Tyrosine Deprotonation and Associated Hydrogen Bond Rearrangements in a Photosynthetic Reaction Center

Photosynthetic reaction centers from Blastochloris viridis possess Tyr-L162 located mid-way between the special pair chlorophyll (P) and the heme (heme3). While mutation of the tyrosine does not affect the kinetics of electron transfer from heme3 to P, recent time-resolved Laue diffraction studies reported displacement of Tyr-L162 in response to the formation of the photo-oxidized P+•, implying a possible tyrosine deprotonation event. pK a values for Tyr-L162 were calculated using the corresponding crystal structures. Movement of deprotonated Tyr-L162 toward Thr-M185 was observed in P+• formation. It was associated with rearrangement of the H-bond network that proceeds to P via Thr-M185 and His-L168.


Introduction
In biological systems, tyrosine residues often play an important role in functioning as a redox active group and mediating electron transfer. In photosystem II (PSII), electronic excitation of the chlorophyll a P680 P D1/D2 pair leads to formation of positively charged P680 +N as a consequence of electron transfer to the secondary quinone via the accessory chlorophyll a, a pheophytin a, and the primary quinone. The resulting P680 +N is reduced by D1-Tyr161 (Y Z ) through electron transfer events from the Mn 4 CaO 5 cluster [1]. The PSII reaction center that consists of D1 and D2 subunits has considerably large structural similarity with photosynthetic reaction centers from purple bacteria (bRC) [2]. In bRC from Blastochloris viridis, the corresponding chlorophyll pair is the bacteriochlorophyll b (BChlb) P L/M pair P960 (P). The photooxidized P +N state that is generated as a consequence of electronic excitation of P960 can be reduced by electron transfer from the nearest heme group (heme3) in the adjacent tetraheme subunit. The role of a highly conserved residue, Tyr-L162, has been long discussed [3,4] due to its unique position halfway between heme3 and P ( Figure 1). Nevertheless, in kinetic studies, the electron transfer rate from heme3 to P was not altered significantly in the Tyr-L162 mutations. Thus, it was concluded that neither tyrosine nor aromaticity is required for fast electron transfer from heme3 to P [5,6]. Hence, functionally dominant electron transfer pathways may not proceed via Tyr-L162.
On the other hand, displacement of Tyr-L162 by 1.3 Å toward P +N was very recently reported in the light-exposed crystal structure (light structure) with respect to the dark-state structure (dark structure) in time-resolved Laue diffraction analysis. Wöhri et al. interpreted that negatively charged and deprotonated Tyr-L162 was attracted to the P +N positive charge [7]. Furthermore, they proposed that Tyr-L162 deprotonation may be important for the mechanism of electron transfer from heme3 to P via stabilization of heme3 in the oxidized state. A simple free energy calculation on the basis of molecular dynamics simulation is useful as an initial survey to roughly estimate the energetics of the tyrosine deprotonation. However, the energy profile is generally calculated in the fixed protonation pattern of the protein titratable residues. In particular, bRC possesses a number of titratable residues that can alter the protonation states in response to changes in redox states or protonation states of the cofactors or residues [8,9,10,11,12]. Apparently, the pK a value of Tyr-L162 (pK a (Tyr-L162)) is neither experimentally measured nor explicitly calculated in Ref. [7], without considering the equilibrium in the strongly coupled protonation states of titeatable residues in the bRC protein environment.
Although there are crystal structures of bRC from Blastochloris viridis at higher resolutions, so far only the crystal structure by Wöhri et al. [7] was proposed to correspond to the photoactivated form. Notably, in their original structural studies [7], they discussed subtle differences in the orientation of the tyrosine side chain between the photoactivated form (PDB; 2X5V) and the dark form (PDB 2X5U), irrespective of the resolutions at ,3 Å . Thus, it is a request from the community, at least once to evaluate i) pK a (Tyr-L162) in the original protein geometry of the photoactivated form and ii) what residues/groups contribute to downshift pK a (Tyr-L162). As a driving force of the tyrosine deprotonation, the P +N state formation is definitely a key factor. However, there are also other titratable residues in the neighborhood of P. It is unclear whether protonation state changes of other titratable residues may occur in response to the P +N formation, or whether deprotonation of other titratable residues compensate for the influence of P +N on pK a (Tyr-L162).
To evaluate the energetics of Tyr-L162 deprotonation in the P +N state formation, pK a (Tyr-L162) were calculated using the corresponding protein crystal structures, by solving the linear Poisson-Boltzmann equation with consideration of the protonation states of all titratable sites in the entire bRC protein. Using this approach, one will be able to sufficiently consider the equilibrium in protonation states of all titratable groups in bRC [9,11] and clarify the factors (e.g., residues, cofactors, atomic charges, or hydrophobicity of the protein environment) that shift pK a (Tyr-L162) in the protein environment.

Movement of deprotonated tyrosine
To investigate the possible presence of deprotonated tyrosine, Tyr-L162 was treated in its deprotonated form, and its geometry was energetically optimized with CHARMM in the P +N state. As a consequence, deprotonated Tyr-L162 moved further toward Thr-M185 ( Figure 2): the H-bond distance between Tyr-L162 and Thr-M185 (O Tyr-L162 -O Thr-M185 ) was 2.7 Å in the resulting geometry with deprotonated Tyr-L162 (Y deprot position), which was 0.6 Å shorter than that in the light structure (Y light position). Although the resulting Y deprot position was not exactly identical to the Y light position, this result implies that Tyr-L162 deprotonation leads to tyrosine movement from the one in the dark structure (Y dark position) to the Y light position. Interestingly, the O Tyr-L162 -O Thr-M185 distance obtained with deprotonated Tyr-L162 is 2.7 Å (in the Y deprot structure).

Tyr-L162 deprotonation induces H-bond network rearrangements
In addition to Tyr-L162 movement, a striking rearrangement in the H-bond network containing P and Tyr-L162 was observed in the transition from the initial uncharged P 0 and protonated Tyr-L162 state (P 0 Y) to the photo-oxidized P +N and deprotonated Tyr- In contrast to the P 0 Y dark state, orientation of the H-bond network is completely different in the P +N Y deprot 2 state, since the hydroxyl OH group of Thr-M185 is subject to forming an H-bond with the deprotonated Tyr-L162 (O Tyr-L162 -O Thr-M185 distance = 2.7 Å ) to stabilize the negative charge (Figure 3, right). The absence of the hydroxyl H atom near His-L168 promotes protonation of the His-L168 Nd site (Table 1). In accordance with reorientation of the Thr-M185 hydroxyl group, H atoms of a water molecule at an H-bonding distance with Tyr-L162 were also reoriented toward the deprotonated Tyr-L162. As a consequence, the OH dipole orientations were altered, forming the O Tyr-L162 NNN H-O Thr-M185 NNN H-N His-L168 network (Figure 3, right). pK a (Tyr-L162) value shift from the P 0 Y dark to the P +N Y deprot 2 state pK a (Tyr-L162) was calculated to be 22 in the P 0 Y dark state ( Table 1), indicating that this residue will never be deprotonated in the dark structure. The significantly high pK a (Tyr-L162) value of 22, which is even higher than that in aqueous solution (,10), is mainly due to the presence of acidic residues in the bRC that upshift pK a (Tyr-L162), e.g., Asp-M182, Glu-C254, Asp-L155, and Glu-M171 ( Table 2). The presence of these negatively charged acidic residues upshifts pK a (Tyr-L162) and thus does not energetically allow deprotonated Tyr-L162 formation.
In contrast to the P 0 Y dark state, P +N Y 2 state formation leads to a drastic decrease in pK a (Tyr-L162). In particular, the P +N Y deprot 2 state possesses the deprotonated Tyr-L162 since pK a (Tyr-L162) = 6.7 (Table 1. Two major factors contribute to decreased pK a (Tyr-L162): i. H-bond pattern change. The most crucial groups that decrease pK a (Tyr-L162) are Thr-M185 and a water molecule. They alter the H-bond pattern with respect to Tyr-L162 in response to the P +N Y deprot 2 state formation ( Figure 3). As a consequence, H-bond alternation in Thr-M185 and a water molecule decrease pK a (Tyr-L162) by 8 and 4 in the P +N Y deprot 2 state (relative to the P 0 Y dark state), respectively (Table 3).
ii. Direct electrostatic influence of a positive charge in the photo-oxidized P +N state. The positive charge on P +N  contributes to stabilization of the deprotonated Tyr-L162 form, downshifting pK a (Tyr-L162) by 4.2 (2.7 from P A and 1.5 from P B ) in the P +N Y deprot 2 state (Table 3). The influence of P +N on pK a (Tyr-L162) did not essentially differ in the Y dark , Y light , and Y deprot positions (Table 3).

Concluding Remarks
Deprotonation of Tyr-L162 resulted in the displacement of the side chain, lowering the pK a value to 6.7. Movement of deprotonated Tyr-L162 toward Thr-M185 was observed in P +N formation. It was associated with rearrangement of the H-bond network that proceeds to P via Thr-M185 and His-L168.

Atomic coordinates and charges
For performing computations of bRC from Blastochloris viridis, crystal structures in the photoactivated form (protein data bank (PDB); 2X5V) [7] were used. A crystal structure corresponding to the dark state is available (PDB 2X5U), but this crystal structure does not contain water molecules that can be seen in the photoactivated crystal structure. Furthermore, the conformer labeled with A in the photoactivated crystal structure is essentially identical to the dark state crystal structure in terms of the Tyr-L162 position while the conformer labeled with B in the photoactivated crystal structure is considered to correspond to the photoactivated state. Thus, in the present study, atomic coordinates for the A and B conformers (PDB 2X5V) were used as the dark and light structures, respectively.
The atomic coordinates were obtained using the same procedures used in previous studies (e.g., Refs. [11,13,14]). The positions of H atoms were energetically optimized with CHARMM [15] by using the CHARMM22 force field. While carrying out this procedure, the positions of all non-H atoms were fixed, and the standard charge states of all the titratable groups were maintained, i.e., basic and acidic groups were considered to be protonated and deprotonated, respectively. All of the other atoms whose coordinates were available in the crystal structure were not geometrically optimized. To investigate a possible movement of deprotonated Tyr-L162 (i.e., to yield the Y deprot position, see the later part), atomic coordinates for the minimum set of relevant residues, i.e., Tyr-L162, Thr-M185, and a water molecule (HOH M 2001 in PDB: 2X5V) were released and geometrically optimized (Table S1 for atomic coordinates). As a general and uniform strategy, other crystal waters are removed in our computations [16] because of the lack of experimental information for hydrogen atom positions. Cavities resulting after removal of crystal water are uniformly filled with solvent dielectric of e = 80.  Table 1. Calculated pK a (Tyr-L162, His-L168, and Glu-C254) and redox potential (Tyr-L162) values in mV and pK a units, respectively.  Table 2. Main residues that contribute to increase of pK a (Tyr-L162) in pK a units (i.e., residues that stabilize the Tyr-L162 protonation state). Atomic partial charges of the amino acids were adopted from the all-atom CHARMM22 [15] parameter set. The charges of protonated acidic oxygen atoms in Asp and Glu were both increased symmetrically by +0.5 unit charges to account implicitly for the presence of a proton. Similarly, instead of removing a proton in the deprotonated state, the charges of all protons of the basic groups of Arg and Lys were diminished symmetrically by a total unit charge. For residues whose protonation states are not available in the CHARMM22 parameter set, appropriate charges were computed [17]. The atomic charges for the redox-active tyrosine (Tyr-L162) were adopted from the previous applications [16,18] (deprotonated with negative charge (Y 2 ), and protonated with neutral charge (YH)). The atomic charges of BChlb and bacteriopheophytin b (BPheob) were determined from the electronic wave functions obtained with the density functional (DFT) module (B3LYP) in Gaussian03 [19] with 6-31G** basis set by fitting the resulting electrostatic potential in the neighborhood of these molecules by the RESP procedure [20] (Tables S2 and  S3). To represent the charge states of the light-induced oxidized special pair P +N , a unit positive charge was distributed with a ratio of P A +N /P B +N = 2/1 derived from ENDOR studies [21] as done in the previous application [22].

pKa and protonation pattern
The present computation is based on the electrostatic continuum model by solving the linear Poisson-Boltzmann (LPB) equation with the MEAD program [23]. To facilitate a direct comparison with previous computational results, identical computational conditions and parameters were used (e.g., Refs. [11,13,14]) such as atomic partial charges and dielectric constants. To obtain absolute pK a values of a target site (e.g. pK a (Tyr-L162)), the electrostatic energy difference was calculated between the two protonation states, protonated and deprotonated, in a reference model system using a known experimentally measured pK a value. The difference in the pK a value of the protein relative to the reference system was added to the known reference pK a value. Experimentally measured pK a values employed as references are 12.0 for Arg, 4.0 for Asp, 9.5 for Cys, 4.4 for Glu, 10.4 for Lys, 9.6 for Tyr [25], and 7.0 and 6.6 for deprotonation/ protonation at Ne and Nd atoms of His, respectively [26,27,28]. All of the other titratable sites were fully equilibrated to the protonation state of the target site during the titration. The Monte Carlo sampling for a titratable residue yielded the probabilities [protonated] and [deprotonated] of the two protonation states of the molecule. The pK a value was evaluated using the Henderson-Hasselbalch equation. A bias potential was applied to obtain an equal amount of both protonation states ([protonated] = [deprotonated]), yielding the pK a value as the resulting bias potential.

Error estimation
The procedures to compute pK a of titratable residues are equivalent to those of the redox potential for redox-active groups, although in the latter case, the Nernst equation is applied instead of the Henderson-Hasselbalch equation [29]. Therefore, the accuracy of the present pK a computations is directly comparable to that obtained for recent computations [16]. From the analogy, the numerical error of the pK a computation can be estimated to be about 0.2 pH units. Systematic errors typically relate to specific conformations that may differ from the given crystal structures.

Supporting Information
Table S1 Energetically minimized atomic coordinates of Tyr-L162 (Y deprot ), Thr-M185, and a water molecule. (DOC)  Table 3. Main residues that contribute to decrease of pK a (Tyr-L162) in the P +N Y 2 state formation in pK a units (i.e., residues that promote the Tyr-L162 deprotonation).