Report

Light-Induced Structural Changes in Photosynthetic Reaction Center: Implications for Mechanism of Electron-Proton Transfer

See allHide authors and affiliations

Science  02 May 1997:
Vol. 276, Issue 5313, pp. 812-816
DOI: 10.1126/science.276.5313.812

Abstract

High resolution x-ray diffraction data from crystals of theRhodobacter sphaeroides photosynthetic reaction center (RC) have been collected at cryogenic temperature in the dark and under illumination, and the structures were refined at 2.2 and 2.6 angstrom resolution, respectively. In the charge-separated D+QAQB state (where D is the primary electron donor (a bacteriochlorophyll dimer), and QA and QB are the primary and secondary quinone acceptors, respectively), QB is located approximately 5 angstroms from the QB position in the charge-neutral (DQAQB) state, and has undergone a 180° propeller twist around the isoprene chain. A model based on the difference between the two structures is proposed to explain the observed kinetics of electron transfer from QA QB to QAQB and the relative binding affinities of the different ubiquinone species in the QBpocket. In addition, several water channels (putative proton pathways) leading from the QB pocket to the surface of the RC were delineated, one of which leads directly to the membrane surface.

The primary processes of photosynthesis, the conversion of electromagnetic energy (light) into chemical energy, are mediated by an integral membrane protein-pigment complex called the reaction center (RC) in which a sequence of photoinduced electron and proton-transfer reactions take place (1, 2). Our knowledge of these processes was greatly enhanced through the determination of the three-dimensional structure of the RC from two purple photosynthetic bacteria:Rhodopseudomonas viridis (3, 4) andRhodobacter sphaeroides (5-8). In previous structure determinations, the primary reactants were in the neutral state, that is, no electron transfer (charge separation) had taken place. Several independent experimental findings, however, point toward a structural change accompanying charge separation (9-15). A particularly dramatic effect was observed (11) when the rate of electron transfer in RCs that were frozen under illumination was compared with that of those frozen in the dark. The rate of the electron transfer from the primary ubiquinone QA to the secondary ubiquinone QBwas increased by several orders of magnitude when RCs were frozen under illumination, that is, in the charge separated state, as compared to RCs frozen in the dark.

We have now determined the structural changes accompanying charge separation in the RC and used them as a basis for a model to explain changes in the kinetics of electron transfer observed on freezing. The changes were obtained by comparing the structure of RCs in single crystals cooled to cryogenic (∼90 K) temperatures under illumination (the light structure) with the structure of RCs cooled to cryogenic temperatures in the dark (the dark structure). In our experiments, we used tetragonal crystals of Rb. sphaeroides R-26, which had been obtained earlier by Allen (16). Our crystals diffracted at cryogenic temperatures to higher resolution (1.9 Å in the dark state) than previously reported RC crystals. Data collection and refinement to 2.2 Å resolution has led to the determination of the positions of a number of water molecules that provide several possible pathways for protons from the aqueous phase to the QBpocket. Two of these proton “channels” are delineated in detail; one of these has been reported (8). The importance of water molecules for proton transfer to reduced QB is discussed below.

RCs from Rb. sphaeroides R-26 were isolated and purified (17). Crystallization conditions were similar to those described (16). Crystals grew in 1 to 3 weeks to a thickness of 0.1 to 0.2 mm in space groupP4 3 2 1 2 (unit cell dimensions a = b = 140.1 Å and c = 271.6 Å), with two RCs in the asymmetric unit (Table1). Typical crystals showed a 60 to 70% QBoccupancy as assayed with a microspectrophotometer (18). The QB occupancy, was increased by soaking the crystals in 1 mM ubiquinone-2 (19) for 2 days before data were collected. The dark-adapted crystals were plunged into liquid nitrogen and transferred via cryo-transfer tongs (20) to a goniostat cooled with a stream of nitrogen to ∼90 K (21). Formation of the charge separated state D+QAQB (where D+ is the primary donor, a bacteriochlorophyll dimer) was accomplished by illuminating a crystal with a filtered tungsten light source (bandpass 550 to 900 nm). The illumination conditions necessary to create D+QAQB were deduced from light saturation curves (22) in control crystals (Fig. 1). A 0.2-mm thick crystal required 0.4 W/cm2 (150-ms pulse) to reach 90% charge separation. The crystals investigated by x-ray diffraction were illuminated for 150 ms with 0.5 W/cm2 light directly above a liquid nitrogen bath and plunged into the bath under illumination to trap ∼90% of the RCs in the charge separated state. The sample was transferred to the goniostat, and a constant illumination of 10 mW/cm2 was maintained. Dark-adapted crystals were kept in dim light (<1 mW/cm2) during manipulations and data collection. From the saturation curves described above, the dark crystals were >95% in the neutral state DQAQB.

Figure 1

Dependence of the extent of charge separation (which is proportional to the absorbance change ΔA 480) on light intensity in a single, 0.2- mm-thick crystal of RCs from Rb. sphaeroidesat 295 K. The left ordinate is the reciprocal of the optical absorbance change; the intercept gives ΔA MAX for 100% charge separation. The right ordinate shows the percentage charge separation. A light intensity of 0.5 W/cm2 (see arrow) was used for the crystals investigated by x-ray diffraction, which corresponds to 92% charge separation. A tungsten projector lamp was used with band pass filters from 550 to 900 nm.

Table 1

Data collection and refinement statistics. Data were collected at beamline 7-1 of the Stanford Synchrotron Radiation Laboratory (λ = 1.08 Å) with a 30-cm MAR Research imaging plate system. The resolution of the dark state crystal was limited to 2.2 Å by the detector setting. Data were processed with the DENZO/HKL package (43) and merged and scaled with Rotavata-Agrovata of the CCP4 program suite (44). Molecular replacement was performed with MERLOT (45). The molecular replacement solution without ubiquinones was used to calculate Hendrickson-Lattman coefficients (46) with PHASES (47) using Sim’s weighting. These starting phase probabilities were utilized in a series of solvent flattening and twofold averaging cycles as implemented in SOLOMON (48). The SOLOMON electron density maps were of excellent overall quality and allowed straightforward rebuilding of the starting model with TOM/FRODO and O (49). Iterative cycles of model building, X-PLOR refinement (50), without NCS twofold restrictions and inspection of 2F oF c andF oF c electron density maps lead to the present models. The final models contain two RCs, four ubiquinones, eight bacteriochlorophylls, four bacteriopheophytins, two irons, and four LDAO molecules for a total of 13929 protein atoms. The dark-adapted structure contains 468 waters and the light-adapted structure contains 60 waters. Residual electron density that has not yet been modeled in the present structures is indicative of additional detergent molecules surrounding the RCs, as well as a large number of unmodeled waters. Molecular representations were prepared with the program SETOR (51).

View this table:

The overall structure of the Rb. sphaeroides RC in the DQAQB state (dark structure), including the cofactors and the polypeptide folds of the three (H, M, and L) subunits, follows, with only minor differences, the structure that has been previously described (1, 5-8). We focus, therefore, on the location of QB (and its surroundings) where the largest changes occur upon charge separation. The electron density map (Fig.2A) shows the position of the ubiquinone molecule in the QB binding site (QB1). It is located in a pocket (Fig. 3) with the O1 of QB 7.2 Å from the Nδ of HisL190. The carbonyl oxygen O4 forms a single hydrogen bond with the amide backbone of IleL224, in a manner similar to that described (8). This is in contrast with room temperature FTIR (Fourier transform infrared) data which suggested that the two carbonyl oxygens form weak and equivalent hydrogen bonds (23). The ubiquinone ring stacks directly on the conserved PheL216 in a parallel manner (Fig 3B), suggesting that this interaction contributes to the binding affinity of the ubiquinone. Further experimental support for this interaction comes from a herbicide resistant mutant of the Rps. viridis RC, in which PheL216 is replaced with serine, resulting in an RC with reduced affinity (60 μM versus 4.5 μM for wild type) for ubiquinone (24). The atomic displacement factor of QB is greater than that of QA (43 Å2 compared to 29 Å2), an indication of a more disordered structure or a lowered occupancy at the QBsite as compared to the QA site, which is fully occupied. The electron density between the bound ubiquinone and HisL190 and GluL212 is somewhat problematic. It can be modeled either by three water molecules, or by a partially occupied ubiquinone at a second binding site (QB2) or a combination of both. A partially occupied ubiquinone site is likely, and is assumed in the model discussed below.

Figure 2

(A) Dark structures. Stereoview of the final 2F o – F comit, refined electron density at the QB binding pocket in the DQAQB (dark) state, contoured at 1σ. The ubiquinone is bound in the position referred to as the QB1 site. The electron density between the QB1site and GluL212 and HisL190 is postulated to be due to partial occupancy of ubiquinone at the second site QB2. Side chain residues from the dark RC structure are indicated. (B) Light structures. Stereoview of the final 2F o – F c omit refined electron density at the QB binding pocket in the D+QAQB (light) state, contoured at 1σ. The ubiquinone is now in a position referred to as the QB2 site, which is ∼4.5 Å removed from the QB1 site shown in (A). In addition, the ubiquinone has undergone a 180° propeller twist with respect to the QB1site observed in the dark structure. Side chain residues from the light RC structure are indicated. Oxygen, nitrogen, and carbon atoms are colored red, blue, and gray, respectively.

Figure 3

Comparison of the dark (DQAQB) and the light (D+QAQB ) RC structures in the QB binding pocket. The dark and light RC structures are colored red and blue, respectively. Hydrogen bonds are indicated with dotted lines, where bond distances of less than and greater than 2.9 Å are indicated by large and small dots, respectively. In the light RC structure, the ubiquinone has moved ∼4.5 Å and undergone a 180° propeller twist. Side chain residues from the light RC structure are indicated. Oxygen, nitrogen, and carbon atoms are colored red, blue, and gray, respectively. Orthogonal views of this region are provided in panels A andB.

The positions of ubiquinone in the QB pocket of dark adapted crystals reported by various groups differ significantly from each other (Fig. 4). There are several possible sources for this discrepancy. (i) There may be differences in the extent of dark adaptation. (ii) The position of ubiquinone in the QBpocket has been more difficult to establish because of the lower occupancy at the QB pocket and the lower resolution of previous structure determinations. (iii) Variability in sample preparations, specifically in regard to the detergent, has been shown to influence the binding of ubiquinone in the QB pocket (25). (iv) Electrons created by the x-ray irradiation during data collection may reduce ubiquinone in the QB pocket. At room temperature, the reduced quinone can move to the QB2position, whereas at 90 K it remains frozen in the QB1position. This could explain why the position of QB in several of the room temperature structures are closer to that observed in the charge separated state discussed below.

Figure 4

Superposition of previously reported positions of ubiquinone in the QB binding pocket [see also (4)]. PDB (Protein Data Bank, Brookhaven) entries and color code: 1PCR, Ermler et al. (8), yellow; 2RCR, Chang et al. (6), dark blue; 1YST, Arnouxet al. (7), green; 4RCR, Allen et al. (5), red; present model of the dark structure, light blue. Superposition was performed by the method of Kabsch (52). Side chain residues from the current dark structure are indicated. Oxygen, nitrogen, and carbon atoms are colored red, blue, and gray, respectively.

The higher resolution of this work enabled us to determine the location of water molecules. In particular, two distinct water channels, P1 and P2, that connect the QB pocket with the surface of the protein are shown in Fig. 5. P1 is essentially identical to the water chain reported earlier (8).

Figure 5

Water channels P1 and P2 observed in the dark RC structure leading from the QB pocket to the surface of the protein on the cytoplasmic side of the RC. The coloring scheme is: H subunit (green), L subunit (yellow), M subunit (blue), water molecules (red), bacteriochlorophylls (green), and bacteriopheophytin (purple). QA and QB are colored orange-red, while the nonheme iron is rust-colored. The positions of the quinone tails past carbon C16 are less well defined because of poor electron density. The approximate location of the membrane is indicated by the shaded region. The details of the P1 and P2 pathways are shown in Fig. 6, A and B.

The structure of the charge separated D+QAQB state (light structure) is approximately the same as that of the DQAQB (dark) structure. The O4 carbonyl of the QB semiquinone hydrogen bonds to HisL190 and the O1 carbonyl hydrogen bonds to the backbone amide nitrogen of IleL224 (Figs. 2B and 3). This binding site is similar to the partially occupied QB2 site postulated for the dark structure. The refined atomic displacement parameters for QB are similar to those of QA, indicative of a nearly fully occupied ubiquinone at this site. In addition to the L224 amide nitrogen, SerL223 and the backbone amide of L225 are both ∼3.1 Å from the O1 carbonyl of the ubiquinone and may form a second set of hydrogen bonds (albeit longer) to the ubisemiquinone anion. The presence of these three hydrogen bonds to the O1 carbonyl of the ubisemiquinone at the QB2 site, especially the longer interactions with SerL223 and L225 amide, are consistent with ENDOR studies on native and SerL223 → Ala mutant RCs (26).

The most striking observation in the light-adapted structure is a 4.5 Å (center to center distance) movement of the ubisemiquinone toward the cytoplasm with an accompanying 180° propeller twist about the isoprene tail (Fig. 3). There are fewer well-ordered water molecules than in the DQAQB structure. Particularly striking is the observation that GluH173, located along the P2 water channel, is disordered compared to either GluH173 in the dark structure or to the surrounding residues in the light structure. This suggests movement of water (protons) within the P1 and P2 channels that is concomitant with formation of the D+QAQB state.

A model of the electron transfer mechanism from QA QB to QAQB needs to explain the observation that at low temperatures (∼90°K) the electron transfer from QA QB to QAQB is completely blocked in RCs cooled in the dark, whereas the electron transfer proceeds readily when RCs are frozen under illumination (11). The differences between the light and dark structures of the RCs offer a straightforward and simple explanation of these observations. We postulate a model in which the ubiquinone can assume two positions; in one, electron transfer from QA is inhibited, in the other it is not. In the dark-adapted RCs, the position QB1 (Figs. 2 and 3) is thermodynamically favored. In this position, the distance between the carbonyl oxygen O1 of QB and Nδ of HisL190 is 7.2 Å. It is, therefore, disconnected from the most direct pathway for electron transfer from QA and hence, electron transfer from QA QB to QAQB is inhibited. However, a fraction of RCs exist in an activated state QB2corresponding to a quinone position that is ∼5 Å removed from QB1. In this position, the ubiquinone is hydrogen-bonded to HisL190 and electron transfer from QA can readily proceed. This position corresponds to the residual patch of electron density of the dark structure (Fig. 2A). We propose that the movement of the quinone from QB1 to the QB2 position is a necessary prerequisite for electron transfer from QA QB to QAQB . The observed activation energy (27) represents the barrier between these two states. The main contribution to the activation energy is probably the breaking of the hydrogen bond from O1 to the backbone amide of IleL224 (Figs. 2 and 3) and the energy associated with the 180° propeller twist. The above model is also consistent with the observation that the measured electron transfer rate from QA QB to QAQB is independent of the redox potential (that is, the driving force) of different ubiquinones substituted in the QA site, demonstrating that the rate-limiting step is not electron transfer (28).

In the light-adapted structure of the RC, the ubiquinone forms a hydrogen bond with HisL190, thereby favoring the QB2 site (Figs. 2B and 3). In this position, electron transfer can readily take place, as experimentally observed. The ground state (dark) x-ray crystal structure that has been universally reported is not the kinetically active structure for electron transfer. Contributions from possible light-induced structural changes upon forming D+QA QB(29) to the electron transfer from D+QA QB to D+QAQB require further study.

After the one-electron reduction of QB discussed above, a second electron is transferred to QB . The doubly reduced ubiquinone is protonated to form the ubiquinol (QBH2), which leaves the RC (30), initiating the formation of the proton gradient across the plasma membrane that drives ATP synthesis (31). Thus, the protonation of QB and the release of quinol are two fundamentally important processes. The main question associated with the protonation concerns the mechanism by which protons are transferred from the outside, aqueous phase, to the QB site which is buried inside the RC (32). The generally accepted view is that protons move along a chain of proton donor and acceptor groups. These groups could be either side chains of protonatable amino acids or water molecules. Several of the amino acids in the chain have been identified in Rb. sphaeroides by site-directed mutagenesis (33-38). The role of the water molecules has been less well established. In one of the Rb. sphaeroides RC structures, the positions of water molecules have been described (8); the locations of water molecules were also inferred by a computational approach (39).

The higher resolution of this work made it possible to determine the positions of a large number of water molecules, giving rise to several possible water channels (proton paths) connecting QB to the aqueous phase. Two of these are indicated as P1 and P2 (Fig. 5). Pathway P1 (Fig. 6A) proceeds ∼23 Å from the QB site via GluL212through the H subunit to the cytoplasm. This pathway is approximately normal to the membrane surface and is similar to the one previously described (8). The second pathway, P2, which had been previously proposed (32) has now been identified (Fig. 6B). It leads from SerL223 to AspL213 via the interface between the H and M subunits, parallel to the membrane surface at approximately the depth of the nonheme iron. This pathway traverses a number of residues that have been identified by mutational studies to be involved in proton uptake by QB. These include SerL223 (33), AspL213(34, 35), and GluH173 (36, 37). The terminus of this pathway is near the surface of the negatively charged membrane (40) where the proton concentration is expected to be substantially greater than that in bulk water (41).

Figure 6

Details of the two water channels P1 (A) and P2 (B). Water oxygens are shown in red, with positions of ubiquinone and iron provided for reference. The cytoplasmic side of the RC is at the bottom of the figure. Side chain positions from the dark RC structure are indicated. Oxygen, nitrogen, and carbon atoms are colored red, blue, and gray, respectively.

For the release of quinol and its replacement by ubiquinone, the binding affinities of the three species QBH2, QB , and QB must meet certain criteria for proper function. The affinity for QB must be greater than that for QBH2 and QB to prevent the loss of the functionally important QB intermediate and to avoid potentially detrimental release of the chemically reactive ubisemiquinone radical anion. The affinity of the RC for QBshould be greater than for QBH2 to thermodynamically favor the replacement of the quinol with ubiquinone. This ensures the presence of an electron acceptor for continuing ubiquinone reduction. Indeed, the observed binding affinities correspond to these expectations (19, 30, 42), that is, the affinities decrease in the order QB > QB > QBH2. These results are readily explained by the current structures. As discussed above, both carbonyl oxygens of QB form multiple close (<2.9 Å) hydrogen bonds, whereas in the dark-adapted structure, only one of the hydrogen bonds to QB (NH of IleL224) is present (Figs. 2 and 3). For ubiquinol, both carbonyl oxygens are protonated, resulting in weaker binding. The release of ubiquinol may be further facilitated by the presence of the water channels.

By cooling crystals of the RC to low temperature under illumination, we have determined the structural changes around the QB site accompanying light-induced charge separation in bacterial RCs. These changes provide an explanation for the observed temperature dependence of the electron transfer kinetics from QA QB to QAQB . In addition, these results account for the relative binding affinities of the different ubiquinone species in the QB binding pocket and identify several proton transfer pathways to the QB pocket that are critical for efficient function of the RC in photosynthesis.

  • * Present address: Medical Research Council, Laboratory of Molecular Biology, Cambridge CB2 2QH, UK.

  • Present address: Stanford Synchrotron Radiation Laboratory, P.O. Box 4349, Bin 69, Stanford, CA 94309, USA.

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article