Efficient exchange of the primary quinone acceptor Q_A in isolated reaction centers of Rhodopseudomonas viridis

Abstract

A key step in the conversion of solar energy into chemical energy by photosynthetic reaction centers (RCs) occurs at the level of the two quinones, Q_A and Q_B, where electron transfer couples to proton transfer. A great deal of our understanding of the mechanisms of these coupled reactions relies on the seminal work of Okamura et al. [Okamura, M. Y., Isaacson, R. A., & Feher, G. (1975) Proc. Natl. Acad. Sci. USA 88, 3491–3495], who were able to extract with detergents the firmly bound ubiquinone Q_A from the RC of Rhodobacter sphaeroides and reconstitute the site with extraneous quinones. Up to now a comparable protocol was lacking for the RC of Rhodopseudomonas viridis despite the fact that its Q_A site, which contains 2-methyl-3-nonaprenyl-1,4-naphthoquinone (menaquinone-9), has provided the best x-ray structure available. Fourier transform infrared difference spectroscopy, together with the use of isotopically labeled quinones, can probe the interaction of Q_A with the RC protein. We establish that a simple incubation procedure of isolated RCs of Rp. viridis with an excess of extraneous quinone allows the menaquinone-9 in the Q_A site to be almost quantitatively replaced either by vitamin K₁, a close analogue of menaquinone-9, or by ubiquinone. To our knowledge, this is the first report of quinone exchange in bacterial photosynthesis. The Fourier transform infrared data on the quinone and semiquinone vibrations show a close similarity in the bonding interactions of vitamin K₁ with the protein at the Q_A site of Rp. viridis and Rb. sphaeroides, whereas for ubiquinone these interactions are significantly different. The results are interpreted in terms of slightly inequivalent quinone–protein interactions by comparison with the crystallographic data available for the Q_A site of the two RCs.

In the reaction center (RC) of photosynthetic purple bacteria, two quinone molecules (Q_A and Q_B) play an essential role in coupling the electron- and proton-transfer reactions leading to the conversion of light energy into chemical energy (for a review, see ref. 1). Following absorption of a photon, transmembrane electron transfer occurs between a dimer of bacteriochlorophyll molecules and Q_A in ≈200 ps and then to Q_B in 20–100 μs. Q_A acts as a one-electron acceptor only, whereas Q_B plays the role of a two-electron gate and can accept two protons before leaving the RC.

The crystal structure of the RCs of the two species of photosynthetic bacteria that have been investigated the most, namely Rhodobacter sphaeroides and Rhodopseudomonas viridis, has been solved in several laboratories (2–9). The best atomic resolution for which structural data on the Q_A binding sites are presently available is 2.65 Å for the former species and 2.3 Å for the latter. For these two highly homologous proteins, the structural models reveal that the bonding interactions of Q_A, which are 2-methyl-3-nonaprenyl-1,4-naphthoquinone (menaquinone-9) in Rp. viridis and 2,3-dimethoxy-5-methyl-6-decaprenyl-1,4-benzoquinone (ubiquinone-10) in Rb. sphaeroides, are relatively well defined (within the ±0.25- to 0.5-Å precision on the position of the nonhydrogen atoms in the x-ray data), although the details vary among the various structures. Notably, the identity of some of the residues involved in hydrogen bonding and the relative strength of the hydrogen bonds to the two carbonyls of Q_A are still not totally clear (2–9).

The precise knowledge of the interactions of the two carbonyls of the quinones with the protein binding sites, which is a prerequisite for a more detailed understanding of the function of these cofactors, can be revealed by applying various techniques of structural spectroscopy such as electron nuclear double resonance (ENDOR) spectroscopy, EPR, Fourier transform infrared (FTIR) difference spectroscopy, or NMR that can probe the close environment of the quinones. Often these techniques require the Q_A site to be depleted of its native quinone and reconstituted with an isotopically labeled quinone. The key extraction/reconstitution procedure, using detergent instead of organic solvents, was first achieved by Okamura et al. (10) for the RC of Rb. sphaeroides and has since been widely used in a number of laboratories. This approach, combined with isotope labeling of the quinones, has been essential to characterize the bonding interactions of the ubiquinone in the Q_A site of Rb. sphaeroides first by ENDOR (11) and, more recently, by solid-state NMR (12), EPR (13, 14), and FTIR difference spectroscopy (15–21). Furthermore, by allowing experiments in which the chemical nature and the in situ redox potential of the primary quinone acceptor are varied, the extraction/reconstitution procedure has already provided much of our present knowledge on the structural factors affecting the binding of various quinones to the Q_A site as well as on the driving force for electron transfer to Q_A and Q_B (22, 23).

An efficient quinone reconstitution method in Rp. viridis is a prerequisite to extend to the Q_A site of Rp. viridis our investigations of the bonding interactions of menaquinone and ubiquinone in Rb. sphaeroides RCs by light-induced FTIR difference spectroscopy (15–20). Although previous studies have indicated that extraction/reconstitution of Q_A should also be applicable to Rp. viridis RCs (24, 25), it is noteworthy that no report on the use of this technique to investigate the Q_A site of this species has yet appeared. In view of the higher resolution of the crystallographic data for the Q_A site of Rp. viridis compared with that of Rb. sphaeroides and of the fact that the chemical nature of Q_A differs for the RCs of the two species, this absence of data for the former species is unexpected and may stem from the difficulties that we have encountered to reproduce the quinone depletion/reconstitution results previously reported for Rp. viridis (24, 25). This has prompted us to assess the potential of other reconstitution techniques and, notably, of the direct exchange of the quinone that has proven successful in the case of the oxygenic photosynthetic organisms. In photosystem I, this technique has been used to exchange the 2-methyl-3-phytyl-1,4-naphthoquinone (vitamin K₁) playing the role of an early electron acceptor (26). Vitamin K₁ is a close analogue of the menaquinone-9 of Rp. viridis and differs from the latter only by the replacement of the isoprenoid chain by a phytol chain (see Inset of Fig. 1). In photosystem II, methods for exchanging the native plastoquinone in the Q_A site as well as its replacement by ubiquinone have been established (27).

Figure 1.

Absorption spectra of films of isolated vitamin K₁. (a) Unlabeled vitamin K₁. (b) Uniformly ¹⁸O-labeled vitamin K₁. (c) Uniformly ¹³C-labeled vitamin K₁. (Inset) Structural formula of vitamin K₁.

In the present report, it is demonstrated for the first time that the native menaquinone-9 in the Q_A site of Rp. viridis can be almost quantitatively exchanged for vitamin K₁ or for ubiquinone by simply incubating the isolated RCs with an excess of the extraneous quinone. Using light-induced FTIR difference spectroscopy, the bonding interactions of the menaquinone-9 or of the newly introduced vitamin K₁ are shown to be identical for both the neutral and the semiquinone states of Q_A in Rp. viridis. By combining isotope labeling of the quinones and FTIR difference spectroscopy, the interactions of vitamin K₁ or of ubiquinone in the Q_A site of Rp. viridis and of Rb. sphaeroides are compared.

MATERIALS AND METHODS

Quinone labeling and HPLC purification were performed according to ref. 28 following identical protocols for the labeled and unlabeled quinones. The extent of labeling of vitamin K₁, estimated on the basis of the IR absorption spectra, was 70% for ¹⁸O labeling and 97% for ¹³C labeling.

Five batches of Rp. viridis RCs were isolated according to ref. 29. In two of them the lauryldimethylamine N-oxide (LDAO) detergent (0.1%) was exchanged for cholate (0.25%) on a DEAE column. The different batches were used for vitamin K₁ exchange with six sets of experiments made with RCs in LDAO and three with RCs in cholate. For each experiment, RC samples (≈0.4 mM) in 20 mM Tris buffer (pH 7) at 20°C were incubated overnight in parallel with unlabeled and isotopically labeled vitamin K₁. Two sets of 2,3-dimethoxy-5-methyl-6-hexaprenyl-1,4-benzoquinone (ubiquinone-6) exchange experiments were performed with Rp. viridis RCs in each detergent. Rb. sphaeroides RCs were isolated, depleted from Q_A (residual Q_A, <2%), and reconstituted with vitamin K₁ essentially as reported previously (16).

FTIR difference spectra at 5°C were measured according to refs. 30 and 31 except that stigmatellin (2 mM) was used instead of o-phenanthroline as Q_B inhibitor. All the reported spectra have been corrected for the incomplete ¹⁸O labeling of vitamin K₁.

RESULTS

The IR absorption spectra of the isolated vitamin K₁ used for the RC reconstitution experiments are presented in Fig. 1. The single band at 1,662 cm⁻¹ of the unlabeled quinone is downshifted by 29 and 42 cm⁻¹ upon ¹⁸O and ¹³C labeling, respectively. The corresponding calculated shifts for a pure C=O stretching vibration are 40 and 37 cm⁻¹, respectively. These observations indicate some coupling of the predominantly C=O mode at 1,662 cm⁻¹ with C=C vibrations. The bands at 1,618 and 1,597 cm⁻¹ are not much affected by ¹⁸O labeling but are downshifted by 54–58 cm⁻¹ upon ¹³C labeling, consistent with their assignment to predominantly C=C modes from the quinonic and aromatic rings, respectively (16, 32). Almost no isotopic shift is observed for the band at 1,461 cm⁻¹ assigned to δCH₂ modes from the phytyl chain (16). The small shift upon ¹³C labeling observed for the 1,377-cm⁻¹ band, attributed to δCH₃ modes from the phytyl chain and from the methyl group at C₂ on the quinone ring (see Inset of Fig. 1 for atom numbering in vitamin K₁), indicates some coupling to the adjacent C—C vibrations. The band at 1,330 cm⁻¹, downshifted only upon ¹³C labeling, is tentatively assigned to the C₂—CH₃ mode. The band at 1,296 cm⁻¹, attributed to coupled C—C and C=C vibrations from the two cycles (32), shows the expected large ¹³C isotopic shift.

The light-induced Q_A⁻/Q_A spectra of ubiquinone-depleted Rb. sphaeroides RCs after reconstitution with vitamin K₁ of different isotopic compositions are shown in Fig. 2. For unlabeled and ¹⁸O-labeled vitamin K₁, the spectra are essentially identical to those reported previously (16). For ¹³C-labeled vitamin K₁ (Fig. 2c), negative bands appear at 1,596 and 1,253 cm⁻¹ and positive ones, at 1,434, 1,401, and 1,356 cm⁻¹. The positive bands appear related to the three bands at 1,477, 1,444, and 1,394 cm⁻¹ observed with unlabeled quinones (Fig. 2a). The bands above ≈1,660 cm⁻¹ and between 1,560 and 1,500 cm⁻¹, which appear to be affected very little by the isotopic composition of the vitamin K₁ used for the reconstitution, pertain to nonquinone contributions [i.e., to the protein and the other cofactors (16, 33)].

Figure 2.

Light-induced Q_A⁻/Q_A FTIR difference spectra of ubiquinone-depleted Rb. sphaeroides RCs after reconstitution with unlabeled vitamin K₁ (a), uniformly ¹⁸O-labeled vitamin K₁ (b), or uniformly ¹³C-labeled vitamin K₁ (c). Temperature, 5°C; resolution, 4 cm⁻¹; ≈100,000 interferograms added. The frequency of the bands is given with an accuracy of ±1 cm⁻¹.

When Rp. viridis RCs are incubated overnight with ≈10-fold excess of vitamin K₁, the light-induced Q_A⁻/Q_A spectra become strongly dependent on the nature of the isotopic label of the added quinone. Whereas for unlabeled vitamin K₁, the Q_A⁻/Q_A spectrum (Fig. 3a) is essentially identical to that obtained with native RCs (data not shown; see refs. 16, 30, 31, and 34), the Q_A⁻/Q_A spectra recorded after reconstitution with vitamin K₁ uniformly labeled either with ¹⁸O (Fig. 3b) or with ¹³C (Fig. 3c) are distinctly different. The largest changes are observed for positive bands in the region between 1,500 and 1,340 cm⁻¹, with the type of labeling affecting both the amplitude and the frequency of these bands. A negative band at 1,296 cm⁻¹, unaffected by ¹⁸O labeling, appears to downshift to 1,252 cm⁻¹ upon ¹³C labeling. In general, these changes resemble quite closely those observed upon reconstitution of Q_A-depleted Rb. sphaeroides RCs with the same isotopically labeled vitamin K₁ (Fig. 2).

Figure 3.

Light-induced Q_A⁻/Q_A FTIR difference spectra of Rp. viridis RCs after incubation with unlabeled vitamin K₁ (a), uniformly ¹⁸O-labeled vitamin K₁ (b), or uniformly ¹³C-labeled vitamin K₁ (c).

The positive band at 1,478 cm⁻¹ (Fig. 3a) has almost vanished after incubation of the Rp. viridis RCs with ≈10-fold excess ¹³C-labeled vitamin K₁ (Fig. 3c) and thus constitutes a good marker to use to follow the extent of exchange of the quinones. When the RCs are incubated with ≈3-fold excess of ¹³C-labeled quinones/RC (i.e., with only ≈30% of the amount used for the experiment leading to the spectrum in Fig. 3c), the amplitude of the 1,478 cm⁻¹ band remaining in the Q_A⁻/Q_A spectrum increases about three times compared with that present in Fig. 3c (data not shown). This observation qualitatively indicates that the added vitamin K₁ tends to equilibrate with the menaquinone-9 in the binding site. The kinetics of quinone exchange are difficult to follow with our method due to the long time (≈2 hr) required for the preparation and stabilization of the samples for the FTIR measurement. In addition, the kinetics of exchange seem to be somewhat variable. In preliminary experiments aimed at developing the present quinone-exchange protocol, we have observed once that the exchange was complete after only 2 hr of incubation, whereas usually longer incubation times (3, 6, or even 12 hr) were required for complete exchange. No clear correlation could be made with either the batch of RCs, the illumination conditions, or the type of detergent used. Furthermore, preliminary experiments suggest that the same protocol does not work for 2,3-dimethyl-1,4-naphthoquinone in Rp. viridis RCs or for vitamin K₁ in Rb. sphaeroides RCs. There is a definite need for an assay of quinone replacement more suitable than FTIR difference spectroscopy to further investigate and optimize the biochemistry of the exchange.

For the replacement of the native menaquinone-9 in Rp. viridis by ubiquinone, it was observed that both a larger ratio of added quinone/RC (≈30) and a longer incubation time (≈24 hr) were necessary to achieve high quinone exchange compared with those used for vitamin K₁. In one of the four trials made, the exchange was only about 50% after 24 hr and did not increase upon further incubation at 20°C. Furthermore, as previously noticed for reconstitution of ubiquinone in Rb. sphaeroides RCs (16, 35), it was found that the isotopic labeling of the carbonyl oxygen atoms of the ubiquinone in the Rp. viridis RCs tends to equilibrate on a time scale of a few days with the ¹⁸O/¹⁶O composition of the water used to prepare the final RC resuspension buffer. It was thus necessary to perform the quinone replacement in H₂¹⁸O for ¹⁸O-labeled ubiquinone-6 and in H₂¹⁶O for unlabeled ubiquinone-6. The corresponding Q_A⁻/Q_A spectra (Fig. 4) are distinctly different from those recorded with unlabeled and ¹⁸O-labeled vitamin K₁ (Fig. 3 a and b). Notably, a negative band at 1,603 cm⁻¹ for unlabeled ubiquinone-6 appears shifted to 1,585 cm⁻¹ after ¹⁸O labeling. In the 1,500- to 1,300-cm⁻¹ region of absorption of the semiquinone modes, a band of large amplitude at 1,475 cm⁻¹ decreases upon ¹⁸O labeling, whereas a smaller band at 1,441 cm⁻¹ increases in amplitude. The shape and frequency of these anion bands for unlabeled ubiquinone-6 are very different from those peaking at 1,466 cm⁻¹ in the Q_A⁻/Q_A spectrum of native Rb. sphaeroides RCs (16). On the other hand, it is interesting to note that a negative band now appears at 1,262 cm⁻¹ (Fig. 4) with the same position and amplitude as a band present in the Q_A⁻/Q_A spectrum of native Rb. sphaeroides RCs and previously assigned to a ubiquinone mode (16). Another negative band appears at 1,283 cm⁻¹, reminiscent of bands observed at 1,290–1,288 cm⁻¹ in the Q_B⁻/Q_B spectra of Rb. sphaeroides and Rp. viridis (19, 35).

Figure 4.

Light-induced Q_A⁻/Q_A FTIR difference spectra of Rp. viridis RCs after incubation with unlabeled ubiquinone-6 (a) or with ubiquinone-6 ¹⁸O-labeled on both carbonyls (b).

To better characterize the isotope-sensitive vibrations arising from the quinone itself in both the Q_A and Q_A⁻ states, the bands that are not responding to the isotope composition of the added quinone can be canceled by calculating isotopically labeled minus unlabeled double-difference spectra (16). In a light minus dark Q_A⁻/Q_A spectrum, the bands of the neutral quinone appear with a negative sign. Thus, in an isotopically labeled minus unlabeled double-difference spectrum, the bands of the labeled quinone will appear with a negative sign and those of the unlabeled quinone will exhibit a positive sign. The reverse situation is expected for the semiquinone modes.

The ¹⁸O minus ¹⁶O double-difference spectra are shown in Fig. 5 a and b for vitamin K₁ in the Q_A site of Rp. viridis and Rb. sphaeroides, respectively. The corresponding ¹³C minus ¹²C spectra are depicted in Fig. 5 c and d, respectively. Two positive bands of unequal amplitude are found for the C=O of the unlabeled quinone at 1,653 and 1,637 cm⁻¹ for Rp. viridis (Fig. 5 a and c) and at 1,651 and 1,640 cm⁻¹ for Rb. sphaeroides (Fig. 5 b and d). Upon ¹⁸O labeling, they give rise to a negative band at 1,619 cm⁻¹ for Rp. viridis and 1,620 cm⁻¹ for Rb. sphaeroides with a shoulder at ≈1,628 cm⁻¹. Upon ¹³C labeling, the two C=O bands appear well separated at 1,610 and 1,595 cm⁻¹ for Rp. viridis (Fig. 5c) and at 1,607 and 1,597 cm⁻¹ for Rb. sphaeroides (Fig. 5d). For the quinonic C=C band of the unlabeled vitamin K₁, small positive bands at 1,608–1,609 cm⁻¹ appear downshifted by ≈9 cm⁻¹ upon ¹⁸O labeling. The aromatic C=C band of the unlabeled quinone is found at 1,590 cm⁻¹ (Rp. viridis) or 1,588 cm⁻¹ (Rb. sphaeroides) in the ¹⁸O minus ¹⁶O spectra and at 1,583 cm⁻¹ (Rp. viridis) or 1,585 cm⁻¹ (Rb. sphaeroides) in the ¹³C minus ¹²C spectra. The slight difference in the frequency observed upon ¹⁸O or ¹³C labeling in RCs of a given species is due to the overlap of the positive C=C band with negative bands of different origins. It is thus likely that the frequency (1,588 cm⁻¹) previously proposed for this mode in Rb. sphaeroides (16) should be decreased by ≈2 cm⁻¹. This C=C band splits at 1,575 and 1,583 cm⁻¹ for Rp. viridis and, less clearly, at 1577 and ≈1580 cm⁻¹ in Rb. sphaeroides upon ¹⁸O labeling and downshifts to ≈1,535 cm⁻¹ upon ¹³C labeling in both species. In the region of absorption of the semiquinone modes, the double-difference spectra (Fig. 5 a–d) emphasize the remarkable similarity between the RCs of the two species when comparing the pattern of bands corresponding to a given isotope substitution.

Figure 5.

Double-difference spectra (isotopically labeled minus unlabeled) obtained from pairs of Q_A⁻/Q_A FTIR spectra shown in Figs. 2, 3, 4. (a, c, and e) Rp. viridis RCs. (b, d, and f) Rb. sphaeroides RCs. (a and b) ¹⁸O-labeled vitamin K₁. (c and d) ¹³C-labeled vitamin K₁. (e and f) Ubiquinone-6 ¹⁸O-labeled on both carbonyls.

The ¹⁸O minus ¹⁶O double-difference spectra for ubiquinone-6 in the Q_A site of Rp. viridis and Rb. sphaeroides are shown in Fig. 5 e and f, respectively. In general, these two spectra differ much more from each other than do the two equivalent spectra recorded with vitamin K₁ (Fig. 5 a and b). There are, however, several bands of the neutral ubiquinone-6 that are close in frequency in the two sets of spectra. This is notably the case for the positive bands at 1,663, 1,626, and 1,603 cm⁻¹ and the negative bands at 1,615 and 1,585 cm⁻¹ (Fig. 5e) that are all within 3 cm⁻¹ of analogous bands in Fig. 5f. In the region of vibrations of the neutral unlabeled ubiquinone-6, a new band appears at 1,648 cm⁻¹ in the case of Rp. viridis RCs (Fig. 5e). In the semiquinone region, the spectra (Fig. 5 e and f) are very different.

DISCUSSION

Under our experimental conditions where vitamin K₁ is added to the RC sample, the isotopic composition of the quinone is the only parameter that is varied. The Q_A⁻/Q_A FTIR spectrum obtained upon illumination of the Rp. viridis RCs incubated with unlabeled vitamin K₁ (Fig. 3a) is essentially the same as for RCs without added quinones (16, 30, 34), for which controls in the near IR region demonstrate that Q_A photoreduction is the only reaction involved. Thus, the excess of added quinone does not affect the Q_A⁻/Q_A spectrum. The observation that the Q_A⁻/Q_A spectra are strongly modified when the isotopic composition of the added vitamin K₁ is the unique parameter that is changed leaves little room for an explanation other than that implying the displacement of the native quinone from its binding site and its replacement by the isotopically labeled quinone. This further leads to the conclusion that the Q_A⁻/Q_A spectrum of the unlabeled vitamin K₁, which must also exchange with the native menaquinone of Rp. viridis, is essentially the same as for the native quinone.

A striking analogy is observed for the isotopically labeled minus unlabeled double-difference spectra obtained upon ¹⁸O or ¹³C labeling of vitamin K₁ in the RCs of Rp. viridis and Rb. sphaeroides (Fig. 5 a–d). This observation further supports the notion of quinone exchange in the Q_A site of Rp. viridis together with the idea that the bonding pattern of vitamin K₁ in the RCs of the two species must be similar. Thus, the isotopic shifts of the vitamin K₁ vibrations revealed in the FTIR spectra (Figs. 3 and 5 a and c) can be analyzed in terms of the bonding pattern of Q_A and Q_A⁻ in the Rp. viridis RC for the first time. Comparison with the Q_A⁻/Q_A spectra of Rb. sphaeroides RCs reconstituted with the same isotopically labeled quinones (Figs. 2 and 5 b and d) allows any difference in the Q_A site of these two species to be scrutinized.

Vitamin K₁ in the Q_A Site of Rp. viridis and Rb. sphaeroides.

The interaction of vitamin K₁ with the Q_A site of Rb. sphaeroides derived from a ¹⁸O minus ¹⁶O double-difference spectrum similar to that shown in the present study (Fig. 5b) has been described previously (16, 20). The proposed frequency and mode assignment for the various quinone vibrations are strengthened by the ¹³C minus ¹²C double-difference spectrum (Fig. 5d). Notably, the splitting of the single C=O band at 1,662 cm⁻¹ of isolated vitamin K₁ into two bands at 1,651 and 1,640 cm⁻¹ upon binding to the Q_A site, showing different bonding interactions of each quinone C=O with the protein, is confirmed. The C=O bands at 1,651 and 1,640 cm⁻¹ appear downshifted to 1,607 and 1,597 cm⁻¹ upon ¹³C labeling and to ≈1,628 and 1,620 cm⁻¹ upon ¹⁸O labeling, respectively. In the region of the three main semiquinone modes, ¹⁸O labeling essentially shifts only the middle band, whereas ¹³C labeling shifts the three bands by ≈40 cm⁻¹. This observation is consistent with a previous interpretation of the middle anion band predominantly in terms of the C—...O modes with the bands at higher and lower frequency being C—...C modes coupled to the C—...O modes (16). The spectra (Fig. 2) suggest that a derivative signal at 1,354(+)/1,342(−) cm⁻¹ downshifts to 1,315(+)/1,304(−) cm⁻¹ upon ¹³C labeling. A similar signal appears at 1,346(+)/1,342(−) cm⁻¹ in the Q_A⁻/Q_A spectrum of Rb. sphaeroides reconstituted with 2,3-dimethyl-1,4-naphthoquinone (17, 31). This signal probably corresponds to the upshift upon Q_A reduction of the C₂—CH₃ mode of vitamin K₁ absorbing at 1,330 cm⁻¹ in vitro (Fig. 1). In this frame, the ≈12-cm⁻¹ higher frequency of this mode of the neutral quinone in the Q_A site compared with solution suggests some conformational distortion of the C₂—CH₃ bond upon binding of vitamin K₁ to the Q_A site of Rb. sphaeroides.

The isotopic shifts described in the double-difference spectra provide highly specific fingerprints of the quinone vibrational modes (19, 20). Many of these vibrations are affected by interactions at the binding site. Thus, the close similarity of the double-difference spectra between Rp. viridis and Rb. sphaeroides for both the ¹⁸O and ¹³C labeling of vitamin K₁ stresses the similarities of the quinone–protein interactions in the RCs of the two species. The minor differences are interpretable in terms of small variations in the details of the bonding interactions of Q_A and Q_A⁻ in the RCs of the two species. The two quinone C=O modes of Rp. viridis appear at 1,653 and 1,637 cm⁻¹ due to different interactions of the carbonyls with the protein. One of these modes had been previously proposed at 1,636 cm⁻¹ (31). The splitting of the C=O modes is 16 cm⁻¹ in Rp. viridis and 11 cm⁻¹ in Rb. sphaeroides. The frequency shifts induced by the binding of each carbonyl are 9 and 25 cm⁻¹ for Rp. viridis and 11 and 22 cm⁻¹ for Rb. sphaeroides. Although the combined shifts of the two carbonyls (33–34 cm⁻¹) account for a binding energy of ≈5 kcal⋅mol⁻¹ (16) in both species, the asymmetry of the interactions is larger in Rp. viridis than in Rb. sphaeroides. Another notable difference is the position of the semiquinone C—...O modes at 1,438 and 1,444 cm⁻¹ in Rp. viridis and Rb. sphaeroides, respectively. However, the composition of the anion modes in terms of C—...O and C—...C vibrations seems essentially conserved in the two species. Some differences can also be noticed even for small bands such as the differential signal tentatively assigned to a C₂—CH₃ mode at 1,354/1,342 cm⁻¹ in Rb. sphaeroides, which would shift and broaden to 1,356/1,336 cm⁻¹ in Rp. viridis.

Ubiquinone-6 in the Q_A Site of Rp. viridis and Rb. sphaeroides.

The spectra of Rp. viridis incubated with vitamin K₁ (Fig. 3) or with ubiquinone-6 (Fig. 4) are very similar in the region above 1,660 cm⁻¹, where only contributions from nonquinone vibrations are expected. This indicates that spectrum 4a can be primarily assigned to the photoreduction of ubiquinone-6 in the Q_A site of Rp. viridis. Comparison with the spectra obtained for cytochrome photooxidation (34) or Q_B photoreduction (35) in Rp. viridis shows no indication of contribution from these states, notably to the small additional band appearing at 1,559 cm⁻¹ in the region of Amide II modes (Fig. 4). Incubation of Rp. viridis RCs into H₂¹⁸O or H₂¹⁶O does not affect the Q_A⁻/Q_A spectra regardless of whether vitamin K₁ is added (data not shown). This demonstrates the absence of contribution from ¹⁸O-exchangeable OH groups to the reported Q_A⁻/Q_A spectra. Thus, the ¹⁸O-induced isotope effect (Fig. 5 e and f) can be interpreted in terms of isotopic exchange of the carbonyls of ubiquinone-6 in the Q_A site.

In contrast to the very close similarity in the bonding pattern of vitamin K₁ in the Q_A site of Rp. viridis and Rb. sphaeroides, the interactions of ubiquinone-6 in the Q_A site of the two species, as revealed by the double-difference spectra (Fig. 5 e and f), show a number of significant differences. The most pronounced ones are the appearance of a new ubiquinone-6 C=O mode at 1,648 cm⁻¹ and the large alteration of the semiquinone modes in the case of Rp. viridis. As discussed previously (35), the differential signal at 1,360/1,367 cm⁻¹ (Fig. 4) indicates a conformation of the methyl group attached to C₅ of the quinone ring closer to that of Q_B than of Q_A. Similarly, the negative band at 1,283 cm⁻¹ suggests that the conformation of the methoxy groups of ubiquinone-6 in the Q_A site of Rp. viridis is different from that observed for Q_A in Rb. sphaeroides and is more similar to the constrained conformation found for one methoxy group of ubiquinone in the Q_B site of both species (35).

Comparison with X-Ray Structures.

When the x-ray structures of the Q_A binding site of Rp. viridis and Rb. sphaeroides are superimposed, an almost perfect overlap of the quinonic cycle of the menaquinone and of the ubiquinone is observed (8). The proposed residues involved in the hydrogen bonding interactions of the carbonyls of Q_A with the protein are also identical (5, 8). It is thus not too surprising that vitamin K₁, which has the same rigid ring structure as the native menaquinone of Rp. viridis, fits well in the Q_A binding site of both species. In Rb. sphaeroides, FTIR spectroscopy of ubiquinone selectively labeled on either one of the carbonyls in the Q_A site has led to the conclusion that it is the C=O group nearest to the non-heme Fe atom (i.e., at C₄; see ref. 16 for numbering of ubiquinone) that is strongly downshifted in frequency (18, 21). It is thus proposed that the most downshifted mode of vitamin K₁, at 1,641 cm⁻¹ in Rb. sphaeroides and 1,637 cm⁻¹ in Rp. viridis, is also the one closest to the non-heme Fe atom (i.e., at C₁; see Inset of Fig. 1 for numbering of vitamin K₁). This assignment is consistent with the shorter distance between the C₁=O carbonyl of menaquinone-9 and the Nδ of His M217 (2.9 Å) than between the C₄=O carbonyl and the peptide nitrogen atom of Ala M258 (3.2 Å) that is now deduced from the most recent structure of Rp. viridis (5).

The organization of the side chain of Ile M260 of Rp. viridis, equivalent to Met M262 in Rb. sphaeroides, with respect to the aromatic ring of the menaquinone or to the methoxy groups of ubiquinone in the Q_A site of the two species has been described in detail (8). The geometry of these side chains appears relevant to the present FTIR results, notably, those related to the conformation of the methoxy group at C₃ of ubiquinone in the Q_A sites of both species. In Rb. sphaeroides, the side chain of Met M262 is pointing away from this methoxy group. In Rp. viridis, the side chain of Ile M260 is directed more toward the aromatic ring of the menaquinone in the Q_A site and would clash with a methoxy group having the same conformation as in Rb. sphaeroides (8). Methoxy-substituted quinones are notorious for the dependence of their electron affinity on the dihedral angle defining the position of the O—CH₃ bond relative to the quinone ring plane, and it has long been proposed that the protein at the binding sites can tune the redox potential of ubiquinones by modulating the conformation of the methoxy groups (36, 37). The vibrational IR frequency of the carbonyls and the redox properties of ubiquinones in solution as a function of the conformation of the methoxy groups (in-plane or out-of-plane orientation of the O—CH₃ bond with respect to the quinone ring) have been recently investigated in detail (38). In the latter study, the carbonyl frequency observed for ubiquinone in the Q_A site of Rb. sphaeroides (18) and in the Q_B site of both Rb. sphaeroides and Rp. viridis (35) is shown to be consistent with the different orientation of the methoxy groups proposed for the two sites (35). In the present study, the different conformation of the Ile M260 side chain of Rp. viridis compared with that of Met M262 in Rb. sphaeroides provides a rationale for the presence of FTIR bands that is indicative of a methoxy group with a constrained conformation only when ubiquinone is present in the Q_A site of Rp. viridis. When compared with Rb. sphaeroides, the strain on ubiquinone-6 specifically induced by the side chain of Ile M260 in Rp. viridis could also explain the presence in this case of a distorted methyl group at C₅, the evidence for a band of Q_A at 1,648 cm⁻¹ indicative of a rather free carbonyl, and the alteration of the semiquinone modes. In the case of RCs containing vitamin K₁, where the effect of the side chain will differ mostly at the level of the aromatic ring, this difference of strain probably explains the different splitting of the corresponding C=C modes in the RCs of the two species. In turn, the observation of a new protein IR signal at 1,559 cm⁻¹ in the Q_A⁻/Q_A spectra of Rp. viridis when the native menaquinone-9 is exchanged for ubiquinone-6 may well reflect the difference of strain at the level of the Ile M260 side chain.

In summary, the native menaquinone-9 in the Q_A site of Rp. viridis can be exchanged for vitamin K₁ or ubiquinone by incubating the isolated RCs with an excess of these quinones. This simple procedure opens the road to further characterization of the quinone–protein interactions in this species by a variety of biophysical techniques, leading to precise comparison with the high-resolution structural data available for Rp. viridis. As an example, some differences in the bonding interactions of vitamin K₁ or of ubiquinone-6 in the Q_A site of Rp. viridis and of Rb. sphaeroides RCs have been revealed by FTIR spectroscopy in the present work. At least part of these differences can be related to the nature of the side chain at position M260 of Rp. viridis. Beside the possibility of allowing new spectroscopic experiments, the quinone exchange result also raises a number of interesting questions that are beyond the scope of the present work. Among those are the observation that in the native chromatophore membrane of Rp. viridis, where a pool of free ubiquinone is thought to be present, the Q_A site is only occupied by menaquinone-9. Furthermore, the exchange of Q_A in the isolated RC of Rp. viridis, which is reminiscent of the exchange of some of the chlorophyllic cofactors in Rb. sphaeroides RCs (39), demonstrates the existence of large-scale motions of the peptide backbone in isolated RCs at ambient temperature. This observation provides an additional example that the remarkable flexibility of proteins, so frequently overlooked in view of the visual impact of the still pictures used to describe the x-ray structures (40), also extends to membrane proteins.

ABBREVIATIONS

RC

reaction center

Q_A (Q_B)

primary (secondary) quinone acceptor

FTIR

Fourier transform infrared

vitamin K₁

2-methyl-3-phytyl-1,4-naphthoquinone

menaquinone-9

2-methyl-3-nonaprenyl-1,4-naphthoquinone

ubiquinone-10

2,3-dimethoxy-5-methyl-6-decaprenyl-1,4-benzoquinone