The reaction center in photosystem II (PSII) has the O₂ evolving complex, Mn₄CaO₅ cluster ( Figure 1A ) (Shen, 2015; Cardona and Rutherford, 2019). O₂ evolves at the Mn₄CaO₅ cluster, which has five O atoms (O1 to O5), two ligand water molecules at the Mn4 site (W1 and W2), and two additional water molecules at the Ca site (W3 and W4). O1 and O4 form an hydrogen bond to water molecules ( Figure 1A ). To convert two substrate water molecules into O₂, four electrons and four protons must be removed. As electron transfer proceeds, the oxidation state of the Mn₄CaO₅ cluster, S _n , increases, and protons are released with a typical stoichiometry of 1:0:1:2 for the S₀ → S₁ → S₂ → S₃ → S₀ transitions (Suzuki et al., 2005). O₂ evolves during the S₃ to S₀ transition, followed by the first proton release during the S₀ to S₁ transition. Based on density functional theory (DFT) calculations performed in the absence of the PSII protein environment, it was proposed that the O5 site was protonated in S₀ (Siegbahn, 2013; Krewald et al., 2015; Lohmiller et al., 2017). However, the PSII structure shows that O5 has no H-bond partner, which suggests that the release of the proton from O5 is unlikely to occur in the PSII protein environment. In contrast, the O4 site forms a significantly short hydrogen bond (< 2.5 Å) with the adjacent water molecule (W539) in the PSII structures (Umena et al., 2011; Suga et al., 2015). Quantum mechanical/molecular mechanical (QM/MM) calculations indicated that O4 and W539 form a low-barrier hydrogen bond (LBHB), which facilitates the release of the proton from O4 during the S₀ to S₁ transition (Saito et al., 2015; Takaoka et al., 2016). Time resolved infrared (TRIR) and Fourier transform infrared (FTIR) studies (Shimizu et al., 2018; Yamamoto et al., 2020) suggested that the S₀ to S₁ transition is the fastest among all S state transitions, which is in line with the formation of the LBHB between O4 and W539.

The Mn₄CaO₅ cluster structure was determined by X-ray diffraction (XRD) crystallography at a resolution of 1.9 Å (Umena et al., 2011) on the assumption that it was in the dark-stable S₁ state, as the PSII crystals used for the diffraction experiments were stored in the dark during crystallization and the diffraction intensity measurements. The structure was obtained by using a high-dose X-ray (an average dose of 0.43 MGy) and is referred to as the high-dose structure. Thus, the Mn₄CaO₅ cluster may have been overreduced, including Mn(II) (Yano et al., 2005; Grabolle et al., 2006; Luber et al., 2011; Galstyan et al., 2012; Glockner et al., 2013; Saito and Ishikita, 2019). Using X-ray free-electron laser (XFEL), the “radiation-damage-free” structure was reported (Suga et al., 2015). However, it was suggested that the Mn₄CaO₅ cluster is reduced to S₀ even in the XFEL structures (Askerka et al., 2015). Tanaka et al. determined structures using significantly lower X-ray doses (0.03 and 0.12 MGy) with conventional synchrotron radiation sources (low-dose structures) (Tanaka et al., 2017).

In the low-dose PSII structures, the hydrogen bond distances differ significantly between the two monomer units (A and B) (Tanaka et al., 2017). The main differences were as follows: (1) the hydrogen bond between O3 and D1-His337 was shorter in the A monomer (2.46 Å) than in the B monomer (2.75 Å) ( Figure 1B ); (2) the hydrogen bond between O4 and W539 [W6 in ref. (Tanaka et al., 2017)] was longer in the A monomer (2.66 Å) than in the B monomer (2.44 Å) ( Figure 1B ).

It was speculated that the differences in the hydrogen-bond distance were due to the difference in the protonation states of D1-His337 and W539 (Tanaka et al., 2017). According to Tanaka et al., D1-His337 might be doubly-protonated [HN-His-NH]⁺ in the short O3-His337 hydrogen bond and singly-protonated [HN-His-N]⁰ in the long O3-His337 hydrogen bond ( Figure 2B ). Tanaka et al. also speculated that W539 existed as H₃O⁺ in the short O4-W539 hydrogen bond. However, H₃O⁺ (pK _a = −1.7) can exist only when the binding moiety is stabilized by a cluster of acidic residues (Ikeda et al., 2017). In contrast, there exists the positively charged Mn₄CaO₅ cluster adjacent to W539, which inhibits the formation of H₃O⁺. Thus, either [O4-H...O_W539H₂]⁺ (protonated O4) or [O4...HO_W539H]⁰ (deprotonated O4) can be more relevant (Saito et al., 2015) ( Figure 2A ). The details of the protonation states in the low-dose structure are not reported. Here, we investigate the protonation state in the low-dose structure using a QM/MM approach.

The PSII atomic coordinates were taken from the lower dose structure (0.03 MGy; PDB-ID 5B5E) (Tanaka et al., 2017). The H atom positions were optimized with CHARMM (Brooks et al., 1983), whereas the heavy atom positions were fixed. During the procedure, all titratable groups (e.g., acidic and basic groups) were ionized. Additional counter ions were added to neutralize the entire system in QM/MM calculations. Atomic partial charges of the amino acids were obtained from the CHARMM22 (MacKerell et al., 1998) parameter set, whereas those of cofactors were obtained from previous studies (Saito et al., 2015). D1-Asp61 is fully ionized (Ishikita et al., 2006). Note that other titratable residues (e.g., D1-Glu65 and D2-Glu312) are >13 Å away from the Mn₄CaO₅ cluster.

The QM/MM calculation was performed as done in previous studies (Mandal et al., 2020). Using the QSite (QSite, 2012) program, the unrestricted DFT method was employed with the B3LYP functional and LACVP** basis sets. The atom positions in the QM region were fully relaxed, whereas the H atom positions in the MM region were optimized using the OPLS2005 force field. The Mn₄CaO₅ cluster was considered to be ferromagnetically coupled (i.e., the total spin S = 14/2, 15/2, 16/2, and 17/2 in S₁, S₀, S₋₁, and S₋₂, respectively). Notably, the resulting optimized Mn₄CaO₅ geometry appears not to be crucial to the spin configurations, as demonstrated in previous theoretical studies (Ames et al., 2011; Isobe et al., 2012). Indeed, the calculated distances in the anti-ferromagnetically coupled (i.e. low spin) cases ( Table S1 ) were essentially the same as the ferromagnetically coupled (i.e., high spin) cases. The Mn valence states were determined by spin densities obtained from the Mulliken population analysis ( Table S2 ). O1, O2, O3, and O5 were considered to be unprotonated (O^2–), while O4 was considered to be protonated (OH⁻) in the protonated O4 hydrogen-bond pattern ( Figure 2A ).

The initial-guess wavefunctions were obtained using the ligand field theory (Vacek et al., 1999) implemented in the QSite program. The QM region was defined as the Mn₄CaO₅ cluster (including the ligand side-chains of D1-Asp170, D1-Glu189, D1-His332, D1-Glu333, D1-Asp342, and CP43-Glu354; the ligand carboxy-terminal group of D1-Ala344; and the ligand water molecules, W1–W4), the O4-water chain (W539, W538, and W393) (Saito et al., 2015; Takaoka et al., 2016), the Cl-1 binding site (Cl-1, W442, W446, and the side-chains of D1-Asn181 and D2-Lys317), the second-sphere ligands (side-chains of D1-Asp61 and CP43-Arg357), and the hydrogen-bond network of TyrZ (side-chains of D1-Tyr161, D1-His190, and D1-Asn298), including the diamond-shaped cluster of water molecules (W5, W6, and W7) (Saito et al., 2011; Kawashima et al., 2018). To calculate the distance of the LBHB in the [O3…H-His-H]⁺ pattern in S₋₂, the QM region was extended to include the hydrogen-bond network of three water molecules near D1-His337. All other protein units and cofactors were approximated by the MM force field. See Supporting Information for the atomic coordinates of the resulting QM region.

The O4…O_W539 hydrogen bond is particularly short in the B monomer of the low-dose structure (2.44 and 2.45 Å in the 0.03 and 0.12 MGy structures, respectively). In the QM/MM geometry optimization using the 0.03 MGy low-dose structure, a short hydrogen bond is reproduced (2.48 and 2.44 Å) when O4 is protonated ([O4-H…O_W539H₂]⁺) and D1-His337 is doubly-protonated ([HN-His-NH]⁺) in S₁ in both the A and B monomers ( Table 1 ), as previously reported in a QM/MM study on the high-dose structure (Saito et al., 2015).

The O3…Nε_His337 hydrogen bond is particularly short in the B monomer of the low-dose structure (2.46 and 2.48 Å in the 0.03 and 0.12 MGy structures, respectively). In the QM/MM geometry optimization, the short hydrogen bond cannot be reproduced regardless of the protonation state of D1-His337 ([HN-His-NH]⁺ and [HN-His]⁰) in both S₀ and S₁ ( Table 1 ). Notably, the protonation state of O4 did not substantially affect the O3…Nε_His337 distance. Thus, the short hydrogen bond between O3 and D1-His337 in the A monomer cannot be explained by the difference in the protonation state of D1-His337 in contrast to the mechanism speculated by Tanaka et al. (Tanaka et al., 2017).

A possible reason for the short hydrogen bond between O3 and D1-His337 could be the overreduced state of the Mn₄CaO₅ cluster. A DFT study demonstrated that the proton of D1-His337 can be transferred to O3 during overreduction of the Mn₄CaO₅ cluster (Galstyan et al., 2012). A previous QM/MM study on the high-dose structure also reported that the hydrogen-bond distance between O3 and D1-His337 decreased (2.58 Å for O3…Nε_His337) when the Mn₄CaO₅ cluster was overreduced in S₋₂ (Saito and Ishikita, 2019). Therefore, the short hydrogen bond of the low-dose structures may originate from the overreduced state in the crystal.

The QM/MM optimized geometries indicate that the O3…Nε_His337 distance decreased as the Mn₄CaO₅ cluster is reduced from S₁ to S₋₁ when D1-His337 is doubly-protonated ([HN-His-NH]⁺) ( Figure 3B ). In S₋₂, a proton can move from D1-His337 to O3 (i.e., [O3-H…His-H]⁺) and the O3…Nε_His337 distance increases (2.63 Å) relative to S₋₁ (2.57 Å). In contrast, the O3…Nε_His337 hydrogen bond (2.54 Å) is the shortest when D1-His337 is doubly-protonated (i.e., [O3…H-His-H]⁺). Proton transfer to O3 from D1-His337 during overreduction was previously reported in a DFT study (Galstyan et al., 2012). Proton transfer can easily occur in a short LBHB (~2.5 Å) (Ishikita and Saito, 2014). Accordingly, the short hydrogen-bond (2.46 and 2.48 Å) in the low-dose structure suggests that the LBHB exists in either S₋₁ or S₋₂. It should be noted that the O3…Nε_His337 distance does not change upon oxidation when D1-His337 is singly-protonated ([HN-His]⁰) ( Figure S1a ).

The O4…O_W539 distance decreases in the protonated O4 pattern ([O4-H … O_W539H₂]⁺) as the Mn₄CaO₅ cluster is oxidized from S₋₂ to S₂ ( Figure 3A ). Thus, the O4…O_W539 hydrogen bond in S₁ is the shortest among all S states investigated, which is consistent with a previous QM/MM study showing that [O4-H…O_W539H₂]⁺ forms an LBHB in S₁ ( Figure 1A ) (Saito et al., 2015).

The "standard" O3…Nε_His337 and "short" O4…O_W539 hydrogen bonds ( Figure 1B ) in the B monomer are best fitted to the case with doubly-protonated D1-His337 ([H-His-H]⁺) and deprotonated O4 ([O4…HO_W539H]⁰) in S₁ ( Figure 3 ). In contrast, the short O3…Nε_His337 and standard O4…O_W539 hydrogen bonds in the A monomer are best fitted to the case with doubly-protonated D1-His337 ([H-His-H]⁺) and protonated O4 ([O4H...O_W539H₂]⁺) in S₋₁ or S₋₂ ( Figure 3 ). Note that doubly-protonated D1-His337 was reported in a FTIR study (Nakamura and Noguchi, 2017). These results may indicate that the two monomer units are in different oxidation states (i.e., S₋₂ or S₋₁ for the A monomer and S₁ for the B monomer) as speculated by Tanaka et al. (Tanaka et al., 2017). Notably, a machine learning study suggested that the oxidation state differs between monomer units (e.g., S₀ for the A monomer and S₁ for the B monomer in the 0.03 MGy structure (Amin, 2022).

The difference in the hydrogen-bond distance originates from the difference in the charge of the Mn₄CaO₅ cluster. The O4-H...O_W539 hydrogen bond lengthens due to an increase in pK _a of the Mn₄CaO₅ cluster as the cluster is reduced ( Figure 4A ). Therefore, the short LBHB of [O4-H…O_W539H₂]⁺ in S₁ is lost upon reduction, which results in a standard hydrogen bond ( Figure 4A ). The O3...His337 hydrogen bond shortens due to an increase in pK _a of the Mn₄CaO₅ cluster as the cluster is reduced ( Figure 4B ). Therefore, the short LBHB of [O3…H-His-H]⁺ in S₋₂ is lost upon oxidation, which results in a standard hydrogen bond ( Figure 4B ).

Thus, the [O4-H … O_W539H₂]⁺ hydrogen bond is longer in the overreduced state than that in S₁, whereas the [O3…H-His-H]⁺ hydrogen bond is shorter in the overreduced state than that in the S₁ state ( Figure 3 ). In the low-dose structures, the O3…Nε_His337 hydrogen bond is shorter in the A monomer than that in the B monomer, whereas the O4…O_W539 hydrogen bond is longer in the A monomer than that in the B monomer ( Figure 2A ; Table 1 ). Thus, O4 is protonated and that the Mn₄CaO₅ cluster may be more reduced in the A monomer than in the B monomer ( Figure 5 ).

These results suggest that A monomers are more damaged than B monomers in the crystal, although the X-ray dose received by both monomers should be identical. This may be explained by the difference in radical diffusion between the two monomer units. Water molecules can be involved in the free radical generation induced by X-ray (Ma et al., 2018). The free radical generation is particularly pronounced near metal cofactors (e.g., the Mn₄CaO₅ cluster), as they can absorb free electrons and be reduced (Beyerlein et al., 2018; Nass, 2019) as pointed out in theoretical studies on the XFEL structure (Mandal et al., 2021). Therefore, the arrangement of water molecules near the Mn₄CaO₅ cluster may be a factor for the difference in the radiation damage. In the PSII crystal structure (Umena et al., 2011; Suga et al., 2015), the number and arrangement of water molecules of the O4 channel differ between monomer units [see Table S1 in (Sakashita et al., 2020)]. The structural difference might also be due to the difference in crystal packing (Tanaka et al., 2017).

In the Mn₄CaO₅ cluster of the low-dose structure, the Mn4-O4 distances are 1.87 and 2.27 Å and the Mn3-O3 distances are 2.27 and 1.96 Å in the A and B monomers, respectively (Tanaka et al., 2017). The calculated geometries did not reproduce the differences ( Table S3 ).

In summary, the difference in the hydrogen-bond distance between the two homogeneous (A and B) monomer units in the low-dose structure originates from the difference in the D1-His337/O4 protonation state and the Mn₄CaO₅ oxidation state. The short O4-W539 hydrogen bond in the B monomer can be reproduced when O4 is protonated in S₁ ( Table 1 and Figure 3A ). The short O3-His337 hydrogen bond in the A monomer indicates that the LBHB forms between O3 and doubly-protonated D1-His337 in S₋₁ or S₋₂ ( Table 1 and Figure 3B ). These results suggest that the Mn₄CaO₅ oxidation state differs between the two monomer units (i.e., S₋₁ or S₋₂ for the A monomer and S₁ for the B monomer; Figure 5 ).

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

HI designed research. KS and SN performed research. KS and HI wrote the manuscript. All authors contributed to the article and approved the submitted version.