Induced conformational changes upon Cd²⁺ binding at photosynthetic reaction centers

Computational Procedures

Atomic Coordinates. We used crystal structures of the bRC from Rhodobacter sphaeroides for WT-bRC [Protein Data Bank (PDB) ID code 1AIG] (20) and Cd-bRC (PDB ID code 1DV3) (15), obtained under illumination (light-exposed structure), and D(L213)N single mutant with Q_B at the proximal binding position with respect to the nonheme Fe (PDB ID code 1RY5) (8). Atomic coordinates were obtained in the same way as in previous applications (9, 21–23). The positions of hydrogen atoms were energetically optimized with charmm (24) by using the charmm22 force field. During this procedure, the positions of all nonhydrogen atoms were fixed, and all titratable groups were kept in their standard charge state, i.e., basic groups including histidine were considered to be protonated and acidic groups to be ionized. The coordinates of all atoms available in the crystal structure were not optimized.

For both WT-bRC and Cd-bRC, there are also crystal structures available that were obtained under dark conditions (dark-adapted structures PDB ID codes 1AIJ and 1DV6, respectively) (15, 20). The light-exposed and dark-adapted structures were suggested to refer to charge separated and ground PQ_AQ_B states, respectively (20). The most remarkable difference among them is the position of Q_B, being at the proximal binding site in the former and the distal one in the latter with respect to the nonheme iron. Therefore, it was formerly proposed that the movement of Q_B from the distal site to the proximal site upon formation of the state is prerequisite for ET from to Q_B (20). However, the functional relevance of the Q_B movement is still a matter of debate. In recent Fourier transform IR studies, no Q_B movement was found. Hence, both Q_B and seem to be located at the proximal site (25–27). Furthermore, a time-resolved crystallographic study resulted in no quinone motion upon illumination and indicated that Q_B binds only in the proximal binding site (28). Mainly because of these experimental results that question the physiologically relevance of the Q_B distal position, we prefer the light-exposed structures to describe the PQ_AQ_B and states and use them consistently in the present study.

Atomic Charges. Atomic partial charges of the amino acids were adopted from the all-atom charmm22 (24) parameter set. For cofactors and residues whose charges are not available in charmm22, we used atomic partial charges from previous applications (21–23). For the atomic charges for the Cd²⁺–protein complex in Cd-bRC, see Supporting Text and Table 3, which are published as supporting information on the PNAS web site.

Computation of Protonation Pattern and pK_a. The computation of the energetics of the protonation pattern is based on the electrostatic continuum model, in which the linearized Poisson Boltzmann equation is solved by the program mead from Bashford and Karplus (29). To sample the ensemble of protonation patterns by a Monte Carlo method, we used our own program karlsberg (http://agknapp.chemie.fu-berlin.de/karlsberg). The dielectric constant was set to ε_P = 4 inside the protein and ε_W = 80 for solvent and protein cavities corresponding to water (see Supporting Text). All computations were performed at 300 K, pH 7.0, and an ionic strength of 100 mM. The linearized Poisson Boltzmann equation was solved by a three-step grid-focusing procedure with a starting grid resolution of 2.5 Å, an intermediate grid resolution of 1.0 Å, and a final grid resolution of 0.3 Å. Monte Carlo sampling yields the probabilities [A ^–] and [AH] of the deprotonated and protonated state of the titratable residue A, respectively. With the Henderson–Hasselbalch equation, the pK_a can be calculated as the pH where the concentration of [A ^–] and [AH] are equal. The protonation patterns were computed for the redox state, if not otherwise stated. The procedures to obtain 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 (31). Therefore, the accuracy of the present pK_a computations is directly comparable to that obtained for recent computations on redoxactive cofactors (21–23). From the analogy, the numerical error of the pK_a computation can be estimated to be ≈0.2 pH units. Systematic errors, which typically relate to specific conformations that may differ from the given crystal structures, can sometimes be considerably larger. Because the present study was done under the same conditions as our previous pK_a computation for bRC, further details on error estimates and comparisons with the previous results can be obtained in ref. 9.

Results and Discussion

Protonation Pattern Coupled to PT in Kinetic Phase 1 and 2. To study PT inhibition in bRC upon Cd²⁺ binding, we calculated the protonation pattern of titratable residues coupling to the following charge states: (i) and for PT in kinetic phase 1 (Eq. 1a) and (ii) and for PT in kinetic phase 2 (Eq. 2a). With the electrostatic method used in the present study we can evaluate energetic differences in protonation states between initial states before PT and the corresponding final states, while ignoring activation barriers. In WT-bRC, both PT events considered here, the one to Glu-L212 belonging to kinetic phase 1 and the one belonging to kinetic phase 2, did not result in essential changes of protonation pattern (Table 1). This finding contrasts with the PT to Glu-L212 in Cd-bRC, which is accompanied by deprotonation of Asp-L213 by 0.74 H⁺ and Glu-M236 by 0.37 H⁺. Astonishingly, even after removing Cd²⁺ in the Cd-bRC structure, the same amount of deprotonation for these two residues was obtained (Table 1), which indicates that the Cd²⁺ ionic effect on the protonation pattern is small (see Main Electrostatic Effects of Cd²⁺ Binding). These results imply that to protonate Glu-L212 during kinetic phase 1 Cd-bRC requires an additional energy to deprotonate Asp-L213 and Glu-M236, which is not needed in the corresponding PT in WT-bRC. Deprotonation of Asp-L213 (by 0.89 H⁺) was also observed for PT to in Cd-bRC belonging to kinetic phase 2, although the protonation state of Glu-M236 remained unchanged (Table 1).

It is known that mutation of Asp-L213 to Asn in bRC inhibits PT events more drastically [ and decrease by a factor of 10 and 6,000, respectively] than Cd²⁺ binding, resulting in a photosynthetically incompetent bRC (5, 6). The calculated protonation pattern for the D(L213)N mutant showed deprotonation of Glu-M236 by 0.57 H⁺ or 0.10 H⁺ to yield PT to Glu-L212 or to , respectively (Table 1). These observed deprotonations of Glu-M236 in the D(L213)N mutant were larger than those computed for Cd-bRC (Table 1). We assume that in vivo the corresponding deprotonation requires considerable energy, which consequently might prevent deprotonation. We interpret deprotonation of these residues as their “resistance” to PT. The computational results in the present study imply energetically unfavorable conditions for PT involving Asp-L213 or Glu-M236 in these PT-inhibited bRCs [Cd-bRC and D(L213)N mutant bRC].

Proton Release from His-H126 and His-H128 upon Cd²⁺ Binding. The crystal structure of Cd-bRC revealed the Cd²⁺ binding site, which consists of Asp-H124, His-H126, and His-H128 (11, 15). Based on the measurement for the dissociation constant of Cd²⁺, Paddock et al. (18) suggested that below pH 6.0 two protons (2 H⁺) were released upon binding of Cd²⁺ from His-H126 and His-H128. This result is comparable to the proton release by 1.5 H⁺ at pH 6.5 upon Ni²⁺ binding studied by Gerencsér and Maróti (14). Based on the WT-bRC structure we computed the protonation of His-H126 and His-H128 to be 0.9 and 0.9 H⁺, respectively, at pH 6.0 in the Q_AQ_B state. Removing Cd²⁺ from the Cd-bRC structure, we obtained under the same conditions practically the same protonation, namely 0.9 and 0.8 H⁺ for His-H126 and His-H128, respectively. If Cd²⁺ binds at these two histidines, these protons will be released, which corroborates the experimentally observed proton release of 1.5–2 H⁺ at pH ≈6. Indeed, His-H126 and His-H128 are the residues that changed their protonation states most dramatically upon Cd²⁺ deletion for Cd-bRC in the present study.

PT Pathways: pK_a Decrease upon Cd²⁺ Binding. It is commonly accepted that Cd²⁺ binding has little impact on the charge recombination rate of despite the positive Cd²⁺ charge near Q_B (10, 11, 13, 14). To elucidate the electrostatic influence of Cd²⁺ binding on residues involved in PT events, we calculated the pK_a for those residues along the PT pathways for the charge state. With Cd²⁺ binding at bRC, several acidic residues in the PT pathway showed significant decrease in pK_a by up to 3.9 units: for instance, 1.1, 1.3, 1.5, 1.8, 2.5, and 3.9 units for Glu-H230, Asp-L213, Asp-M17, Glu-M236, Asp-H170, and Asp-H119, respectively (Table 2).

These residues are of interest for two reasons. First, most of these residues are those that experience pK_a increases upon revertant mutations in bRC (9). The recovery of PT in revertant mutant bRC suggests that an alternative PT pathway involving Glu-H173 is operative in this case (2, 8, 9, 32). In comparison with WT-bRC (20), both Cd-bRC (15) and revertant mutant bRC (8) revealed significant structural changes at Glu-H173 and the near surroundings, which seemingly relates to variation in the activity of an additional PT pathway (i.e., being inhibited in WT-bRC and operative in both Cd-bRC and revertant mutant bRC). The present study suggests that residues affected by revertant mutants or Cd²⁺ binding are closely associated with each other, which points to the importance of these residues for the revertant mutant bRC, as proposed in a previous study (9).

Second, Gerencsér and Maróti (14) observed that , the rate of kinetic phase 1 [], is constant at low pH but starts to decrease above a critical pH value. This effect occurs at pH 9.2 in WT-bRC and is thus connected to a moiety of titratable residues with corresponding apparent pK_a, which was down-shifted by 1.8 pH units to 7.4 in Cd-bRC. On the other hand, we calculated a pK_a of 8.9 and 7.6 for Asp-L213 in WT-bRC and in Cd-bRC, respectively (Table 2). Taking into account the experimental evidence for the importance of Asp-L213 in PT (reviewed in ref. 1), we tentatively conclude that one of the main titratable groups observed in the study of Gerencsér and Maróti (14) is Asp-L213.

Identification of the -related pK_a is still an open question. The measured pH limit, above which starts to decrease, varies slightly with the experimental conditions used by Paddock et al. (33), Gerencsér and Maróti (14), or Takahashi and Wraight (34), being 8.5, 9.2, and 9.5, respectively. From the midpoint of the pH-dependent region, Paddock et al. (33) deduced an apparent pK_a value of 9.5 related to . On the other hand, Takahashi and Wraight (34) obtained two pK_a values, 9.5 and 11, which they deduced from both edges of the pH-dependent region and assigned to Q_B and , respectively. In the present study, we determine pK_a values at the protonation midpoint of a titratable group, which might be more related to the determination of the pK_a used by Paddock et al. (33). Mainly by mutational studies for either E(L212)Q or D(L213)N bRC, this apparent pK_a was believed to refer to Glu-L212 as a single residue (33, 34). The calculated pK_a values for Glu-L212 are 6.1 for Q_B and 9.4 for , indicating a proton uptake of this residue at pH 7, as observed in IR (35) and Fourier transform IR (36) studies. Nevertheless, the assignment of Glu-L212 alone to the apparent pK_a seems oversimplified, because Q_B is located in a region of strongly interacting titratable residues. The bRC of the revertant mutant E(L212)A/D(L213)A/R(M231)L from Rhodobacter capsulatus lacks the acidic residue Glu-L212. Nevertheless, the same pH dependence was observed (7, 37), implying that Glu-L212 cannot be the only residue determining the apparent pK_a for (14, 38, 39). Indeed, if the two pK_a values of 9.5 and 11 for Q_B and , respectively, were assigned just to Glu-L212 (34), a straightforward interpretation leads to the conclusion that this residue is not involved in proton uptake upon formation of the state at pH 7, which is in conflict with the stoichiometry of proton uptake observed in IR (35) and Fourier transform IR (36) studies and our calculated pK_a. Hence, a cluster of strongly interacting titratable residues near Q_B hinders a clear assignment of this apparent pK_a to a specific residue (discussed in refs. 14, 35, 38, and 39). Consistently, recent FTIR studies for E(L212)D/D(L213)E double mutant bRC also demonstrated a strong interaction between carboxylic acids specifically in the Q_B region, which cannot be interpreted with a simple, conservative concept of amino acid replacement effect between Glu and Asp (40). More recently, electron nuclear double resonance studies suggested a significant role of Asp-L213 in the rate-limiting step for kinetic phase 1 [] by rotating the hydroxyl group of Ser-L223 to ,† as predicted in electrostatic computations (22, 42). Unlike Glu-L212, Asp-L213 is generally considered not to be involved in proton uptake upon formation of the state, but this idea does not imply that Asp-L213 cannot participate in PT (reviewed in ref. 1).

We conclude that Asp-L213 is one of the main contributors to the apparent pK_a shift for upon Cd²⁺ binding, without excluding a substantial contribution of nearby interacting residues such as Glu-L212 or Asp-M17. This finding agrees with the deduced conclusion in former studies that not only Glu-L212 but also a cluster of strongly interacting titratable residues contribute to the pH dependence of (14, 38, 39).

Asp-L213, a Residue in PT Pathways. Formerly the decrease of upon Zn²⁺ binding was interpreted as being caused by the inhibition of the conformational gating step (10). However, more recent kinetic studies suggest that the decrease of upon metal binding (by 10-fold with respect to WT-bRC) is caused by the inhibition of PT to Glu-L212 rather than the conformational gating step (13). An equal amount of reduction in for the D(L213)N mutant was previously attributed to the inhibition of PT to Glu-L212 via Asp-L213 (i.e., proton uptake by Glu-L212 belonging to kinetic phase 1) (5) (see also Fig. 1a for the location of Asp-L213 and Glu-L212). From these experimental observations, we believe that Cd²⁺ binding inhibits the function of Asp-L213 as PT mediator in kinetic phase 1, which slows down the PT to Glu-L212. Because Asp-L213 is directly involved in the PT pathway to Q^– _B of kinetic phase 2, the inhibition of Asp-L213 function in PT should also reduce dramatically, thus changing the energetics of kinetic phase 2 from a ET driving force-dependent to driving force-independent reaction. The calculated pK_a down-shift of 1.5 for Asp-M17 upon Cd²⁺ binding (Table 2) may also partially contribute to the inhibition of PT to in kinetic phase 2. However, the influence of Asp-M17 on the PT rates is suggested to be smaller [only 2-fold decrease in upon the mutation from Asp-M17 to Asn], i.e., ET is still the rate-limiting step (12).

The drastic 10-fold decrease in and 6,000-fold decrease in upon mutation from Asp-L213 to Asn indicates the importance of the role of Asp-L213 in efficient PT events of WT-bRC (5, 6). In turn, this large impact of Asp-L213 on PT events implies already that, even in WT-bRC, only a small perturbation of this residue might result in serious retardation of the PT for both kinetic phases. Thus, we interpret the decrease in pK_a for Asp-L213 upon Cd²⁺ binding in the present study as being one of the important factors of the mechanism of PT inhibition. On the other hand, the distance between Cd²⁺ and Asp-L213 is ≈12 Å, which may be too large for a significant direct electrostatic interaction (Cd²⁺ ionic effect) (see Main Electrostatic Effects of Cd²⁺ Binding).

Remarkably, similar long-range electrostatic interactions were formerly proposed by Sebban et al. (39) with their study of the revertant mutant bRC from R. capsulatus, which possesses Arg-M231 and Asn-M43 at distances of 9 and 15 Å from Q_B, respectively. Each single mutation of Arg-M231 to Leu or Asn-M43 to Asp shifted the pK_a of titratable residues with respect to WT-bRC and resulted in a similar level of electrostatic influence on Q_B. They interpreted these long-range electrostatic influences as the consequence of a rearrangement of a combined salt-bridge/H-bond network propagated between the mutation site and Q_B, as an electrostatic domino effect (39). The comparison of the side-chain orientation in the crystal structures between Cd-bRC and WT-bRC clearly indicates the electrostatic domino effect among a number of residues along the water channels to Q_B (Fig. 1a). The side-chain reorientation of Asp-L213 upon Cd²⁺ binding is likely to propagate via Asp-H124, His-H126, His-H128, Asp-H170, Glu-H173, Asp-L210, and Asp-M17 (Fig. 1a). The majority of these residues were proposed to participate in the main PT pathway (1) to Glu-L212 or Q_B. These rearrangements of side chains are accompanied by displacement of crystal water molecules, which can be clearly seen near those residues participating in an H-bond network involving Asp-H124, His-H126, His-H128, Glu-H173, Asp-L210, Glu-L212, and Asp-M17 (Fig. 2, which is published as supporting information on the PNAS web site). The absence of two crystal water molecules near Glu-M173 in WT-bRC compared with Cd-bRC is probably related to its disordered side chain in the former structure, which is fixed upon Cd²⁺ binding (discussed in ref. 15). Because of its strong polarization effect, a water molecule is able to stabilize newly oriented side-chain positions efficiently. If there is an H-bond network nearby, such a movement of water molecule leads, in turn, to an immediate rearrangement of the H-bond pattern along this network. Therefore, the electrostatic domino effect can probably occur more preferentially if the residues are located near channel water molecules. We propose an electrostatic influence of Cd²⁺ binding on PT inhibition as follows. Cd²⁺ binding causes reorientation of the Cd²⁺-ligating residues Asp-H124, His-H126, and His-H128, which in turn results in side-chain reorientations of Asp-H170, Asp-M17, Asp-L210, and Glu-H173 mediated by water molecules along the water channel. In this way, rearrangement of the Asp-L213 side-chain can be induced by Cd²⁺ even at a distance of 12 Å, leading to its pK_a shift and inhibition of PT.

Main Electrostatic Effects of Cd²⁺ Binding. Despite the well known inhibition of PT by Cd²⁺ binding, the exact mechanism is a matter of debate. In the present study, we observe significant shifts of pK_a for residues along the PT pathways of WT-bRC or revertant mutant bRC, which are consistent with a pK_a inhibition mechanism (14). But, taking into account current criticisms against the pK_a inhibition mechanism (see discussions in ref. 17), we further examined the origin of the pK_a shifts obtained in the present study.

Quite generally, pK_a shifts of titratable groups in proteins, computed relative to the corresponding model compounds in solution, originate from electrostatic interactions with the protein environment. Cofactors like Cd²⁺ are part of this environment in the case of Cd-bRC. The total electrostatic interaction of Cd²⁺ with a titratable residue can be split into (i) the bare influence from the positive charge on Cd²⁺, which also includes changes in protonation pattern (Cd²⁺ ionic effect) and (ii) the remaining part, which involves conformational changes induced by Cd²⁺ binding (the conformation effect). Accordingly, we defined the Cd²⁺ ionic effect on the pK_a of a titratable group as ΔpK_a(Cd²⁺ charge) = pK_a(with Cd²⁺ in the Cd-bRC structure) – pK_a(without Cd²⁺ in the Cd-bRC structure) and the conformation effect as ΔpK_a(Cd²⁺ conformation) = pK_a(without Cd²⁺ in the Cd-bRC structure) – pK_a(WT-bRC structure) (Table 2). Notably, the two light-exposed crystal structures of bRC (WT-bRC and Cd-bRC) were obtained in the same laboratory (15, 20). Comparison of the atomic coordinates of the two bRC structures yields a rms difference of 0.65 Å, which is considerably larger than uncertainties in the atomic coordinates (see ref. 15). Therefore, the difference of calculated pK_a values for the same residue between the two structures refers directly to their structural differences. One pK_a unit, being equal to 60 mV, is the same magnitude as the direct influence of His ligand on E _m(Chla) (43), which is sufficiently larger than the error in the present computation.

Evaluating these pK_a we conclude that the pK_a shifts can predominantly be attributed to conformational changes induced by Cd²⁺ binding and not by Cd²⁺ ionic effect (Table 2). In the present study, this conclusion holds true for all residues that showed significant pK_a shifts upon Cd²⁺ binding. Indeed, those residues experiencing pK_a shifts upon Cd²⁺ binding also display considerable rearrangements of their own side chain and nearby surroundings (Fig. 1a).

The direct-charge effect of Cd²⁺ binding was small and, in most cases, even negligible with respect to the effect caused by conformational changes. It is of interest to understand how the direct-charge influence of the cationic Cd²⁺ is weakened (Table 2). Asp-L210, located proximal to the Cd²⁺–protein complex, is the only residue whose pK_a is significantly affected (down-shift of 2 units) by the Cd²⁺ ionic effect (Table 2). However, the pK_a of this residue is simultaneously up-shifted by the same amount on account of the conformational change effect, resulting in a vanishing pK_a shift. Thus, to resist electrostatic perturbations applied to the protein environment upon Cd²⁺ binding, some residues are apt to neutralize the Cd²⁺ ionic effect on their pK_a by a conformational change (i.e., conformational change effect). This remarkable compensation of the direct-charge effect of Cd²⁺ found in the present study is consistent with the small effect on the rate of charge recombination for that is commonly observed with Cd²⁺ binding (10, 11, 13, 14).

Concluding Remarks

Our results can be summarized as follows: Cd²⁺ binding induces rearrangements of side chains of titratable groups along the PT pathways, leading to significant pK_a shifts. Especially, inhibitions of residues in the immediate vicinity of Q_B, namely of Asp-L213, are assumed to affect the PT kinetics most dramatically. The electrostatic influence of Cd²⁺ over a large distance (≈12 Å) can be understood by the electrostatic domino effect (39), in which successive reorientation of residues (Asp-H124, His-H126, His-H128, Asp-H170, Glu-H173, Asp-M17, and Asp-L210) propagate over a large distance. The direct electrostatic influence of the Cd²⁺ ionic charge on these pK_a shifts is small and, in most cases, even negligible. The results of the present study are consistent with the pK_a inhibition mechanism (14) and provide theoretical detail for understanding this mechanism.

The present study may also provide a key to understanding the Cd²⁺ binding effect on the Q_A/B site in photosystem II (PSII). PSII is an oxygen-evolving protein–pigment complex functioning in cells ranging from cyanobacteria to higher plants. Especially, the D1/D2 protein in PSII resembles considerably subunit L/M in bRC, which suggests that both PSII and bRC derive from a common ancestor (44). It has been proposed that D1-His-252 may be a Cd²⁺ binding site in PSII (45). In the present study of bRC, the pK_a of Asp-L213 was found to decrease upon Cd²⁺ binding. According to the protein sequence alignment [based on analysis with clustal (46)], Asp-L213 and Ser-L223 in bRC from R. sphaeroides (Fig. 1a) correspond to D1-His-252 and D1-Ser-264 in PSII (Fig. 1b), which suggests their importance in ET and PT to Q_B. The most significant difference between the two proteins near the Q_A/B binding site is that PSII lacks a polypeptide corresponding to subunit H in bRC. Thus, Asp-H124, His-H126, and His-H128 of the Cd²⁺ binding site and Glu-H173 (Fig. 1a), all of which play an important role in the PT events of WT-bRC or PT-revertant mutants, are not conserved in PSII. Therefore, D1-His-252 in PSII, which corresponds to Asp-L213 in bRC, is exposed to solvent (30) and may be the Cd²⁺ binding site. In our computation we observed significant proton uptake at D1-His-252 upon formation of in PSII (41), which suggests its importance in the coupled ET and PT events of PSII. The conservation of Asp-L213/D2-His-252 seems to be thus important for both Q_B function and Cd²⁺ binding in bRC/PSII.

In the present study, we could not directly evaluate the His-entry inhibition mechanism for Cd²⁺ binding (17, 18), partly because of the unavailability of the crystal structure for the H(H126)A/H(H128)A double mutant. Thus, it remains unclear whether this double mutant is PT-inhibited by disabling His-H126/128 as a proton entry point or whether it also supports an electrostatic domino effect by removal of protonated His-H126/128. The decrease of and by 10- and 4-fold, respectively, in the double mutant H(H126)A/H(H128)A relative to WT-bRC (17) undoubtedly indicates their importance in the PT events. Nevertheless, that the 4-fold decrease in for this double mutant is still smaller than the 20-fold reduction of upon Cd²⁺ binding (11, 12) implies the presence of a missing link between the His-entry inhibition mechanism and the actual mechanism for PT inhibition by Cd²⁺ binding. This discrepancy might, however, be interpreted as a possible weaker electrostatic domino effect in the double mutant compared with Cd-bRC. We do not exclude a potential contribution of the proposed delocalized PT network (7, 14, 37) to the overall PT events, although we assume that the main PT pathway probably proceeds from the Cd²⁺ binding site of His-H126/128 (18). In fact, the majority of residues belonging to delocalized PT network in bRC whose pK_a were affected by the revertant mutants (9) are in common with those influenced by Cd²⁺ binding.