Protein dynamics in the bacteriorhodopsin photocycle - Europe PMC

Proc. Natl. Acad. Sci. USA Vol. 88, pp. 2388-2392, March 1991 Biophysics

Protein dynamics in the bacteriorhodopsin photocycle: Submillisecond Fourier transform infrared spectra of the L, M, and N photointermediates (purple membrane/active transport/energy transduction/kinetic)

MARK S. BRAIMAN*t, OLAF BOUSCHOt, AND KENNETH J. ROTHSCHILDt *Department of Biochemistry, University of Virginia Health Sciences Center, Charlottesville, VA 22908; and tPhysics Department and Program in Cellular Biophysics, Boston University, Boston, MA 02215

Communicated by Walther Stoeckenius, December 7, 1990 (received for review September 6, 1990)

ABSTRACT The usefulness of stroboscopic time-resolved Fourier transform IR spectroscopy for studying the dynamics of biological systems is demonstrated. By using this technique, we have obtained broadband JR absorbance difference spectra after photolysis of bacteriorhodopsin with a time resolution of z50 ps, spectral resolution of 4 cm.1, and a detection limit of AA 10-4. These capabilities permit observation of detailed structural changes in individual residues as bacteriorhodopsin passes through its L, M, and N intermediate states near physiological temperatures. When combined with band assignments based on isotope labeling and site-directed mutagenesis, the stroboscopic Fourier transform IR difference spectra show that on the time scale of the L intermediate, Asp-96 has an altered environment that may be accompanied by change in its protonation state. On the time scale of the L -* M transition, this Asp-96 perturbation/deprotonation is largely reversed, and Asp-85 becomes protonated. During the M -* N transition, Asp-85 appears to remain protonated but undergoes a change in its environment as evidenced by a shift of VC=O from 1761 to 1755 cm' . The retention of a proton on Asp-85 in the N state indicates that the proton transferred from the Schiff base to this residue in the L -* M step is not released to the extracellular medium during the same photocycle, but rather during a subsequent one. Also during the M- N transition, Asp-96 undergoes a deprotonation (possibly for the second time in a single photocycle). Bands in the amide I and amide H spectral regions in the M N difference spectrum indicate the occurrence of a conformational change involving one or more peptide groups in the protein backbone.

With spectral assignments from isotope labeling (13-15) and site-directed mutagenesis (16, 17), FTIR difference spectra were used previously to develop a model for the protonpumping mechanism that involved proton transfers among the retinal Schiff base and residues Asp-96, Tyr-185, Asp212, and Asp-85 (17). Along with a specific sequence of these proton transfers, this model included a detailed 3-dimensional structure for the retinal binding pocket and helices C, F, and G. This structural model took into account existing low-resolution information from both EM (18) and neutron diffraction (19). As it turns out, the detailed structural model deduced from FTIR spectroscopy (20) is very similar to that recently proposed (1) on the basis of electron cryomicroscopy with improved resolution. The EM results thus lend support to the general features of the mechanism previously proposed from FTIR spectroscopy. The mechanism proposed earlier (17) is thus a useful starting point for further investigations, although it is probably incorrect in a number of details and it is certainly incomplete. Very little FTIR spectral information was available about the later intermediates (N, 0) because they cannot be trapped in significant steady-state concentrations at low temperatures. However, these late intermediates are probably of great importance for the proton transport mechanism because uptake of a proton from the cytoplasmic medium is thought to occur on the time scale of M decay (21). To obtain a clear picture of IR spectral changes occurring during the photocycle under physiologically relevant conditions, submillisecond time resolution is needed. Previous kinetic IR spectra of bR on a submillisecond time scale were obtained from a series of single-wavelength measurements (15, 22). However, the technique known as stroboscopic time-resolved FTIR (TR-FTIR) (23) can provide broadband spectral information with a time resolution down to =-10 jus or better. Recent technical improvements (24) have reduced a number of problematic spectral artifacts that can arise with the stroboscopic technique. In this communication, we demonstrate that stroboscopic TR-FTIR can be applied to a biological system. The results are a series of evenly spaced IR difference spectra of bR and its photoproducts after an actinic flash. The TR-FTIR data confirm the general features of the previously proposed mechanism (17) and provide additional information about protein structural changes that occur when the N intermediate is formed at room temperature.

-

The recent publication of a high-resolution structure for bacteriorhodopsin (bR) based on EM (1) has focused attention on. relating the bR structure to its mechanism of lightdriven proton transport. Visible absorption spectroscopy (2, 3) originally established the cycle of transitions that occurs after light absorption by the retinal chromophore and showed that at room temperature the time scales of these reactions are in the range of 10 ps for bR -* K, 1 ,us for K -i L, 50 Us for L -+ M, and 5-10 ms for the M -+ N -O 0 -i bR steps. However, most of what is known about the actual structural changes corresponding to these transitions has come from vibrational spectroscopy. Resonance Raman spectroscopy has provided information selectively about the retinal chromophore (4-7) and more recently about aromatic residues (8). IR spectroscopy, on the other hand, is sensitive to changes throughout the protein. By trapping bR photoproducts through partial dehydration (9) or cooling (10-12), it has been possible to obtain very precise Fourier transform IR (FTIR) difference spectra corresponding to the bR -* K, bR -* L, and bR -* M transitions.

EXPERIMENTAL PROCEDURE Sample. As described (25), purple membrane pellets were sealed between 32-mm-diameter x 3-mm-thick CaF2 win-

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: bR, bacteriorhodopsin; FTIR, Fourier transform IR; TR-FTIR, time-resolved Fourier transform IR; sh, shoulder.

tTo whom reprint requests should be addressed.

2388

Biophysics: Braiman et al. dows. The IR spectrum showed the water content of the samples to be 50-70%o by weight. For the high pH samples, the membranes were first washed several times in 10 mM sodium borate buffer (pH 9.2).

uri

0I 0 0

Proc. Natl. Acad. Sci. USA 88 (1991)

2389

Photolysis Laser. The 532-nm frequency-doubled output of a Spectra-Physics DCR-11 or DCR-3 Nd'-yttrium/aluminum-garnet laser was used to illuminate the entire =4-mmdiameter region of the sample intersected by the IR measuring beam. The measured light intensity at the sample was -25 mJ'cm-2 per pulse. Stroboscopic Time-Resolved Difference Spectroscopy. Nicolet 60SX spectrometers with time-resolved option were used for all FTIR measurements. A version of the manufacturer's TR-FTIR program (TIMRES) was used to control both the spectrometer and the external photolysis laser. The TIMRES software was modified as described elsewhere (24) to permit accumulation of difference interferograms. For all experiments the minimum time between each flash-on spectrum and the subsequent no-flash reference spectrum was 2 s. A long-pass interference filter with 5-,um cutoff was placed in front of the detector to protect it from the scattered photolysis flash. The resulting decreased bandpass also allowed the interferogram to be sampled at fewer points. One-sided difference interferograms were collected with 4-cm-1 spectral resolution. The difference interferograms were apodized and Fourier transformed and then phase-corrected using the phase-array calculated for the no-flash reference interferogram. A number ofintensity difference spectra from the same sample were then averaged together and finally converted to absorbance difference spectra (25). See figure captions for additional data collection and plotting parameters used for each set of spectra.

RESULTS AND DISCUSSION Timing of Asp-96 Protonation Changes. TR-FTIR differ-

1700 1600 1500 1400 1300 1200 wavenumber (cm-') FIG. 1. TR-FTIR difference spectra of bR at 16.5°C. The indicated times represent the median delay between the photolysis flash and the digitized interferogram points used to calculate each spectrum. This figure combines two data sets obtained with different digitization timing parameters to cover two time ranges. To obtain both sets, a total of -3 x 105 flashes were used. For spectra A-K, the retardation velocity of the interferometer was 12.56 cm/s, and the interferogram was sampled once every 2 wavelengths of the He-Ne reference laser. Before digitization, the analog signal from the detector was subjected to a low-pass filter with a 111-kHz cutoff frequency. The first 98 interferogram points sampled after each flash were sorted one by one, resulting in a temporal spacing of 10.1,us and a total of 98 separate time-point spectra. Of these 98, spectrum A represents the average of 1-5; spectrum B represents 6-10; spectrum C, 11-15; spectrum D, 16-20; spectrum E, 21-25; spectrum F, 26-30; spectrum G, 31-35; spectrum H, 36-40; spectrum I, 41-45; spectrum J, 46-72; and spectrum K, 73-98. For spectra L-P, the velocity was 11.72 cm/s; the interferogram was sampled once every 4 He-Ne wavelengths; the low-pass frequency was 16 KHz; and the first 784 points digitized after each flash were sorted in groups of 16, resulting in a temporal spacing of 0.69 ms and a total of 49 spectra. Of these 49, spectrum L represents the average of 2-6; spectrum M, 11-20;

ence spectra covering a time range of :0.03-30 ms after photolysis are shown in Fig. 1. The earliest spectra show a clear negative band at 1741 cm-1. This feature loses intensity within several hundred As (see Fig. 2)-i.e., on the time scale of L -* M decay at 16.50C. This difference peak can thus be clearly identified as arising from formation of the L intermediate, and can furthermore be assigned to Asp-96 based on the similarity of the 30-ps time-resolved spectrum (Fig. 1A) to the low-temperature static bR -* L difference spectrum, in which the 1742-cm-1 peak was assigned to Asp-96 based on effects of mutations at this residue (17, 26). Additional support for this assignment comes from TR-FTIR measurements on Asp-96 mutants, to be published separately. The time scale of the disappearance of the 1741-cm-1 band in these TR-FTIR data agrees well with that seen previously in single-wavelength measurements at room temperature (28). These kinetics support the hypothesis that Asp-96 undergoes a change in its environment and deprotonation upon L formation, followed by a nearly complete reprotonation upon M formation (17). Gerwert et al. (26) previously came to different conclusions regarding the timing of Asp-96 protonation changes, based on their interpretation of low-temperature FTIR data. According to their conclusions, Asp-96 undergoes an alteration in its environment but no change in protonation state during the K -- L step. Instead, deprotonation of Asp-96 occurs in the time range of the M intermediate, and the subsequent reprotonation occurs when bR is reformed. However, the room temperature kinetic data shown in Figs. 1 and 2 permit a spectrum N, 22-30; spectrum 0, 31-40; spectrum P, 41-49. All spectra in this figure were obtained from the same sample and are plotted on the same scale, except for the bottom 5 spectra, which were expanded by the following factors as a crude correction for damping of the fastest spectral changes by the low-pass electronic filter of the spectrometer: spectrum A, 1.4x; spectrum B, 1.2x; spectrum C, 1.15x; spectrum D, 1.1x; spectrum E, 1.05x. a.u., absorbance units.

2390

Biophysics: Braiman et al. 3

r1)A

Proc. Natl. Acad. Sci. USA 88 (1991) If .3-LtK 4U9U9 )

0

o3

(.0 C

[( I-

0

~~~~~~~~8. .0

I

~~~~~~~~~~6. 6

7. 3

.9 2 4. 5 8

1)1

5.

CU0 0

AS

I 0

--2

I3

1

4 ~~~~~~~~~2.,

I

-7

.0

4

0 2 4 6 8 10 20 30 time after photolysis (ms) FIG. 2. Kinetic behavior of the positive band at -1400 cm-' (A and B) and the negative band at -1741 cm- (C and D). To improve the signal/noise ratio, the absorbance change was averaged over the range 1405-1390 cm-1 for A and B, 1743-1737 cm-1 for C, and 1745-1739 cm-1 for D. (Slightly different ranges were used in C and D to minimize contributions from neighboring bands.) The integrations were performed on the original data sets used to generate Fig. 1 before the temporal averaging was applied. Spacing of time points in A and C is thus 10.1 As; in B and D spacing is 0.69 ms. No correction for the low-pass electronic filter (see Fig. 1 caption) was

applied.

comparison of the intensity of the negative 1740-cm-1 band in the L and M states. They show quite clearly that this band reaches its maximum (negative) intensity during the L state. This result clearly rules out a deprotonation of Asp-96 when M intermediate is formed. Our data are, instead, consistent with the hypothesis that at physiological temperatures, Asp-96 is protonated in bR but deprotonated in the L intermediate, and this deprotonation is reversed during the L -- M transition. A possible alternative explanation that cannot be ruled out at this time is that Asp-96 remains protonated in L but undergoes a change in its environment that decreases the intensity of its C=O vibrational band relative to the bR state. To distinguish these two possibilities, it is necessary to assign either a positive COOH (1700-1760 cm-') or COO-(1400 cm-') band to Asp-96 in the bR -* L difference spectrum at room temperature. Neither such band can be identified with certainty in our current data, although a small absorbance increase at -1400 cm-1 possibly occurs on the L time scale (see Fig. 2A). In addition to the early changes in Asp-96, our kinetic data show another change correlated temporally with the M -+ N transition. There is a renewed increase in the intensity of the -1741-cm-1 negative band on the 5- to 10-ms time scale at 16.50C (Fig. 2D). Concomitantly, there is an intensity increase in a broad positive band at -1400 cm-' (see Figs. 2B and 3), which is in the appropriate frequency range for a COO- symmetric stretch. Both the negative 1741-cm-1 and positive 1400-cm-1 features are more easily discerned in TR-FTIR spectra obtained at high pH (see Fig. 4), which is known to favor the buildup of N (3). Other features that appear under similar conditions and confirm the presence of N are the positive bands at -1530 [shoulder (sh)] and 1186 cm-1, which are close in frequency to the strongest bands in the resonance Raman spectrum of N (7). By scaling and subtracting TR-FTIR spectra having different relative contributions of bR, M, and N, it is possible to generate a difference spectrum in which the predominant features correspond to the M -- N transition (Fig. 4 B and C). These differences of difference spectra show more clearly the

1 750 1 550

1 400

11 50

wavenumber (ecrs1) FIG. 3. TR-FTIR difference spectra of washed unbuffered purple membrane covering a range of 0-8 ms, obtained using data collection parameters similar to those described for Fig. 1, spectra U-P but at -20TC; same set of 12 spectra were used for each panel. In each panel spectrum 1 (corresponding to digitization times of 0.043-0.690 Ins) is at the bottom, and each upper spectrum corresponds to a range of times 0.690 ms later than the one immediately below it; within each panel all spectra are plotted on the same scale. a.u., absorbance units.

features ascribed to the N intermediate. It is evident that the small size of the intensity change near 1742 cm-' during the M -- N transition is a result of cancellation by overlapping positive difference bands at -1735 and 1753 cm-'. This cancellation can be avoided by substituting Asp-96 with glutamic acid (O.B., M.S.B., Y.-W. He, T. Marti, H. G. Khorana, K.J.R., unpublished data). This shifts the COOH frequency from 1742 to -1720 cm-1. In the Asp-96 -- Glu mutant, the -1720-cm-1 negative band shows a considerably larger intensity increase during the M -- N transition than occurs at 1742 cm-' with wild-type bR. In this mutant, therefore, Glu-96 clearly undergoes a deprotonation on the time scale of the M -+ N transition; this result supports our conclusion that Asp-96 undergoes a similar reaction in the wild-type protein. Our current FTIR results are thus in substantial agreement with the generally accepted interpretations of the effects of Asp-% mutations on conductivity changes (29) and displacement currents (30-32) that are related to the uptake of a proton from the cytoplasmic medium. These measurements showed that proton uptake (as well as reprotonation of the Schiff base during M decay) is slowed as a result of mutations at Asp-96 and led to the hypothesis that this residue donates a proton to the Schiff base during the M -- N transition and picks up another proton from the cytoplasmic medium during the N -* bR decay. The TR-FTIR method now permits a direct spectroscopic observation of Asp-96 protonation states that could previously only be inferred based on pHdependent effects of mutations at this residue. Protein Structural Changes Involving Asp-85. The 1761cm-1 positive band that appears in the bR -* M difference spectrum was previously assigned to Asp-85 (17), which undergoes a protonation during the photocycle. The kinetic FTIR spectra (Fig. 1) show that this band appears on the same time scale as M intermediate formation-i.e., on the time scale of Schiff base deprotonation. This and other recent FTIR evidence (33) supports the model that Asp-85 serves as the acceptor group for the Schiff base proton when M is formed (17). Previous measurements using a single-

Proc. Natl. Acad. Sci. USA 88 (1991)

Biophysics: Braiman et A

I i I

1 700

1 300 1 500 wavenumber (cm-1)

1100

FIG. 4. (Spectrum A) TR-FTIR difference spectrum of purple membrane prepared in pH 9.2 buffer, obtained using conditions similar to those of Fig. 3. However, to improve the signal/noise ratio of this bR

-*

(M

+

N) difference spectrum all of the time-point spectra

covering a range of 0.690 ms-34.5 ms were averaged together. (SpecN difference spectrum obtained by subtracting a bR -+ trum B) M M difference spectrum (Fig. 1 spectrum L) from the high-pH bR (M + N) spectrum (spectrum A). The relative scaling was selected to minimize the magnitude of positive or negative absorbance at 1169 and --

cm-'.

M -* N difference spectrum obtained by 1526 (Spectrum M difference spectrum obtained with 2-ms delay subtracting the bR from a bR -* (M + N) spectrum obtained by time (Fig. 1 spectrum -+

averaging spectra from the same time series with 11- to 30-ms delay times (Fig. 1 spectra M-P). The former was scaled by a factor of 0.79 relative to the latter. The negative band at 1569 cm-' and the positive band at 1530 cm-l correspond to the C=C stretch frequencies of M and N, respectively (7).

wavelength technique also demonstrated clearly that the rise of the

1761-cm-1 peak was associated with the L

--

M

decay,

although its appearance was attributed to the uptake by a carboxylate group of a proton from the cytoplasmic medium rather than from the Schiff base (15). On a time scale of -1 ms at 20TC, the 1761-cm-1 band undergoes a partial shift of intensity to 1754 cm-1 (Fig. 3). Subsequently, both 1761- and 1754-cm-1 bands appear to decay in parallel. The presence of two positive bands in this region with different kinetic behavior was demonstrated previously using single-wavelength measurements (28) as well as FTIR spectroscopy at temperatures near freezing (27, 34, 35), but these bands were attributed to two distinct aspartic acid residues gaining a proton during M formation (15, 35). However, our TR-FTIR data suggest an alternative interpretation-namely, that the two peaks are both due to the same residue but in different photocycle intermediates. (i) TR-FTIR spectra obtained under conditions known to favor N formation (Fig. 4) show the down-shifted band (at -1754 cm-'). This demonstrates that this band is most likely due to N intermediate (rather than to M or 0). (ii) The fact that the rise of the 1754-cm-1 band is accompanied by the decay of that at 1761 cm-1 (Fig. 3) indicates that the former could be evolving from the latter. The presence of both 1754and 1761-cm-1 components in the later spectra in Fig. 3 is consistent with the fact that M and N are in equilibrium on this time scale at room temperature (31, 37). (iii) We considered the possibility that the ==1754-cm-1 band is due to Asp-96, which is expected to reprotonate late in the photocycle (31). However, this assignment is ruled out by recently obtained TR-FTIR spectra, which show that neither the 1761-

2391

nor the 1754-cm-1 band is affected by mutations at Asp-96 (O.B., M.S.B., Y.-W. He, T. Marti, H. G. Khorana, K.J.R., unpublished data). Thus, it appears that Asp-85 is protonated in both M and N intermediates but has a decreased frequency in the latter intermediate. An important implication of this is that the proton transferred from the Schiff base to Asp-85 cannot be the proton that gets released to the external medium during the same photocycle, because the proton release has been measured to occur on a time scale similar to (or faster than) M formation (21). Hence, at least two photocycles are needed to transfer the proton from the Schiff base to the external medium. The well-established correlation of COOH carbonyl frequency with pKa (36) suggests that during the M -* N transition the pKa of Asp-85 is increased, by as much as -0.5 pH units. Such a change in pKa could be caused by the removal of a positive charge from the vicinity of this group. The protonation of the nearby Schiff base during the M -* N step (7) would be expected to cause the opposite effect and is therefore unlikely to explain the observed Asp-85 COOH band shift. Instead, a local structural alteration in this region of the C helix might occur, affecting the environment of Asp-85. Protein Structural Changes Involving the Peptide Backbone. A positive band at -1555 cm-1 in the amide II region and a negative band at -1670 cm-' in the amide I region both appear to increase in intensity during the M -* N transition as seen in Figs. 1, 3, and 4. Thus, structural alterations in the protein backbone do not reach their maximum extent at the M state but continue to increase at least until the N state is formed. The observation of the amide II peak growing during N formation prompts a reinterpretation of earlier rapid-sweep FTIR results obtained with 5-ms time resolution (25). The 1555-cm-1 peak observed in these spectra was seen to increase in intensity during the first -1 ms after photolysis, and this change was attributed to a conformational change occurring within the M intermediate. However, the kinetic behavior and pH dependence of the -1555-cm-1 band in Figs. 1, 3, and 4, as well as the observation of the associated positive band at 1535 cm-1 (made possible by the improved spectral resolution), all indicate that the previously observed 1555-cm-1 band was probably due to the N intermediate. The conformational change that we now attribute to the M -* N conversion appears likewise to be related to that observed (via static FTIR difference spectroscopy) on a time scale of several minutes at temperatures in the 240-260 K range (35). A shift in (negative) intensity from 1659 to 1674 cm-1, observable during the first few minutes after photolysis at these low temperatures, was also attributed to an M decay process distinct from the M -- N conversion. However, the similarity of the spectral changes measured at 240 K (35) to those observed in our room-temperature data (especially in those obtained at high pH) suggests that the transition observed at low temperature may, in fact, correspond to the M -N decay. The presence of a protein conformational change during the M -- N transition argues for modifying the "C-T" model for proton pumping by bR (7). Although our data indicate that a protein backbone conformational change occurs during the L -* M transition (since difference peaks in both amide I and amide II regions appear on the time scale of M formation), there appears to be another backbone conformational change at least as great in magnitude during the M -+ N transition. The C-T model as proposed by Fodor et al. (7) thus represents an oversimplification of the protein conformational changes occurring during the photocycle. The protein conformational changes during the M -- N transition are probably the rate-limiting process for the Schiff

2392

Biophysics: Braiman et al.

base reprotonation, which occurs during this step. Further analysis of these conformational changes with TR-FTIR spectroscopy could provide important information about how protons are transferred over the -E10-A hydrophobic gap between Asp-96 and the Schiff base (1). The conformational rearrangement in N may also play an important role in the subsequent reisomerization of the chromophore-e.g, by bringing a catalytic residue into an appropriate position to decrease the C13=C14 bond order and/or by providing steric constraints on the chromophore-binding pocket that favor the formation of the trans isomer. To test such possibilities, additional TR-FTIR studies are needed.

CONCLUSIONS Our TR-FTIR data are consistent with Asp-96 undergoing two distinct transient structural changes during the photocycle. The first occurs when L is formed and reverses when L decays to M; this could reflect a transient deprotonation of Asp-96 in L as proposed previously (17). However, without a clear assignment of either a positive COO- or COOH band due to Asp-96 in the L state, it is impossible to be certain whether Asp-96 is deprotonating or merely undergoing a change in its environment as proposed by other workers (26). The second change in Asp-96 occurs at the N state. Our data demonstrate clearly that Asp-96 loses its proton when the N intermediate is formed-i.e., on the same time scale that the Schiff base group is protonated. A similar conclusion was drawn from recent FTIR spectra with 7-ms time resolution (27). These conclusions are consistent with the previously postulated proton transfer from Asp-% to the Schiff base group during the M -- N transition (31). Asp-85, meanwhile, accepts a proton during M formation and appears to remain protonated in the N state, although in an environment different from that in M. The observed retention of a proton on Asp-85 throughout the lifetime of M and N suggests that the proton-release mechanism is more complicated than a simple sequential transfer of a proton from the Schiff base to Asp-85 and then to the external medium. Instead, our TR-FTIR data are consistent with a previously postulated mechanism in which Asp-85 protonation triggers proton ejection to the external medium through unshielding of a nearby cationic residue (17). According to this hypothesis, the protonation of Asp-85 disrupts its salt bridge with nearby Arg-82. The unshielded positive charge of the latter group causes a nearby water molecule to lose a proton and form a hydroxide counter-ion for Arg-82, and the released proton moves into the external medium. In summary, our results support, in general, the proposed models (1, 17, 30-32, 37) that show Asp-96 as a temporary acceptor for protons passing from the cytoplasmic medium to the Schiff base and Asp-85 as a temporary acceptor for protons passing from the Schiff base to the external medium. However, none of these previously proposed models can fully account for the timing of structural changes of these two residues as observed in the current work, indicating that our understanding of the bR proton pump mechanism is still incomplete. We are grateful to Pal Ormos for providing us a copy of his manuscript before publication, and to M. Dufiach, M. Heyn, H. G. Khorana, T. Marti, P. Rath, and S. Subramanian for helpful discussions. K.J.R. acknowledges the technical assistance of Y.-W. He and R. Dagostino. M.S.B. thanks A. Klinger and Dr. K. J. Wilson for technical assistance, Drs. J. Hockensmith and L. Andrews for loans of Nd+-yttrium/aluminum-garnet lasers, and the State of Virginia Equipment Trust for the purchase of a Nicolet 60SXR spectrometer. M.S.B. is a Lucille P. Markey Scholar, and this work was supported by a grant from the Lucille P. Markey Charitable Trust to M.S.B. and

Proc. Natl. Acad. Sci. USA 88 (1991) by grants from the National Science Foundation (DMB-8806007), National Institutes of Health (EY05499), and Office of Naval Research (N00014-88-K-0464) to K.J.R. 1. Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E. & Downing, K. H. (1990) J. Mol. Biol. 213, 899-929. 2. Lozier, R. H., Bogomolni, R. A. & Stoeckenius, W. (1975) Biophys. J. 15, 955-962. 3. Kouyama, T., Nasuda-Kouyama, A., Ikegami, A., Mathew, M. K. & Stoeckenius, W. (1988) Biochemistry 27, 5855-5863. 4. Braiman, M. S. & Mathies, R. A. (1982) Proc. Natl. Acad. Sci. USA 79, 403-407. 5. Argade, P. V. & Rothschild, K. J. (1983) Biochemistry 22, 34603466. 6. Smith, S. O., Lugtenburg, J. & Mathies, R. A. (1985) J. Membr. Biol. 85, 95-109. 7. Fodor, S. P. A., Ames, J. B., Gebhard, R., van den Berg, E. M. M., Stoeckenius, W., Lugtenburg, J. & Mathies, R. A. (1988) Biochemistry 27, 7097-7101. 8. Harada, I., Yamagishi, T., Uchida, K. & Takeuchi, H. (1990) J. Am. Chem. Soc. 112, 2443-2445. 9. Rothschild, K. J., Zagaeski, M. & Cantore, W. A. (1981) Biochem. Biophys. Res. Commun. 103, 483-489. 10. Rothschild, K. J. & Marrero, H. (1982) Proc. Natl. Acad. Sci. USA

79, 4045-4049. 11. Bagley, K., Dollinger, G., Eisenstein, L., Singh, A. K. & Zimanyi, L. (1982) Proc. Natl. Acad. Sci. USA 79, 4972-4976. 12. Siebert, F. & Mantele, W. (1983) Eur. J. Biochem. 130, 565-573. 13. Roepe, P., Ahl, P. L., Das Gupta, S. K., Herzfeld, J. & Rothschild, K. J. (1987) Biochemistry 26, 6696-6706. 14. Dollinger, G., Eisenstein, L., Lin, S.-L., Nakanishi, K. & Termini, J. (1986) Biochemistry 25, 6524-6533. 15. Engelhard, M., Gerwert, K., Hess, B., Kreutz, W. & Siebert, F. (1985) Biochemistry 24, 400-407. 16. Braiman, M. S., Mogi, T., Stern, L. J., Hackett, N. R., Chao, B. H., Khorana, H. G. & Rothschild, K. J. (1988) Proteins Struct. Funct. Genet. 3, 219-229. 17. Braiman, M. S., Mogi, T., Marti, T., Stern, L. J., Khorana, H. G. & Rothschild, K. J. (1988) Biochemistry 27, 8516-8520. 18. Henderson, R. & Unwin, P. N. T. (1975) Nature (London) 257, 28-32. 19. Heyn, M. P., Westerhausen, J., Wallat, I. & Seiff, F. (1988) Proc. Natl. Acad. Sci. USA 85, 2146-2150. 20. Rothschild, K. J., Braiman, M. S., Mogi, T., Stern, L. J. & Khorana, H. G. (1989) FEBS Lett. 250, 448-452. 21. Holz, M., Lindau, M. & Heyn, M. P. (1988) Biophys. J. 53, 623-634. 22. Siebert, F., Mintele, W. & Kreutz, W. (1980) Biophys. Struct. Mech. 6, 139-146. 23. Durana, J. F. & Mantz, A. W. (1979) in Fourier Transform Infrared Spectroscopy, eds. Ferraro, J. R. & Basile, L. J. (Academic, New York), Vol. 2, pp. 1-77. 24. Braiman, M. S. & Wilson, K. J. (1989) Proc. SPIE Int. Soc. Opt. Eng. 1145, 397-399. 25. Braiman, M. S., Ahl, P. L. & Rothschild, K. J. (1987) Proc. Nat!. Acad. Sci. USA 84, 5221-5225. 26. Gerwert, K., Hess, B., Soppa, J. & Oesterhelt, D. (1989) Proc. Natl. Acad. Sci. USA 86, 4943-4947. 27. Gerwert, K., Souvignier, G. & Hess, B. (1990) Proc. Natl. Acad. Sci. USA 87, 9774-9778. 28. Siebert, F., Mantele, W. & Kreutz, W. (1982) FEBS Lett. 141, 82-87. 29. Marinetti, T., Subramanian, S., Mogi, T., Marti, T. & Khorana, H. G. (1989) Proc. Nat!. Acad. Sci. USA 86, 529-533. 30. Butt, H. J., Fendler, K., Bamberg, E., Tittor, J. & Oesterhelt, D. (1989) EMBO J. 8, 1657-1663. 31. Otto, H., Marti, T., Holz, M., Mogi, T., Lindau, M., Khorana, H. G. & Heyn, M. P. (1989) Proc. Natl. Acad. Sci. USA 86, 9228-9232. 32. Holz, M., Drachev, L. A., Mogi, T., Otto, H., Kaulen, A. D., Heyn, M. P., Skulachev, V. P. & Khorana, H. G. (1989) Proc. Natl. Acad. Sci. USA 86, 2167-2171. 33. Rothschild, K. J., Braiman, M. S., He, Y.-W., Marti, T. & Khorana, H. G. (1990) J. Biol. Chem. 265, 16985-16991. 34. Gerwert, K. (1988) Ber. Bunsenges. Phys. Chem. 92, 978-982. 35. Ormos, P. (1991) Proc. Natl. Acad. Sci. USA 88, 473-477. 36. Bellamy, L. J. (1968) The Infrared Spectra of Complex Molecules: Advances in Infrared Group Frequencies (Chapman & Hall, London), Vol. 2. 37. Ames, J. B. & Mathies, R. A. (1990) Biochemistry 29, 7181-7190.