Energetics of Proton Transfer Pathways in Reaction Centers from Rhodobacter sphaeroides

Electron transfer between the primary and secondary quinones (QA, QB) in the bacterial photosynthetic reaction center (bRC) is coupled with proton uptake at QB. The protons are conducted from the cytoplasmic side, probably with the participation of two water channels. Mutations of titratable residues like Asp-L213 to Asn (inhibited mutant) or the double mutant Glu-L212 to Ala/Asp-L213 to Ala inhibit these electron transfer-coupled proton uptake events. The inhibition of the proton transfer (PT) process in the single mutant can be restored by a second mutation of Arg-M233 to Cys or Arg-H177 to His (revertant mutant). These revertant mutants shed light on the location of the main proton transfer pathway of wild type bRC. In contrast to the wild type and inhibited mutant bRC, the revertant mutant bRC showed notable proton uptake at Glu-H173 upon formation of the QB– state. In all of these mutants, the pKa of Asp-M17 decreased by 1.4–2.4 units with respect to the wild type bRC, whereas a significant pKa upshift of up to 5.8 units was observed at Glu-H122, Asp-H170, Glu-H173, and Glu-H230 in the revertant mutants. These residues belonging to the main PT pathway are arranged along water channel P1 localized mainly in subunit H. bRC possesses subunit H, which has no counterpart in photosystem II. Thus, bRC may possess alternative PT pathways involving water channels in subunit H, which becomes active in case the main PT pathway is blocked.

؊ state. In all of these mutants, the pK a of Asp-M17 decreased by 1.4 -2.4 units with respect to the wild type bRC, whereas a significant pK a upshift of up to 5.8 units was observed at Glu-H122, Asp-H170, Glu-H173, and Glu-H230 in the revertant mutants. These residues belonging to the main PT pathway are arranged along water channel P1 localized mainly in subunit H. bRC possesses subunit H, which has no counterpart in photosystem II. Thus, bRC may possess alternative PT pathways involving water channels in subunit H, which becomes active in case the main PT pathway is blocked.
The primary event in the bacterial photosynthetic reaction center (bRC) 1 after electronic excitation of the bacteriochlorophyll a dimer, the special pair, is a charge-separation process. As a result, the special pair becomes oxidized while an electron is transferred along the A-branch cofactors from an accessory bacteriochlorophyll a via bacteriopheophytin to ubiquinone Q A in the A-branch and subsequently to Q B in the B-branch. After the first ET process, Q B Ϫ is protonated and forms Q B H that is stabilized by a second ET and proton transfer (PT) event, resulting in the formation of the doubly protonated dihydroquinone Q B H 2 . Seven residues (namely His-H126, His-H128, Asp-M17, Asp-L210, Glu-L212, Asp-L213, and Ser-L223) were suggested to be involved in these PT events (reviewed in Ref. 1) (Fig. 1). The single mutant Asp-L213 to Asn (D(L213)N) decreases the rate of these PT events by about a factor of 10 6 (2). The decreased PT rate for the single mutant can be recovered by an additional mutation of Asn-M44 to Asp, since the side chain of Asp at M44 can substitute the removed carboxylate at L213 in the mutant bRC (3)(4)(5). On the other hand, the PT in the D(L213)N mutant bRC was also restored by mutations of Arg-M233 to Cys or Arg-H177 to His, namely the double mutant D(L213)A/R(M233)C or D(L213)A/R(H177)H, respectively (2,6,7). Similar revertants were also observed in the AA mutant bRC from Rhodobacter capsulatus (8). However, in these two mutants, no carboxylate is reintroduced, and, based on the wild type bRC structure, the corresponding two mutated sites M233 and H177 have a distance of more than 10 Å from residue L213 (i.e. 13 and 17 Å, respectively) (9) (Fig. 1). Based on the observed structural changes close to Glu-H173 from wild type to revertant mutant bRC, the proposed mechanism to recover PT in the revertants involves Glu-H173 in the PT pathway (7,10).
In experimental investigations on ET-coupled proton uptake at Q B , it is suggested that the addition of Cd 2ϩ and Zn 2ϩ ions decreases the rates of ET from Q A to Q B (11,12). The crystal structure confirms that these metal ions bind at Asp-H124, His-H126, and His-H128 (13). Thus, these two histidines were proposed as the proton entry point. However, the rates of the first and second ET process in the H(H126)A/H(H128)A double mutant were reduced by only a factor of 10 and 4 relative to the wild type bRC (14). Alternatively, the metal binding effect on the ET rates was suggested to be due to electrostatic interactions rather than blocking of the proton entry point (15). In this context, it is interesting to note that the metal binding decreases proton uptake by 2 pH units in response to the formation of Q A Ϫ and Q B Ϫ symmetrically, as suggested in the same study (15).
Based on spectroscopic measurements, seven titratable/polar residues His-H126, His-H128, Asp-M17, Asp-L210, Glu-L212, Asp-L213, and Ser-L223 were suggested to be involved in PT events of proton uptake at Q B (reviewed in Ref. 1). These seven residues are located along three water channels, P1-P3 (reviewed in Ref. 16), found in the crystal structure of bRC at 2.2-Å resolution (9). Two of these water channels, P1 and P2 (9), were proposed to connect Q B with the solvent on the cytoplasm side and are considered to participate in proton uptake at Q B . Water channel P1 extends about 23 Å from Q B via Glu-L212, Lys-H130, Glu-H122, Glu-M236 to Arg-H70, and His-H68 at the cytoplasmic side in an approximately orthogo-nal orientation to the membrane surface. It connects Q B mainly through subunit H with Asp-H224 (P1 a ) or Asp-M240 (P1 b ) as surface residue. Water channel P2, which is oriented essentially parallel to the membrane surface, is 20 Å, slightly shorter than P1. It connects Q B via Ser-L223, Asp-L213, Asn-M44, Glu-H173, and Gln-H174 with the surface residue Gln-M11 or Tyr-M3 (9,16). Water channel P3, proposed more recently, connects Q B via Asp-L213 with the surface-exposed Asp-M17 (16). Since there exists a patch of closely interacting acidic residues (Asp-L210, Asp-L213, Asp-M17, Asp-H124, Asp-H170, and Glu-H173) and an extended water cluster in the neighborhood, the water channels can be partially delocalized (16).
In the present study, we report the changes in protonation pattern and pK a values of titratable residues with respect to the wild type bRC, focusing on the Q A 0 Q B Ϫ state for inhibited (D(L213)N single mutant) and revertant (D(L213)A/R(M233)C or D(L213)A/R(H177)H double mutant) mutant bRC (10). The double mutant Glu-L212/Asp-L213 to Ala (AA mutant) is known to interrupt the ET-coupled PT reaction after the first ET event at Q B (6,17). To elucidate the PT pathway in bRC, we also investigated the AA mutant bRC, and the results were compared with those calculated for the revertant or wild type bRC.  (17), and the wild type light-exposed structure (Protein Data Bank code 1AIG) (9) in the P ϩ Q B Ϫ state. The atomic coordinates were prepared in the same way as in previous applications (18 -20). The position of hydrogen atoms were energetically optimized with CHARMM (21) 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 were considered to be protonated, and acidic groups were considered to be ionized). All of the other atoms whose coordinates were available in the crystal structure were not geometry optimized.
Atomic partial charges of the amino acids were adopted from the all-atom CHARMM22 (21) parameter set. For cofactors and residues whose charges are not available in CHARMM22, we used atomic partial charges from previous applications (18 -20).
Dielectric Volume-The dielectric volume of a protein complex is the spatial area and shape covered by molecular components of the protein that are polypeptide backbone, side chains, and cofactors but not water molecules. To facilitate a direct comparison with our past computational results, we used uniformly the same computational conditions and parameters such as atomic partial charges and dielectric constants. As a general and uniform strategy, all of the crystal waters are removed in our computations (18 -20, 22-25) because of the lack of experimental information for hydrogen atom positions. Cavities resulting after the removal of crystal water are uniformly filled with a solvent dielectric of ⑀ ϭ 80. Accordingly, the effect of the removed water molecules was considered implicitly by the high value of the dielectric constant in these cavities. A discussion on the appropriate value of the dielectric constant in proteins can be found in Refs. 26 -30. Computation of Protonation Pattern and pK a -The computation of the energetics of the protonation pattern is based on the electrostatic continuum model by solving the linear Poisson Boltzmann equation with the program MEAD from Bashford and Karplus (31). To sample the ensemble of protonation patterns by a Monte Carlo method, we used our own program Karlsberg (by B. Rabenstein; Karlsberg online manual available on the World Wide Web at agknapp.chemie.fu-berlin.de/ karlsberg/). For the first 3000 Monte Carlo scans, random protonation changes were applied for all individual titratable residues. For the remaining 7000 Monte Carlo scans, titratable residues whose protonation probability deviated by less than 10 Ϫ6 from zero or unity were fixed at the corresponding pure protonation state. The detailed procedure is described in Refs. 18,19,22, and 23. The dielectric constant was set to ⑀ P ϭ 4 inside the protein and ⑀ W ϭ 80 for water. All computations were performed at 300 K with pH 7.0 and an ionic strength of 100 mM. The linear Poisson Boltzmann equation was solved using a three-step gridfocusing 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 Ϫ 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 instead of Equation 1 (33). Therefore, the accuracy of the present pK a computations is directly comparable with that obtained for recent computations on redox-active cofactors in bRC (18 -20), photosystem I (PSI) (34), and PSII (35). 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 differ from the given crystal structures. They sometimes can be considerably larger. Since for different bRC the agreement between computed (18,19) and measured quinone redox potentials (36 -39) is about 50 mV, we expect a similar accuracy for computed pK a values. Note that 1.0 pK a unit corresponds to 60 mV in redox potential.

RESULTS AND DISCUSSION
pK a Shift of Glu-H173 in Revertant bRC Mutants-As observed in the wild type bRC (18, 19, 40 -42), we found a large proton uptake at Glu-L212 upon formation of the Q B Ϫ state in both revertant and inhibited mutant bRC (Table I). Remarkably, the proton uptake at Glu-L212 in the revertant mutants was slightly smaller than in the wild type and inhibited mutant Residues that showed significant changes in the calculated pK a upon mutation were labeled in boldface red letters. Colored side chains and boldface letters of the same color were also used for the mutated arginines.
bRC. The revertant mutants showed a small but significant increase of protonation also at Glu-H173, whereas the wild type bRC does not show a protonation at this residue. In experiments, the mutation of Glu-H173 to Gln was found to slow down the first and second ET process from Q A to Q B , presumably by affecting the kinetics of PT to Q B , where Glu-H173 may participate (43). On the other hand, in steady-state FTIR measurements (44) and our previous computations (19,23), Glu-H173 in wild type bRC remains deprotonated regardless of the redox state of Q B . The latter fact implies a small pK a for Glu-H173, whereas at the same time, it does not exclude transient protonation that may be required for the PT events at Q B . Hence, the protonation state at Glu-H173 in the revertant mutants may suggest also a participation of Glu-H173 in the PT process coupled with formation of Q B Ϫ (7, 10, 45). Interestingly, in the revertant mutants, the computed proton uptake at Glu-L212 upon formation of the Q A 0 Q B Ϫ state was lower than in the wild type and inhibited mutant bRC (Table I). This reduced proton uptake at Glu-L212 was approximately compensated by additional proton uptake at Glu-H173.
The observed proton uptake at Glu-H173 in revertant mutant bRC implies a pK a increase upon formation of the Q A 0 Q B Ϫ state. Thus, with respect to wild type bRC, we observed a considerable increase of the calculated pK a for Glu-H173 by about 3.5 units in both revertant mutants (Table II). It has been suggested that a rearrangement of the side chains of charged residues like Arg-H177 increases the pK a for Glu-H173 such that it can function equally well as proton donor and acceptor as needed in the PT chain connecting the solvent with Q B (10,45). Indeed, our computations showed that the pK a for Glu-H173 is larger than 6 in both revertant mutants. Contrary to the revertant mutants, the inhibited mutants and the AA mutants showed a pK a for Glu-H173 that is 1.1 and 3.2 units lower than in the wild type bRC, respectively (Table II).
In the wild type bRC, the calculated pK a of 5.4 for Asp-M17 is significantly larger than that of 2.7 for Glu-H173, which favors an involvement of Asp-M17 rather than Glu-H173 in the PT pathway (44,46). Notably, all mutant bRC considered here showed a pK a decrease for Asp-M17 by 1.4 -2.4 units with respect to wild type bRC (Table II). This is predominantly due to Asp-L213 that is mutated to a nontitratable residue of vanishing total charge. Since only the revertant mutant bRCs have a significantly larger pK a of 6.2-6.3 for Glu-H173 as compared with the pK a of 3.5-3.8 for Asp-M17, it is unlikely that the other mutant bRC can utilize Glu-H173 for the PT chain.
One might anticipate that the main effect of pK a shifts observed with mutations in bRC is due to changes in the net charge of a residue. Accordingly, the elimination of a positive charge at Arg-M233 or Arg-H177 should result in a pK a upshift for Glu-H173 and a pK a downshift for Asp-M17. To mimic the action of the D(L213)N/R(M233)C double mutant, we con-strained Asp-L213/Arg-M233 to be protonated/deprotonated in using the crystal structure for the wild type bRC and calculated the pK a for Glu-H173. Surprisingly, this constraint resulted in a decrease of the pK a for Glu-H173 by 1.8. At the same time, this constraint shifted the pK a for Glu-H230 from 0.9 in the original wild type bRC to 11.9, resulting in a fully protonated Glu-H230. Therefore, the increase of pK a calculated for Glu-H173 in revertant mutants is not merely due to a decrease of the net charge in the neighborhood of this residue but requires also a reorientation of the corresponding amino acid side chains, as stated by Paddock et al. (7). The pK a for both Asp-M17 and Glu-L212 remained unchanged in using the constraint where Asp-L213/Arg-M233 are protonated/deprotonated. Hence, together with a change of net charge around Glu-H173 in the revertant mutants, a shift from Asp-M17 to Glu-H173 in the role of the key residue for proton uptake requires also a suitable rearrangement of the participating amino acid side chains. Indeed, the crystal structures of revertant mutant bRC show a drastic rearrangement of amino acid side chains and change in salt bridge pattern around Glu-H173 relative to either wild type or inhibited mutant bRC (10).
Activation of the Alternative Proton Transfer Pathway in Subunit H-For the revertant mutants, there is a surprisingly large pK a upshift of the acidic residues Asp-H170, Glu-H122, and Glu-H230, which amounts to 2.8 -5.8 units in the D(L213)N/R(M233)C mutant and 0.9 -5.4 units in the D(L213)N/R(H177)H mutant relative to the wild type bRC (Table II). Based on the crystal structure (9), these three acidic residues together with Glu-M232 are suggested to interact with Arg-M233 and Arg-H177 (2).
All of these charged residues are located in a polar region containing a large number of crystal water (9). In the revertant mutant bRC, the pK a of Asp-H170, the acidic residue proximal to Glu-H173, is upshifted by 5.4 -5.8 with respect to a pK a of Ϫ4.7 in the wild type bRC. The pK a of 2.7 for Glu-H173 calculated for the wild type bRC indicates also a large energy barrier to protonate Glu-H173. The very low pK a for Glu-H173 in the wild type bRC corroborates the result of the steady-state FTIR studies (44), which revealed that Glu-H173 is deprotonated for all charge states of Q B . Hence, in the wild type bRC, this residue is less likely to be involved in proton uptake events at the Q B site. For revertant mutants, however, the larger pK a of this residue is considerably enhanced to allow Glu-H173 to participate in the PT pathway.
Interestingly, the three acidic residues Asp-H170, Glu-H122, and Glu-H230 bridge the gap between the acidic group Glu-H230 at the protein surface and Glu-H173 in the neighborhood of Q B . Glu-H230 belongs together with Glu-H229, Glu-H224, and Asp-H119 to the surface-exposed patch of acidic residues in subunit H and may couple to the water channel P1 (9, 16). The majority of these acidic residues was proposed to serve possibly as an internal "proton reservoir" facilitating fast protonation of Q B Ϫ (16) (Fig. 1).
Titratable residues, which act as mediators for proton transport, require suitable pK a values. These values should be in a range that they can play the role of a proton donor and acceptor in PT chains energetically equally well. The calculated pK a values for the residues known to participate in PT processes of the wild type bRC (i.e. Asp-M17, Asp-L210, Glu-L212, and Asp-L213) are in the range of 3.0 -9.4 (Table II). The largest value of 9.4 refers to Glu-L212, which agrees well with the value of 9.5 suggested by experimental measurements (47,48) and recent electrostatic computations (49).
The four buried acidic residues (Glu-H173, Asp-H170, Glu-H122, and Glu-H230) with increased pK a in revertant mutants are likely to function as PT mediators together with the three solvent-exposed acidic residues (Glu-H229, Glu-H224, and Asp-H119) whose pK a values are in the range of 3-4. Thus, besides the seven residues His-H126, His-H128, Asp-M17, Asp-L210, Glu-L212, Asp-L213, and Ser-L223 that are suggested to be PT-active in the main PT pathway of the wild type bRC (1), there is a network of acidic residues Glu-H173, Asp-H170, Glu-H122, Glu-H230, Asp-H119, Glu-H224, and Glu-H229 in subunit H that may be involved in PT of the revertant mutants. Indeed, diminishing of proton uptake by Q B Ϫ upon Cd 2ϩ binding in revertant mutants is less pronounced than in wild type bRC. Here, the ratio of ET rates k AB (2) (without Cd 2ϩ )/ k AB (2) (with Cd 2ϩ ) of the second ET process from Q A to Q B is 10 in the wild type and 4 -5 in revertant mutant bRC (10), indicating that an alternative PT pathway may be operative for proton uptake.
Interestingly, the effect of metal binding on PT efficiency in P(L209)Y or P(L209)F mutants, which have modified water chains, is similar to that in D(M17)N or D(L210)N mutants (50). This result suggests that the actual PT pathway in wild type bRC is more delocalized (6,15,51), involving not only the seven residues considered before (His-H126, His-H128, Asp-M17, Asp-L210, Glu-L212, Asp-L213, and Ser-L223). To interpret the metal binding influence and our calculated pK a values, we assume that bRC possesses a delocalized PT network as proposed in Refs. 6, 8, 15, 50, and 51. The residues in subunit H could become more active in the PT process when the main PT pathway involving His-H126 and His-H128 is inhibited.
This view corroborates the suggestion that residues near P1 and the other water channels P2 and P3 can serve as internal "proton reservoir" (16).
Indeed, a single mutation of Glu-H173 to Gln leads to a decrease of the first/second ET rates coupled with PT, suggesting a significant contribution of Glu-H173 to an efficient PT in the wild type bRC (43). This result from kinetic measurements seems to be in conflict with steady-state FTIR measurements (44) and computational studies (19,23), since the latter two indicate an absence of proton uptake at Glu-H173 forming Q B Ϫ in the wild type bRC. However, the kinetic method focuses on transient states only as opposed to the latter methods that probe the steady state. Therefore, Glu-H173 may be able to transiently bind a proton even in the wild type bRC, and this process is likely to be important for the kinetics of PT events at Q B , although the direct electrostatic interaction between Glu-H173 and Q B Ϫ is rather weak. A similar kinetic behavior with decreasing ET rates was measured depleting the H-chain in bRC (52). Interestingly, the reaction center D1-D2 complex of PSII, structurally and functionally similar to the L-M complex of bRC (53), does not possess a polypeptide corresponding to subunit H of bRC (54,55). Therefore, subunit H in bRC may, in principle, not be necessary for the photosynthetic reaction but could be important to transfer protons with high efficiency or to give the flexibility to purple bacteria in overcoming a detrimental mutation of the L-M complex as suggested in Ref. 10. For PSII, the latter ability may not be crucial, because the photo-damaged D1 subunit of the D1-D2 complex can be removed and reconstituted by an intact polypeptide, possibly as a result of an evolutionary process emerging from purple bacteria. The primary cleavage site of D1 degradation is proposed to be the glutamate-enriched loop near the nonheme iron at the stromal side (56,57). Thus, the presence of a subunit like the H chain that covers the stromal region of the D1-D2 core complex would hinder degradation of a photo-damaged D1 subunit. On the other hand, the ET rate between quinones in PSII is roughly by a factor of 10 smaller than in bRC (32,58), which might be associated with the absence of a PSII subunit corresponding to the H chain in bRC.
Conclusion-A comparison of the calculated pK a for titratable residues along the suggested PT pathways at Q B between wild type and mutant bRC provides insight in energetics and function of proton uptake at Q B Ϫ . The elimination of a positive charge at Arg-M233 or Arg-H177 and a suitable rearrangement of side chains around these basic residues in the revertant mutants readjust the energetics of the nearby residues. As a consequence, the pK a values for these residues are upshifted by up to 5 units, facilitating their participation in proton uptake events occurring at Q B Ϫ . This influence upon the mutation at Arg-M233 or Arg-H177 is predominant for the residues Glu-H122, Asp-H170, Glu-H173, and Glu-H230, belonging to subunit H. Especially, in revertant mutants Glu-H173 showed a notable proton uptake upon the formation of Q B Ϫ . We conclude that bRC potentially possesses a delocalized PT network along the water channel involving subunit H and that the PT through subunit H can become more active if the main PT pathway consisting of the seven residues His-H126, His-H128, Asp-M17, Asp-L210, Glu-L212, Asp-L213, and Ser-L223 is partially blocked. The slowing down of the ET between quinones in bRC after removal of the H-chain relates to the comparatively slow ET in PSII. The reason may an inhibition of PT by the absence of the H-chain.