Introduction

In photosystem II (PSII), the Mn4CaO5 cluster catalyzes the water-splitting reaction: 2H2O→O2 + 4H++4e− (reviewed in refs. (Dau et al. 2012; Cox and Messinger 2013; Shen 2015)). Protons are released in response to changes in the oxidation state (the S n state, where the subscript represents the number of oxidation steps accumulated) of the oxygen-evolving complex, and it occurs with a typical stoichiometry of 1:0:1:2 for the S0→S1→S2→S3(→S4)→S0 transitions, respectively. Candidates for the relevant proton transfer pathways (e.g., (Renger 2001; Murray and Barber 2007; Ho and Styring 2008; Vassiliev et al. 2012; Ogata et al. 2013; Linke and Ho 2014) and proton-releasing sites (e.g., (Saito et al. 2015)) have been reviewed recently. The energetically lowest process for proton release from the Mn4CaO5 cluster is the S0-to-S1 transition, where the electron transfer occurs prior to H+ release and is thus rate limiting (Dau and Haumann 2008). It has been argued that the exchangeable μ-oxo bridge is possibly either O4 (linking Mn4 and Mn3 in Mn3CaO4-cubane) or O5 (in one of the corners of the cubane linking Mn4 and the cubane) (Yamanaka et al. 2011; Rapatskiy et al. 2012; McConnell et al. 2012; Galstyan et al. 2012; Cox et al. 2013) (see also relevant articles published prior to the detailed crystal structure (Robblee et al. 2002; Messinger 2004; Kulik et al. 2007)). The PSII crystal structures (Umena et al. 2011; Suga et al. 2015) possess a chain of strongly H-bonded water molecules (O4-water chain) directly linked to O4. The cyanobacterial PSII crystal structure also identified a chain of water molecules near the O1 site of the Mn4Ca cluster (O1-water chain). The potential energy profile of the H-bonds along the O4-water chain indicated that the proton released from OH− at O4 could be transferred along the O4-water chain (Saito et al. 2015). The activation energy of proton transfer is the lowest when all the water molecules are strongly H-bonded in the H-bond network (Stuchebrukhov 2009; Takaoka et al. 2016). In the cyanobacterial PSII crystal structure (Umena et al. 2011), the O4-water chain is connected with a channel that extends to the protein bulk surface via the PsbU protein subunit (O4-PsbU channel) (Takaoka et al. 2016). This channel may correspond to a previously proposed possible proton channel of PSII (Gabdulkhakov et al. 2009; Linke and Ho 2014). However, membrane-extrinsic protein subunits PsbU and PsbV in cyanobacteria are replaced with PsbP and PsbQ in higher plants and green algae (Bricker et al. 2012; Ifuku and Noguchi 2016). These extrinsic proteins are associated with the efficiency of O2-evolving activity in PSII (Ifuku et al. 2005). They are also involved in the retention of Ca2+ and Cl− (Ghanotakis et al. 1984a; Shen and Inoue 1993) or play a role in protecting the Mn4Ca cluster (Ghanotakis et al. 1984b). Notably, the crystal structures of the isolated PsbP (Ifuku et al. 2004) and PsbQ (Calderone et al. 2003) proteins show that the plant PsbP and PsbQ proteins and the cyanobacterial PsbU and PsbV proteins (Ferreira et al. 2004; Umena et al. 2011) have no structural similarities. This raises questions regarding the function of the O4-water chain and the continuous PsbU-like channel in plant PSII.

Recently, the cryo-electron microscopy (cryo-EM) structure of the plant PSII from spinach was reported at a resolution of 3.2 Å (Wei et al. 2016). The plant cryo-EM structure also shows membrane-extrinsic protein subunits, including PsbP and PsbQ. On the other hand, the existence of the channels in PsbP and PsbQ is unclear, because the positions of water molecules are not reported in the plant cryo-EM structure. Here, we present the structural properties of the channels that proceed from the Mn4Ca cluster toward the protein bulk surface in plant PSII, using the plant cryo-EM structure.

Methods

Coordinates and atomic partial charges

As a basis for the computations, the X-ray crystal structure of cyanobacterial PSII (PDB ID: 3ARC) (Umena et al. 2011) and cryo-EM structure of plant PSII (PDB ID: 3JCU) (Wei et al. 2016) were used. Hydrogen atoms were generated and energetically optimized using CHARMM version 40b (Brooks et al. 1983), where the positions of all non-hydrogen atoms were fixed and all titratable groups were kept in their standard protonation states (i.e., acidic and basic groups were ionized). Atomic partial charges of the amino acids were adopted from the all-atom CHARMM22 (MacKerell et al. 1998) parameter set. The atomic charges of cofactors were taken from our previous studies on PSII (Saito et al. 2011).

Analysis of the channel space and distribution of water molecules

The channel space was analyzed using the program CAVER (Petrek et al. 2006). To analyze the possible distribution of water molecules and Cl− in the channels, including the region where these molecules and ions are not identified in the crystal structure, we used a three-dimensional reference interaction site model (3D-RISM) with Placevent analysis (Beglov and Roux 1997; Kovalenko and Hirata 1999; Luchko et al. 2010; Case et al. 2012; Sindhikara et al. 2012), as previously used for the inner channel of channelrhodopsin (Watanabe et al. 2013; Wietek et al. 2014) and PSII (Takaoka et al. 2016). It should be noted that the distribution pattern of water molecules obtained from the 3D-RISM with Placevent analysis was consistent with the positions of the water molecules identified in the PSII crystal structure (Takaoka et al. 2016).

MD simulations

The PS II assembly described above was embedded in a lipid bilayer consisting of 546 1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine (POPC) and soaked in 80,353 flexible water models (SPC-Fw) (Wu et al. 2006). After structural optimization with position restraints on heavy atoms of the PSII assembly, the system was heated from 0.001 to 300 K during 5.0 ps, and the restraints were gradually released over 1.0 ns. After an equilibrating MD run for 45 ns, a production run was conducted over 5.0 ns for B-factor evaluation. All the equilibrating simulations were conducted by MD engine AMBER 14 (Case et al. 2014) with SHAKE algorithm for hydrogen constraint (Ryckaert et al. 1977), while NAMD version 2.10 (Phillips et al. 2005) was used for the production run with an MD time step of 0.5 fs without hydrogen constraint. For temperature and pressure control, Berendsen thermostat and barostat were employed in the equilibrating process (Berendsen et al. 1984), while Langevin thermostat and piston were used in the production run (Kubo et al. 1991; Feller et al. 1995).

Protonation pattern and pK a values

The present computation was based on the electrostatic continuum model created by solving the linear Poisson–Boltzmann equation with the MEAD program (Bashford and Karplus 1990). To obtain the protonation pattern of all titratable residues in PSII, we calculated the differences in electrostatic energies between protonated and deprotonated states in a reference model system using an experimentally measured pK a value. The experimentally measured pK a values employed as references were 12.0 for Arg, 4.0 for Asp, 9.5 for Cys, 4.4 for Glu, 10.4 for Lys, 9.6 for Tyr (Nozaki and Tanford 1967), and 7.0 and 6.6 for the Nε and Nδ atoms of His, respectively (Tanokura 1983a, b, c). The ensemble of the protonation patterns was sampled using a Monte Carlo method with Karlsberg (Rabenstein and Knapp 2001). The dielectric constants were set to ε p = 4 for inside the protein and ε w  = 80 for water. All computations were performed at 300 K, pH 7.0 at an ionic strength of 100 mM. The linear Poisson–Boltzmann equation was solved using a three-step grid-focusing procedure at the resolutions of 2.5, 1.0, and 0.3 Å. The Monte Carlo sampling yielded the probabilities of the two protonation states, [protonated] and [deprotonated], for each titratable site.

Results and discussion

The O4-water chain and the connecting PsbU and PsbP channels

The channel space of the O4-water chain, originally reported in the cyanobacterial PSII structure (Saito et al. 2015; Takaoka et al. 2016), was also present in the D1 and CP43 proteins of the plant cryo-EM structure (Fig. 1). Channel space analysis suggested that the plant O4-water chain also proceeds from the Mn4Ca cluster toward CP43-Pro334, CP43-Thr335, and D1-Asn338, similar to the reported cyanobacterial O4-water chain (Saito et al. 2015; Takaoka et al. 2016). Analysis of the plant cryo-EM structure using 3D-RISM with Placevent suggested that water molecules could be present along the O4-water chain, as previously reported in the cyanobacterial structure (Takaoka et al. 2016). The distribution patterns of water molecules in the plant cryo-EM structure were consistent with the distribution patterns and the positions of water molecules in the cyanobacterial PSII structure (Takaoka et al. 2016) (Fig. 1). The shapes of O4-water chains of the cyanobacterial and plant PSII are not remarkably different, as the D1 and CP43 proteins, where the Mn4Ca cluster and the H-bonded O4-water chain are located, are highly conserved in the two PSII proteins.

Fig. 1
figure 1

Channel space of the O4-water chain for the cyanobacterial (blue surface, left) and plant (green surface, right) PSII, where 3D distribution of water oxygen is shown as red mesh obtained via 3D-RISM (the threshold of 3D distribution function is 4.5)

Full size image

In the cyanobacterial PSII, the O4-water chain in the D1 and CP43 proteins is further linked with the O4-PsbU channel (Takaoka et al. 2016). We found that the corresponding channel was also present along the interface of PsbP (O4-PsbP channel) in the plant cryo-EM structure, irrespective of the low structural similarity of PsbP with PsbU (Fig. 2a). The cyanobacterial O4-PsbU channel is mainly composed of D1, D2, CP43, CP47, PsbO, and PsbU (Takaoka et al. 2016), whereas the plant O4-PsbP channel is composed of only four protein subunits, D1, D2, CP43, and PsbP (Table 1). Both the O4-PsbP and O4-PsbU channels commonly start at the terminus of the O4-water chain (e.g., the D1-Asn338 moiety). On the other hand, the orientations of the channels toward the protein bulk surface are different. Superposition of the two channel spaces suggests that the plant O4-PsbP channel would not exist in cyanobacterial PSII due to the presence of PsbU-Tyr98 (Fig. 2c), whereas the cyanobacterial O4-PsbU channel would not exist in plant PSII due to the presence of PsbP-Val108 and PsbP-Asp137 (Fig. 2b). Notably, near the channel branch, there exists a salt-bridge between CP43-Lys339 and PsbP-Asp137 in plant PSII and between CP43-Lys339 and PsbU-Asp96 in cyanobacterial PSII. The two salt-bridges are most likely conserved and associated with the channel orientation (Fig. 2c).

Fig. 2
figure 2

Location of the cyanobacterial O4-PsbU (blue mesh) and plant O4-PsbP (green mesh) channels. a Overview. b Superposition of the cyanobacterial PsbU channel space into the plant cryo-EM structure (Wei et al. 2016). c Superposition of the plant PsbP channel space into the cyanobacterial structure (Umena et al. 2011). Residues that interrupt the channels are circled in red

Full size image
Table 1 Protein subunits that form the water chains in the cyanobacterial (Umena et al. 2011) and plant (Wei et al. 2016) PSII structures
Full size table

The entire lengths and radii of the O4-PsbU and O4-PsbP channels were similar (Fig. 3). The radii of both the channels were smaller than ~1.4 Å in most of the regions, implying that water molecules are arranged as a single chain. These analyses suggest that plant PsbP may correspond to cyanobacterial PsbU in terms of connecting the O4-water chain in the D1 protein with the PSII protein bulk surface.

Fig. 3
figure 3

Cavity radii along the cyanobacterial O4-PsbU (upper left), plant O4-PsbP (upper right), cyanobacterial O1-PsbU (bottom left), and plant O1-PsbP (bottom right) channels (in Å)

Full size image

The channel space of the O4-water chain and the connecting O4-PsbU channel comprise several loop regions located in the protein–subunit interface with D1, CP43, and PsbU, e.g., residues D1−167 to 170, D1−333 to 337, CP43−333 to 339, CP43−353 to 357, PsbU−13 to 17, PsbU−21 to 22, PsbU−30 to 34, and PsbU−91 to 102. To evaluate the structural stability of the loop regions, MD simulations were performed for the cyanobacterial PSII complex (Umena et al. 2011). The calculated B-factors in the D1, CP43, and PsbU in the cyanobacterial PSII complex were highly associated with the original B-factors of the crystal structure (Fig. 4). Remarkably, the MD simulations suggested significantly high structural stability in the loop regions of D1, CP43, and PsbU proteins (Fig. 4). The calculated B-factors of the backbone Cα atoms in the loop regions were as low as those in the helices (e.g., trans-membrane helices D1−34 to 53, D1−110 to 136, D1−143 to 165, D1−196 to 221, and D1−268 to 292). Although MD simulations were not performed for the plant cryo-EM structure (Wei et al. 2016), high structural stability would be expected for the plant O4-water chain because of the structural similarity between the D1 and CP43 protein subunits in the cyanobacterial and plant PSII structures.

Fig. 4
figure 4

Calculated B-factors of the backbone Cα atoms in the cyanobacterial D1, CP43, and PsbU proteins, using the cyanobacterial 1.9 Å structure (closed circles) (Umena et al. 2011). The protein-loop regions of D1, CP43, and PsbU (D1−167 to 170, D1−333 to 337, CP43−333 to 339, CP43−353 to 357, PsbU−13 to 17, PsbU−21 to 22, PsbU−30 to 34) are colored red, while the other regions are colored black. For comparison, the B-factors stated in the PDB file (PDB ID: 3ARC) are shown (black open circles). Calculated B-factors are highly correlated with the original B-factors of the crystal structure

Full size image

The O1-water chain and the connecting PsbU/V and PsbP channels

The cyanobacterial PSII crystal structure shows that the O1 site of the Mn4Ca cluster has a chain of H-bonded water molecules, the O1-water chain (Fig. 5) (Umena et al. 2011). Channel space analysis suggested that the O1-water chain in D1, which terminated near D1-Glu329, was linked with the channel that initiated near the CP43-Glu413…PsbV-Lys47 salt-bridge (NPsbV−Lys47–OCP43−Glu413 = 2.8 Å) and proceeded toward the bulk surface via PsbU and PsbV (O1-PsbU/V channel, Fig. 5). Remarkably, the corresponding water chain and the linked channel were conserved in the plant PSII; the O1-water chain was linked with the channel that initiated near the CP43-Glu413…PsbP-Lys170 salt-bridge (NPsbP−Lys170–OCP43−Glu413 = 2.7 Å) and proceeded toward the bulk surface via PsbP (O1-PsbP channel, Fig. 5). From the comparison, it seems likely that the salt-bridge between CP43-Glu413 and PsbP-Lys170 in plant PSII is structurally conserved as the salt-bridge between CP43-Glu413 and PsbV-Lys47 in cyanobacterial PSII.

Fig. 5
figure 5

Cyanobacterial O1-PsbU/V (blue surface, left) and plant O1-PsbP (green surface, right) channels. Dotted squares indicate the common D1-protein region. The PsbQ subunit is colored gray

Full size image

PsbV-Lys47 seems to be involved in the region that affects the redox potential (E m) of heme in PsbV (i.e., cytochrome c550), in response to binding of the isolated PsbV protein to the PSII complex (Ishikita and Knapp 2005). The exact role of cytochrome c550 in the PSII complex is unclear, but the E m value of the heme differs significantly, by ~160 mV between the isolated cytochrome c550 form (Kerfeld et al. 2003) (–240 mV) and the PSII-bound cytochrome c550 form (–80 mV) (Roncel et al. 2003). It has been pointed out that the difference in the protein backbone C=O orientations in the two PsbV forms could explain 113 mV of the total E m difference, the majority of which (~80 mV) originated from the region of PsbV− 43 to 53 (Ishikita and Knapp 2005). In particular, the backbone C=O orientation differs significantly at PsbV-Lys47 in the two cytochrome c550 forms (Ishikita and Knapp 2005), which is obviously due to the formation of the CP43-Glu413…PsbV-Lys47 salt-bridge upon binding of the PsbV subunit to the PSII complex. The cyanobacterial CP43-Glu413…PsbV-Lys47 salt-bridge seems to correspond to the plant CP43-Glu413…PsbP-Lys170 salt-bridge (Fig. 5).

The cyanobacterial O1-PsbU/V and plant O1-PsbP channels have larger channel radii and shorter channel lengths than the cyanobacterial O4-PsbU and O4-PsbP channels (Fig. 3). In the narrow pore of the proton-conducting O4-water chain, 8 or 9 water molecules, in particular the initial four O moieties on the proton donor side (i.e., O4…OW539…OW538…OW393), are tightly H-bonded (Takaoka et al. 2016). The larger channel radii in the O1-PsbU/V and O1-PsbP channels imply that water molecules are probably more mobile than those in the O4-PsbU and O4-PsbP channels. Therefore, the O1-PsbU/V and O1-PsbP channels are likely to serve as a water-intake channel rather than a proton-conducting channel.

Protonation states near the channels in extrinsic proteins

The extrinsic proteins, cyanobacterial PsbU and PsbV proteins and plant PsbP and PsbQ proteins, are rich in charged residues. The calculated protonation states of titratable residues obtained using the cyanobacterial PSII structure indicated that in cyanobacterial PsbU and PsbV, all charged residues were in the standard protonation states, i.e., protonated basic and deprotonated acidic residues. However, the calculated protonation states of titratable residues obtained using the plant cryo-EM structure (Wei et al. 2016) indicated that 9 charged residues in PsbP and 7 charged residues in PsbQ were not in the standard protonation states, i.e., protonated acidic residues and deprotonated basic residues (Table 2). None of these residues were located along the O4-PsbP channel, while three of them, PsbP-Lys166, PsbP-Lys173, and PsbQ-Asp28, were located along the O1-PsbP channel.

Table 2 Protonated acidic residues and deprotonated basic residues obtained using the cryo-EM structure (Wei et al. 2016) and solving the Poisson–Boltzmann equation
Full size table

PsbP-Lys166 is exposed to the protein bulk surface (i.e., protonated) in the isolated form of the PsbP crystal structure (isolated PsbP structure) (Ifuku et al. 2004), whereas it was calculated to be deprotonated in the plant cryo-EM structure. In the plant cryo-EM structure, PsbP-Asp165, next to PsbP-Lys166, is involved in the salt-bridge with CP43-Arg320 (OAsp165–NArg320 = 3.1 Å) at the interface of PsbP with CP43 (Wei et al. 2016). This could be the underlying reason why in the plant cryo-EM structure PsbP-Lys170 needs to be oriented toward PsbP-Lys166 (NLys166–NLys170 = 3.9 Å) in the interface with D1 and CP43, leading to deprotonation of PsbP-Lys166 (Table 2). The side-chain orientation of PsbP-Lys170 differs significantly between the isolated PsbP (Ifuku et al. 2004) and plant cryo-EM (Wei et al. 2016) structures, whereas the side-chain orientations of PsbP-Lys166 and PsbP-Asp165 resemble in the two structures. PsbP-Asp165 forms an H-bond with PsbP-His144 in the two structures, the presence of which was already suggested by Ido et al. (Ido et al. 2012; 2014) before the plant cryo-EM structure was reported.

The following three cases may be considered as the plausible reasons for the apparently energetically unstable face-to-face side-chain orientation of the PsbP-Lys166/PsbP-Lys170 pair:

(i) Case 1: the side-chain orientations are unreasonable in the plant cryo-EM structure

The apparently energetically unstable face-to-face side-chain orientation of the PsbP-Lys166/PsbP-Lys170 pair might possibly be due to structural uncertainties in the plant cryo-EM structure analyzed at 3.2 Å resolution (Wei et al. 2016). To test this hypothesis, we optimized the side chains of the D1, CP43, PsbP, and PsbQ proteins in the plant cryo-EM structure at the molecular mechanical level (MM-optimized structure, see SI for the atomic coordinates). In the MM-optimized structure, PsbP-Lys166 formed a salt-bridge with D1-Glu329 and PsbP-Lys173 formed a salt-bridge with PsbQ-Asp28 (Fig. 6). These results imply that some side-chain orientations are energetically unstable in the plant cryo-EM structure (Wei et al. 2016). Notably, the shape of the plant O1-PsbP channel essentially remained unchanged upon geometry optimization of the side chains (Fig. 6). The plant O1-PsbP channel seems to exist in plant PSII, irrespective of structural uncertainties in the plant cryo-EM structure.

Fig. 6
figure 6

Side-chain orientations in the original plant cryo-EM structure (Wei et al. 2016) (cyan) and in the MM-optimized geometry (magenta) near the O1-PsbP channel. In contrast to the original plant cryo-EM structure, the PsbP-Lys166/D1-Glu329 pair and the PsbP-Lys173/PsbQ-Asp28 pair formed salt-bridges (dotted lines) in the MM-optimized geometry. The shape of the plant O1-PsbP channel was essentially the same in the original plant cryo-EM structure (green mesh) and the MM-optimized geometry (magenta mesh). See SI for the atomic coordinates of the MM-optimized structure

Full size image

(ii) Case 2: the side-chain orientations are reasonable in the plant cryo-EM structure, and the basic residues listed in Table 2 are actually deprotonated in PsbP of the PSII complex

Face-to-face side-chain orientation of the PsbP-Lys166/PsbP-Lys170 pair might be the case for the plant cryo-EM structure. However, it should also be noted that chemical modification of PsbP-Lys166 and PsbP-Lys170 affected the PsbP binding to the PSII complex (Tohri et al. 2004), implying that the two Lys residues form salt-bridges with acidic residues in the other protein subunits, in contrast to the geometry of the plant cryo-EM structure (Wei et al. 2016).

(iii) Case 3: the side-chain orientations are reasonable in the plant cryo-EM structure, but the basic residues are not deprotonated

This would be possible if anions (e.g., Cl−), which have not been identified in the plant cryo-EM structure, are present near the basic residues. The presence of Cl− would also stabilize the positively charged PsbP-Lys166/PsbP-Lys170 pair in the plant cryo-EM structure. This case seems to be energetically more favorable than case 2 and might possibly be associated with a role of PsbP in Cl− retention (Ghanotakis et al. 1984a). Based on observations of the isolated PsbP crystal structure, Ifuku et al. pointed out that structural rearrangement, possibly including these Lys residues that may occur upon the PsbP binding to the PSII complex, could be required for Cl− retention (Ifuku et al. 2004).

The calculated distribution pattern of Cl− covers the two Cl− binding sites, Cl-1 (Fig. 7c) and Cl-2 (Fig. 7d), identified in the cyanobacterial PSII crystal (Umena et al. 2011) and plant cryo-EM (Wei et al. 2016) structures. However, the Cl− distribution pattern of the plant cryo-EM structure obtained using 3D-RISM with Placevent suggested that Cl− is unlikely to exist in the PsbP-Lys166/PsbP-Lys170 moiety in the plant cryo-EM structure (Fig. 7b). The absence of the Cl− distribution in the PsbP-Lys166/PsbP-Lys170 moiety seems to be due to the insufficient binding space for Cl−. Possible Cl− binding sites in PsbP could be elsewhere, including the inner space of the O1-PsbP channel, as indicated in Fig. 7b. These results suggest that Cl− binding is determined not only by the positive charges in the protein environment, but also by the binding space size.

Fig. 7
figure 7

3D distributions of Cl− are shown as orange mesh obtained via 3D-RISM a in the cyanobacterial O1-PsbU/V and b the plant O1-PsbP channels. The focusing space between the PsbP-Lys166 and PsbP-Lys170 side chains in the plant cryo-EM structure (Wei et al. 2016) is indicated using red dotted circle c near the Cl-1 and d Cl-2 ions identified in the cyanobacterial PSII crystal structure (Umena et al. 2011). The backbone N atoms that provide the Cl− binding sites are labeled in brackets. The threshold of 3D distribution functions is 1.0 in all cases

Full size image

Considering the three cases, it seems most likely that the unusual protonation states found for the residues listed in Table 2 are highly associated with structural uncertainties in the cryo-EM structure at a resolution of 3.2 Å (i.e., case 1), because the orientations of side chains cannot be determined unambiguously at this level of resolution. Indeed, the energetically optimized side-chain orientations in the MM-optimized structure show that protonated PsbP-Lys166 forms a salt-bridge with ionized D1-Glu329 and protonated PsbP-Lys173 forms a salt-bridge with ionized PsbQ-Asp28 (Fig. 6). It can be concluded that the unusual protonation states of PsbP-Lys166, PsbP-173, and PsbQ-Asp28 (Table 2) originate from the energetically unfavorable side-chain orientations that prevent the formation of salt-bridges in the original geometry of the plant cryo-EM structure.

Conclusions

The results of the present analysis suggested that the O4-PsbU channel in cyanobacterial PSII that proceeds from the O4 site of the Mn4Ca cluster toward the protein bulk surface was structurally conserved as the O4-PsbP channel in plant PSII (Figs. 1, 2). The salt-bridge between CP43-Lys339 and PsbP-Asp137 in the plant O4-PsbP channel corresponds to the salt-bridge between CP43-Lys339 and PsbU-Asp96 in the cyanobacterial O4-PsbU channel (Fig. 2). The radii of both channels were smaller than ~ 1.4 Å in most of the regions, implying that water molecules are arranged as a single chain (Fig. 3). The O1-PsbU/V channel in cyanobacterial PSII was also structurally conserved as the O1-PsbP channel in plant PSII (Fig. 5). The salt-bridge between CP43-Glu413 and PsbP-Lys170 near the plant O1-PsbP channel corresponds to that between CP43-Glu413 and PsbV-Lys47 near the cyanobacterial O1-PsbU/V channel. These salt-bridges seem to contribute to the observed high structural stability of the channel regions (Fig. 4). None of these salt-bridges exist in the channel inner spaces, as demonstrated in Fig. 3.

In the calculated protonation states, PsbP and PsbQ showed several deprotonated basic residues and protonated acidic residues in the original side-chain orientations of the plant cryo-EM structure (Table 2). The protonation states of the deprotonated basic residues and protonated acidic residues in the two extrinsic proteins (e.g., deprotonated PsbP-Lys166, Table 2) would be altered in the presence of Ca2+ and Cl−. This effect might possibly be associated with the retention of Cl− (e.g., deprotonated PsbP-Lys166, Table 2) and Ca2+ in PsbP and PsbQ, which are the known roles of the two proteins (Ghanotakis et al. 1984a). However, Cl− is unlikely to exist in the PsbP-Lys166/PsbP-Lys170 moiety in the plant cryo-EM structure due to the insufficient binding space for Cl− (Fig. 7b). The unusual protonation states of the residues listed in Table 2 are most likely to originate from structural uncertainties in the cryo-EM structure at a resolution of 3.2 Å (Wei et al. 2016). The energetically optimized side-chain orientations suggest that protonated PsbP-Lys166 forms a salt-bridge with ionized D1-Glu329 and protonated PSbP-Lys173 forms a salt-bridge with ionized PsbQ-Asp28 (Fig. 6), in contrast to the original geometry of the plant cryo-EM structure.

The existence of the O4-PsbP channel in plant PSII and the O4-PsbU channel in cyanobacterial PSII, irrespective of the low structural similarities between PsbP and PsbU, is remarkable. If PsbP were not present in plant PSII, the O4-water chain might be able to release protons directly toward the protein bulk surface. However, the removal of the extrinsic, highly charged PsbP subunit also affects the PSII protein assembly. Indeed, thermoluminescence studies showed that the Mn4Ca cluster was significantly unstable in the absence of PsbP (Ifuku et al. 2005). Remarkably, in cyanobacterial PSII, the Mn4Ca cluster was also significantly unstable in the absence of PsbU (Inoue-Kashino et al. 2005). The properties of the proton-conducting O4-water chain (Saito et al. 2015; Takaoka et al. 2016), which directly forms a short H-bond with the Mn4Ca cluster (Umena et al. 2011; Suga et al. 2015), would also be affected by removal of these extrinsic proteins. To maintain both the stability of the Mn4Ca cluster and the activity in the proton transfer along the O4-water chain, employing PsbP (or PsbU) that possesses the O4-PsbP channel (or the O4-PsbU channel) would be important. Tomita et al. reported that PsbP, neither PsbQ nor PsbO, affects the properties of the Mn4Ca cluster, possibly protein conformation near the Mn4Ca cluster, without altering the ligand structure (Tomita et al. 2009), e.g., assumingly, with altering the H-bond network or the ion localizations. The PsbP channels identified in the present study might be associated with the observed direct functional link of PsbP (Tomita et al. 2009) with the Mn4Ca cluster.