Proton translocation mechanism and energetics in the light-driven pump bacteriorhodopsin

241

Biochimica et Biophysica Acta, 1183 (1993) 241-261 © 1993 Elsevier Science Publishers B.V. All rights reserved 0005-2728/93/$06.00

Review

B B A B I O 43935

Proton translocation mechanism and energetics in the light-driven pump bacteriorhodopsin Janos K. Lanyi

*

Department of Physiology and Biophysics, University of California, lrvine, CA 92717 (USA) (Received 23 July 1993)

Key words: Bacteriorhodopsin; Retinal protein; Retinal isomerization; Proton pump; Proton transiocation; Photocycle; (Halobacteria)

Contents I.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . .

~...................................

242

II.

Overview of the transport m e c h a n i s m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

242

III.

T h e chromophore reaction cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Spectroscopic properties of the photocycle intermediates . . . . . . . . . . . . . . . . . . . . . . . . . . B. Kinetic description of the reactions of the photocycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

243 243 244

IV.

Proton transfer pathways suggested by the structure of bacteriorhodopsin and static interactions a m o n g residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. T h e extracellular proton release domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. T h e cytoplasmic proton uptake domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

245 245 247

V.

Deprotonation of the retinal Schiff base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Proton transfer to D85 follows decrease of the proton affinity of the Schiff base . . . . . . . . . . B. M e c h a n i s m of the protonation of D85 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

247 247 248

VI.

Proton release at the extracellular surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Proton affinity of the extracellular proton release group . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Composition of the extraceilular proton release complex . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Sequence of internal to external proton transfers in the photocycle . . . . . . . . . . . . . . . . . . . D. Change of the environment of D96 in the L intermediate . . . . . . . . . . . . . . . . . . . . . . . . . .

248 249 249 250 250

VII.

T h e Schiff base reprotonation switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. There are two plausible mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. T h e reprotonation switch is the M 1 ~ M 2 reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Does the switch reside in the retinal or the protein? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

251 251 251 252

VIII. Reprotonation of the Schiff base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Proton conduction between D96 and the Schiff base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

253 253

* Corresponding author. Fax: + 1 (714) 8568540. Abbreviations: J, K, L, M, N and O are the photointermediates of the bacteriorhodopsin photocycle; B R is the initial state. Superscripts, where given, refer to the net protonation of the protein relative to the initial state. Subscripts for M substates refer to pre and post-switch states (i.e., M 1 and M2), or an M with changed N-like protein conformation (i.e., MN). D, E, N, R, Q, T, S, V, Y and F designate aspartate, glutamate, asparagine, arginine, glutamine, threonine, serine, valine, tyrosine and phenylalanine residues, respectively. Proteins with residue replacements are designated with the wild-type residue, its n u m b e r and the replacement, e.g., D96N.

242 B. Chromophore reaction kinetics in the M to BR pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Proton transfer to the Schiff base is the consequence of a decrease of the proton affinity of D96 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

253

IX.

Proton uptake on the cytoplasmicsurface, and recoveryof the initial state . . . . . . . . . . . . . . . . . A. Relationship of proton uptake and the reisomerizationof the retinal . . . . . . . . . . . . . . . . . . B. The charge state of D96 influences proton uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Recoveryof the initial BR state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

254 255 255 255

X.

Thermodynamicsof the photocycle; coupling between chromophore and proton transfer reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Dissipation of free energy in the cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Enthalpy and entropy cycles:conversionof enthalpy into entropy at the reprotonation switch.

256 256 257

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

257

XI.

254

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

258

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

258

I. Introduction The various ion pumps in cells and organelles that generate transmembrane electrochemical potential for protons are driven in various ways: changes in substrate-binding energy during chemical reactions (ionmotive ATPases and N A D H / N A D P transhydrogenase), electron transfer after photoexcitation and during redox reactions (photosynthetic reaction centers, cytochrome oxidase and the cytochrome bc-complex), or directly through p K a changes set off by lightd e p e n d e n t b o n d - r o t a t i o n s in retinal (bacteriorhodopsin). The last of these is the simplest. It is based on easily visualized acid-base reactions that require no more than what is probably the smallest imaginable functional membrane-spanning protein. Bacteriorhodopsin, the best understood ionic pump, has become the paradigm for both the structure of the transmembrane core of membrane proteins and the internal workings of proton transport systems. Bacteriorhodopsin is an integral membrane protein (26 kDa) whose seven transmembrane helical segments ( A - G ) enclose a binding pocket for the all-trans-retinal chromophore [76], bound via a protonated Schiff base to K216 near the center of helix G and inclined about 20 ° [79,99] from the plane of the membrane. A naturally occurring extended two-dimensional hexagonal lattice ('purple membrane') comprised of trimers of this protein has made it possible to determine its three-dimensional electron density map at 3.5-7 A resolution [12,76,77]. From this map and the primary and predicted secondary structures, a structural model with nearly atomic resolution was constructed [76]. Although the exact spatial dispositions of some of the residues, particularly at the membrane surfaces, are still uncertain, this model is the point of departure for all attempts to describe the transport mechanism. Illumination of bacteriorhodopsin initiates a multistep reaction cycle that begins with isomerization of the retinal from the 13-trans-15-anti configuration (the

'light-adapted' chromophore) to 13-cis-15-anti [6,22], and proceeds through a series of thermal steps that translocate a proton across the membrane. Spectroscopic changes have identified the intermediates that accumulate in photostationary states when this 'photocycle' is arrested at cryogenic temperatures (e.g. Refs. 15,102,164). Photoexcitation with light pulses shorter than the lifetimes of the intermediate states made it possible to follow the interconversions of these states at ambient temperature and during a single turnover (e.g., Refs. 4,21,64,118,195,200). The retinal and its configurational transformations have been described in such studies by a variety of spectroscopic methods, including visible, UV, resonance Raman, FTIR, and N M R spectroscopy, and replacement of the retinal with analogues, while changes in the protein have been revealed by F T I R and UV spectroscopy and by studies utilizing site-specific residue replacements. The timecourse of proton exchange between the protein and the bulk has been followed, in turn, using pH-indicator dyes and photoelectric measurements. During the last few years a considerable body of work utilizing these approaches firmly established the basic elements of the proton transport. The purpose of this review is to assemble a coherent mechanistic and thermodynamic model from the diverse findings, and to indicate what information is missing or appears contradictory. Particular emphasis is placed on those residue interactions in bacteriorhodopsin that modulate proton affinities along the pathway of translocation because these should be relevant for proton pumps in general. For additional information on bacteriorhodopsin and other bacterial rhodopsins, as well as other pints of view, the reader is referred to many recent reviews [48,76,86,94,111,135, 151,170,182]. II. Overview of the transport mechanism It should be emphasized that the only light-dependent event in the 'photocycle' of bacteriorhodopsin is

243

cytoplasmic side

Fig. 1. An approximate structure for bacteriorhodopsin containing all-trans-retinal and showing residues important for proton transport. It is based on a structural model derived primarily from electron diffraction [76]. The seven helices span the lipid bilayer, not shown. Helices F and G are partly cut away to show the interior. The proton trajectory is shown as proton transfer from the Schiff base to D85, proton release on the extracellular side, reprotonation of the Schiff base from D96, and proton uptake on the cytoplasmic side that reprotonates D96. Thus, from a functional standpoint the protein may be divided into an extraceUular proton release domain, the Schiff base domain, and a cytoplasmic proton uptake domain (indicated by dashed lines).

cess of the Schiff base to D85 is then blocked by a reaction postulated to be the central event in all ionic pumps: the transfer of the access of the active center from one membrane side to the other. In bacteriorhodopsin it is termed the reprotonation switch. Reprotonation of the Schiff base occurs therefore not from the extracellular side but from D96 located on the cytoplasmic side. The latter residue is reprotonated from the cytoplasmic aqueous phase in a reaction loosely linked to reisomerization of the retinal. The initial state recovers as a proton is transferred from D85 to the group in the extracellular domain which had released the proton. Under some conditions the order in which these steps occur is different. Regardless of their sequence, however, the net result of these proton transfers is that a proton appears in the extracellular phase and a proton disappears from the cytoplasmic phase. Thus, a proton (and a positive charge) is translocated from one side to the other. The reaction cycle is most conveniently followed by measuring changes in the absorption spectrum of the centrally located retinal chromophore because these reflect most of the events that occur in this small protein. This review will therefore begin with a discussion of the chromophore reaction cycle. It will be followed by considerations of the possible routes for proton transfer suggested by the protein structure, the proton transfer reactions as reported by the retinal chromophore and by the more direct FTIR measurements, in the sequence they occur, and finally the energetics of the pump.

III. The chromophore reaction cycle the initial isomerization of the retinal. All subsequent steps are thermal relaxations as in enzyme reaction cycles and in the translocation cycles of other ion pumps. In this sense the fact that bacteriorhodopsin is a light-driven pump is not essential in understanding its transport mechanism. As must be the case in any pump, the translocation of the proton during the transport cycle comprises sequential transfer steps between pairs of donor and acceptor groups extending across the protein. Fig. 1 shows a rough sketch of the structure of the protein, with the approximate positions of the retinal and some of the groups that play part in the transport, as well as the path of the transported proton. Three of the groups are critical: the protonated Schiff base, the anionic D85 near the extracellular side, and the protonated D96 near the cytoplasmic side. They divide the protein into three domains, the Schiff base domain, the proton release domain and the proton uptake domain. The first proton transfer is from the Schiff base to D85. Protonation of D85 is followed by release of a proton on the extraceUular side into the aqueous phase. Ac-

The photocycle contains numerous intermediate states characterized by absorption maxima in the visible and UV, as well in the infrared, designated as J, K, L, M, N and O. In most cases substates of these have also been described (K and KL, M 1 and M 2, etc.). The intermediates arise and decay in a roughly linear sequence after absorption of the photon. The lifetimes of these states range from femtoseconds to milliseconds; near physiological conditions the turnover of the cycle of the wild-type protein is under a few tens of milliseconds.

IliA. Spectroscopic properties of the photocycle intermediates As the initial BR state, each of the intermediates has a single broad asymmetric absorption band in the visible. The maxima * of all but M, that has a strongly * The photointermediates are often designated with the measured (or estimated) wavelength maxima as subscripts, e.g., K600, L550, M41o, N560, and 0640.

244 blue-shifted maximum at about 410 nm, are between about 550 and 620 nm. J, K and O are red-shifted and L and N are blue-shifted relative to BR. In the UV, all intermediates between the initial few and the last state in the cycle exhibit a band near 335 nm, the 'c/s-peak,' that reflects the 13-c/s isomeric state of the retinal (e.g., Ref. 92). FTIR difference spectra between the photointermediates and the unphotolysed protein contain many bands that reveal the state of ionizable protein residues. The most obvious infrared features that identify the intermediates are the following. The L state is characterized by a complex spectral feature near 1740 cm-1 due to mostly to perturbation of D96 [21,23,62,107]. The spectrum of the M state contains a positive band at 1761 cm -~ due to the protonation of D85, and exhibits greatly decreased chromophore band intensities [21,23,169]. In the N state the 1761 cm -~ band shifts to 1755 cm-1 and a negative band at 1742 cm -1 appears from deprotonation of D96 [21,23,107, 147], while strong bands due to amide I and II vibrations reflect protein backbone changes [20,21,139,140, 146]. In the O state the 1755 cm -~ positive band persists but the 1742 cm -1 negative band is absent because D96 is reprotonated. The downshifted ethylenic stretch band at 1505 cm-~ is useful in following the O intermediate [169].

III-B. Kinetic description of the reactions of the photocycle The kinetics of a reaction sequence with more than five intermediate states would be difficult to solve in any case [126], but in a practical sense an exact solution is virtually precluded by the considerable overlap of the spectra of most of the intermediates with one another and the spectrum of the initial state, and the fact that many of the interconversions have similar rate constants. In general, the solution of the rate equations of a complex reaction is a matrix of the measurable relaxation constants and amplitudes that are model-dependent functions of the desired elementary rate constants. Global analyses of the time-course of absorption changes at selected wavelengths have identified these phenomenological parameters [104,112,125, 126,128,169] but so far have not uniquely defined a photocycle model. However, it is clear that a single sequence containing only unidirectional reactions is inconsistent with the data [128]; the models proposed over the last decade to account for the observed multiple rise and decay time-constants, particularly for the M intermediate, include parallel photocycles [10,16,32, 41,50,71,148], branched photocycles [14,41,162], single unbranched photocycles but with one or more reversible reactions [4,64,184,189,191,201], and a twophoton cycle in which the slower decaying M is produced by photoreaction of the N intermediate [57,90].

It was suggested recently [104] that spectroscopic data alone are not sufficient to decide fully among these alternatives, and solution of the kinetics will require additional and independent information, e.g., on the structure and ionization state of residues and the proton release and uptake kinetics, during the reactions of the photocycle. Indeed, this kind of pragmatic approach seems to work. Even with incomplete data, a degree of concensus based on time-resolved FTIR [64,169] and resonance Raman [4] spectra, as well as kinetics derived from visible spectroscopy [118,189,191, 201,202], is developing in favor of the simplest of the viable photocycle alternatives: a single reaction sequence with reversible reactions, hv

BR~K ~ L,--,M ~ N ~ O ~ BR. However, this scheme is clearly an oversimplification. A model with one M intermediate leads to unacceptable discrepancies in the 10 tzs to 100 ~s time range where the L to M interconversion takes place. Introducing another intermediate between L and M yielded calculated spectra for some of the putative intermediates that were obviously mixtures, and thus did not solve the problem [104]. On the other hand, postulating two sequential M substates (M 1 and M 2) connected by a unidirectional reaction gave a simple and reasonable explanation for the fact that near neutral pH the concentration of L tends to zero at the time when M reaches its maximal concentration in spite of the L to M equilibrium [189,201]. As discussed below, the existence of more than one M state is now supported by more direct observations. The persistence of increasing amounts of L during the lifetime of M as the pH is lowered from 7 to 4, as well as the pH-dependent reversal from net proton release to net proton uptake (see below), suggested to at least one research group that the ph.otocycle proceeds along one of two alternative pathways after the L to M reaction [202]. As shown in Fig. 2, the events after L are somewhat different in the proposed two pH-dependent alternative branches. The sequence of proton release followed by uptake at high pH is reversed in the low pH pathway (the net protonation state of the protein relative to BR is shown in Fig. 2 with superscripts). In the high pH pathway mainly N accumulates, while in the low pH pathway mainly O [27]. This feature of the model is a possible explanation for the observation that the transient accumulation of O is strongly pH-dependent [31,50,97,191] even though the rates of its rise and decay are relatively pH-independent [50,137]. Very rapid and pH-dependent equilibrium between N and O in a single sequence would be another explanation for this [30], but according to several recent reports the equilibrium is not rapid

245

0 (°).

BR

O(+1 )

K (o)

H+

\

N(O)H+-'~ (o) 0

H+

L(°)

(/owpH)

a(l)

(highpH)

(o)

(-1)

M2

I (-I)

M (o)

~ / "-'1 H+

Fig. 2. A photocycle model for bacteriorhodopsin, including both protonation reactions in the protein and chromophore reactions. Intermediates before the state labeled K (referred to as KL by some authors) are not shown. The superscripts denote the net protonation of the protein relative to the BR state. Because the sequence of proton release and uptake depend on the pH, the cycle is shown with a branch at M~°). At this intermediate the pH relative to the pK a of an extracellular group XH decides the course of events [202]. At pH > 6 a proton is released promptly from this group and thus release precedes the proton uptake on the cytoplasmic side, but at pH < 6 a proton is not released by this group and uptake occurs before the later release probably directly from D85. M 1 and M E (regardless of their superscripts) refer to access of the Schiff base to either D85 or D96, respectively [191]. The intermediate M N thought to arise after M2 [156] is not shown for the sake of simplicity. The sequence of proton uptake and retinal reisomerization (the latter being the N to O reaction) is different in the low and high pH pathways, and result in two O states at low pH and two N states at high pH. The accumulation of N at high pH and O at low at pH, respectively, is small or even insignificant.

enough, since the rise and decay times of N and O do not coincide [4,169]. Given this kind of a branched scheme, the photocycle will be difficult to study in the p H region between 5 and 8. From the point of view of the c h r o m o p h o r e the transformations are described by a linear sequence, although only approximately. However, it cannot be ignored that from the point of view of the protonation reactions the cycle proceeds along two alternative M to B R pathways. Exact descriptions of the kinetics are therefore simpler at the two p H extremes where there are single pathways. In addition, such an approach has the advantage that at p H > 8 the O state and at p H < 5 the N state does not accumulate significantly in the cycle, and the data are simpler. Recent studies under these two conditions have yielded some insights into the details of the two proposed M to B R pathways. At high pH, strong evidence for a significant N ~ M back-reaction was provided by two-flash perturbation kinetics for M [44,199]. This feature, as well as the

kinetics of M, N, and the proton uptake, require the existence of two sequential N substates that differ in protonation [199]. The high p H pathway is shown in Fig. 2 in this way. At low pH, comparison of the c h r o m o p h o r e and proton kinetics suggested, in turn, the existence of two sequential O states that differ in protonation [27]. Although there is much recent evidence that favors various parts of the scheme in Fig. 2, general agreement on a photocycle model has not been reached by any means (e.g., Refs. 57,181). The details of the photocycle thus continue to be under intense study and are matters of controversy. Finding the correct kinetic model is important because the photointermediates represent not only distinct states of the chromophore, but also distinct stages in the proton translocation process. This is most obvious in the case of the M state(s). Convincing evidence, such as lack of deuterium effect on the C = N stretch frequency demonstrating that coupling with N - H bending is absent [96], indicates that in M the Schiff base is unprotonated. The initial and crucial proton transfer event is therefore the loss of the Schiff base proton during the L to M reaction. A clear correlation exists, however, between several of the other chromophore states and the protonation state of protein groups. F T I R spectra show that residue D96 is deprotonated in N but again protonated in O [21,147,169]. D85 is protonated in both of these states, as well as in M, but not in L and the initial bacteriorhodopsin [21,23,64,125,147,169].

IV. Proton transfer pathways suggested by the structure of bacteriorhodopsin and static interactions among residues The seven t r a n s m e m b r a n e helices enclose space that spans the m e m b r a n e and is occupied by the retinal and nearly all of the buried protonable residues. The Schiff base is about midway across the protein, and divides the interhelical cavity into an extracellular region containing numerous charged residues and a cytoplasmic region containing mostly hydrophobic residues [76]. The two peripheral regions are sometimes referred to as the extracellular and cytoplasmic 'channels,' respectively. From a functional point of view these can be termed also proton release and proton uptake domains. Together with the Schiff base, they constitute the potential proton trajectory across the protein as shown in Fig. 1. IV-A. The extraceUular proton release domain The extracellular domain has a complex structure. The small magnitude of the 15N-NMR isotropic shift of the Schiff base, as c o m p a r e d to those measured in model systems with various added anions, can be ex-

246 plained only with the existence of a complex counterion in which the negative charge is not localized on a single group but distributed over a hydrogen-bonded network [33,34]. This network probably contains the charged residues D85, D212 and R82, and at least one bound water molecule [40,80,108], which together provide a diffuse delocalized negative charge to balance the positive charge of the Schiff base [39]. The net charge of this Schiff base-extracellular complex region would be zero, consistent with its buried location in the protein. That the interaction between the protein counterion and the Schiff base is unusually weak is indicated by two additional facts: first, the absorption maximum in the visible is considerably red-shifted from the 440 nm observed in model compounds where the counterion, e.g., chloride, is at van der Waals distance from a protonated Schiff base (the 'opsin shift') [84,130,193], and second, the C = N stretching frequency of the Schiff base is unusually low [2,166]. The presence of bound water, hydrogen-bonded to both Schiff base and D85, is suggested by the kinetics of h y d r o g e n / d e u t e r i u m exchange on the Schiff base [40] and by comparison of the 3450 to 3750 c m - 1 region of F T I R spectra (that contains the O - H stretch frequencies of water) for the L intermediates of wild-type and D85N proteins [108]. The nature of interactions among the polar residues in the extracellular region is the least understood part of the structure. Strong coulombic influence of D85, R82 and D212 on the Schiff base is indicated by color changes in the chromophore, change of the Schiff base p K a, and the deprotonation kinetics upon site-directed mutations of the three amino acid residues. Replacement of either D212 [132] or D85 [25,116,122,144, 176,183,198] with a neutral residue causes a red-shift in the absorption maximum of the chromophore and similar replacement of R82 a blue-shift [25,176], as expected from electrostatic through-space interaction of these negatively and positively charged residues with the Schiff base [7,8,82]. However, D85 and D212 are clearly not equivalent. R e p l a c e m e n t of D85 with asparagine shifts the absorption maximum of the chrom o p h o r e by 47 nm [25,198] to the red and lowers the p K a of the Schiff base from well above 10 to about 7 [144] or 8.4 [25], but similar replacement of D212 results in a red-shift * of only about 15 nm and the p K a remains unchanged [132,176]. The importance of the negative charge of D85 as the principal component of the Schiff base counterion is underlined by the p H - d e p e n d e n t spectral properties of a recombinant protein in which this residue is replaced by histidine [174]. Participation of Y185, Y57 and other nearby residues capable of hydrogen-bonding (e.g., T89), in the extracellular counterion complex is likely, but data on their effects on the Schiff base are sparse or generally not

well understood. A great deal of F T I R evidence had suggested that, by becoming protonated in the K state and deprotonated again in the M state, Y185 plays a key role in the early events of the photocycle [24,154]. Recent NMR, U V and U V - R a m a n experiments contradict this, as they indicate that the protein does not contain any tyrosinates up to p H 9 [10] or 10 [3,78], and none form during the photocycle [5]. A more recent suggestion is that the phenolic proton of Y185 forms a polarizable hydrogen-bond with D212, and it shifts toward D212 during the photocycle [152]. This is based partly on the appearance of a positive band at 1738 cm - I in the F T I R spectrum of M of the wild-type protein [23,152], interpreted ** as partial protonation of D212, and partly on the disappearance of F T I R bands assigned to Y185 in D212 mutants [152]. On the other hand, while replacing D212 with asparagine [132], or Y185 with phenylalanine [121], was found to decrease transport activity it was by not more than about 2 / 3 , indicating that although located at the active site these residues are not indispensable components of the proton transfer system. The strongly blue-shifted absorption maximum of the deprotonated Schiff base provides the means for spectroscopic titration of the unphotolysed chromophore. The p K a of a Schiff base in model compounds is near 7 when in solution [8], but in bacterio-

* The maximum of the D212N chromophore is blue-shifted by 9-13 nm from wild-type when the protein is expressed in Escherichia coli [122,171,176], but red-shifted by 16 nm when expressed in Halobacterium halobium [132]. Such differences between phenotypes have been observed with a number of directed mutations. For example, the light-adapted absorption maxima of Dll5N and D85N, expressed in H. halobium [192,198], are virtually unshifted and red-shifted by 46 nm, respectively, rather than blue-shifted by 17-28 nm and red-shifted by only 22-35 nm, respectively when expressed in E. coli [122,144,176,179]. These and many other H. halobium expressed mutated bacteriorhodopsins show red-shifts on light-adaptation [91,132,192,198] as the wild-type, rather than the anomalous blue-shifts reported for the E. coli expressed proteins [47]. The differences are probably explained by a more native-like in vivo folding of the protein and its assembly into purple membrane lattice in the homologous H. halobium expression system and the more negative surface charge in the membranes than in the detergent micelles used to solubilize and renature the E. coli expressed proteins. ** The assignment of the positive 1738 cm -1 vibrational band to D212 was on the basis of its absence in the photoproducts of the D212N protein [23]. Since the photocycle is changed considerably by this residue replacement [29,132], secondary effects on the source of the band which originate from lack of the negative charge of D212 cannot be excluded. In fact, the 1738 cm -1 band is absent also in the more wild-type like photocycle of the Dll5N protein (Maeda, A., personal communication). Since protonation of D212 during the photocycle is now excluded by investigations of the NMR chemical shifts of aspartate residues [114] as well as recent FTIR spectroscopy [53], the most likely assignment of the band is to Dl15.

247 rhodopsin titration shows it to be about 13 [45]. The strongly elevated p K a reflects the high free energy of the uncompensated charge of the counterion when the Schiff base is deprotonated [163], and might be influenced by water molecules between them that form hydrogen bonds [61]. The marked lowering of the pK~ upon replacement of D85 with neutral residues is consistent with this. As expected from the critical role of D85 in determining this p K a, the Schiff base of halorhodopsin, a related retinal protein that lacks an anionic residue equivalent to D85, has a nearly unperturbed p K a of 7.5 [93]. Protonation of D85 at lowered pH, similarly to its replacement with uncharged residues, results in a substantial red-shift of the chromophore maximum, transforming the purple protein into a blue one [49,54,124,185]. When the ionic strength is sufficient to eliminate major effects of surface charge on the local pH, the purple-to-blue shift occurs at a bulk pH of about 2 (e.g., Ref. 83). This anomalously low pKa for an aspartic acid is the consequence of stabilization of the aspartate anion by R82, because replacement of R82 with glutamine or alanine raises the apparent pKa of D85 to about 7 [25,176,179]. As expected from the different lengths of aspartate and glutamate residues, the p K a for the color transition is very different (higher) than in wild-type in the D85E protein also [67,95]. The p K a of D212 is less certain, but appears to be below that of D85, because D212 remains anionic at a pH low enough to form the blue chromophore [114].

IV-B. The cytoplasmic proton uptake domain The cytoplasmic half of the protein contains mostly non-protonable or uncharged residues, with the important exceptions of D96 and R227. Since D96 is the internal proton donor to the Schiff base during the photocycle, the nearly unchanged rate of the Schiff base reprotonation up to alkaline p H indicates that the pK~ of this residue is at least 10 [199]. An unusually high p K a for D96 is suggested also by the fact that the typical 1742 c m - ~ negative F T I R band due to changes in the C O O H group of D96 persists up to p H 9 or 10 [21,51,62,107,147], and it is confirmed by N M R [114]. This high proton affinity must be due to the hydrophobic environment of the cytoplasmic domain. It ensures that D96 will stay protonated and function as proton donor throughout the physiological p H range [119]. On the other hand, according t o the structural model of bacteriorhodopsin [76] D96 and the positively charged R227 may be in position to interact, and the approach of these groups toward one another [172] could be one of the reasons for the observed lowering of the p K a of D96 late in the photocycle [29,199]. Interaction of one or both of these residues with the neighboring residues T46 and $226 seems to provide further regulation of

the p K a of D96 (Ref. 110; Brown, Cao, Needleman and Lanyi, unpublished data). V. Deprotonation of the retinal Schiff base Destabilization of the Schiff base proton in the L intermediate leads to its transfer to the anionic residue D85. There is strong evidence that D85 is the proton acceptor. First, loss of protonation of the Schiff base is accompanied by the simultaneous appearance of a positive F T I R band at 1761 cm-1 [165] assigned to the C = O stretching vibration of the protonated D85 [21,23]. Second, the Schiff base does not deprotonate during the photocycle when D85 is replaced with a nonprotonable group [144,171], or at such low pH that D85 is protonated from the bulk [124,185]. Third, proton transport is reactivated in the otherwise inactive D85C protein when a carboxylate residue is introduced by reaction of C85 with iodoacetic acid [67].

V-A. Proton transfer to D85 follows decrease of the proton affinity of the Schiff base The free energy difference between the L state (protonated Schiff base and deprotonated D85) and the M state (deprotonated Schiff base and protonated D85) can be estimated by summing the AGs of the protonation of D85 in the wild-type protein and the deprotonation of the Schiff base in the D85N or D85T proteins [25]. This calculation yields a AG of 30-33 k J / m o l , a value well within the amount of free energy retained after absorption of the photon [17]. It corresponds to an expected effective A p K a of 5.3-5.7 pH units. The equilibrium constant K = [L]/[M] calculated from the kinetics at room temperature is in fact about 4 [189], indicating that the p K a difference between proton donor and acceptor in L is only 0.6. Thus, either the pKa of the Schiff base is lowered by about 5 pH units or the pK~ of D85 is raised in L so as to create the conditions where the proton transfer which produces M can take place. It is very likely that both occur, although the p K a of D85 during the photocycle is less certain * The C = O stretch FTIR band in M identified as originating from the protonated D85 is at the unusually high frequency of 1761 cm-1, which argued that the p K a of this residue may remain as low as 2.5, i.e., essentiallyunchanged from the unphotolysed state [155]. On the other hand, the empirical relationship between the C = O frequency of a non-interacting carboxylicacid and its p K a might not be applicable here. The 1742 cm-t band for D96 would predict, for example, a p K a of 4 instead of 9-10 as suggested by a large amount of evidence (see above). The observed high C = O frequency for D85 may be the result of hydrogen-bondingof the carboxylic acid OH in a trans configuration instead. Such interaction would stabilize the protonated form, and in effect raise the p K a of D85 for the proton transfer in the L to M reaction.

248 There are several possible reasons why the Schiff base pK a would be lowered in L. First, calculations suggest that the trans-to-cis rotation of the C13-C14 double bond evident in the L state will have disrupted the r-system of the retinal chain. This decreases the electron density on the Schiff base nitrogen and destabilizes the proton [138,178]. Second, because the distal part of the chain and the fl-ionone ring of the retinal are fixed by several flanking tryptophan residues [76], isomerization of the C13-C14 bond forces the Schiff base to move [37] to a different, possibly more electronegative environment. Indeed, an increased deuterium shift of the Schiff base C = N frequency (22-24 cm -1 in L vs. 16 cm -1 in BR) indicates that the hydrogen-bond between the Schiff base and its counterion becomes stronger [2,166]. Third, ab initio calculations of the magnitude and sign of the energy difference between a model protonated Schiff base-aspartate ion pair (i.e., before proton transfer) and the corresponding neutral pair (i.e., after proton transfer) indicate [157,158] that it is criticially dependent on the geometry of the hydrogen bond and the polarizability of the environment. Both will have changed after photoexcitation. Consistent with this, the pK a of the Schiff base in model compounds is strongly dependent on the orientation of the (sterically fixed) counterion [61]. This kind of regulation of the Schiff base pK a was suggested also for octopus rhodopsin [89]. Finally, a decrease in the amplitude of the C15-H in-plane vibration at 1303 cm -1 suggests that in the L state the retinal skeleton is twisted so as to remove a steric conflict between the hydrogens on C12 and C15 [105,147]. Such a twist would further disrupt the extended 7r-system along the retinal skeleton and contribute to lowering the pK a of the Schiff base [52,138,161].

V-B. Mechanism of the protonation olD85 How does the geometry of the Schiff base change relative to D85 and the other nearby residues in the L intermediate? Comparison of the linear dichroism of bacteriorhodopsins containing either retinal or 3,4-dehydroretinal defined the initial direction of the N ~ H bond as pointing toward the exterior surface [99]. A weak hydrogen-bond between the Schiff base and D85 is suggested by a 3 cm -1 down-shift of the C = N frequency in the D85N protein relative to wild-type [98]. If proton transfer is to take place from the Schiff base to D85, the interaction between the Schiff base and its complex counterion (see above) must change in L so as to reorient the C = N - H bond more directly toward D85. Molecular dynamics calculations suggested that the interaction of the Schiff base with its counterion is stabilized by water molecules in the retinal binding pocket [197]. The presence of bound water near the Schiff base was suggested from Raman spec-

troscopy [40,80]. FTIR spectra indicated that the hydrogen-bonding properties of one or a few bound water molecules in L do indeed depend on D85. The L state is normally characterized by a negative band at 3642 cm -1 that is sensitive to both D20 and H~So and indicates disappearance of weakly bound water [106]. It is replaced by a small positive band at about 3652 cm-1 (free water) and a large broad absorption increase in the 3450 to 3560 cm-1 region (in part strongly bound water). These changes disappear in the M state, thus implicating the protonated Schiff base as a participant in hydrogen bonding with the water. They are absent also in L state of D85N, and the remaining features in this spectral region are no longer changed in H180, thus implicating also D85 in the hydrogen bonding with water [108]. These findings, together with perturbation of D96 in L (see below) and the increase of the deuterium effect on the C = N frequency [2,166], suggest large-scale rearrangement of the hydrogen bonds of D85, the Schiff base, bound water and D96 in the L intermediate of the wild-type protein. The result appears to be stronger hydrogen bonding within both proton channels, but in particular on the extracellular side so as to create the conditions for proton transfer from the Schiff base to D85. VI. Proton release at the extracellular surface

The transfer of the Schiff base proton to D85 during the L to M reaction takes place inside the protein, but at approximately the same time a proton appears on the extracellular surface of the membrane. The release of protons after flash excitation, as well as the subsequent uptake on the cytoplasmic side, have been followed by measuring transient absorption changes of pH-indicator dyes either in the bulk or covalently bound at the protein surface. Examples of dyes in the bulk are nitrophenol [42,103], pyranine [29,69,73,75,143,144], bromocresol green [38,120], and phenol red and chlorophenol red [101,186,202]. The covalently bound dyes used are fluorescein conjugated to residue K129 (on the extracellular end of helix D) as succinimidyl ester or isothiocyanate [75], and iodoacetamido fluorescein conjugated to a cysteine residue introduced to the location of interest by site-specific mutagenesis [159]. Dyes located on the surface show the proton release to be at 20-100 /zs, i.e., roughly concurrent with the L ~ M reaction. Dyes located in the bulk, however, respond on a time-scale of several hundred /~s. The transfer of the proton from the surface to the bulk appears to be therefore delayed in the bound water layer near the membrane [74,75,159]. As expected from model studies [70], proton exchange between the surface layer and the bulk is measurably accelerated, for kinetic reasons, when a mobile buffer is added [42,69], but the long residency time for the transported protons

249 at the membrane surface, and its causes, remain interesting questions. Another approach to measure proton release has been to follow those photoelectric signals from oriented purple membranes that originate from the proton leaving the protein [100,101].

VIA. Proton affinity of the extracellular proton release group At p H > 6 the direction of the absorption changes of pyranine and other dyes show that proton release precedes proton uptake. Since the release on the extracellular side occurs during the rise of M and the uptake on the cytoplasmic side roughly (although not exactly, see below) during the decay of N, the finding that there is approximately one proton lost transiently from the protein as it passes through the M and N states is readily interpreted as the vectorial release and uptake of the transported proton. This proton loss can be observed also during steady illumination of purple membrane sheets. A p H decrease measured with a glass electrode detects protons released in the medium during the photostationary state that develops, and in amounts consistent with the accumulation of M and N [54,59,177,186]. Since, according to F T I R evidence, D85 remains protonated until the recovery of BR at the end of the photocycle [19,21,23,64,125,147,165,169], the proton deficit in the M and N states does not arise from release of the same proton that was transferred from the Schiff base to D85. Rather, the proton must originate from another group with access to the extracellular surface. This group, termed X H [202], seems to be influenced by several residues (see below). Its p K a must be high enough in BR to keep it protonated up to at least p H 9 in order to maintain it as a source of protons, but will be lowered in L so as to cause its timely deprotonation. It is an important clue to the p K a of X H at the time of the proton release that at p H below 6 the proton kinetics in the medium are dramatically changed. Bromocresol green [38], chlorophenol red [186,202], as well as pyranine [27], detected not proton release but transient proton uptake and at a later time in the photocycle, roughly (although not exactly, see below) coincident with the accumulation of the O state. One report [120] is at odds with these findings, as it detected only proton release with bromocresol green, regardless of the pH. On the other hand, conductivity change after flash excitation had also revealed the delayed proton disappearance at lower p H rather than the rapid release observed at higher p H [109]. Consistent with this, the direction of measured p H change during photostationary states was reversed at lower p H [54,59,177,186]. Under these conditions therefore, the protein appears to gain rather than lose a proton

transiently in the photocycle. The dye kinetics indicate that the proton uptake step is at about the same time as at higher p H but the release step is delayed until the O ~ BR reaction. The proton release occurs under these conditions not before but after the proton uptake. This effect of pH on the proton release must reflect the protonation state of XH. More quantitative information about proton release from X H is obtained from kinetic analysis of the chromophore reactions between p H 4 and 8 [202]. Modeling the kinetics with the measured rate constants indicated that if proton release from X H were obligatory for the next chromophore reaction the rates of all of the reactions that follow would be significantly slowed as the p H is decreased. This is not found; because the rate of M decay is only slightly dependent of proton concentration between pH 4 and 8, reprotonation of the Schiff base evidently proceeds independent of the protonation state of XH. It is implicit in the model in Fig. 2 that the chromophore reactions and the proton transfer steps in the photocycle are not strictly coupled to one another. Because proton release at pH > 6 occurs roughly at the same time as the rise of M, it has long been thought to be associated with the L ~ M reaction also in a mechanistic sense. However, lowering the pH to 6 and below caused the appearance of an M 2 ~ M 1 back-reaction rather than an increase in the rate of the M 1 ~ L back-reaction [202]. This suggests that the protonation change of X H is connected to interconversion of M substates rather than directly to deprotonation of the Schiff base. According to this mechanism the proton release should be concurrent with the second phase of the M rise (with the time-constant of the M~I°) ~ M~-I) reaction) but not with the first (with the time-constant of the L (°~ ,~, M~I°~ reaction). The timecourse of the absorbance change of a covalently surfacebound dye [74] confirmed that the proton release exhibits zero amplitude during the first phase of M rise, and occurs concurrently with the second phase, even though the response time of the dye would have been fast enough to detect an early phase. The scheme in Fig. 2 shows proton release from X H in this way *

VI-B. Composition of the extracellular proton release complex The most obvious candidate for X H is R82. One of the two likely dispositions of R82 locates it approxi-

* Although the proton release from the extracellular domain is placed just before the M1~ M2 reaction, the data do not distinguish between this possibilityand the alternative in which it occurs just afterward. We do not know, therefore, whether the proton release triggers the M 1~ M2 reprotonation switch (see below),or the switch triggers the proton release.

250 mately in the right place, near the Schiff base and D85 [13,66,76]. On the basis of the observed coulombic interaction between the anionic D85 and the protonated R82 (see above) it might be expected that protonation of D85 would lower the p K a of R82 [9], as required if R82 were XH. Indeed, replacement of R82 with glutamine or alanine changes the early proton release near pH 7 to delayed proton uptake [9,27,144] that is similar to the behavior of the wild-type protein at lower pH. Presumably, when X H is rendered nonfunctional, proton exchange with the bulk proceeds at any p H as it would normally only in the low pH pathway. On the other hand, the pKa of arginine in proteins is usually > 12. Titration of recombinant bacteriorhodopsins with various site-specific replacements of D85, R82, and both, allowed calculation of the PKa of R82 during protonation changes of the Schiff base and D85 [25]. If the assumption is made that replacing R82 with glutamine or alanine will have the same effect on the pK~ values of the Schiff base and D85 as its deprotonation, the result of this calculation is that the p K , of R82 is 13.8 in the initial state, and decreases to only 11.5 in the M state. This would preclude R82 being XH. In fact, replacement of Y57 with phenylalanine has the same consequence on proton kinetics as replacement of R82 (Brown, Cao, Needleman and Lanyi, unpublished data), which indicates that XH is not a simple entity. It seems likely that X H is a water molecule in hydrogen-bonded complex that includes Y57, as well as R82 as suggested earlier [23].

VI-C. Sequence of internal to external proton transfers in the photocycle The vectoriality of the proton transfers in this branched chromophore reaction cycle is illustrated in Fig. 3. The protein is shown schematically as the three adjoining domains in Fig. 1 that traverse the bilayer: the cytoplasmic proton uptake domain, the Schiff base and the extracellular proton release domain. At low pH two sequential proton transfers occur inside the protein, and they redistribute the mobile protons in the three domains so as to create a deficit on the cytoplasmic side and an excess on the extracellular side. These result in first uptake and then release on the respective sides, regenerating the initial state and effectively moving a proton across the membrane. At high pH the internal and external proton transfer steps alternate: the Schiff base to the extracellular domain transfer is followed by release, and the cytoplasmic to the Schiff base domain transfer is followed by uptake. The two pathways in Fig. 2 thus differ only in the sequential order of the two internal and two external proton transfers; both result in the net cytoplasmic-to-extracellular translocation of the proton.

at highpH

at lowpH

M (°)

N(O)dO)

M (-1)

o1+1)

N (-1)

BR

N, O, BR

(0)

(0)

Fig. 3. Schematic representation of the vectorial proton transfers associated with the branched chromophore reaction cycle in Fig. 2. Bacteriorhodopsin is shown as three adjoining domains that span the membrane: the cytoplasmicproton uptake domain (upper), the Schiff base (middle), and the extracellular proton release domain (lower). Light arrows refer to intra-protein, and dark arrows to protein-bulk proton transfers. Each domain is shown with its number of mobile protons. The cytoplasmicdomain (with D96 as the protonable group) and the Schiff base have capacities for a single proton only, but the extracellular domain has two protonable groups: D85 and XH (see text) and thus it may contain two protons. The designations of the intermediate states are as in Fig. 2, but unlike in that figure the M state is shown without subscripts because the reprotonation switch is implicit in these diagrams.

VI-D. Change of the environment of D96 in the L intermediate Understanding the internal and external proton transfers during the L (°) ,~, M(x°) ~ M~- 1) sequence (Fig. 2) is potentially complicated by the fact that changes in the C O O H vibrational bands show D96 to be affected in L. The shape of the F T I R spectrum in this region suggested at first [23] that D96 deprotonates during the K to L transition and reprotonates as M is formed, before its subsequent deprotonation in the M--, N reaction that reprotonates the Schiff base in the proton transport mechanism. On the other hand, from similar data it could be also argued [62] that the F T I R changes reflect perturbation rather than deprotonation of D96. Dissection of the overlapping negative and positive

251 F T I R bands in the carboxylic acid region of several aspartate-mutated proteins indicated [107], that at 170 K at least it is clearly the hydrogen bonding state of D96 that changes in the L intermediate. Whether D96 dissociates at room temperature [21] is still an open question. VII. The Schiff base reprotonation switch Vectorial translocation in a transport cycle requires at least three distinct states [135], in which a 'conformational' factor determines access to one or the other side of the membrane but never to both at the same time [85]. In bacteriorhodopsin access refers specifically to proton exchange between the Schiff base and either D85 on the extracellular side or D96 on the cytoplasmic side. The change of access from one of these residues to the other is referred to as the reprotonation switch [55,76,85,111,127,135,160,191,198]. I/II-A. There are two plausible m e c h a n i s m s

Although the requirement for a switch in active transport is clear, its nature is not well defined in any of the pumps studied. The semiotic rationale of the switch in bacteriorhodopsin is to ensure that once the Schiff base is deprotonated its reprotonation will not be from D85. A minimal switch in this system will therefore fulfill the single purpose of transiently closing access to D85. In principle this might be accomplished in two alternative ways (as discussed in Ref. 85): first, by a switch based on proton affinity in which proton transfer is controlled solely by changes in the pKa's of the respective groups, and second, by a switch based on geometrical factors in which proton transfer is controlled solely by changing barriers through altered bond angles and distances between donor and acceptor. It is likely that in bacteriorhodopsin the second alternative, either alone or in combination with the first, is utilized. The second alternative is suggested by specific blue-shifts of the absorption maxima of the proposed pre-switch and post-switch states M 1 and M 2 (see below) when D85 or D96 are replaced, respectively, with asparagine [198]. If these shifts reflect disruptions of hydrogen bonds between the Schiff base and the two groups on the extracellular and cytoplasmic sides in M 1 and M2, respectively, as suggested, the geometric aspect of the switch would be firmly established. Calculations indicate that geometry is important: transfer rates are changed by many orders of magnitude through small alterations in bond angles along the proton trajectory [158]. I/II-B. The reprotonation switch is the M 1 ~ M 2 reaction

Although some models place the switch at the N [55-57,140] or O [117] intermediates, the logical place

for it in the photocycle would be after deprotonation of the Schiff base but before its reprotonation, i.e., in the M state [85,160,161,191]. It is important, therefore, to establish the identities of the immediate pre-switch and post-switch states. There is a considerable amount of evidence for the existence of M substates, and some of it argues for their functional role in the reprotonation switch. First, fitting time-resolved difference spectra in the visible with a kinetic scheme containing a single reaction sequence and reversible reactions required postulating two sequential M substates, termed M 1 and M 2 and connected by a unidirectional reaction [187,189]. The introduction of the M 1 ~ M 2 reaction ( f o r w a r d / reverse rate >__200 at pH 7, although much less at lower pH, see above) solved the problem that the concentration of L tended to zero as the concentration of M reached a maximum, even though the calculated L ,~, M equilibrium with a single M would have predicted the persistence of considerable amounts of L as long as M was present. Determining the concentration of L at this time in the photocycle is complicated by the accumulation of N because L and N have similar absorption [189] and resonance Raman [4] spectra. Confirming the results with the D96N protein, under conditions where for kinetic reasons the accumulation of N was eliminated, gave therefore important support to this interpretation of the wild-type kinetics [189,201]. Second, while in purple membrane containing wildtype protein the maximum of M is at 411-412 nm regardless of whether the putative M 1 o r M 2 is present, under some conditions the kinetically defined M substates do exhibit different absorption maxima. The maximum of M 2 is blue-shifted by a few nanometers from that of M 1 in the detergent-solubilized (monomeric) wild-type protein [118,175,189]. Upon the D115N residue replacement the difference in Areax under these conditions becomes as much as 18 nm [192]. Since this shift in the absorption maximum of M occurs at a time in the photocycle when absorbance is nearly constant at both 410 and 570 nm, a scheme with parallel M intermediates with different maxima does not explain the data as well as one with sequential M states. Distinguishing the maxima of M substates in the purple membrane lattice is easier at pH 10, where the rise of M is much faster and M~ is expected to accumulate to concentrations comparable to that of M 2. Determined under these conditions, the maximum of M 1 of D96N is like in wild-type M 1 but that of M 2 is blue-shifted by 7 nm [198]. The opposite is found in D85N. A quasi-stable M state with a lifetime of 2 min, produced by sustained illumination of the D85N protein, has a maximum at 400 nm [198]. This is an Ml-like state, since by definition in M 2 the Schiff base would be reprotonated within a few milliseconds by D96. Thus, it appears that replacing D85 or D96 affect

252 the maxima of M 1 and M 2 specifically, and as expected if M 1 and M 2 were respectively pre-switch and postswitch states. Third, the quantum yield and rate of a blue-flash induced photo-backreaction of M to BR changes during the lifetime of M in such a way as to suggest the existence of two distinct M substates [43]. The rate of the conversion of the first M into the second agrees well with that predicted from the kinetics of L, suggesting that these experiments detect the proposed M~ and M 2 states. Fourth, photoacoustic measurements had shown that a large decrease of enthalpy occurs between proton release and uptake [58,141]. It argued for a strong decrease of entropy and thus a large protein conformational change during the lifetime of M [191]. Recent studies with better time-resolution localized a large part of this enthalpy decrease at 80-90/~s [150], i.e., at about the time-constant of the proposed M 1 ~ M 2 transition. Fifth, FTIR spectra of bacteriorhodopsin films where the decay of M was greatly slowed, i.e., in D96N at pH 10 and 276 K [156] or in glucose-dehydrated wild-type protein [I46], showed that the amide bands at 1669 and 1558 cm -~ (that originate from peptide bond vibrations) as well as the shift of the C = O stretch frequency of D85 from 1761 to 1755 cm -1, otherwise observed only in N, appear in the virtual absence of chromophore bands due to protonated Schiff base *. The FTIR changes usually associated with protein changes in N can thus occur before the Schiff base is reprotonated, and argue for the existence of a late M state different in its protein conformation from the earlier one. On the other hand, this late M, termed MN [156], is unlikely to be M 2. In the wild-type photocycle the amide bands arise virtually concurrently with the M 2 ~ N reaction [21,64,169] which takes place much later than the proposed M 1 ~ M 2 reaction, and in D96N the M N state coexists with its precursor M state in a constant ratio throughout the recovery of BR [156]. In view of these results, M N is more likely to be a transient state between M 2 and N. The proposed rationale for MN is that in this state the proton of D96 becomes destabilized. This is supported by the findings that (a) in D96N the FTIR band due to N96 shows a shift at about this time in the photocycle consistent with changed hydrogen bonding [156], and (b) in the photocycle of the blue form of the D212N protein the

* F T I R spectra of wild-type bacteriorhodopsin films seemed to indicate at first that the M state produced by illumination at 260 K was different in its changed amide bands from that produced at 240 K [139]. However, u n d e r the conditions used the observed differences were due not to two distinct M states but to N that coexisted with M at the higher temperature [140].

amide bands appear and the p K a of D96 is lowered at the same time, even though in this system a deprotonated Schiff base is not formed [29] (see below). VII-C Does the switch reside in the retinal or the protein? Although the observed shifts in the maxima of M 1 and M 2 upon the D85N and D96N residue replacements suggest that what is detected spectroscopically as the M 1 ~ M 2 reaction could be the reprotonation switch [198], the molecular mechanism by which the switch function is accomplished is not at all clear. In the first of two kinds of proposed models for a switch based on geometrical factors photoexcitation generates a 13-c/s-14-s-c/s chromophore by rotation around both the C13-C14double-bond and the C~4-C15 single-bond of the retinal [63,68,161]. Molecular dynamics simulations strongly favor the 13,14-dic/s configuration over the 13-c/s [197]. Conversion to 13-cis-14-s-trans during the M 1 ~ M 2 reaction would reorient the Schiff base and raise its p K a s o as to make it a proton acceptor [134]. However, this attractive model, as well as another one in which the reorientation of the Schiff base is by C = N anti-syn inversion, is contradicted by the results of Raman investigations of the retinal configuration in the L and N states. According to these * the retinal is 13-cis,14-s-trans,15-anti in both L and N, and by inference therefore also in M [55]. Another proposed alternative [60] invokes bond rotations not only in the retinal but also along the backbone of K216, the residue that binds the retinal. Significant movement of the retinal itself during the time-range of the M 1 ---, M 2 reaction has not been detected. The observation of unchanged linear dichroism during the lifetime of M would appear to limit any reorientation of the retinal transition dipole to less than 3° [142], but under the conditions of this measurement M~ does not accumulate in amounts that would influence the overall behavior of M. In the second proposed model access of the Schiff base to D85 is regulated by protein conformation [55]. The protein adapts to the bent shape of the 13-c/s retinal in either the L ~ M reaction [55], or in the transition from one M substate to the next [111,191] with a changed conformation, as suggested also by a large entropy decrease that takes place during the

* T h e C = N inversion was ruled out by insensitivity of C t 4 - C t 5 stretch to N deuteration. T h e a r g u m e n t against 14-c/s to trans isomerization during the lifetime of M was based on the observed frequency of the C 1 4 - 2 H - C I s - 2 H symmetric rock combination, and assumed that the influence of C14-C15 conformation on this frequency is not compensated by electrostatic influences in the retinal binding pocket. This assumption has been questioned [68,197].

253 lifetime of M [58,141,150,191]. In these models the changed protein conformation would, in a so-far unidentified manner, restrict access of the Schiff base to D85 but allow its reprotonation from D96. Structural change of the protein in M relative to the initial BR state is indeed detected by the FTIR amide bands (see above) and by X-ray, electron and neutron diffraction [35,65,87,173]. The amide bands and the changed electron densities do not necessarily reflect the same conformational transition, however. Embedding the protein in glucose considerably diminishes the structural changes detected by diffraction (compare the data in Ref. 65 with those in Refs. 35,87,131,173), but not the changes detected by the FTIR amide bands [146]. The amide bands are observed also in the photocycle of the blue form of D212N where neither transport nor entropy decrease occurs [29]. Thus, as suggested above, the amide bands might be unrelated to the reprotonation switch (although for a different opinion see Ref. 140). The FTIR bands might correspond to the observed increased electron density on helix G that suggests a more highly organized cytoplasmic domain in M, but not to the differential density changes that suggest tilting of helix F [173]. Better time-resolution might separate these effects: if the tilt for helix F detected by diffraction were found to occur earlier in the cycle than the density increase on helix G and the appearance of the amide bands, it could be a good candidate for causing the kind of of structural distortion near the Schiff base that would rearrange proton conductivity in this region. VIH. Reprotonation of the Schiff base Time-resolved FTIR spectra [18,21,64,169] indicate that the appearance of the negative 1742 cm-1 band, that indicates deprotonation of D96, is coincident with the reprotonation of the Schiff base. Thus, proton loss from D96 and proton gain by the Schiff base are described by a single kinetic process. That D96 is the internal proton donor to the Schiff base is strongly supported also by results with mutated proteins: replacement of D96 with non-protonable residues greatly slows reprotonation of the Schiff base and makes it dependent on pH, but proton transport still occurs [28,62,81,116,119,143,180]. Because a buried proton donor with suitable pK a and access is lacking under these conditions, the Schiff base is reprotonated from the cytoplasmic surface. VIII-A. Proton conduction between D96 and the Schiff base

Since the distance between D96 and the Schiff base is about 12 A [76], proton transfer between the two groups would seem not to be feasible unless facilitated

by an intervening hydrogen-bonded chain. Various observations on the consequences of replacing residues T46 and T89 in the cytoplasmic proton channel were recently rationalized in a model in which the proton trajectory from D96 to the Schiff base extends over the following chain of residues, hydrogen-bonded into a network during the photocycle: D96 ~ T46 ~ T89 Y185 ~ D212 ---, Schiff base [153]. However, the proton transfer from D96 to the Schiff base is not significantly slowed, and in some of the cases becomes accelerated, upon replacement of T46 [110], T89 [110], Y185 [46] or D212 [132] with non-protonable or non-hydrogen bonding residues. A hydrogen-bonded chain of a few water molecules arranged in single file, as suggested for proton conduction in the interior of proteins in general [113,129], is allowed by the present resolution of the structural model [76,115], as well as by calculated energy-minimized structures [133,197]. Others dispute this, and suggest that a hydrogen-bonded chain will be formed only after an appropriate protein conformational change occurs in the photocycle [36,37,57]. Strongly bound water near the Schiff base was detected by neutron diffraction [145], and a role for water is suggested by the finding that dehydration inhibits, before any other reaction in the cycle, the D96 to Schiff base proton transfer [28,88,188]. VIII-B. Chromophore reaction kinetics in the M to BR pathway

A protonation equilibrium between D96 and the Schiff base not far from unity was suggested by the appearance of a second M decay time-constant above pH 8.5 with linear pH-dependency such as expected for reprotonation of D96 [143], and found for the decay of N [90]. Under some conditions the Schiff base deprotonation is therefore described adequately * by the scheme M ~ N ~ BR [4,28,143,169,187]. The existence

An alternative explanation for the slow phase of the decay of M at alkaline pH is that it consists of another M state that arises as the photoproduct of the N intermediate after absorption of a second photon [57,90]. This is argued from the observation that background light increases the amplitude of the slower phase of M decay. Other measurements seemed to indicate, however, that the photoproduct of N is not blue-shifted as M but red-shifted [11,137,187,196]. This model was recently evaluated in the T46V protein, where the chromophore is unperturbed but the M and N decays are so widely separated in time that the photoreaction of N can be studied without interference from the M in the initial photocycle [26]. A second flash indeed produced an M-like species from N with a longer life-time than the M from BR, but only when the photoexcitation was in its main absorption band, i.e., near 560 nm. Photoexcitation near 400 nm, where the kinetics of M are usually followed, produced no M from N. Thus, the two-photon photocycle is confirmed in principle, but under most conditions it can have no effect on the measured M kinetics.

254 of a significant thermal N ~ M back-reaction is supported by the results of two-flash experiments. Depletion of the M state with a second (blue) flash is followed by partial recovery of M with the time-constant expected from the M ~ N equilibration reaction, while two parallel M states with different decay timeconstants are ruled out [26,44,199]. However, there are conditions where the decay of N is measurably slower than the second decay component of M [56,199]. This would be consistent with an M ~ N equilibrium only in a scheme which contained two sequential N states with similar spectra and connected by a unidirectional reaction. The existence of such N substates has been postulated [111] on the grounds that N differs from the O intermediate in both the isomeric configuration of the retinal and the protonation state of D96, and thus the N to O reaction might be in principle resolved into the reprotonation and reisomerization steps. The pH-dependence of the N ~ O reaction supported this [4]. This scheme (Fig. 2) is now confirmed by direct evidence [199] that indicates that the proton uptake is during the interconversion of two N substates (see below).

VIII-C. Proton transfer to the Schiff base is the consequence of a decrease of the proton affinity of D96 The pK a of the Schiff base will have been lowered in L to allow proton transfer to D85. The pK a of D96, on the other hand, is as high as 9 or 10 (see above). An equilibrium between M 2 and N that does not lie far toward M 2 can be established only if the pga's of the Schiff base and D96 approach one another again. The Schiff base pK a will have risen at least to some extent in N, since in this intermediate the retinal skeletal strain observed in L is absent [147]. That the pKa of D96 is lowered at this time in the photocycle, and independently of the protonation state of the Schiff base, is suggested by the photoreaction of D212N [29]. In the blue form of this recombinant protein (at pH >17) the Schiff base remains protonated after photoexcitation, and the photocycle in the neutral pH range is described by the scheme hv

BR~ ~ K ",--,L ,--,N ~ BR. A deprotonated Schiff base is not formed under these conditions. The state N is distinguished from L by a small shift in the absorption maximum in the visible, a negative 1742 cm -1 band indicative of deprotonation of D96, and the appearance of amide I and II bands indicative of a protein backbone change, as in N of the wild-type. The pH indicator dye pyranine detects the release of a proton at a time well after the L ~ N reaction but before the decay of N. Neither the FTIR changes nor the proton release are observed in the

double mutant D212N/D96N. The results thus indicate that D96 deprotonates even in the absence of its normal proton acceptor, the unprotonated Schiff base. Net proton translocation does not occur, however, and the kinetics of the photovoltage produced [123] indicates that the subsequent proton uptake is on the same side as the release, i.e., most likely the cytoplasmic side where D96 is located. It is reasonable to suppose that such lowering of the pK a of D96 takes place in the wild-type photocycle also, but that here the proton is captured directly by the Schiff base. The destabilization of this proton must be caused by a change in the dipole environment of D96. The protein backbone conformation change, indicated by the amide I and II bands which appear concurrently with the deprotonation of D96, might cause either a rearrangement of bound water near D96 [28] or move R227 on helix G nearer to D96 [172]. Either of these alternatives would in principle destabilize the proton of D96; it is probable that both contribute to the lowering of the pK a of D96. Indeed, replacement of R227 with glutamine shifts the [ME]/[N] equilibrium toward M 2 as expected, although only to an extent consistent with raising the pK a of D96 by about 1 pH units [199]. Replacement of the neighboring residues T46 with valine or $226 with alanine causes marked acceleration of the reprotonation of the Schiff base and the slowing of the subsequent proton uptake (Ref. 110; Brown, Cao, Needleman and Lanyi, unpublished data), suggesting that interaction with these residues modulates the pK a of D96 also, but in the opposite sense than R227. The barrier that determines the rate of proton transfer from D96 to the Schiff base appears to consist largely of the enthalpy cost of separating the proton from the aspartate anion [28]. The rate is about seven orders of magnitude slower than predicted for proton conduction via a string of water molecules, as e.g., in gramicidin. This would correspond to an additional barrier of 40 kJ/mol. The activation enthalpy for M decay after replacing D96 with asparagine, which eliminates the transition state ion-pair, is indeed lowered by about 40 k J / m o l [28,119,180]. As expected from such a model, the hydration state of the protein has strong influence on the proton transfer in the wild-type protein but not in D96N [28]. Withdrawal of bound water from the protein, either by lowering the vapor pressure over deposited films [188] or by adding osmotically active solutes to purple membrane suspensions [28], strongly inhibits the M E,~ N equilibration reaction in the photocycle. IX. Proton uptake on the cytoplasmic surface, and recovery of the initial state

From FTIR and resonance Raman spectra it is evident that reprotonation of D96 and reisomerization

255 of the retinal from 13-c/s to all-trans are both associated with the N ~ O chromophore reaction. Results that resolve of this transition into the N (-1) --) N (°) --) O (°) --)BR sequence at high pH [199], and into the N (°) --)O (°) --)O (+1) --)BR sequence at low pH [27] indicate, however, that the two processes are not necessarily coincident.

IX-A. Relationship of proton uptake and the reisomerization of the retinal The kinetics of proton uptake at pH 8.5 measured with pyranine indicated that it occurs after the rise of N but well before its decay, i.e., during the process described as the N (-1) to N (°) reaction in the chromophore kinetics [199]. The time-dependent absorption changes at 410 and 570 nm between pH 9 and 11 allowed calculation of the pH-dependencies of the apparent rate constants. As expected, the N (-1) ~ N (°) reaction was found to be pH-dependent, while its back-reaction was not [199]. The calculated pK a for the proton uptake is about 11. Thus, in contrast with the proton release that has a pK a within the physiological range and allows for two alternative pathways (Fig. 2), the pK a for proton uptake is well above the pH range for the proton transport. Since it is not certain that the proton taken up directly reprotonates D96, this pK a might refer to the high proton affinity of a cytoplasmic proton transfer complex comprising probably R227, T46 and bound water, rather than the regained proton affinity of D96 in N (- 1). The existence of a direct proton acceptor in the cytoplasmic domain that is not D96 is suggested by the finding that in R82Q where there is net proton uptake FTIR spectra show D96 to be deprotonated throughout the lifetime of N (Brown, Cao, Yamazaki, Maeda, Needleman and Lanyi, unpublished results). At lower pH (e.g., at pH 4) the proton uptake measured with pyranine lags behind the rise of the O state [27]. Therefore, the same kind of argument that suggests two N states at high pH requires two O states at low pH. The N decay pathway under these conditions will be thus described by the scheme:

ton release complex XH, i.e., of the net charge in the region of the Schiff base, in the two pathways. Indeed, theory suggests [178,194] that a more negative environment for the Schiff base should raise the barriers to bond rotations in the retinal. Recent observations established such a connection between the barrier for isomerization and the charge environment of the Schiff base [9]. When the pH was low enough to protonate D85 [9,136], or the anionic D85 was replaced with a neutral residue [183], equilibration of the 13-cis and all-trans chromophores became unusually rapid.

IX-B. The charge state of D96 influences proton uptake The course of proton uptake from the cytoplasmic surface is changed when D96 is replaced with a noriprotonable group. The M decay is described by a single exponential with a pH-dependent rate, because proton transfer to the Schiff base is now directly from the bulk, and the pK a of the Schiff base at this stage is apparently high enough to preclude the development of a protonation equilibrium. The M decay is so slow that N scarcely accumulates, or not at all [29], i.e., the rate of reisomerization is not slowed correspondingly by replacement of D96. Proton transfer to the Schiff base is hindered by what appears to be an increased entropic barrier [28,119,180]. Resolution of the proton trajectory into two segments, between the Schiff base and residue 96 (about 12 A) and between residue 96 and the cytoplasmic surface (about 6 A), and comparison of the rates and the activation parameters indicated that the entropically unfavorable consequence of replacing D96 is on the capture of the proton at the opening of the cytoplasmic channel [28]. This suggests another role for D96: its negative charge in the N state might sustain a protein conformation appropriate for effective entry of a proton into the cytoplasmic channel. This function of D96 is strongly influenced by the nearby residues T46 and $226, since their replacement slows proton uptake by about 2 orders of magnitude (Ref. 110; Brown, Cao, Needleman and Lanyi, unpublished data).

IX-C. Recovery of the initial BR state

N(O) ~ o(O) ~ O ( + 1) ~ B R .

These results show that the relationship of the proton uptake a n d / o r the reprotonation of D96 to the reisomerization of the retinal to all-trans is not the same under all conditions. The reisomerization, as reflected in the photocycle by the N to O chromophore transition, appears to be at different times in the low and high pH pathways (Fig. 2). In the high pH pathway it follows proton uptake, while in the the low pH pathway it precedes it. This might be a consequence of the different protonation state of the extracellular pro-

The initial state is recovered as D85 is deprotonated in the final O ~ BR reaction [19,169]. D85 transfers its proton either to the extracellular proton release complex (at high pH) or to the bulk (at low pH). Resonance Raman [167] indicates that minor relaxations in the retinal, and therefore most likely in the protein also, accompany this internal proton transfer. The key role of the deprotonation of D85 on the equilibrium between O and BR is suggested by the finding [183] that when D85 is replaced with asparagine the chromophore exists as a mixture of states

256 that resemble M, N and O of the wild-type photocycle. These states are thermal states rather than photoproducts, and they are in a pH-dependent equilibrium similar to the pH-dependence of the accumulation of N and O in the proposed transient equilibration during the photocycle (Fig. 2). However, the analogy between these states and the photocycle intermediates is weakened by the fact that the mixture of states in D85N is not shifted toward O by raising the temperature as in the wild-type photocycle (Brown and Lanyi, unpublished experiments). In a somewhat similar way, replacement of Y185 with phenylalanine produces a pHdependent mixture in the light-adapted chromophore, containing at low pH also what appears to be an O-like state [72,149,168]. Its origin is probably the raised pK a of D85. It appears therefore that the O state is distinguished from BR mainly by the negative charge of D85, and this kind of state, whether attained by protonation of D85 after photoexcitation or by the D85N residue substitution, will be in equilibrium with M (presumably M2) and N.

release Kco) dOl M1(o). mdproton r

A

~M~.1)N(.1)Fton uptake

cO

0 (o)

K(o) L(o) M~o) ~Me(°) N(°)0(o)

>, (I) to) (9 4=

hv

E~jR

B proton uptake

/

|

proton

release

X. Thermodynamics of the photocycle; coupling between chromophore and proton transfer reactions The thermal reactions of the photocycle are driven by the free energy retained in the K state after absorption of a photon, that is probably close to the approximately 50 k J / m o l excess enthalpy measured [17]. Understanding how this zIG is transformed and dissipated in the photocycle will reveal how free energy in the retinal is transferred first to protein residues and then to protons so as to produce an electrochemical gradient across the membrane. The thermodynamics of the photocycle between K and the last intermediate O were reconstructed [191] from photoacoustic measurements of the enthalpy changes (i.e., calorimetric enthalpies) and the temperature dependencies of the rate constants (i.e., van't Hoff activation enthalpies and entropies).

reactioncoordinate Fig. 4. Free energy diagram for the bacteriorhodopsin photocycle in Fig. 2. The free energies are approximate and for illustration purposes only. The curved arrows indicate reactions associated with proton release and uptake. At pH > 6 (A) a large part of the excess free energy is dissipated in two events on the protein surface, (M~°) --* M~- 1) which is proton release on the extracellular side, and N (- D ~ N(0) which is proton uptake on the cytoplasmic side). The free energy lost depends on the difference between the pH and the p K a of the protein group involved in the proton exchange with the bulk. At pH < 6 (B) the reprotonation switch is unchanged, but proton uptake on the cytoplasmic side ( O ( ° ) ~ O (+1)) is with a greater loss of free energy, and proton release on the extracellular side is delayed until the 0 (+1) ~ BR reaction. The AG between N and O is shown tentatively as positive at high pH and negative at low pH.

X-A. Dissipation of free energy in the cycle Several of the photocycle reactions appear to proceed near equilibrium, i.e., at close to zero zIG. This is kinetically optimal for the internal reactions of enzymes in general [1]. As shown in Fig. 4A, under physiological conditions the two reactions associated with proton exchange between the protein and the aqueous phase dissipate most of the free energy. At a physiological pH (about 7) the overall Mr1°) --->M<2-]) reaction exhibits an apparent equilibrium constant K = [M~-l)]/[M~x°)]not less than 200, which corresponds to a zIG of at least - 1 5 k J / m o l [189]. Of this amount about 7.5 k J / m o l is calculated to be lost at the reprotonation switch, and the rest, that is pH-dependent, is

accounted for by proton release to the bulk on the extracellular side, i.e., 5.7 k J / m o l for every pH unit above the pK a of the release complex XH [202]. The other pH-dependent step at which free energy is lost is the N (-1) ~ N (°) (or at low pH the O (°) --->0 (+1)) reaction [199]. Here, AG will be dissipated by proton uptake on the cytoplasmic side, because the pH under most conditions is well below the pK a of the proton uptake group. This p K a is about 11 at this time in the cycle [199], and zIG will amount to - 5 . 7 k J / m o l for every pH unit below 11. The difference between the pK a values for release and uptake is 5 pH units, which amounts to almost 30 kJ/mol, and by itself could drive

257 a transmembrane proton potential of 300 mV. The rest of the AG available is used to recover the initial state. From the initial 50 k J / m o l available in K and the three calculated AGs, the free energy that remains for driving the O ~ BR reaction is estimated to be no more than 15 kJ/mol. This corresponds to an [BR]/[O] equilibrium constant of several hundred. Consistent with this, very small amounts of an O-like state were detected in BR at temperatures above 20°C [57]. When the pH is substantially lower than the pK a of the extracellular proton release complex (e.g., at pH 4), the pattern of free energy changes is different (Fig. 4B). No proton is released during the lifetime of M and there is no decrease in free energy at this time, but the proton uptake later in the cycle will occur with a correspondingly greater AG because the pK a minus pH difference for the uptake is greater. The total magnitude of the free energy difference between the uptake and release processes thus remains at about 30 kJ/mol, unaffected by the pH.

X-B. Enthalpy and entropy cycles: conversion of enthalpy into entropy at the reprotonation switch The activation enthalpies associated with the observable rate constants, and the photoacoustic determination of an approximately 80 k J / m o l enthalpy decrease * between proton release and uptake [58,141], allowed reconstruction of the enthalpy cycle [191]. The assumption that the entropy of the K state does not differ greatly from that of BR, and a reasonable although somewhat arbitrary equal apportion of the free energy changes between the M 1 ~ M 2 and O ---, BR reactions allowed, in turn, reconstruction of the entropy cycle [191]. The two cycles indicate that the retained free energy in the system changes from A H to - T . A S (i.e., from excess enthalpy to negative entropy) at the reprotonation switch. Enthalpy is con-: verted to entropy at the switch because until M 1 the excess free energy resides in the chromophore, mainly as retinal bond torsions and the lowered proton affinity of the Schiff base relative to D85, but in M 2 and the subsequent states the high Schiff base pK a recovers and the remaining free energy is transformed into the negative entropy of a restricted protein conformation a n d / o r increased organization of bound water. Relaxation of these drives the completion of the photocycle. Consistent with this, removing the protein from the motionally restricted environment of the purple membrane lattice caused little or no change in the enthalpies and entropies of the photocycle reactions be-

* This is greater than the entire enthalpygain of the systemafter absorption of the photon,i.e., the enthalpydrops be/owthat of the initial state, and the subsequent events must be endothermic.

fore the switch, but resulted in large changes after the switch [190]. It is significant that these changes in the detergent-solubilized protein, as well as in purple membranes which contain residue replacements that perturb non-covalent bonds assumed to play roles in proton transfer between the Schiff base and D96 (e.g., R227Q or T46V) (Brown, Cao, Needleman and Lanyi, unpublished data), always include decreased activation enthalpies and more negative activation entropies. It appears, therefore, that the protein under these conditions becomes more flexible but less ordered ( a n d / o r bind less ordered water). Although the transition states of the reactions are reached at less enthalpic cost, the protein must pass through a greater number of conformational states ( a n d / o r organize more water). Since at the M 1 ---, M 2 reaction the enthalpy of the system decreases below the initial level, the M 2 to BR portion of the photocycle contains mainly endothermic reactions *. This explains the well-known fact that the transient accumulation of the O intermediate, unlike the others, is greater at higher temperatures [31,97,191].

Xl. Summary In spite of many still unsolved problems, the mechanism and energetics of the light-driven proton transport are now basically understood. Energy captured during photoexcitation, and retained in the form of bond rotations and strains of the retinal, is transformed into directed changes in the pK a values of vectorially arranged proton transfer groups. The framework for the spatial and temporal organization of these changes is provided by the protein near the retinal Schiff base. The transport is completed by proton transfer among three essential groups in three domains lying roughly parallel with the membrane plane (Fig. 1): (a) the anionic D85 that is included in a complex of residues on the extracellular side containing also R82, D212, Y57 and bound water; (b) the protonated Schiff base; and (c) the protonated D96 that is included in a complex of residues on the cytoplasmic side containing also R227, T46, $226, and bound water. Other neighboring polar groups and water bound elsewhere which play a role in the transport do so either by further influencing the pK a values of the three protonable groups, or by providing passive pathways for proton transfer. The Schiff base proton, destabilized after photoexciration, is transferred to the low pK a group D85 located on the extracellular side. The access of the deproto-

* The enthalpyand entropydifferencebetween the N and O intermediates was determined also in temperature jump experiments [31]; the numericalvalues confirmremarkablywell those obtained from the temperature dependenceof rates [191].

258 nated Schiff base then changes to the cytoplasmic side (the 'reprotonation switch') and its proton affinity increases. Finally, the proton of the high p K a group D96, with access to the cytoplasmic side, is destabilized by a protein conformational change through rearrangement of R227, T46, $226 and bound water, and becomes transferred to the Schiff base. As shown schematically in Fig. 3, these internal events are coupled to proton release and uptake at the two aqueous surfaces. The charge of the extracellular hydrogenbonded complex is redistributed upon protonation of D85, and if the p H is above the PKa of the complex a proton is released to the bulk. After reprotonation of the Schiff base the p K a of the cytoplasmic hydrogenbonded complex is raised well above the pH, and D96 regains a proton from the bulk. If the p H is lower than the p K a of the extracellular complex the proton release is delayed until the end of the photocycle. In either sequence there is net transfer of a proton from the cytoplasmic to the extracellular phase. The transfer of excess free energy from the chromophore to the protein, and finally to the transported proton, is described by a characteristic thermodynamic cycle. At physiological p H the excess enthalpy retained in the form of local perturbation near the active site (the retinal Schiff base) drives proton transfer to D85 and release on the extracellular side. During the reprotonation switch the active site loses enthalpy, and the negative entropy retained now globally by a distorted protein conformation provides for perturbation of D96 and reprotonation of the Schiff base. Relaxation of the protein drives proton uptake on the cytoplasmic side. AG is transformed into proton electrochemical potential as a proton is released at a p H higher than the p K a of the extracellular proton release complex and a proton is taken up at a p H lower than the p K a of the cytoplasmic proton uptake complex.

Acknowledgements The author is supported by grants from the National Institutes of Health (GM 29498), the U.S. Department of Energy (DE-FG03-86ER13525) and the National Science Foundation (MCB-9202209, principal investigator R. Needleman). He is grateful to Drs. L.S. Brown, A. Maeda, R. Needleman, S.H. White, L. Zimfinyi, and to two anonymous reviewers for their critical reading of this manuscript.

References 1 Albery, W.J and Knowles, J.R. (1976) Biochemistry 15, 56315640. 2 Alshuth, T. and Stockburger, M. (1986) Photochem. Photobiol. 43, 55-66. 3 Ames, J.B., BoRon, S.R., Netto, M.M. and Mathies, R.A. (1990) J. Am. Chem. Soc. 112, 9007-9009.

4 Ames, J.B. and Mathies, R.A. (1990) Biochemistry 29, 71817190. 5 Ames, J.B., Ros, M., Raap, J., Lugtenburg, J. and Mathies, R.A. (1992) Biochemistry31, 5328-5334. 6 Aton, B., Doukas, A.G., CaUender, R.H., Becher, B. and Ebrey, T.G. (1977) Biochemistry. 16, 2995-2999. 7 Baasov, T., Friedman, N. and Sheves, M. (1987) Biochemistry 26, 3210-3217. 8 Baasov, T. and Sheves, M. (1986) Biochemistry25, 5249-5258. 9 Balashov, S., Govindjee, R., Kono, M., Lukashov, E., Ebrey, T.G., Feng, Y., Crouch, R.K. and Menick, D.R. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 111-114, John Libbey Eurotext, Montrouge. 10 Balashov, S.P., Govindjee, R. and Ebrey, T.G. (1991) Biophys.J. 60, 475-490. 11 Balashov, S.P., Imasheva, E.S., Litvin, F.F. and Lozier, R.H. (1990) FEBS Lett. 271, 93-96. 12 Baldwin, J.M., Henderson, R., Beckmann, E. and Zemlin, F. (1988) J. Mol. Biol. 202, 585-591. 13 Bashford, D. and Gerwert, K. (1992) J. Mol. Biol. 224, 473-486. 14 Beach, J.M. and Fager, R.S. (1985) Photochem. Photobiol. 41, 557-562. 15 Becher, B., Tokunaga, F. and Ebrey, T.G. (1978) Biochemistry. 17, 2293-2300. 16 Birge, R.R. (1990) Biochim. Biophys. Acta 1016, 293-327. 17 Birge, R.R., Cooper, T.M., Lawrence, A.F., Masthay, M.B., Zhang, C.-F. and Zidovetzki, R. (1991) J. Am. Chem. Soc. 113, 4327-4328. 18 Bousch6, O., Braiman, M., He, Y.-W., Marti, T., Khorana, H.G. and Rothschild, K.J. (1991) J. Biol. Chem. 266, 11063-11067. 19 Bousch6, O., Sonar, S., Krebs, M.P., Khorana, H.G. and Rothschild, K.J. (1992) Photochem. Photobiol. 56, 1085-1095. 20 Braiman, M.S., Ahl, P.L. and Rothschild, K.J. (1987) Proc. Natl. Acad. Sci. USA 84, 5221-5225. 21 Braiman, M.S., Bousch6, O. and Rothschild, K.J. (1991) Proc. Natl. Acad. Sci. USA 88, 2388-2392. 22 Braiman, M.S. and Mathies, R.A. (1982) Proc. Natl. Acad. Sci. USA 79, 403-407. 23 Braiman, M.S., Mogi, T., Marti, T., Stern, L.J., Khorana, H.G. and Rothschild, K.J. (1988) Biochemistry27, 8516-8520. 24 Braiman, M.S., Mogi, T., Stern, L.J., Hackett, R.D., Chao, B.H. and Khorana, H.G. (1988) Proteins 3, 219-229. 25 Brown, L.S., Boner, L., Needleman, R. and Lanyi, J.K. (1993) Biophys. J., in press. 26 Brown, L.S., Zim~inyi, L., Ottolenghi, M., Needleman, R. and Lanyi, J.K. (1993) Biochemistry65, 124-130. 27 Cao, Y., Brown, L.S., Needleman, R. and Lanyi, J.K. (1993) Biochemistry 32, 7679-7685. 28 Cao, Y., Vfir6, G., Chang, M., Ni, B., Needleman, R. and Lanyi, J.K. (1991) Biochemistry30, 10972-10979. 29 Cao, Y., V~ir6, G., Klinger, A.L., Czajkowsky, D.M., Braiman, M.S., Needleman, R. and Lanyi, J.K. (1993) Biochemistry 32, 1981-1990. 30 Chernavskii, D.S., Chizhov, I.V., Lozier, R.H., Murina, T.M., Prokhorov, A.M. and Zubov, B.V. (1989) Photochem. Photobiol. 49, 649-653. 31 Chizhov, I., Engelhard, M., Chernavskii, D.S., Zubov, B. and Hess, B. (1992) Biophys. J. 61, 1001-1006. 32 Dancsh~izy,Z., Govindjee, R. and Ebrey, T.G. (1988) Proc. Natl. Acad. Sci. USA 85, 6358-6361. 33 De Groot, H.J.M., Harbison, G.S., Herzfeld, J. and Griffin, R.G. (1989) Biochemistry28, 3346-3353. 34 De Groot, H.J.M., Smith, S.O., Courtin, J., Van den Berg, E., Winkel, C., Lugtenburg, J., Griffin, R.G. and Herzfeld, J. (1990) Biochemistry 29, 6873-6883. 35 Dencher, N.A., Dresselhaus, D., Zaccai, G. and Biildt, G. (1989) Proc. Natl. Acad. Sci. USA 86, 7876-7879.

259 36 Dencher, N.A., Heberle, J., Biildt, G., H61tje, H.-D. and H61tje, M. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 213-216, John Libbey Eurotext, Montrouge. 37 Dencher, N.A., Heberle, J., Biildt, G., H61tje, H.-D. and H61tje, M. (1992) in Membrane Proteins: Structures, Interactions and Models (Pullman, A., Jortner, J. and Pullman, B., eds.), pp. 69-84, Kluwer, Dordrecht. 38 Dencher, N.A. and Wilms, M. (1975) Biophys. Struct. Mech. 1, 259-271. 39 D6r, A., Sz~iraz, S., T6th-Bocon~di, R., Tokaji, Z., Keszthelyi, L. and Stoeckenius, W. (1991) Proc. Natl. Acad. Sci. USA 88, 4751-4755. 40 Doukas, A.G., Pande, A., Suzuki, T., Callender, R.H., Honig, B. and Ottolenghi, M. (1981) Biophys. J. 33, 275-279. 41 Drachev, L.A., Kaulen, A.D. and Komrakov, A.Y. (1992) FEBS Lett. 313, 248-250. 42 Drachev, L.A., Kaulen, A.D. and Skulachev, V.P. (1984) FEBS Lett. 178, 331-335. 43 Druckmann, S., Friedman, N., Lanyi, J.K., Needleman, R., Ottolenghi, M. and Sheves, M. (1992) Photochem. Photobiol. 56, 1041-1047. 44 Druckmann, S., Heyn, M.P., Lanyi, J.K., Ottolenghi, M. and Zim~inyi, L. (1993) Biophys. J., in press. 45 Druckmann, S., Ottolenghi, M., Pande, A., Pande, J. and Callender, R.H. (1982) Biochemistry. 21, 4953-4959. 46 Dufiach, M., Berkowitz, S., Marti, T., He, Y.-W., Subramaniam, S., Khorana, H.G. and Rothschild, K.J. (1990) J. Biol. Chem. 265, 16978-16984. 47 Dufiach, M., Marti, T., Khorana, H.G. and Rothschild, K.J. (1990) Proc. Natl. Acad. Sci. USA 87, 9873-9877. 48 Ebrey, T.G. (1993) in: Thermodynamics of membranes, receptors and channels (ed. Jackson, M.), pp. 353-387, CRC Press, New York. 49 Edgerton, M.E., Moore, T.A. and Greenwood, C. (1980) Biochem.J. 189, 413-420. 50 Eisfeld, W. and Stockburger, M. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 139-142, John Libbey Eurotext, Montrouge. 51 Engelhard, M., Gerwert, K., Hess, B., Kreutz, W. and Siebert, F. (1985) Biochemistry 24, 400-407. 52 Fahmy, K., Siebert, F., Grossjean, M.F. and Tavan, P. (1989) J. Mol. Struct. 214, 257-288. 53 Fahmy, K., Weidlich, O., Engelhard, M., Sigrist, H. and Siebert, F. (1993) Biochemistry 32, 5862-5869. 54 Fischer, U. and Oesterhelt, D. (1979) Biophys. J. 28, 211-230. 55 Fodor, S.P., Ames, J.B., Gebhard, R., van der Berg, E.M., Stoeckenius, W., Lugtenburg, J. and Mathies, R.A. (1988) Biochemistry 27, 7097-7101. 56 Fukuda, K. and Kouyama, T. (1992) Photochem. Photobiol. 56, 1057-1062. 57 Fukuda, K. and Kouyama, T. (1992) Biochemistry 31, 1174011747. 58 Garty, H., Caplan, S.R. and Cahen, D. (1982) Biophys. J. 37, 405-415. 59 Garty, H., Klemperer, G., Eisenbach, M. and Caplan, S.R. (1977) FEBS. Lett. 81, 238-242. 60 Gat, Y., Grossjean, M., Pinevsky, I., Takei, H., Rothman, Z., Sigrist, H., Lewis, A. and Sheves, M. (1992) Proc. Natl. Acad. Sci. USA 89, 2434-2438. 61 Gat, Y. and Sheves, M. (1993) J. Am. Chem. Sot., in press. 62 Gerwert, K., Hess, B., Soppa, J. and Oesterhelt, D. (1989) Proc. Natl. Acad. Sci.USA 86, 4943-4947. 63 Gerwert, K. and Siebert, F. (1986) EMBO J. 5, 805-811. 64 Gerwert, K., Souvignier, G. and Hess, B. (1990) Proc. Natl. Acad. Sci. USA 87, 9774-9778.

65 Glaeser, R.M., Baldwin, J.M., Ceska, T.A. and Henderson, R. (1986) Biophys. J. 50, 913-920. 66 Greenhalgh, D.A., Altenbaeh, C., Hubbell, W.L. and Khorana, H.G. (1991) Proc. Natl. Acad. Sci. USA 88, 8626-8630. 67 Greenhalgh, D.A., Subramaniam, S., Alexiev, U., Otto, H., Heyn, M.P. and Khorana, H.G. (1992) J. Biol. Chem. 267, 25734-25738. 68 Grossjean, M., Tavan, P. and Schulten, K. (1989) Eur. Biophys. J. 16, 341-349. 69 Grzesiek, S. and Dencher, N.A. (1986) FEBS Lett. 208, 337-342. 70 Gutman, M., Nachliel, E. and Gershon, E. (1985) Biochemistry 24, 2937-2941. 71 Hanamoto, J.H., Dupuis, P. and EI-Sayed, M.A. (1984) Proc. Natl. Acad. Sci. USA 81, 7083-7087. 72 He, Y., Krebs, M.P., Fischer, W.B., Khorana, H.G. and Rothschild, K.J. (1993) Biochemistry 32, 2282-2290. 73 Heberle, J. and Dencher, N.A. (1990) FEBS Lett. 277, 277-280. 74 Heberle, J. and Dencher, N.A. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 221-224, John Libbey Eurotext, Montrouge. 75 Heberle, J. and Dencher, N.A. (1992) Proc. Natl. Acad. Sci. USA 89, 5996-6000. 76 Henderson, R., Baldwin, J.M., Ceska, T.A., Zemlin, F., Beckmann, E. and Downing, K.H. (1990) J. Mol. Biol. 213, 899-929. 77 Henderson, R., Baldwin, J.M., Downing, K.H., Lepault, J. and Zemlin, F. (1986) Ultramicroscopy 19, 147-178. 78 Herzfeld, J., Das Gupta, S.K., Farrar, M.R., Harbison, G.S., McDermott, A.E., Pelletier, S.L., Raleigh, D.P., Smith, S.O., Winkel, C., Lugtenburg, J. and Griffin, R.G. (1990) Biochemistry 29, 5567-5574. 79 Heyn, M.P., Cherry, R.J. and Muller, U. (1977) J. Mol. Biol. 117, 607-620. 80 Hildebrandt, P. and Stockburger, M. (1984) Biochemistry 23, 5539-5548. 81 Holz, M., Drachev, L.A., Mogi, T., Otto, H., Kaulen, A.D., Heyn, M.P., Skulachev, V.P. and Khorana, H.G. (1989) Proc. Natl. Acad. Sci. USA 86, 2167-2171. 82 Honig, B., Ebrey, T.G., Callender, R.H., Dinur, U. and Ottolenghi, M. (1979) Proc. Natl. Acad. Sci. USA 76, 2503-2507. 83 Jonas, R. and Ebrey, T.G. (1991) Proc. Natl. Acad. Sci. USA 88, 149-153. 84 Kakitani, H., Kakitani, T., Rodman, H. and Honig, B. (1985) Photochem. Photobiol. 41, 471-479. 85 Kalisky, O., Ottolenghi, M., Honig, B. and Korenstein, R. (1981) Biochemistry. 20, 649-655. 86 Khorana, H.G. (1993) Proc. Natl. Acad. Sci. USA 90, 1166-1171. 87 Koch, M.H.J., Dencher, N.A., Oesterhelt, D., Plfhn, H.-J., Rapp, G. and Biildt, G. (1991) EMBO J. 10, 521-526. 88 Korenstein, R. and Hess, B. (1977) Nature 270, 184-186. 89 Koutalos, Y., Ebrey, T.G., Gilson, H.R. and Honig, B. (1990) Biophys. J. 58, 493-501. 90 Kouyama, T., Nasuda-Kouyama, A., Ikegami, A., Mathew, M.K. and Stoeckenius, W. (1988) Biochemistry 27, 5855-5863. 91 Krebs, M.P., Mollaaghababa, R. and Khorana, H.G. (1993) Proc. Natl. Acad. Sci. USA 90, 1987-1991. 92 Kuschmitz, D. and Hess, B. (1982) FEBS Lett. 138, 137-140. 93 Lanyi, J.K. (1986) Biochemistry 25, 6706-6711. 94 Lanyi, J.K. (1992) J. Bioenerg. Biomembr. 24, 169-179. 95 Lanyi, J.K., Tittor, J., V~r6, G., Krippahl, G. and Oesterhelt, D. (1992) Biochim. Biophys. Acta 1099, 102-110. 96 Lewis, A., Spoonhower, J.P., Bogomolni, R.A., Lozier, R.H. and Stoeckenius, W. (1974) Proc. Natl. Acad. Sci. USA 71, 44624466. 97 Li, Q., Govindjee, R. and Ebrey, T.G. (1984) Proc. Natl. Acad. Sci. USA 81, 7079-7082. 98 Lin, S.W., Fodor, S.P.A., Miercke, L.J.W., Shand, R.F., Betlach,

260 M.C., Stroud, R.M. and Mathies, R.A. (1991) Photochem. Photobiol. 53, 341-346. 99 Lin, S.W. and Mathies, R.A. (1989) Biophys. J. 56, 653-660. 100 Liu, S.Y. (1990) Biophys. J. 57, 943-950. 101 Liu, S.Y., Govindjee, R. and Ebrey, T.G. (1990) Biophys. J. 57, 951-963. 102 Lozier, R.H., Bogomolni, R.A. and Stoeckenius, W. (1975) Biophys. J. 15, 955-963. 103 Lozier, R.H., Niederberger, W., Bogomolni, R.A., Hwang, S. and Stoeckenius, W. (1976) Biochim. Biophys. Acta 440, 545556. 104 Lozier, R.H., Xie, A., Hofrichter, J. and Clore, G.M. (1992) Proc. Natl. Acad. Sci. USA 89, 3610-3614. 105 Maeda, A., Sasaki, J., Pfefferld, J.-M., Shichida, Y. and Yoshizawa, T. (1991) Photochem. Photobiol. 54, 911-921. 106 Maeda, A., Sasaki, J., Shichida, Y. and Yoshizawa, T. (1992) Biochemistry 31, 462-467, 107 Maeda, A., Sasaki, J., Shichida, Y., Yoshizawa, T., Chang, M., Ni, B., Needleman, R. and Lanyi, J.K. (1992) Biochemistry 31, 4684-4690. 108 Maeda, A., Sasaki, J., Yamazaki, Y., Needleman, R. and Lanyi, J.K. (1993) Biochemistry (submitted). 109 Marinetti, T. and Mauzerall, D. (1983) Proc. Natl. Acad. Sci. USA 80, 178-180. 110 Marti, T., Otto, H., Mogi, T., R6sselet, S.J., Heyn, M.P. and Khorana, H.G. (1991) J. Biol. Chem. 266, 6919-6927. 111 Mathies, R.A., Lin, S.W., Ames, J.B. and Pollard, W.T. (1991) Annu. Rev. Biophys. Biophys. Chem. 20, 491-518. 112 Maurer, R., Vogel, J. and Schneider, S. (1987) Photochem. Photobiol. 46, 247-253. 113 Merz, H. and Zundel, G. (1981) Biochem. Biophys. Res. Commun. 101, 540-546. 114 Metz, G., Siebert, F. and Engelhard, M. (1992) FEBS Lett. 303, 237-241. 115 Meyer, E. (1992) Protein Sci. 1, 1543-1562. 116 Miercke, L.J.W., Betlach, M.C., Mitra, A.K., Shand, R.F., Fong, S.K. and Stroud, R.M. (1991) Biochemistry 30, 3088-3098. 117 Milder, S.J. (1991) Biophys. J. 60, 440-446. 118 Milder, S.J., Thorgeirsson, T.E., Miercke, L.J.W., Stroud, R.M. and Kliger, D.S. (1991) Biochemistry 30, 1751-1761. 119 Miller, A. and Oesterhelt, D. (1990) Biochim. Biophys. Acta Bio-Energetics 1020, 57-64. 120 Mitchell, D. and Rayfield, G.W. (1986) Biophys. J. 49, 563-566. 121 Mogi, T., Stern, L.J., Hackett, N.R. and Khorana, H.G. (1987) Proc. Natl. Acad. Sci. USA 84, 5595-5599. 122 Mogi, T., Stern, L.J., Marti, T., Chao, B.H. and Khorana, H.G. (1988) Proc. Natl. Acad. Sci. USA 85, 4148-4152. 123 Moltke, S., Heyn, M.P., Krebs, M.P., Mollaaghababa, R. and Khorana, H.G. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 201-204, John Libbey Eurotext, Montrouge. 124 Mowery, P.C., Lozier, R.H., Chae, Q., Tseng, Y.W., Taylor, M. and Stoeckenius, W. (1979) Biochemistry 18, 4100-4107. 125 Miiller, K.-H., Butt, H.J., Bamberg, E., Fendler, K., Hess, B., Siebert, F. and Engelhard, M. (1991) Eur. Biophys. J. 19, 241251. 126 Nagle, J.F. (1991) Biophys, J. 59, 476-487. 127 Nagle, J.F. and Mille, M. (1981) J. Chem. Phys. 74, 1367-1372. 128 Nagle, J.F., Parodi, L.A. and Lozier, R.H. (1982) Biophys. J. 38, 161-174. 129 Nagle, J.F. and Tristram-Nagle, S. (1983) J. Membr. Biol. 74, 1-14. 130 Nakanishi, K., Balogh-Nair, V., Arnaboldi, M., Tsujimoto, K. and Honig, B. (1980) J. Am. Chem. Soc. 102, 7945-7947. 131 Nakasako, M., Kataoka, M., Amemiya, Y. and Tokunaga, F. (1991) FEBS Lett. 292, 73-75.

132 Needleman, R., Chang, M., Ni, B., Vhr6, (3., Fornes, J., White, S.H. and Lanyi, J.K. (1991) J. Biol. Chem. 266, 11478-11484. 133 Nonella, M., Windemuth, A. and Schulten, K. (1991) Photochem. Photobioi. 54, 937-948. 134 Oesterhelt, D., Hegemann, P., Tavan, P. and Schulten, K. (1986) Eur. Biophys. J. 14, 123-129. 135 Oesterhelt, D., Tittor, J. and Bamberg, E. (1992) J. Bioenerg. Biomembr. 24, 181-191. 136 Ohno, K., Takeuchi, Y. and Yoshida, M. (1977) Biochim. Biophys. Acta 462, 575-582. 137 Ohtani, H., Itoh, H. and Shinmura, T. (1992) FEBS Lett. 305, 6-8. 138 Orlandi, G. and Schulten, K. (1979) Chem. Phys. Lett. 64, 370-374. 139 Ormos, P. (1991) Proc. Natl. Acad. Sci. USA 88, 473-477. 140 Ormos, P., Chu, K. and Mourant, J. (1992) Biochemistry 31, 6933-6937. 141 Ort, D.R. and Parson, W.W. (1979) Biophys. J. 25, 355-364. 142 Otto, H. and Heyn, M.P. (1991) FEBS Lett. 293, 111-114. 143 Otto, H., Marti, T., Holz, M., Mogi, T., Lindau, M., Khorana, H.G. and Heyn, M,P. (1989) Proc. Natl. Acad. Sci. USA 86, 9228-9232. 144 Otto, H., Marti, T., Holz, M., Mogi, T., Stern, L.J., Engel, F., Khorana, H.G. and Heyn, M.P. (1990) Proc. Natl. Acad. Sci. USA 87, 1018-1022. 145 Papadopoulos, G., Dencher, N.A., Zaccai, G. and Biildt, G. (1990) J. Mol. Biol. 214, 15-19. 146 Perkins, G.A., Liu, E., Burkard, F., Berry, E.A. and Glaeser, R.M. (1992) J. Struct. Biol. 109, 142-151. 147 Pfefferld, J.-M., Maeda, A., Sasaki, J. and Yoshizawa, T. (1991) Biochemistry 30, 6548-6556. 148 Pusch, C., Diller, R., Eisfeld, W., Lohrmann, R. and Stockburger, M. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 143-146, John Libbey Eurotext, Montrouge. 149 Rath, P., Krebs, M.P., He, Y., Khorana, H.G. and Rothschild, K.J. (1993) Biochemistry 32, 2272-2281. 150 Rohr, M., Schulenberg, P., G~irtner, W. and Braslavsky, S.E. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 151-154, John Libbey Eurotext, Montrouge. 151 Rothschild, K.J. (1992) J. Bioenerg. Biomembr. 24, 147-167. 152 Rothschild, K.J., Braiman, M.S., He, Y.-W., Marti, T. and Khorana, H.G. (1990) J. Biol. Chem. 265, 16985-16991. 153 Rothschild, K.J., He, Y.-W., Sonar, S., Marti, T. and Gobind Khorana, H. (1992) J. Biol. Chem. 267, 1615-1622. 154 Rothschild, K.J., Roepe, P,, Ahl, P.L., Earnest, T.N., Bogomolni, R.A., Das Gupta, S.K., Mulliken, C.M. and Herzfeld, J. (1986) Proc. Natl. Acad. Sci. USA 83, 347-351. 155 Rothschild, K.J., Zagaeski, M. and Cantore, W.A. (1981) Biochem. Biophys. Res. Commun. 103, 483-489. 156 Sasaki, J., Shichida, Y., Lanyi, J.K. and Maeda, A. (1992) J. Biol. Chem. 267, 20782-20786. 157 Scheiner, S. and Duan, X. (1991) Biophys. J. 60, 874-883. 158 Scheiner, S. and Hillenbrand, E.A. (1985) Proc. Natl. Acad. Sci. USA 82, 2741-2745. 159 Scherrer, P., Alexiev, U., Otto, H., Heyn, M.P., Marti, T. and Khorana, H.G. (1992) in Structures and Functions of Retinal Proteins (Rigaud, J.L., ed.), pp. 205-211, John Libbey Eurotext, Montrouge. 160 Schulten, K., Schulten, Z. and Tavan, P. (1984) in Information and Energy Transduction in Biological Membranes (Bolis, A., Helmreich, H. and Passow, H., eds.), pp. 113-131, Alan R. Liss, New York, 161 Schuiten, K. and Tavan, P. (1978) Nature 272, 85-86. 162 Sherman, W.V., Eicke, R.R., Stafford, S.R. and Wasaez, F.M. (1979) Photochem. Photobiol. 30, 727-729.

261 163 Sheves, M., Albeck, A., Friedman, N. and Ottolenghi, M. (1986) Proc. Natl. Acad. Sci. USA 83, 3262-3266. 164 Shichida, Y., Matuoka, S., Hidaka, Y. and Yoshizawa, T. (1983) Biochim. Biophys. Acta 723, 240-246. 165 Siebert, F., Mantele, W. and Kreutz, W. (1982) FEBS Lett. 141, 82-87. 166 Smith, S.O., Myers, A.B., Pardoen, J.A., Winkel, C., Mulder, P.P.J., Lugtenburg, J. and Mathies, R.A. (1984) Proc. Natl. Acad. Sci. USA 81, 2055-2059. 167 Smith, S.O., Pardoen, J.A., Mulder, P.P.J., Curry, B., Lugtenburg, J. and Mathies, R.A. (1983) Biochemistry 22, 6141-6148. 168 Sonar, S., Krebs, M.P., Khorana, H.G. and Rothschild, K.J. (1993) Biochemistry 32, 2263-2271. 169 Souvignier, G. and Gerwert, K. (1992) Biophys. J. 63, 1393-1405. 170 Spudich, J.L. and Bogomolni, R.A. (1988) Annu. Rev. Biophys. Biophys. Chem. 17, 193-215. 171 Stern, L.J., Ahl, P.L., Marti, T., Mogi, T., Dufiach, M., Berkovitz, S., Rothschild, K.J. and Khorana, H.G. (1989) Biochemistry 28, 10035-10042. 172 Stern, L.J. and Khorana, H.G. (1989) J. Biol. Chem. 264, 1420214208. 173 Subramaniam, S., Gerstein, M., Oesterhelt, D. and Henderson, R. (1993) EMBO J. 12, 1-8. 174 Subramaniam, S., Greenhaigh, D.A. and Khorana, H.G. (1992) J. Biol. Chem. 267, 25730-25733. 175 Subramaniam, S., Greenhalgh, D.A., Rath, P., Rothschild, K.J. and Khorana, H.G. (1991) Proc. Natl. Acad. Sci. USA 88, 6873-6877. 176 Subramaniam, S., Marti, T. and Khorana, H.G. (1990) Proc. Natl. Acad. Sci. USA 87, 1013-1017. 177 Takeuchi, Y., Ohno, K., Yoshida, M. and Nagano, K. (1981) Photochem. Photobiol. 33, 587-592. 178 Tavan, P., Schulten, K. and Oesterhelt, D. (1985) Biophys. J. 47, 415-430. 179 Thorgeirsson, T.E., Milder, S.J., Miercke, L.J.W., Betlach, M.C., Shand, R.F., Stroud, R.M. and Kliger, D.S. (1991) Biochemistry 30, 9133-9142.

180 Tittor, J., Soell, C., Oesterhelt, D., Butt, H.-J. and Bamberg, E. (1989) EMBO J. 8, 3477-3482. 181 Tokaji, Z. and Dancsh~izy, Z. (1992) FEBS Lett. 311, 267-270. 182 Trissl, H.W. (1990) Photochem. Photobiol. 51, 793-818. 183 Turner, G.J., Miercke, L.J.W., Thorgeirsson, T.E., Kliger, D.S., Betlach, M.C. and Stroud, R.M. (1993) Biochemistry 32, 13321337. 184 V~r6, G., Duschl, A. and Lanyi, J.K. (1990) Biochemistry 29, 3798-3804. 185 V~ir6, G. and Lanyi, J.K. (1989) Biophys. J. 56, 1143-1151. 186 V~ir6, G. and Lanyi, J.K. (1990) Biochemistry 29, 6858-6865. 187 V~ir6, G. and Lanyi, J.K. (1990) Biochemistry 29, 2241-2250. 188 V~r6, G. and Lanyi, J.K. (1991) Biophys. J. 59, 313-322. 189 V~ir6, G. and Lanyi, J.K. (1991) Biochemistry 30, 5008-5015. 190 V~ir6, G. and Lanyi, J.K. (1991) Biochemistry 30, 7165-7171. 191 V~ir6, G. and Lanyi, J.K. (1991) Biochemistry 30, 5016-5022. 192 V~.r6, G., Zim~nyi, L., Chang, M., Ni, B., Needleman, R. and Lanyi, J.K. (1992) Biophys. J. 61,820-826. 193 Warshel, A. (1978) Proc. Natl. Acad. Sci. USA 75, 2558-2562. 194 Warshel, A. and Ottolenghi, M. (1979) Photochem. Photobiol. 30, 291-293. 195 Xie, A.H., Nagle, J.F. and Lozier, R.H. (1987) Biophys. J. 51, 627-635. 196 Yamamoto, N., Naramoto, S. and Ohtani, H. (1992) FEBS Lett. 314, 345-347. 197 Zhou, F., Windemuth, A. and Schulten, K. (1993) Biochemistry 32, 2291-2306. 198 Zimfinyi, L., Cao, Y., Chang, M., Ni, B., Needleman, R. and Lanyi, J.K. (1992) Photochem. Photobiol. 56, 1049-1055. 199 Zim~nyi, L., Cao, Y., Needleman, R., Ottolenghi, M. and Lanyi, J.K. (1993) Biochemistry 32, 7669-7678. 200 Zim~inyi, L., Keszthelyi, L. and Lanyi, J.K. (1989) Biochemistry 28, 5165-5172. 201 Zimfinyi, L. and Lanyi, J.K. (1993) Biophys. J. 64, 240-251. 202 Zim~inyi, L., V~r6, G., Chang, M., Ni, B., Needleman, R. and Lanyi, J.K. (1992) Biochemistry 31, 8535-8543.