Identifying carboxylate ligand vibrational modes in photosystem II with QM/MM methods

The light-driven oxidation of water in photosystem II (PSII) produces nearly all of the molecular oxygen on Earth and drives the production of nearly all of Earth’s biomass. Determining the mechanism of O₂ formation in PSII is one of the hottest topics in photosynthesis and has stimulated much interest in the development of artificial photosynthesis. The catalytic water-oxidizing center (WOC) consists of a Mn₄CaO₅ cluster, nearby amino acid residues, and numerous surrounding water molecules (Fig. 1). Rapid progress in understanding the mechanism of O₂ formation has been made in the last 5 years because of developments in crystallography (e.g., the development of free electron laser sources) and the interplay between new structural information, computational studies, and advanced biophysical methods such as pulsed EPR spectroscopy, X-ray absorbance spectroscopy, and membrane inlet mass spectrometry (1–4). FTIR spectroscopy is an additional biophysical method that has the advantage of being sensitive to the amino acid residues and water molecules that surround the Mn₄CaO₅ cluster. However, the interpretation of much of the FTIR literature on PSII has been hampered by an inability to assign spectral features to specific residues or functional groups. In PNAS, Nakamura and Noguchi (5) address this problem by simulating FTIR difference spectra of the WOC in the symmetric carboxylate stretching region on the basis of quantum mechanics/molecular mechanics (QM/MM) methods.

Fig. 1.

The Mn₄CaO₅ cluster and its ligation environment from the 1.95-Å X-ray crystallographic structure (7) (Protein Data Bank ID code 4UB6). For clarity, only selected residues are shown. Except as noted otherwise, all residues are from the D1 polypeptide. Purple spheres, manganese ions; yellow sphere, calcium; large red spheres, μ-oxo bridges; blue sphere, chloride; small red spheres, water molecules including the four water molecules bound to Mn4 (W1 and W2) and Ca (W3 and W4).

PSII is located in the thylakoid membranes of plants, algae, and cyanobacteria. It is a homodimer in vivo, having a total molecular weight of ∼700 kDa with each monomer containing at least 20 different subunits and nearly 60 organic and inorganic cofactors. The largest membrane-spanning subunits include CP47 (56 kDa), CP43 (52 kDa), D2 (39 kDa), and D1 (38 kDa). The Mn₄CaO₅ cluster is ligated by six carboxylate groups, one histidine side chain, and four water molecules. The histidine residue (H332) and five of the six carboxylate groups (D170, E189, E333, D342, and the C terminus at A344) are from D1, and one (E354) is from CP43. Recent 1.9- and 1.95-Å crystallographic structures of PSII show that the Mn₄CaO₅ cluster is arranged as a distorted Mn₃CaO₄ cube that is linked to a fourth “dangling” Mn ion (denoted Mn4) by two oxo bridges (6, 7). Two water molecules (W1 and W2) coordinate to Mn4, and two (W3 and W4) coordinate to the Ca²⁺ ion. Networks of hydrogen bonds in the Mn₄CaO₅ cluster’s environment efficiently transport protons away from the cluster and permit the access of water. Light-induced separations of charge within PSII drive the accumulation of oxidizing equivalents on the Mn₄CaO₅ cluster. During each catalytic cycle, the cluster accumulates four oxidizing equivalents, cycling through five oxidation states termed S_n, where “n” denotes the number of oxidizing equivalents stored (n = 0–4). The S₁ state predominates in dark-adapted samples. The S₄ state is a transient intermediate whose formation triggers the utilization of the four stored oxidizing equivalents, the formation of the O–O bond, the release of O₂ and two protons, the rebinding of at least one substrate molecule, and the regeneration of the S₀ state. The Mn₄CaO₅ cluster thus serves as the interface between one-electron photochemistry and the four-electron/four-proton process of water oxidation.

FTIR spectroscopy is an extremely sensitive tool for characterizing structural changes that occur during an enzyme’s catalytic cycle and is particularly suited for analyzing protonation/deprotonation reactions, pK_a shifts, and changes in hydrogen bonded structures. In PSII, the vibrational modes of numerous functional groups shift as the Mn₄CaO₅ cluster is oxidized through the S-state cycle, including many that are attributable to carboxylate residues or hydrogen-bonded water molecules (8–10). The identification of these functional groups would impose additional constraints on proposals for the mechanism of O–O bond formation. However, only a single carboxylate group has been clearly identified to date: the symmetric carboxylate stretching mode of the C terminus of the D1 polypeptide at A344 was identified by the incorporation of l-[1-¹³C]alanine (11). Nakamura and Noguchi address this problem by using QM/MM methods to simulate the S₂-minus-S₁ FTIR difference spectrum in the symmetric carboxylate stretching region (1,480–1,250 cm⁻¹). The MM region included all atoms within 20 Å of the Mn₄CaO₅ cluster, whereas the QM region included the cluster, Y_Z (Y161, the cluster’s immediate oxidant during each S-state transition), the cluster’s protein ligands, additional residues that interact with these ligands or Y_Z (D61, H190, H337, CP43-R357), and 15 water molecules including W1, W2, W3, W4, and others that form a network of hydrogen bonds stretching to Y_Z. Previously, Nakamura, Noguchi, and coworkers simulated the S₂-minus-S₁ FTIR difference spectrum in the O−H stretching region (3,800−2,500 cm⁻¹), finding that (i) bands between 3,700 and 3,500 cm⁻¹ correspond to the coupled vibrations of weakly hydrogen-bonded water molecules that are mostly hydrogen-bonded with amino acid residues or Cl⁻ and (ii) a broad positive feature between 3,200 and 2,500 cm⁻¹ corresponds to the coupled vibrations of strongly hydrogen-bonded water molecules, including W1 and W2 (12). In the current study, Nakamura and Noguchi’s simulations reveal that the symmetric stretching modes of the Mn₄CaO₅ cluster’s carboxylate ligands are significantly coupled, with most spectral features having contributions from multiple carboxylate groups. However, the most prominent feature, a negative band at 1,401 cm⁻¹, arises mainly from D170, the residue that bridges Mn4 with the Ca²⁺ ion. E333, the residue that bridges Mn4 and Mn3, contributes to the low-frequency side of this band. Both bands shift to near 1,360 cm⁻¹ in the S₂ state, reflecting changes in C−O and Mn−O distances during the S₁-to-S₂ transition. The assignment of the most prominent feature in the S₂-minus-S₁ FTIR difference spectrum to D170 and E333 is consistent with another QM/MM study that concluded that the small structural changes that accompany the S₁-to-S₂ transition affect mostly the position of Mn4 relative to the rest of the cluster (13). Nakamura and Noguchi also examine the consequence of deprotonating W2 to OH⁻, but the simulations having W2 deprotonated do not fit the experimental data as well as the simulations having W2 fully protonated.

Nakamura and Noguchi also successfully simulated the symmetric carboxylate stretching region in the ¹²C-minus-¹³C S₂-minus-S₁double-difference spectrum of PSII having D1 specifically labeled at its C terminus (at A344) (11) and simulated their earlier spectra of Ca-depleted PSII (14, 15). Their simulation of the ¹²C-minus-¹³C spectrum accounts for the asymmetry of A344 as a bridging ligand between Mn2 and the Ca²⁺ ion and the increased asymmetry of the bridge in the S₂ state. In the spectrum of Ca-depleted PSII, the

Nakamura and Noguchi provide an excellent framework for understanding the vibrational modes of the carboxylate ligands of the Mn₄CaO₅ cluster in PSII, but more work will be necessary to understand the discrepancy between their spectral simulations and the earlier mutagenesis studies.

symmetric stretching modes of several carboxylate groups are altered in the absence of Ca²⁺, but the largest change is a downshift of the negative band of D170 from at 1,401 cm⁻¹ to near 1,350 cm⁻¹, consistent with the coordination mode of D170 changing from bidentate to unidentate after the removal of Ca²⁺.

In addition to their simulation of the FTIR data of native, l-[1-¹³C]alanine-labeled, and Ca-depleted PSII, Nakamura and Noguchi also provide additional support for the high oxidation state model of the Mn ions [e.g., that the Mn₄CaO₅ cluster in its S₁ state contains two Mn(III) and two Mn(IV) ions]: simulations having the Mn ions in lower oxidation states fit the experimental data much less well. Nakamura and Noguchi also point out an important puzzle: their predicted spectra disagree with mutagenesis studies that found that individually mutating four of the Mn₄CaO₅ cluster’s six carboxylate ligands, D170, E189, E333, and D342, produced little or no changes in any of the S_n+1-minus-S_n FTIR difference spectra (e.g., refs. 9, 16, and 17). The lack of mutation-induced changes was not the result of mutants losing their mutation: in all studies, the integrity of each 21-L culture harvested for the purification of mutation-bearing PSII core complexes was verified by extracting genomic DNA from an aliquot of cells, and then amplifying and sequencing the target gene. In contrast, mutations constructed at residues located in the hydrogen-bond networks that surround the Mn₄CaO₅ cluster, 5–11 Å from the nearest Mn ion and outside the QM region examined by Nakamura and Noguchi (e.g., E65, Q165, N181, R334, D2-E312, and D2-K317), produced numerous changes in the FTIR difference spectra (e.g., refs. 9 and 18–20). Nakamura and Noguchi provide an excellent framework for understanding the vibrational modes of the carboxylate ligands of the Mn₄CaO₅ cluster in PSII, but more work will be necessary to understand the discrepancy between their spectral simulations and the earlier mutagenesis studies.