Communication pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. 2017, 139, 14340-14343
Differences in the Active Site of Water Oxidation among Photosynthetic Organisms Marius Retegan† and Dimitrios A. Pantazis*,‡ †
European Synchrotron Radiation Facility, 71 Avenue des Martyrs, 38000 Grenoble, France Max Planck Institute for Chemical Energy Conversion, Stiftstrasse 34−36, 45470 Mülheim an der Ruhr, Germany
‡
S Supporting Information *
the oxygen-evolving complex (OEC), and all electron transfer components. Amino acid ligands to the OEC are identical in cyanobacteria, plants, and algae. On the other hand, spectroscopic studies on the Mn4CaO5 cluster of the OEC have documented important differences between cyanobacteria and higher plants,4 both in the accessibility of substrate analogues, suggesting differences in channel properties, and in the electron paramagnetic resonance (EPR) signatures of the cluster, indicating intrinsic differences in electronic structure despite their having the same coordination sphere. Here we determine the precise structural origins of these differences. We use the PSII crystallographic model of T. vulcanus1a as a template to construct membrane-embedded, allatom PSII models and employ computational mutagenesis to create chimeric forms with core protein sequences of S. oleracea, in order to produce atomistic representations of the higher-plant active-site environment. The protein complex is embedded in a lipid bilayer that conforms to the thylakoid membrane composition. Appropriate combinations of all-atom force fields were used for protein, cofactors, explicit water solvent, lipid tails, and carbohydrate heads (see Supporting Information for details). Unbiased molecular dynamics (MD) simulations at 300 K with 2 fs time-step were used in production runs totaling hundreds of nanoseconds. Sequence alignments of the core proteins of T. vulcanus and spinach show 41 substitutions in D1 (PsbA), 34 in D2 (PsbD), 73 in CP43 (PsbC), and 114 in CP47 (PsbB). Nonconserved substitutions are mostly found in the CP proteins. Homology is 96% for the D1 and D2 chains with 88% and 91% identity. Corresponding values for CP43 and CP47 are 94%/84% and 89%/78%. Projection of substitution sites on cyanobacterial PSII (Figure 1) locates most of them in the outer loops of D1/ D2 or the periphery of CP transmembrane helices and the lumenal side of CP47. Differences in EPR phenomenology between cyanobacterial and higher plant OEC are suggestive of a local causality.4 In the S2 state of the catalytic cycle, the Mn4CaO5 cluster in plants displays two ground-state signals: a low spin S = 1/2 multiline signal at g ≈ 2.0 and an S = 5/2 broad signal at g ≈ 4.1. These signals have been attributed to isomeric forms of the cluster,5 one with an open and one with a closed Mn3CaO4 cubane core.5,6 In cyanobacteria the S2-state low-spin signal is essentially the same, but the g ≈ 4.1 is absent in native PSII7 and similar signals around this value can only be produced by
ABSTRACT: The site of biological water oxidation is highly conserved across photosynthetic organisms, but differences of unidentified structural and electronic origin exist between taxonomically discrete clades, revealed by distinct spectroscopic signatures of the oxygen-evolving Mn4CaO5 cluster and variations in active-site accessibility. Comparison of atomistic models of a native cyanobacterial form (Thermosynechococcus vulcanus) and a chimeric spinach-like form of photosystem II allows us to identify the precise atomic-level differences between organisms in the vicinity of the manganese cluster. Substitution of cyanobacterial D1-Asn87 by higher-plant D1-Ala87 is the principal discriminating feature: it drastically rearranges a network of proximal hydrogen bonds, modifying the local architecture of a water channel and the interaction of second coordination shell residues with the manganese cluster. The two variants explain species-dependent differences in spectroscopic properties and in the interaction of substrate analogues with the oxygen-evolving complex, enabling assignment of a substrate delivery channel to the active site.
A
s a foundation for understanding function, the structural characterization of membrane proteins involved in oxygenic photosynthesis has been a grand challenge for protein crystallography. In recent years, significant progress has been made in resolving the structure of photosystem II (PSII) from thermophilic cyanobacteria, culminating in sub-2.0-Å resolution models of PSII from Thermosynechococcus vulcanus.1 These studies revealed in atomic detail the connectivity of the inorganic Mn4CaO5 cluster that oxidizes water to dioxygen,2 and recently exploited femtosecond X-ray free-electron laser pulses to minimize the radiation damage that compromised previous X-ray diffraction studies.1b So far it has proven impossible to replicate these successes with the PSII of higher plants. Single-particle cryo-electron microscopy recently provided a view into the organization of the PSII/lightharvesting complex II supercomplex from spinach (Spinacia oleracea),3 but no conclusive information on the active site of water oxidation could be derived. Despite extensive differences in the extrinsic proteins of PSII and the light-harvesting apparatus, the central region is highly conserved so higher plants and cyanobacteria have strong similarities in the four core proteins (D1, D2, CP43, and CP47) that accommodate the photoactive chlorophyll reaction center, © 2017 American Chemical Society
Received: June 19, 2017 Published: September 26, 2017 14340
DOI: 10.1021/jacs.7b06351 J. Am. Chem. Soc. 2017, 139, 14340−14343
Communication
Journal of the American Chemical Society
bridges the two central Mn ions (Mn2 and Mn3 in crystallographic numbering). Although D1-D170 and D1-D61 are also perturbed, the H-bond arrangement at this site principally modifies: (1) the nature of the interaction between R357 and the OEC, and (2) the end point of a water channel that connects the active site to the lumen. In cyanobacteria the side-chain carbonyl of N87 forms short and strong hydrogen bonds with the Hε and one of the Hη protons of R357, while the N87 amido group bonds with the backbone carbonyl of E354 (Figure 2). Trajectory analysis of Figure 1. (a) Membrane-embedded PSII model in a simulation box of ca. 150 Å edge length. (b) PSII core proteins with substitution sites marked in orange. The circle indicates the region included in the chimeric spinach model.
mutations or chemical treatments.8 These observations are suggestive of variations in magnetic interactions between Mn ions, but probably too small to arise from intracluster or firstsphere structural differences. Cyanobacteria and plants also respond differently to the substrate analogue methanol, which stabilizes low-spin states and affects relative intensities of EPR signals in plant OEC, whereas cyanobacteria are almost insensitive.7 This is similarly suggestive of local variations in the methanol interaction mode. In the following, we focus on sequence differences within ca. 20 Å of the OEC.9 No variations exist in the first coordination sphere of the cluster and around the redox active tyrosine D1-Y161.10 Similarly invariant are protein parts that interact with the first coordination sphere. Within 10 Å of the cluster two different D1 residues are identified: cyanobacterial serine S85 and asparagine N87 are threonine and alanine in spinach. Within 15 Å substitutions in D1 also include V82I, V83I, Q113E, I116V, and F155T. One substitution is found in the D2 protein, L319I. Finally, within 20 Å there are four additional D1 (T79S, F117L, S153A, and A286T) and one D2 substitution (L182I). CP47 is rather distant from the OEC; only two residues (F383 and R384) appear within 20 Å, both conserved. CP43 is identical up to 15 Å, but eight differences appear beyond this radius: T287C, M315F, I319V, I349V, F361L, A392S, F419Y, and A427S. All substitutions were used to construct a “spinach-like” model. Following the initial setup and equilibration steps, the native (T. vulcanus) PSII MD simulations showed stability over long time scales, which allowed us to collect data for long runs. No large-scale movements were observed in either the protein or the membrane. The above substitutions were performed simultaneously and individually to better identify their structural effects. The spinach-like models were allowed to run for at least 100 ns to enable sufficient sampling. Comparison of MD snapshots from spinach-like and native T. vulcanus models showed that only one substitution, D1N87A, has a marked structural effect. The change from cyanobacterial asparagine to alanine introduces a residue of distinctly smaller size and different chemical properties. Comparison of many D1 (psbA) sequences indicates that A87 is common to higher plant and N87 to cyanobacterial proteins, so the implications of the present analysis are general. Intriguingly, A87 is present in the sequence of a “rogue” D1 variant discovered in the genome of certain diazotrophic cyanobacteria.11 The substitution is important because N87 participates in a strongly coupled H-bonded triad with CP43R357 and CP43-E354, a first-sphere ligand to the cluster that
Figure 2. (a) Snapshot showing the hydrogen bond network around the cyanobacterial N87 site, with part of the related water channel. (b) H-bond distance distribution from trajectory analysis of the T. vulcanus model (left); comparison showing the enhancement of R357-E354 interaction in plant-like PSII (right). (c) Two S2-state isoforms of the Mn4CaO5 cluster related to distinct spin states and EPR features, with indication of antiferromagnetic (green) and ferromagnetic (red) pathways.5
5000 snapshots over a 500 ns MD run (Figure 2) confirms the persistence of bonds between the Oδ of N87 and the Hε/Hη of R357 (average distances Rav = 2.05 and 2.12 Å with standard deviations σ = 0.21 and 0.23 Å), and of two additional H-bonds bonds between N87-E354 (Rav = 2.25 Å, σ = 0.23 Å) and the backbone functionalities of R357-E354 (Rav = 2.32 Å, σ = 0.15 Å). This dense network defines the orientation of R357, whose side-chain Hη atoms interact with the O4 bridge of the inorganic core, the Mn4 ligand D1-D170, and the secondsphere residue D1-Q165. In the spinach-like analogue the replacement of carboxamido by the methyl of alanine eliminates the associated H-bonding network. Only the bond between the backbone NH of R357 and CO of E354 is retained at almost the same average distance (2.35 Å), locking this CP43 loop in the same way as in cyanobacteria. In the absence of other H-bond acceptors R357 tilts in the spinach-like model so that Hε establishes a weak but persistent interaction with the peptide carbonyl of E354. As a result, this distance is shortened from an average of 3.23 Å in cyanobacteria to 2.76 Å in spinach (Figure 2). 14341
DOI: 10.1021/jacs.7b06351 J. Am. Chem. Soc. 2017, 139, 14340−14343
Communication
Journal of the American Chemical Society These structural conclusions do not depend on the electronic state of the manganese cluster, but only on differences in protein sequence. Conversely, the protein surrounding is known to remain unperturbed by the presence of distinct OEC forms.1c An obvious consequence of the R357 tilt on the electronic structure of the Mn4CaO5 cluster relates to the O4 and O2 bridges that can H-bond with the Hη protons of guanidinium. O2 bridges Mn2 and Mn3, while O4 bridges the pendant Mn4 and the Mn3 ions, modulating the sign and magnitude of their exchange coupling.5,12 Depending on the isomeric form, the Mn3−Mn4 coupling may be the unique antiferromagnetic interaction (Figure 2c) and hence controls the total spin of the OEC and the energetic separation between low-lying spin states.12b,13 The arginine tilt resulting from the Hε-E354 interaction in spinach allows Hη to H-bond with O2 instead of O4 as in cyanobacteria. The consequences of these effects were investigated through quantum mechanical optimizations and spectroscopic calculations of the S2 state of the OEC using large species-specific models that include the substitution region. Our models follow the “high-valent” paradigm, i.e., Mn oxidation states (III)(IV)3.12b An alternative “low-valent” scheme with oxidation states (III)3(IV) for the S2 state is being discussed in the literature.14 In the Supporting Information we compare the two possibilities and explain why we favor the present choice. Optimized geometries confirm that “open” and “closed” isomers are stable minima in both organisms, and hence the N87A substitution does not alter the intrinsic bonding properties of the cluster. The major difference is that CP43R357 optimizes into an H-bonding interaction with O4 in the cyanobacterial versus O2 in the spinach-like model. Calculation of exchange coupling constants and spin states shows that in the spinach-like system the correspondence is as previously deduced from smaller OEC models: the open form has a spin doublet and the closed form a sextet ground state, associated with g ≈ 2 and g ≈ 4.1 signals, respectively. The cyanobacterial open form also displays an S = 1/2 ground state, but the closed cubane isomer differs from the plant model. Enhancement of the ferromagnetic Mn2−Mn3 and attenuation of the antiferromagnetic Mn3−Mn4 pathways, consistent with the changes in R357-O2/O4 interactions, lead to S ≥ 7/2 and hence the g ≈ 4.1 signal is not predicted for cyanobacteria. Such higher-spin models may be relevant to higher-g signals,15 but a more thorough investigation would be required for specific correlations. The results indicate that S2-state differences between organisms arise not from differences in the structure of the Mn4CaO5 core, but from second-shell induced differences in exchange coupling pathways, particularly in the high-spin isomer. This type of H-bond mediated effect is reminiscent of other second-sphere perturbations such as the D2-K317R mutation that led to a g ≈ 4 signal in cyanobacteria.8a Another important consequence is the modification of the local channel architecture at a critical constriction point (Figure 3). The N87A substitution in plant PSII enlarges the cavity of the water channel that terminates at the O4/Mn4 side of the cluster and the D1-D61 residue, i.e., right past N87 in cyanobacterial PSII. Simultaneously, the A87 analogue loses the ability to hydrogen-bond with channel water molecules. Several water channels with undefined roles are connected to the OEC. This specific channel (see SI for nomenclature in earlier literature) connects the OEC with the lumen at the interface of cyanobacterial PsbU and PsbO proteins.16 It has
Figure 3. Comparison of the channel end point in the cyanobacterial (left) and the spinach-like model (right).
been suggested to function as a proton egress pathway,17 but also had the lowest barrier for water entry to the OEC in steered MD simulations.18 In addition, interpretations of ammonia and methanol effects on the spectroscopy of the OEC suggested interaction with the Mn4 or O4 sites. NH3 was shown to substitute the W1 ligand of Mn4,19 while based on quantum chemical interpretation of data obtained with 13Clabeled methanol,20 MeOH was shown to replace the terminal water of this channel in spinach (the H2O molecule with crystallographic ID 539 or 511 in PSII structures 3ARC or 3WU2)1a affecting the Mn3−Mn4 coupling.21 Our simulations show D1-N87A to be the only channel-related substitution close to the OEC, therefore differences in interaction of substrate analogues must be attributed to variations in this channel. The present results thus affirm the assignment of the methanol interaction site in spinach by providing a natural explanation for the distinct effect in cyanobacteria and plants: MeOH, a molecule with a larger effective radius than either H2O or NH3, does not fit in the smaller local cavity of cyanobacteria or is trapped in interaction with N87, similar to H-bonded water molecules seen in MD trajectories. Either would be consistent with a limited perturbation of the cyanobacterial versus the higher-plant cluster, in line with experiment. In tandem with recent models for water binding in the S2−S3 transition,22 these results support the assignment of this channel as the transport pathway of substrate analogues, and hence as a substrate delivery pathway. In summary, we compared the environment of the water oxidation active site in cyanobacterial and spinach-like PSII models using all-atom molecular dynamics. The OEC of higher plants is differentiated by a single substitution, D1-N87A. This rearranges the local hydrogen bond network, perturbing the second coordination shell of the Mn4CaO5 cluster and affecting the interaction mode with CP43-R357, which results in distinct magnetic properties between organisms. Concomitant modification of a water channel associated with the O4 bridge and the pendant Mn rationalizes substrate analogue accessibility differences. The results anticipate advances in atomic-level characterization of plant PSII and genetic manipulation of the D1 protein. They also support the assignment of a water transport channel with implications for the mechanism of substrate incorporation and formation of terminal intermediates prior to O2 evolution.22a,23
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b06351. 14342
DOI: 10.1021/jacs.7b06351 J. Am. Chem. Soc. 2017, 139, 14340−14343
Communication
Journal of the American Chemical Society
■
2013, 110, 15561−15566. (b) Schraut, J.; Kaupp, M. Chem. - Eur. J. 2014, 20, 7300−7308. (20) Oyala, P. H.; Stich, T. A.; Stull, J. A.; Yu, F.; Pecoraro, V. L.; Britt, R. D. Biochemistry 2014, 53, 7914−7928. (21) Retegan, M.; Pantazis, D. A. Chem. Sci. 2016, 7, 6463−6476. (22) (a) Retegan, M.; Krewald, V.; Mamedov, F.; Neese, F.; Lubitz, W.; Cox, N.; Pantazis, D. A. Chem. Sci. 2016, 7, 72−84. (b) Capone, M.; Narzi, D.; Bovi, D.; Guidoni, L. J. Phys. Chem. Lett. 2016, 7, 592− 596. (23) (a) Cox, N.; Retegan, M.; Neese, F.; Pantazis, D. A.; Boussac, A.; Lubitz, W. Science 2014, 345, 804−808. (b) Askerka, M.; Brudvig, G. W.; Batista, V. S. Acc. Chem. Res. 2017, 50, 41−48. (c) Li, X.; Siegbahn, P. E. M. Phys. Chem. Chem. Phys. 2015, 17, 12168−12174.
Methodological details and additional results and analysis (PDF)
AUTHOR INFORMATION
Corresponding Author
*
[email protected] ORCID
Dimitrios A. Pantazis: 0000-0002-2146-9065 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank the Max Planck Society for financial support. Computing time was granted by the John von Neumann Institute for Computing (NIC) at the Jülich Supercomputing Centre (JSC) under NIC project No. 7056. D.A.P. acknowledges network support by COST action CM1305 “Explicit Control Over Spin-states in Technology and Biochemistry”.
■
REFERENCES
(1) (a) Umena, Y.; Kawakami, K.; Shen, J.-R.; Kamiya, N. Nature 2011, 473, 55−60. (b) Suga, M.; Akita, F.; Hirata, K.; Ueno, G.; Murakami, H.; Nakajima, Y.; Shimizu, T.; Yamashita, K.; Yamamoto, M.; Ago, H.; Shen, J.-R. Nature 2015, 517, 99−103. (c) Tanaka, A.; Fukushima, Y.; Kamiya, N. J. Am. Chem. Soc. 2017, 139, 1718−1721. (2) Krewald, V.; Retegan, M.; Pantazis, D. A. Top. Curr. Chem. 2015, 371, 23−48. (3) Wei, X.; Su, X.; Cao, P.; Liu, X.; Chang, W.; Li, M.; Zhang, X.; Liu, Z. Nature 2016, 534, 69−74. (4) Haddy, A. Photosynth. Res. 2007, 92, 357−368. (5) Pantazis, D. A.; Ames, W.; Cox, N.; Lubitz, W.; Neese, F. Angew. Chem., Int. Ed. 2012, 51, 9935−9940. (6) Bovi, D.; Narzi, D.; Guidoni, L. Angew. Chem., Int. Ed. 2013, 52, 11744−11749. (7) Su, J.-H.; Cox, N.; Ames, W.; Pantazis, D. A.; Rapatskiy, L.; Lohmiller, T.; Kulik, L. V.; Dorlet, P.; Rutherford, A. W.; Neese, F.; Boussac, A.; Lubitz, W.; Messinger, J. Biochim. Biophys. Acta, Bioenerg. 2011, 1807, 829−840. (8) (a) Pokhrel, R.; Service, R. J.; Debus, R. J.; Brudvig, G. W. Biochemistry 2013, 52, 4758−4773. (b) Boussac, A.; Rutherford, A. W.; Sugiura, M. Biochim. Biophys. Acta, Bioenerg. 2015, 1847, 576−586. (9) Retegan, M.; Neese, F.; Pantazis, D. A. J. Chem. Theory Comput. 2013, 9, 3832−3842. (10) Vogt, L.; Vinyard, D. J.; Khan, S.; Brudvig, G. W. Curr. Opin. Chem. Biol. 2015, 25, 152−158. (11) Murray, J. W. Photosynth. Res. 2012, 110, 177−184. (12) (a) Ames, W.; Pantazis, D. A.; Krewald, V.; Cox, N.; Messinger, J.; Lubitz, W.; Neese, F. J. Am. Chem. Soc. 2011, 133, 19743−19757. (b) Krewald, V.; Retegan, M.; Cox, N.; Messinger, J.; Lubitz, W.; DeBeer, S.; Neese, F.; Pantazis, D. A. Chem. Sci. 2015, 6, 1676−1695. (13) Krewald, V.; Retegan, M.; Neese, F.; Lubitz, W.; Pantazis, D. A.; Cox, N. Inorg. Chem. 2016, 55, 488−501. (14) Jin, L.; Smith, P.; Noble, C. J.; Stranger, R.; Hanson, G. R.; Pace, R. J. Phys. Chem. Chem. Phys. 2014, 16, 7799−7812. (15) Boussac, A.; Un, S.; Horner, O.; Rutherford, A. W. Biochemistry 1998, 37, 4001−4007. (16) (a) Ho, F. M.; Styring, S. Biochim. Biophys. Acta, Bioenerg. 2008, 1777, 140−153. (b) Ogata, K.; Yuki, T.; Hatakeyama, M.; Uchida, W.; Nakamura, S. J. Am. Chem. Soc. 2013, 135, 15670−15673. (17) Saito, K.; Rutherford, A. W.; Ishikita, H. Nat. Commun. 2015, 6, 8488. (18) Vassiliev, S.; Zaraiskaya, T.; Bruce, D. Biochim. Biophys. Acta, Bioenerg. 2012, 1817, 1671−1678. (19) (a) Pérez Navarro, M.; Ames, W. M.; Nilsson, H.; Lohmiller, T.; Pantazis, D. A.; Rapatskiy, L.; Nowaczyk, M. M.; Neese, F.; Boussac, A.; Messinger, J.; Lubitz, W.; Cox, N. Proc. Natl. Acad. Sci. U. S. A. 14343
DOI: 10.1021/jacs.7b06351 J. Am. Chem. Soc. 2017, 139, 14340−14343
