evolution (39). In order to exclude artifacts introduced by the presence of 2,6-dichloro-1,4-benzoquinone, the variable Chla fluorescence of PSIIcc, after excitation with a single laser flash, was recorded in the presence of terbutryn without an artificial electron ⫺ acceptor. In this approach, the reoxidation of QA is tracked via the decay of the Chla fluorescence yield, excited by weak probe flashes (29). This fluorescence decay is composed of at least three kinetic components (see Ref. 29 and references therein). The two fast processes on the microseconds to milliseconds ⫺ time scale are ascribed to the electron transfer between QA and ⫺ QB/QB and are referred to as fast phase (see “Experimental Procedures”). The addition of terbutryn leads to an inhibition ⫺ of the oxidation of QA by electron transfer to QB, because of its competitive binding to the QB site. Therefore, the time decay of the Chla fluorescence is drastically slowed down with increasing terbutryn concentration. Fig. 1 shows the change of the amplitude of the fast phase as a function of the terbutryn concentration. At a terbutryn concentration of 2 ⫻ 10⫺7 M, halfinhibition of the electron transfer to QB is achieved in accordVOLUME 286 • NUMBER 18 • MAY 6, 2011

Crystal Structure of Terbutryn-bound Photosystem II TABLE 1 Data collection and refinement statistics Parameter

a

Value

Data collection for PSIIcc-terbutryn co-crystals Space group Unit cell parameters a (Å) b (Å) c (Å) ␣ ⫽ ␤ ⫽ ␥ (degrees) Resolution (Å) Measured reflections Unique reflections 具I/␴(I)典 Redundancy Rsym Completeness (%)

126.9 225.0 304.9 90 50-3.2 (3.3-3.2)a 711,691 137,557 7.76 (2.34)a 4.93 (5.07) 0.129 (0.588)a 94.1 (90.9)a

Refinement Resolution (Å) No. of reflections Rwork/Rfree

20-3.2 (3.83-3.2)a 136,912 26.9/29.9 (46.2/48.6)

Root mean square deviations Bond lengths (Å) Bond angles (degrees)

0.013 2.2

P212121

Data in the highest resolution shell.

ance with measurements of oxygen evolution using a Clarktype electrode (21). Above 10 ␮M terbutryn, the fast fluorescence decay drops down to almost zero, indicating a ⫺ completely blocked electron transfer from QA to QB. Taking into account the 2 mM Chla concentration (equal to 28.6 ␮M dimeric PSIIcc), as it is used for the crystallization setups, we conclude that a terbutryn concentration of 1 mM is sufficient to ensure full saturation of herbicide binding to the QB site. Crystallographic Analysis—Crystals grown in the presence of the terbutryn feature the same space group and cell parameters as crystals of native PSIIcc dimer, and it was possible to collect and process a data set to a resolution of 3.2 Å (Table 1). The structure was solved by molecular replacement using the 2.9 Å resolution structure (PDB code 3BZ1 (4)) as search model. The Fo ⫺ Fc map revealed one terbutryn molecule in the QB binding site in each monomer. The positions of the terbutryn molecules were refined with full occupancy. The obtained final model contains the following per monomer: 20 protein subunits, 35 Chla, two pheophytins, 12 ␤-carotenes, two heme groups, the non-heme iron, plastoquinone QA, 25 lipids, and the Mn4Ca-cluster. All of these proteins and cofactors are virtually identical to the 2.9 Å crystal structure of native PSIIcc (4). In addition to the model of native PSIIcc, four DMSO molecules could be assigned in the obtained electron density. Two are located within a putative oxygen channel, as described previously (28), and two DMSO molecules reside at the periphery. One ␤-DM molecule was also found at the periphery of the complex, in addition to the seven ␤-DM positions in the native structure. Because no electron density was observed at the position of QC, it was not possible to model either PQ or terbutryn at this site. Close to the Mn4Ca cluster, two positions for chloride ions were found with different occupancies (see below). In PSIIcc, the QB binding pocket in subunit D1 is formed by the C-terminal part of transmembrane ␣-helix (TMH) d, the cytosolic surface helix de, the following loop region and the N-terminal part of TMH e. The headgroup of the mobile cofactor QB in native PSIIcc is placed between the residues Phe-255A MAY 6, 2011 • VOLUME 286 • NUMBER 18

FIGURE 2. Plastoquinone and terbutryn binding in PSIIcc. A, structural formula of terbutryn. Nitrogen atoms are numbered according to PDB entry 3PRQ (also used in Ref. 25, differing from IUPAC rules). B, superposition of the structural models with plastoquinone or terbutryn bound to the QB site. PSIIcc subunit D1 (PsbA) is shown as a yellow ribbon with selected amino acids participating in hydrogen bonding, plastoquinone (colored magenta; coordinates taken from PDB entry 3BZI (4)), and terbutryn (colored green) drawn as sticks. The blue sphere indicates the non-heme iron.

and Leu-271A. Hydrogen bonds between the keto-oxygens of the QB head group and Ser-264A (hydrogen of the ␥-oxygen), His-215A (hydrogen of the ␦-nitrogen), and Phe-265A (amide hydrogen) could possibly be formed (Fig. 2). The distance of the QB headgroup to the non-heme iron is 7.5 Å, bridged by His215A. Eight lipids are arranged in a bilayer structure next to the QB site, most likely providing a hydrophobic environment for fast PQ-PQH2 exchange (4, 40). Terbutryn Binding to the QB Site of PSIIcc—The binding mode of the terbutryn molecule in the QB pocket at 3.2 Å resolution revealed that terbutryn is bound via four potential hydrogen bonds. These bonds could be formed by the three nitrogens of terbutryn at the side pointing away from the nonheme iron (Fig. 3A). N5 of the terbutryn ring could form a hydrogen bond with the backbone amide of Phe-265A (hydrogen bond distance 3.1 Å), the ethylamino nitrogen with the ␥-oxygen of Ser-264A (hydrogen bond distance 3.3 Å), and with the backbone carbonyl oxygen of Phe-265A (hydrogen bond distance 3.2 Å). A fourth hydrogen bond between the t-butylamino nitrogen and the carbonyl oxygen of residue Ala-263A with a hydrogen bond distance of 4.3 Å seems to be possible. The amino acid residues Phe-265A and Ser-264A are involved in both PQ and terbutryn binding (Fig. 2). Note that the maximum coordinate error of the model is 0.3 Å (as calculated by SFCHECK). Besides the above described hydrogen bonds, terbutryn binding is stabilized through non-polar interactions. The aromatic ring of Phe-255A is found in close contact to the triazine ring of terbutryn, but no reliable information about the orientation of its ring plane could be obtained due to the limited resolution. Other non-polar interactions are provided by the residues Met-214A and Leu-271A, which are in van der Waals JOURNAL OF BIOLOGICAL CHEMISTRY

15967

Crystal Structure of Terbutryn-bound Photosystem II

FIGURE 3. Terbutryn bound to the QB site of either PSIIcc from T. elongatus (A and B; colored gray) or the pbRC from B. viridis (C and D; colored pink) as derived from the crystal structures at 3.2 Å (PDB codes 3PRQ and 3PRR) and 2.0 Å resolution (PDB code 1DXR (25)), respectively. For each RC, two orientations are shown, viewing either onto (A and C) or along (B and D) the ring plane of terbutryn. Terbutryn and the surrounding amino acid residues of subunit D1 (PsbA of PSIIcc) or L (pbRC) are indicated, and possible hydrogen bonds between the herbicide and the protein are shown as red dashed lines (distances are given in Å). For PSIIcc (A and B), the electron density of terbutryn is shown as green mesh at a contour level of 1.2 ␴. The non-heme iron cofactor and two water molecules (W1 and W2) are depicted as blue and red spheres, respectively.

distance with the methylthio group of terbutryn (Fig. 3, A and B). Crystal structures of terbutryn bound to pbRC have been published at resolutions of 2.5 and 2.0 Å for Rhodobacter sphaeroides and B. viridis, respectively (25, 41). In all structures, terbutryn is bound in the QB pocket via three hydrogen bonds to subunit L (sequence homologous with D1 of PSII). Similar to PSIIcc, the hydrogen bonds are formed by the three nitrogen atoms: N5 of the terbutryn ring, the t-butylamino nitrogen, and the ethylamino nitrogen (Fig. 3C). The involved amino acid residues are Ile-224L (backbone amide, hydrogen bond distance 3.1 Å), Ser-223L (␥-oxygen, hydrogen bond distance 3.0 Å), and Tyr-222L (backbone carbonyl oxygen, hydrogen bond distance 3.3 Å). Between terbutryn and the non-heme iron, two tightly bound water molecules are located. One (W1; see Fig. 3C) forms an additional hydrogen bond with the nitrogen atom N1 of the terbutryn ring and with His-190L (␦-nitrogen). The other (W2; see Fig. 3C) is in a bridging position between W1 and the ⑀-oxygen of Glu-212L. The aromatic ring of the Phe216L is oriented almost parallel (14°) to the ring plane of terbutryn (Fig. 3D), thereby additionally stabilizing the complex through ␲-␲ interaction (25). Despite the similar arrangement of the QB binding site in both RCs, distinct differences in the hydrogen bonding pattern are found. The hydrogen bonds to the side chain of the conserved serines (Ser-264A and Ser-223L) and to the following


backbone amide are present in both PSIIcc and pbRC. In contrast, the length of the hydrogen bond of the t-butylamino nitrogen to the respective backbone carbonyl is significantly enlarged in PSIIcc. Moreover, a second hydrogen bond partner for the ethylamino nitrogen is provided by the backbone carbonyl of Phe-265A, which cannot be formed in pbRC. Comparing the hydrophobic environment of terbutryn in both RCs revealed that the residues Met-214A and Leu-271A in PSIIcc replace the pbRC residues Leu-189L and Ile-229L. These nonpolar groups are of similar size, indicating a comparable steric surrounding. Chloride Positions at the Donor Side—The metal ions of the Mn4Ca cluster in the PSIIcc-terbutryn complex structure produce a pattern of high electron density of extended shape as observed earlier for native PSIIcc at 2.9 Å resolution (4). At the position of the chloride ion found in Ref. 4, which is associated with the Mn4Ca cluster, a patch of electron density is visible that was assigned to a chloride ion with an occupancy of only ⬃30% (Cl1A; see Fig. 4). Interestingly, an additional electron density is present in the vicinity (⬃6.7 Å) of this chloride position and separated by the side chain of Lys-317D. This pronounced patch of electron density, which is absent in the electron density map at 2.9 Å resolution, is assigned to a chloride ion with an occupancy of ⬃70% (Cl1B; see Fig. 4). Other small molecules or ions present in the crystallization buffer were tested as well, but only the assignment to chloride gave reliable VOLUME 286 • NUMBER 18 • MAY 6, 2011

Crystal Structure of Terbutryn-bound Photosystem II

FIGURE 4. Chloride positions found according to the structural model at 3.2 Å resolution derived from PSIIcc-terbutryn co-crystals. A, chloride ions are drawn as green spheres, and positions are labeled either Cl1A (occupancy of ⬃30%, corresponding to the chloride position in native PSIIcc) or Cl1B (occupancy of ⬃70%). Distances between the chloride positions (green dashed line) and to Mn4 (pink dashed lines; manganese shown as red spheres) are given in Å. The dashed red lines indicate possible interaction of the anions with the surrounding amino acids (distances are given under “Results”). B, arrangement of metal and chloride ions as well as selected amino acids according to the structural model at 3.2 Å resolution, relative to channels C and G (28) calculated on the basis of the 2.9 Å resolution model. The calculated trajectories of the channels are drawn as meshes (channel C colored blue, channel G colored green).

results. At this position, the chloride ion is surrounded by the positively charged side chains of Lys-317D (3.5-Å distance to the ⑀-amino group), Arg-334A (3.6-Å distance to the guanidinium group), and the side chain amide group of Asn-335A (3.9 Å). The carboxylate groups of Asp-61A and Glu-65A are also found in close vicinity to the anion (3.2 and 2.8 Å, respectively). Within the error limit of our structural data, it is not possible to define exactly the orientations of all amino acid side chains. Thus, it is possible that the amide group of Lys-317D occurs in different orientations, depending on the respective chloride positions. Taking these uncertainties into account, we found no indication for an altered protein environment at the second putative chloride position (Cl1B). The minimum distance from Cl1B to one of the metal ions of the Mn4Ca cluster was found for Mn4 with ⬃8.7 Å (Fig. 4A).

DISCUSSION This work presents the first direct structural information about terbutryn binding to cyanobacterial PSIIcc. To achieve full occupancy of terbutryn in the QB site, it was necessary to MAY 6, 2011 • VOLUME 286 • NUMBER 18

illuminate PSIIcc prior to incubation with herbicide and crystallization. Without preillumination, binding of terbutryn to the QB pocket occurs, as is evident from the fluorescence measurements (Fig. 1). Nonetheless, we observed that the occupancy in the QB site is not sufficient for a reliable crystallographic analysis unless an illumination period precedes the terbutryn binding (data not shown). The reduction of PQ to PQH2, which has a lower affinity to the QB pocket, may be required to prevent a slow exchange of bound terbutryn against oxidized quinone during crystallization and thus allows a full binding of the herbicide. An implication of this result is that PQ competes with terbutryn for binding to the QB site or at least disturbs terbutryn binding during crystal growth. Earlier structural studies of PSIIcc inhibition by herbicides had to resort to homology modeling based on a comparison with the pbRC (9). In particular, theoretical models employing energy minimization techniques predicted the interaction between triazine derivatives and the QB site in PSIIcc in considerable detail (26, 42– 44). Several groups proposed that Ser264A and Phe-265A, which are homologous to Ser-223L and Ile-224L of pbRC, form hydrogen bonds to the ethylamino nitrogen and the aromatic ring nitrogen, respectively (13, 44, 45). These results are essentially confirmed by the present structural data. An additional hydrogen bond between Ala263A and the t-butylamino nitrogen was considered by Egner et al. (44) by analogy with Tyr-222L in pbRC but was abandoned after energy minimization. Indeed, our crystal structure shows a rather long distance between the backbone carbonyl oxygen of Ala-263A and the t-butylamino nitrogen (Fig. 3A), suggesting no binding or weak binding at this site. A new result from the crystal structure is that the ethylamino nitrogen of terbutryn could be engaged in hydrogen bonding with the backbone carbonyl oxygen of Phe-265A, albeit at suboptimal geometry. This additional binding interaction could compensate for the weaker binding at the t-butylamino nitrogen. The changed backbone conformation of Phe-265A with respect to its counterpart Ile-224L in pbRC could result from different steric constraints introduced by a variation of the amino acid sequence. In PSIIcc, Phe-265A is followed by Asn-266A, whose side chain provides a hydrogen bond to the phospholipid PG22 (phosphatidylglycerol 22), whereas in pbRC, the corresponding Ile224L is followed by Gly-225L (not shown). In PSIIcc, the t-butylamino moiety of terbutryn is surrounded by phenylalanine residues (Phe-211A, Phe-255A, Phe-265A, and Phe-274A, all within a distance of less than 4.2 Å; see Fig. 3, A and B). Van der Waals interactions with these groups could contribute to the stabilization of inhibitor binding. This idea is supported by mutagenesis studies of both Synechococcus sp. PCC 7002 and Synechocystis PCC 6714, where the replacement of Phe-211A with Ser results in an increased resistance to atrazine (16). In pbRC, two water molecules participate in the hydrogen bonding network of terbutryn (Fig. 3C). At the present resolution, it is not possible to identify water molecules in PSIIcc. However, there is sufficient space between terbutryn and the side chains of His-215A and Tyr-246A as well as the backbone of Ala-251A to accommodate water molecules in an analogous way to stabilize the herbicide in the binding pocket. Because Ala-251A is found at a homologous position to Glu-212L, TyrJOURNAL OF BIOLOGICAL CHEMISTRY

15969

Crystal Structure of Terbutryn-bound Photosystem II 246A is the only amino acid that could form a possible hydrogen bond to a putative W2 position in PSIIcc. According to the Fourier transform infrared (FTIR) studies of Takahashi et al. (46), Tyr-246A is hydrogen-bonded to bicarbonate and is believed to participate in a hydrogen bond network around the non-heme iron. Based on our structural data, it is finally possible to rationalize the inhibitory effect of terbutryn on the electron transfer to QB in PSIIcc. Similar to the pbRC, the position of terbutryn partially overlaps with that of quinone. In contrast to the quinone, terburtyn does not directly interact with His-215A. Besides blocking the electron transfer, binding of herbicides to the QB pocket is known to affect the redox potential of the remote QA. As found by several groups, changes in the redox potential of QA correlate with the suppression or stimulation of photoinhibition (47). Phenolic herbicides were found to lower the redox potential of QA by ⬃45 mV, whereas DCMU increases the potential by ⬃50 mV (22). The upshift of the QA redox potential observed for DCMU binding seems to be representative for all urea-type and also for triazine-type herbicides (22, 48). It was suggested that phenolic herbicides (e.g. bromoxynil and ioxynil) interact directly with His-215A instead of Ser-264A due to their hydroxyl group (16, 45). Recent results of FTIR studies in combination with density functional theory and docking calculations by Takahashi et al. (49) strongly indicate that phenolic herbicides are bound in their deprotonated state to the imidazole NH group of His-215A. It is also proposed that this strong hydrogen bond to His-215A weakens the hydrogen bond between His-214D and QA mediated by the non-heme iron. The decreased hydrogen bond strength might cause the downshift of the QA redox potential. Consequently, the opposite potential shift induced by DCMU and other urea-type herbicides as well as by triazines was ascribed to the lack of a direct hydrogen bond to His-215A. For DCMU, such a binding mode was postulated based on docking calculations using the 2.9 Å resolution structure of PSIIcc (49). Our crystallographic data on terbutryn binding to PSIIcc are consistent with this proposal. The third PQ binding site, QC, found in native PSIIcc (4) is occupied neither by PQ nor by terbutryn in the PSIIcc-terbutryn complex. In contrast to the QB pocket, the QC site offers no polar contacts for the binding of a quinone headgroup or terbutryn. PQ is held in the QC site most probably by van der Waals contacts mainly between its isoprenoid chain and surrounding cofactors (CarD2, fatty acids of SQDG4 (sulfoquinovosyldiacylglycerol 4), MGDG7 (monogalactosyldiacylglycerol 7), the isoprenoid chain of QB, and the phytol chain of ChlD2). Because terbutryn would probably form fewer van der Waals contacts with this environment, it is not expected to have a significant affinity to the QC site. The reason for the lack of PQ in the QC site is presently unknown but could be related to the illumination of the samples. An interesting side product of the present study is the occurrence of an additional chloride binding site, Cl1B, close to the Mn4Ca cluster. This chloride is different from the site found in the structure of native PSIIcc at 2.9 Å resolution, here referred to as the Cl1A site (Fig. 4). However, it is also distinct from the second halide site identified in crystallographic work on bro-


mide-substituted PSIIcc (50, 51). In contrast to these studies, we found no electron density at this halide site, which might result from different preparation procedures. In the present work, the Cl1A site exhibits an occupancy of only ⬃30%, whereas the newly identified Cl1B site shows ⬃70% occupancy. In contrast, the Cl1A site was fully occupied in native PSIIcc (4). The most straightforward interpretation is that Cl1A and Cl1B represent two different positions of one chloride ion. In principle, there are two possible reasons for this position change of the chloride: (i) the illumination of the PSIIcc sample prior to addition of the herbicide or (ii) the incubation with terbutryn in the dark. As indicated by our fluorescence measurements, most of the RCs contain a reducible QB. Therefore, illumination prior to herbicide treatment will cause at least one turnover of the Mn4Ca cluster. Nevertheless, the subsequent dark adaptation period probably allows for a relaxation in the S1 state (about 40 min), even in the presence of terbutryn (52). This excludes the possibility that the observed shift of the chloride positions is caused by redox state changes of the Mn4Ca cluster. Hence, it seems that binding of terbutryn to the QB site affects the chloride position at the donor side. A mutual interaction between the donor and acceptor side of PSII was demonstrated by the shift of the redox potential of the primary plastoquinone QA after depletion of Ca2⫹ from the Mn4Ca cluster. This finding led to the proposal of a structural linkage of these two functional areas in PSII (47). Studies of chloride binding to PSII from spinach using radioactive 36Cl⫺ suggest the presence of one chloride ion per Mn4Ca cluster (53). This chloride was found to bind in two modes corresponding either to high affinity binding with a slow exchange rate or to low affinity binding with a high exchange rate. Interconversion between the two modes was found to be faster than exchange at the high affinity site. This finding was interpreted with two alternative models: (i) a one-site two-state model (53) or (ii) a two-site model (54). The present structural data suggest that two binding sites for one chloride ion exist close to the Mn4Ca cluster, between which the anion can move. This favors the two-site model. Calculations based on the native PSIIcc structure at 2.9 Å resolution (4) identified possible channels connecting the Mn4Ca cluster with the lumen (28). The chloride Cl1A is located in a region where two channels (referred to as C and G; see Fig. 4B) branch out. These two channels were recommended as proton exit pathways, and the chloride ion was suggested to be involved in proton transfer in agreement with earlier proposals (55). Interestingly, Cl1B is found to be shifted into channel C close to the side chains of Asp-61A and Glu-65A (Fig. 4). These two residues are likewise discussed as being involved in proton transfer (56 –58). The close proximity of the carboxylate groups to the anion Cl1B is rather untypical for a chloride binding site within a protein. Stable binding of chloride to this site strongly suggests that both carboxylate groups are protonated. This implies that either these groups have to be protonated to allow the chloride to move to the Cl1B position or, vice versa, the chloride movement promotes protonation of these groups. The possible interdependence of chloride position and protonation states could be related to the presumed role of Cl⫺ in proton transfer and suggests that chloride movement may indeed VOLUME 286 • NUMBER 18 • MAY 6, 2011

Crystal Structure of Terbutryn-bound Photosystem II occur in the catalytic cycle of water oxidation. However, the present structural data only give a hint of such a mechanism, and further work is required to provide evidence.

CONCLUSION The first crystal structure of PSIIcc inhibited by the triazinetype herbicide terbutryn was presented here. A prerequisite for obtaining this structure was the preillumination of the samples to reduce the native PQ substrate, facilitating full binding of terbutryn to the QB site. According to the new structural information, the triazine forms at least two hydrogen bonds to the protein, in keeping with earlier theoretical models and in analogy to pbRC. Terbutryn does not directly bind to His215A, which is a ligand to the non-heme iron. This result supports proposed models to explain the effect of herbicide binding to the QB site on the redox potential of QA. To further evaluate these models, it is of interest to also co-crystallize PSIIcc with phenolic herbicides, which are known to have a distinct effect on the QA redox potential, possibly caused by a different binding mode. These studies may provide the structural basis for a deeper understanding of the influence that the herbicide exerts on photoinhibition. Another important result of the present work is the identification of an additional chloride binding position close to the water-oxidizing Mn4Ca cluster, suggesting mobility of the Cl⫺ cofactor. The different chloride positions are probably coupled to protonation states of amino acid residues probably involved in proton transfer away from the catalytic center of water oxidation (28, 59). Future studies will have to evaluate the functional significance of the proposed chloride movement. Acknowledgments—We acknowledge beam time for x-ray data collection at the ESRF (Grenoble) and SLS (Villigen) and competent beamline support.

14.

15.

16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.

REFERENCES 1. Wydrzynski, T. J., and Satoh, K. (eds) (2005) Photosystem II: The Lightdriven Water: Plastoquinone Oxidoreductase, Springer, Dordrecht, The Netherlands 2. Kern, J., and Renger, G. (2007) Photosynth. Res. 94, 183–202 3. Renger, G. (2008) in Primary Processes of Photosynthesis: Principles and Apparatus (Renger, G., ed) pp 237–290, Royal Society of Chemistry, Cambridge, UK 4. Guskov, A., Kern, J., Gabdulkhakov, A., Broser, M., Zouni, A., and Saenger, W. (2009) Nat. Struct. Mol. Biol. 16, 334 –342 5. Guskov, A., Gabdulkhakov, A., Broser, M., Glo¨ckner, C., Hellmich, J., Kern, J., Frank, J., Mu¨h, F., Saenger, W., and Zouni, A. (2010) ChemPhysChem 11, 1160 –1171 6. Deisenhofer, J., and Michel, H. (1989) Science 245, 1463–1473 7. Crofts, A. R., and Wraight, C. A. (1983) Biochim. Biophys. Acta 726, 149 –185 8. Renger, G., and Holzwarth, A. (2005) in Photosystem II: The Light-driven Water: Plastoquinone Oxidoreductase (Wydrzynski, T., and Satoh, K., eds) pp. 139 –175, Springer, Dordrecht, The Netherlands 9. Michel, H., Epp, O., and Deisenhofer, J. (1986) EMBO J. 5, 2445–2451 10. Deisenhofer, J., Epp, O., Miki, K., Huber, R., and Michel, H. (1984) J. Mol. Biol. 180, 385–398 11. Michel, H., and Deisenhofer, J. (1988) Biochemistry 27, 1–7 12. Zouni, A., Witt, H. T., Kern, J., Fromme, P., Krauss, N., Saenger, W., and Orth, P. (2001) Nature 409, 739 –743 13. Petrouleas, V., and Crofts, A. R. (2005) in Photosystem II: The Light-driven

MAY 6, 2011 • VOLUME 286 • NUMBER 18

36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48.

Water: Plastoquinone Oxidoreductase (Wydrzynski, T. J., and Satoh, K., eds) pp. 177–206, Springer, Dordrecht, The Netherlands Lancaster, C. R. (2008) in Primary Processes of Photosynthesis: Principles and Apparatus (Renger, G., ed) pp. 5–56, Royal Society of Chemistry, Cambridge, UK Parson, W. W. (2008) in Primary Processes of Photosynthesis: Principles and Apparatus (Renger, G., ed) pp. 57–109, Royal Society of Chemistry, Cambridge, UK Oettmeier, W. (1999) Cell. Mol. Life Sci. 55, 1255–1277 Trebst, A. (2007) Photosynth. Res. 92, 217–224 Oettmeier, W. (1992) in Topics in Photosynthesis, Vol. 11 (Barber, J. ed.) pp. 349 – 408, Elsevier Science Publishers B.V., Amsterdam Hirschberg, J., and McIntosh, L. (1983) Science 222, 1346 –1349 Ajlani, G., Kirilovsky, D., Picaud, M., and Astier, C. (1989) Plant Mol. Biol. 13, 469 – 479 Zimmermann, K., Heck, M., Frank, J., Kern, J., Vass, I., and Zouni, A. (2006) Biochim. Biophys. Acta 1757, 106 –114 Krieger-Liszkay, A., and Rutherford, A. W. (1998) Biochemistry 37, 17339 –17344 Rutherford, A. W., and Krieger-Liszkay, A. (2001) Trends Biochem. Sci. 26, 648 – 653 Lancaster, C. R., and Michel, H. (1999) J. Mol. Biol. 286, 883– 898 Lancaster, C. R., Bibikova, M. V., Sabatino, P., Oesterhelt, D., and Michel, H. (2000) J. Biol. Chem. 275, 39364 –39368 Sobolev, V., and Edelman, M. (1995) Proteins 21, 214 –225 Kern, J., Loll, B., Lu¨neberg, C., DiFiore, D., Biesiadka, J., Irrgang, K. D., and Zouni, A. (2005) Biochim. Biophys. Acta 1706, 147–157 Gabdulkhakov, A., Guskov, A., Broser, M., Kern, J., Mu¨h, F., Saenger, W., and Zouni, A. (2009) Structure 17, 1223–1234 Krivanek, R., Kern, J., Zouni, A., Dau, H., and Haumann, M. (2007) Biochim. Biophys. Acta 1767, 520 –527 Grabolle, M., and Dau, H. (2005) Biochim. Biophys. Acta 1708, 209 –218 Buchta, J., Grabolle, M., and Dau, H. (2007) Biochim. Biophys. Acta 1767, 565–574 Kabsch, W. (1993) J. Appl. Crystallogr. 26, 795– 800 McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C., and Read, R. J. (2007) J. Appl. Crystallogr. 40, 658 – 674 Emsley, P., and Cowtan, K. (2004) Acta Crystallogr. D Biol. Crystallogr. 60, 2126 –2132 Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta Crystallogr. D Biol. Crystallogr. 54, 905–921 Clark, L. C. (1956) ASAIO J. 2, 41– 48 Gleiter, H., Haag, E., Inoue, Y., and Renger, G. (1993) Photosynth. Res. 35, 41–53 Reifler, M. J., Szalai, V. A., Peterson, C. N., and Brudvig, G. W. (2001) J. Mol. Recognit. 14, 157–165 Sakurai, I., Shen, J. R., Leng, J., Ohashi, S., Kobayashi, M., and Wada, H. (2006) J. Biochem. 140, 201–209 Loll, B., Kern, J., Saenger, W., Zouni, A., and Biesiadka, J. (2005) Nature 438, 1040 –1044 Katona, G., Snijder, A., Gourdon, P., Andreásson, U., Hansson, O., Andreásson, L. E., and Neutze, R. (2005) Nat. Struct. Mol. Biol. 12, 630 – 631 Tietjen, K. G., Kluth, J. F., Andree, R., Haug, M., Lindig, M., Mu¨ller, K. H., Wroblowsky, H. J., and Trebst, A. (1991) Pestic. Sci. 31, 65–72 Egner, U., Hoyer, G. A., and Saenger, W. (1992) Eur. J. Biochem. 206, 685– 690 Egner, U., Hoyer, G. A., and Saenger, W. (1993) Biochim. Biophys. Acta 1142, 106 –114 Trebst, A. (1987) Z. Naturforsch. C 42, 742–750 Takahashi, R., Boussac, A., Sugiura, M., and Noguchi, T. (2009) Biochemistry 48, 8994 –9001 Krieger-Liszkay, A., Fufezan, C., and Trebst, A. (2008) Photosynth. Res. 98, 551–564 Takano, A., Takahashi, R., Suzuki, H., and Noguchi, T. (2008) Photosynth. Res. 98, 159 –167

JOURNAL OF BIOLOGICAL CHEMISTRY

15971

Crystal Structure of Terbutryn-bound Photosystem II 49. Takahashi, R., Hasegawa, K., Takano, A., and Noguchi, T. (2010) Biochemistry 49, 5445–5454 50. Murray, J. W., Maghlaoui, K., Kargul, J., Ishida, N., Lai, T. L., Rutherford, A. W., Sugiura, M., Boussac, A., and Barber, J. (2008) Energy Environ. Sci. 1, 161–166 51. Kawakami, K., Umena, Y., Kamiya, N., and Shen, J. R. (2009) Proc. Natl. Acad. Sci. U.S.A. 106, 8567– 8572 52. Renger, G. (2007) Photosynth. Res. 92, 407– 425 53. Lindberg, K., and Andreásson, L. E. (1996) Biochemistry 35, 14259 –14267


54. van Vliet, P., and Rutherford, A. W. (1996) Biochemistry 35, 1829 –1839 55. Olesen, K., and Andreásson, L. E. (2003) Biochemistry 42, 2025–2035 56. Barber, J., Ferreira, K., Maghlaoui, K., and Iwata, S. (2004) Phys. Chem. Chem. Phys. 6, 4737– 4742 57. Ishikita, H., Saenger, W., Loll, B., Biesiadka, J., and Knapp, E. W. (2006) Biochemistry 45, 2063–2071 58. Sproviero, E. M., Gascoń, J. A., McEvoy, J. P., Brudvig, G. W., and Batista, V. S. (2007) Curr. Opin. Struct. Biol. 17, 173–180 59. Dau, H., and Haumann, M. (2008) Coord. Chem. Rev. 252, 273–295

VOLUME 286 • NUMBER 18 • MAY 6, 2011