Protein conformational changes in the bacteriorhodopsin photocycle

We report a comprehensive electron crystallographic analysis of conformational changes in the photocycle of wild-type bacteriorhodopsin and in a variety of mutant proteins with kinetic defects in the photocycle. Specific intermediates that accumulate in the late stages of the photocycle of wild-type bacteriorhodopsin, the single mutants D38R, D96N, D96G, T46V, L93A and F219L, and the triple mutant D96G/F171C/F219L were trapped by freezing two-dimensional crystals in liquid ethane at varying times after illumination with a light flash. Electron diffraction patterns recorded from these crystals were used to construct projection difference Fourier maps at 3.5 A resolution to define light-driven changes in protein conformation.Our experiments demonstrate that in wild-type bacteriorhodopsin, a large protein conformational change occurs within approximately 1 ms after illumination. Analysis of structural changes in wild-type and mutant bacteriorhodopsins under conditions when either the M or the N intermediate is preferentially accumulated reveals that there are only small differences in structure between M and N intermediates trapped in the same protein. However, a considerably larger variation is observed when the same optical intermediate is trapped in different mutants. In some of the mutants, a partial conformational change is present even prior to illumination, with additional changes occurring upon illumination. Selected mutations, such as those in the D96G/F171C/F219L triple mutant, can sufficiently destabilize the wild-type structure to generate almost the full extent of the conformational change in the dark, with minimal additional light-induced changes. We conclude that the differences in structural changes observed in mutants that display long-lived M, N or O intermediates are best described as variations of one fundamental type of conformational change, rather than representing structural changes that are unique to the optical intermediate that is accumulated. Our observations thus support a simplified view of the photocycle of wild-type bacteriorhodopsin in which the structures of the initial state and the early intermediates (K, L and M1) are well approximated by one protein conformation, while the structures of the later intermediates (M2, N and O) are well approximated by the other protein conformation. We propose that in wild-type bacteriorhodopsin and in most mutants, this conformational change between the M1 and M2 states is likely to make an important contribution towards efficiently switching proton accessibility of the Schiff base from the extracellular side to the cytoplasmic side of the membrane.     

pH-induced structural changes in bacteriorhodopsin studied by Fourier transform infrared spectroscopy.

Previous C13-NMR studies showed that two of the four internal aspartic acid residues (Asp-96 and Asp-115) of bacteriorhodopsin (bR) are protonated up to pH = 10, but no accurate pKa of these residues has been determined. In this work, infrared spectroscopy with the attenuated total reflection technique was used to characterize pH-dependent structural changes of ground-state, dark-adapted wild-type bacteriorhodopsin and its mutant (D96N) with aspartic acid-96 replaced by asparagine. Data indicated deprotonation of Asp-96 at high pH (pKa = 11.4 +/- 0.1), but no Asp-115 titration was observed. The analysis of the whole spectral region characteristic to complex conformational changes in the protein showed a more complicated titration with an additional pKa value (pKa1 = 9.3 +/- 0.3 and pKa2 = 11.5 +/- 0.2). Comparison of results obtained for bR and the D96N mutant of bR shows that the pKa approximately 11.5 characterizes not a direct titration of Asp-96 but a protein conformational change that makes Asp-96 accessible to the external medium.     

Crystal structures of bR(D85S) favor a model of bacteriorhodopsin as a hydroxyl-ion pump

Structural features on the extracellular side of the D85S mutant of bacteriorhodopsin (bR) suggest that wild-type bR could be a hydroxyl-ion pump. A position between the protonated Schiff base and residue 85 serves as an anion-binding site in the mutant protein, and hydroxyl ions should have access to this site during the O-intermediate of the wild-type bR photocycle. The guanidinium group of R82 is proposed (1) to serve as a shuttle that eliminates the Born energy penalty for entry of an anion into this binding pocket, and conversely, (2) to block the exit of a proton or a related proton carrier.     

Study of intermediate N using mutant forms of bacteriorhodopsin at Asp-96]

It has been found that the N(P, R)-type intermediate of the photocycle is formed in the Asp-96-->Asn mutant at acidic pH. Azide, which strongly activates the M decay in this mutant, allows the N intermediate to be shown also at neutral pH. Under these conditions mutant N decays in a pH-independent fashion. In the presence of azide, the H+ uptake by Asp-96-->Asn mutant bacteriorhodopsin follows the M decay, whereas the N decay occurs at a much slower rate. Two electrogenic stages have been shown to be associated with the M--->bR step in the Asp-96--->Asn mutant photocycle. The faster and slower stages correlate with the M--->N and N--->bR transitions, respectively. In the Asp-96--->Asn mutant, high concentrations of azide are found to increase the M decay rate up to the values higher than those in the wild-type protein, both with or without azide. Such an effect is absent for the Asp-96-->Glu mutant. The activation energies for M--->N and N--->bR transitions in the wild-type protein are equal to 18 and 19 kcal x mole-1, respectively. In the Asp-96-->Asn mutant without azide, the activation energy of the M decay is only 5 kcal x mole-1, whereas in the presence of azide in this mutant the activation energies for M and N decays are 8 and 9 kcal x mole-1, respectively. A scheme of events accompanying the Schiff base reprotonation during the photocycle is discussed.     

Proton translocation by bacteriorhodopsin in the absence of substantial conformational changes

Unlike wild-type bacteriorhodopsin (BR), the BR triple mutant D96G/F171C/F219L has been shown to undergo only minor structural rearrangements during its photocycle. Nonetheless, the mutant is capable of transporting protons at a rate of 125(+/-40) H+/BR per minute under light-saturating conditions. Light adaptation of the triple mutant's retinal proceeds in a pH-dependent manner up to a maximum of 63% all-trans. These two findings imply that the transport activity of the triple mutant comprises 66% of the wild-type activity. Time-resolved spectroscopy reveals that the identity and sequence of intermediates in the photocycle of the triple mutant in the all-trans configuration correspond to that of wild-type BR. The only differences relate to a slower rise and decay of the M and O intermediates, and a significant spectral contribution from a 13-cis component. No indication for accumulation of the N intermediate is found under a variety of conditions that normally favor the formation of this species in wild-type BR. The Fourier transform infrared (FTIR) spectrum of the M intermediate in the triple mutant resembles that of wild type. Minor changes in the amide I region during the photocycle suggest that only small movements of the protein backbone occur. Electron microscopy reveals large differences in conformation between the unilluminated state of the mutant protein and wild-type but no light-induced changes in time-resolved measurements. Evidently, proton transport by the triple mutant does not require the major conformational rearrangements that occur on the same time-scale with wild-type. Thus, we conclude that large conformational changes observed in the photocycle of the wild-type and many BR mutants are not a prerequisite for the change in accessibility of the Schiff base nitrogen atom that must occur during vectorial catalysis to allow proton transport.     

Formation of the M(N) (M(open)) intermediate in the wild-type bacteriorhodopsin photocycle is accompanied by an absorption spectrum shift to shorter wavelength, like that in the mutant D96N bacteriorhodopsin photocycle

Maximum of the M intermediate difference spectrum in the wild-type Halobacterium salinarium purple membrane is localized at 405-406 nm under conditions favoring accumulation of the M(N) intermediate (6 M guanidine chloride, pH 9.6), whereas immediately after laser flash the maximum is localized at 412 nm. The maximum is also localized at 412 nm 0.1 msec after the flash in the absence of guanidine chloride at pH 11.3. Within several milliseconds the maximum is shifted to short-wavelength region by 5-6 nm. This shift is similar to that in the D96N mutant which accompanies the M(N) (M(open)) intermediate formation. The main two differences are: 1) the rate of the shift is slower in the wild-type bacteriorhodopsin, and is similar to the rate of the M to N intermediate transition (t1/2 approximately 2 msec); 2) the shift in the wild-type bacteriorhodopsin is observed at alkaline pH values which are higher than pK of the Schiff base (approximately 10.8 at 1 M NaCl) in the N intermediate with the deprotonated Asp-96. Thus, the M(N) (M(open)) intermediate with open water-permeable inward proton channel is observed only at high pH, when the Schiff base and Asp-96 are deprotonated. The data confirmed our earlier conclusion that the M intermediate observed at lower pH has the closed inward proton channel.     

Two forms of N intermediate (N(open) and N(closed)) in the bacteriorhodopsin photocycle.

Glutaraldehyde, aluminum ions and glycerol (that inhibit the M intermediate decay in the wild-type bacteriorhodopsin and azide-induced M decay in the D96N mutant by stabilization of the M(closed)) accelerate the N decay in the D96N mutant. The aluminum ions, the most potent activator of the N decay, induce a blue shift of the N difference spectrum by approximately 10 nm. Protonated azide as well as acetate and formate inhibit the N decay in both the D96N mutant and the wild-type protein. It is concluded that the N intermediate represents, in fact, an equilibrium mixture of the two ('open' and 'closed') forms. These two forms, like M(closed) and M(open), come to an equilibrium in the microseconds range. The absorption spectrum of the N(open) is slightly shifted to red in comparison to that of the N(closed). Again, this resembles the M forms. 13-cis-all-trans re-isomerization is assumed to occur in the N(closed) form only. Binding of 1-2 molecules of protonated azide stabilizes the N(open) form. Existence of the 'open' and 'closed' forms of the M and N intermediates provides the appropriate explanation of the cooperative phenomenon as well as some other effects on the bacteriorhodopsin photocycle. Summarizing the available data, we suggest that M(open) is identical to the M(N) form, whereas M1 and M2 are different substates of M(closed).     

The role of water in the extracellular half channel of bacteriorhodopsin.

The changes in the photocycle of the wild type and several mutant bacteriorhodopsin (D96N, E204Q, and D212N) were studied on dried samples, at relative humidities of 100% and 50%. Samples were prepared from suspensions at pH approximately 5 and at pH approximately 9. Intermediate M with unprotonated Schiff base was observed at the lower humidity, even in the case where the photocycle in suspension did not contain this intermediate (mutant D212N, high pH). The photocycle of the dried sample stopped at intermediate M1 in the extracellular conformation; conformation change, switching the accessibility of the Schiff base to the cytoplasmic side, and proton transport did not occur. The photocycle decayed slowly by dissipating the absorbed energy of the photon, and the protein returned to its initial bacteriorhodopsin state, through several M1-like substates. These substates presumably reflect different paths of the proton back to the Schiff base, as a consequence of the bacteriorhodopsin adopting different conformations by stiffening on dehydration. All intermediates requiring conformational change were hindered in the dried form. The concentration of intermediate L, which appears after isomerization of the retinal from all-trans to 13-cis, during local relaxation of the protein, was unusually low in dried samples. The lack of intermediates N and O demonstrated that the M state did not undergo a change from the extracellular to the cytoplasmic conformation (M1 to M2 transition), as already indicated by Fourier transform infrared spectroscopy, quasielastic incoherent neutron scattering, and electric signal measurements described in the literature.     

Positioning proton-donating residues to the Schiff-base accelerates the M-decay of pharaonis phoborhodopsin expressed in Escherichia coli.

Phoborhodopsin (also called sensory rhodopsin II, sR-II) is a receptor for the negative phototaxis of Halobacterium salinarum (pR), and pharaonis phoborhodopsin (ppR) is the corresponding receptor of Natronobacterium pharaonis. pR and ppR are retinoid proteins and have a photocycle similar to that of bacteriorhodopsin (bR). A major difference between the photocycle of the ion pump bR and the sensor pR or ppR is found in their turnover rates which are much faster for bR. A reason for this difference might be found in the lack of a proton-donating residue to the Schiff base which is formed between the lysine of the opsin and retinal. To reconstruct a bR-like photochemical behavior, we expressed ppR mutants in Escherichia coli in which proton-donating groups have been reintroduced into the cytoplasmic proton channel. In measurement of the photocycle it could be shown that the F86D mutant of ppR (Phe86 was substituted by Asp) showed a faster decay of M-intermediate than the wild-type, which was even accelerated in the F86D/L40T double mutant.     

Chemical and physical evidence for multiple functional steps comprising the M state of the bacteriorhodopsin photocycle

In the photocycle of bacteriorhodopsin (bR), light-induced transfer of a proton from the Schiff base to an acceptor group located in the extracellular half of the protein, followed by reprotonation from the cytoplasmic side, are key steps in vectorial proton pumping. Between the deprotonation and reprotonation events, bR is in the M state. Diverse experiments undertaken to characterize the M state support a model in which the M state is not a static entity, but rather a progression of two or more functional substates. Structural changes occurring in the M state and in the entire photocycle of wild-type bR can be understood in the context of a model which reconciles the chloride ion-pumping phenotype of mutants D85S and D85T with the fact that bR creates a transmembrane proton-motive force.     

Photoexcitation of the O-intermediate in bacteriorhodopsin mutant L93A

During the extended lifetime of the O-state in bacteriorhodopsin (bR) mutant L93A, two substates have been distinguished. The first O-intermediate (OI) is in rapid equilibrium with N and apparently still has a 13-cis chromophore. OI undergoes a photoreaction with a small absorbance change, positive charge transport in the pumping direction, and proton release and uptake. None of these effects was detected after photoexcitation of the late O (OII). The most likely interpretation of the effects seen is an accelerated return of the molecule from the OI- to the bR-state. However, with a lifetime approximately 140 ms, the reaction cannot account for the observed high pumping efficiency of the mutant under continuous illumination. We suggest that OII corresponds to the O-intermediate with a twisted all-trans chromophore seen in the photocycle of wild-type bR, where the 13-cis OI-intermediate under the usual conditions does not accumulate in easily detectable amounts and, therefore, has generally been overlooked. Both the OI- and OII-decays are apparently strongly inhibited in the mutant.     

Photoproducts of bacteriorhodopsin mutants: a molecular dynamics study.

Molecular dynamics simulations of wild-type bacteriorhodopsin (bR) and of its D85N, D85T, D212N, and Y57F mutants have been carried out to investigate possible differences in the photoproducts of these proteins. For each mutant, a series of 50 molecular dynamics simulations of the photoisomerization and subsequent relaxation process were completed. The photoproducts can be classified into four distinct classes: 1) 13-cis retinal, with the retinal N-H+ bond oriented toward Asp-96; 2) 13-cis retinal, with the N-H+ oriented toward Asp-85 and hydrogen-bonded to a water molecule; 3) 13,14-di-cis retinal; 4) all-trans retinal. Simulations of wild-type bR and of its Y57F mutant resulted mainly in class 1 and class 2 products; simulations of D85N, D85T, and D212N mutants resulted almost entirely in class 1 products. The results support the suggestion that only class 2 products initiate a functional pump cycle. The formation of class 1 products for the D85N, D85T, and D212N mutants can explain the reversal of proton pumping under illumination by blue and yellow light.     

Buffer effects on electric signals of light-excited bacteriorhodopsin mutants

The effects of glycyl-glycine and bis-trispropane buffers on the light-excited electric signals due to proton motion in the molecule were studied for the bacteriorhodopsin (bR) mutants D38R, D96N, E204Q, R227Q, D85N, D85T, R82Q/D85N, and D85N/D96N in purple membranes and for delipidated purple membrane containing the wild-type bR. The results show additional charge motion caused by the buffers in all cases. Arrhenius parameters calculated from the temperature dependence of the difference signals (with buffer minus without buffer) are similar to the parameters found for the wild-type bR in the case of these buffers: the values of the activation enthalpies are mostly in the range 25-50 kJ/mol; all the activation entropies are negative. The results are evaluated with the cluster hypothesis outlined previously.     

Pathways of proton release in the bacteriorhodopsin photocycle.

The pH dependencies of the rate constants in the photocycles of recombinant D96N and D115N/D96N bacteriorhodopsins were determined from time-resolved difference spectra between 70 ns and 420 ms after photoexcitation. The results were consistent with the model suggested earlier for proteins containing D96N substitution: BR hv----K----L----M1----M2----BR. Only the M2----M1 back-reaction was pH-dependent: its rate increased with increasing [H+] between pH 5 and 8. We conclude from quantitative analysis of this pH dependency that its reverse, the M1----M2 reaction, is linked to the release of a proton from a group with a pKa = 5.8. This suggests a model for wild-type bacteriorhodopsin in which at pH greater than 5.8 the transported proton is released on the extracellular side from this as yet unknown group and on the 100-microseconds time scale, but at pH less than 5.8, the proton release occurs from another residue and later in the photocycle most likely directly from D85 during the O----BR reaction. We postulate, on the other hand, that proton uptake on the cytoplasmic side will be by D96 and during the N----O reaction regardless of pH. The proton kinetics as measured with indicator dyes confirmed the unique prediction of this model: at pH greater than 6, proton release preceded proton uptake, but at pH less than 6, the release was delayed until after the uptake. The results indicated further that the overall M1----M2 reaction includes a second kinetic step in addition to proton release; this is probably the earlier postulated extracellular-to-cytoplasmic reorientation switch in the proton pump.     

The proton release group of bacteriorhodopsin controls the rate of the final step of its photocycle at low pH

The factors determining the pH dependence of the formation and decay of the O photointermediate of the bacteriorhodopsin (bR) photocycle were investigated in the wild-type (WT) pigment and in the mutants of Glu-194 and Glu-204, key residues of the proton release group (PRG) in bR. We have found that in the WT the rate constant of O --> bR transition decreases 30-fold upon decreasing the pH from 6 to 3 with a pKa of about 4.3. D2O slows the rise and decay of the O intermediate in the WT at pH 3.5 by a factor of 5.5. We suggest that the rate of the O --> bR transition (which reflects the rate of deprotonation of the primary proton acceptor Asp-85) at low pH is controlled by the deprotonation of the PRG. To test this hypothesis, we studied the E194D mutant. We show that the pKa of the PRG in the ground state of the E194D mutant, when Asp-85 is protonated, is increased by 1.2 pK units compared to that of the WT. We found a similar increase in the pKa of the rate constant of the O --> bR transition in E194D. This provides further evidence that the rate of the O --> bR transition is controlled by the PRG. In a further test, the E194Q mutation, which disables the PRG and slows proton release, almost completely eliminates the pH dependence of O decay at pHs below 6. A second phenomenon we investigated was that in the WT at neutral and alkaline pH the fraction of the O intermediate decreases with pKa 7.5. A similar pH dependence is observed in the mutants in which the PRG is disabled, E194Q and E204Q, suggesting that the decrease in the fraction of the O intermediate with pKa ca. 7.5 is not controlled by the PRG. We propose that the group with pKa 7.5 is Asp-96. The slowing of the reprotonation of Asp-96 at high pH is the cause of the decrease in the rate of the N --> O transition, leading to the decrease in the fraction of O.     

Fourier transform infrared evidence for Schiff base alteration in the first step of the bacteriorhodopsin photocycle

The first step of the bacteriorhodopsin (bR) photocycle involves the formation of a red-shifted product, K. Fourier transform infrared difference spectra of the bR570 to K630 transition at 81 K has been measured for bR containing different isotopic substitutions at the retinal Schiff base. In the case of bacteriorhodopsin containing a deuterium substitution at the Schiff base nitrogen, carbon 15, or both, we find spectral changes in the 1600-1610- and 1570-1580-cm-1 region consistent with the hypothesis that the K630 C=N stretching mode of a protonated Schiff base is located near 1609 cm-1. A similar set of Schiff base deuterium substitutions for retinal containing a 13C at the carbon 10 position strongly supports this conclusion. This assignment of the K630 C=N stretching vibration provides evidence that the bR Schiff base proton undergoes a substantial environmental change most likely due to separation from a counterion. In addition, a correlation is found between the C=N stretching frequency and the maximum wavelength of visible absorption, suggesting that movement of a counterion relative to the Schiff base proton is the main source of absorption changes in the early stages of the photocycle. Such a movement is a key prediction of several models of proton transport and energy transduction. Evidence is also presented that one or more COOH groups are involved in the formation of the K intermediate.     

Photovoltage kinetics of the acid-blue and acid-purple forms of bacteriorhodopsin: evidence for no net charge transfer.

Time-resolved photovoltage measurements were performed with the acid-blue (bR605A) and acid-purple (bR565A) forms of bacteriorhodopsin (bR) in the time range from 25 ns to 100 s. The bR605A and bR565A pigments were formed by titration with H2SO4 in the absence and presence of 150 mM KCI, respectively. Qualitatively the kinetics of the charge displacement in these two states are similar and consist of two fast phases in one direction (100 ns bandwidth limited and approximately 1 microsecond) followed by a decay in the opposite direction via one component for bR605A (4.4 +/- 0.6 ms) or two components for bR565A (33 +/- 8 microseconds and 3.6 +/- 0.5 ms). The transient photovoltage signal returns exactly to the initial value after several milliseconds, well before the passive discharge of the electrical measuring system at 2 s. We conclude that no net charge transfer occurs in either bR605A or bR565A. The direction of the fast components is opposite that of net proton translocation in bR at pH 7. So, if the charge that moves back and forth is due to a proton, it moves first in the direction of the cytoplasmic side of the membrane (< 1 microsecond) and returns to its initial position via the 4.4 ms (bR605A) or the 33 microseconds and 3.6 ms (bR565A) decay components. The amplitude of the charge motion in both low pH forms is too large to be due to isomerization alone and is comparable to one of the major components in bR at pH 7.2     

Late events in the photocycle of bacteriorhodopsin mutant L93A

In the photocycle of bacteriorhodopsin (bR) from Halobacterium salinarum mutant L93A, the O-intermediate accumulates and the cycling time is increased approximately 200 times. Nevertheless, under continuous illumination, the protein pumps protons at near wild-type rates. We excited the mutant L93A in purple membrane with single or triple laser flashes and quasicontinuous illumination, (i.e., light for a few seconds) and recorded proton release and uptake, electric signals, and absorbance changes. We found long-living, correlated, kinetic components in all three measurements, which-with exception of the absorbance changes-had not been seen in earlier investigations. At room temperature, the O-intermediate decays to bR in two transitions with rate constants of 350 and 1800 ms. Proton uptake from the cytoplasmic surface continues with similar kinetics until the bR state is reestablished. An analysis of the data from quasicontinuous illumination and multiple flash excitation led to the conclusion that acceleration of the photocycle in continuous light is due to excitation of the N-component in the fast N<-->O equilibrium, which is established at the beginning of the severe cycle slowdown. This conclusion was confirmed by an action spectrum.     

Water is required for proton transfer from aspartate-96 to the bacteriorhodopsin Schiff base.

During the M in equilibrium with N----BR reaction sequence in the bacteriorhodopsin photocycle, proton is exchanged between D96 and the Schiff base, and D96 is reprotonated from the cytoplasmic surface. We probed these and the other photocycle reactions with osmotically active solutes and perturbants and found that the M in equilibrium with N reaction is specifically inhibited by withdrawing water from the protein. The N----BR reaction in the wild-type protein and the direct reprotonation of the Schiff base from the cytoplasmic surface in the site-specific mutant D96N are much less affected. Thus, it appears that water is required inside the protein for reactions where a proton is separated from a buried electronegative group, but not for those where the rate-limiting step is the capture of a proton at the protein surface. In the wild type, the largest part of the barrier to Schiff base reprotonation is the enthalpy of separating the proton from D96, which amounts to about 40 kJ/mol. We suggest that in spite of this D96 confers an overall kinetic advantage because when this residue becomes anionic in the N state its electric field near the cytoplasmic surface lowers the free energy barrier of the capture of a proton in the next step. In the D96N protein, the barrier to the M----BR reaction is 20 kJ/mol higher than what would be expected from the rates of the M----N and N----BR partial reactions in the wild type, presumably because this mechanism is not available.     

Arginine-82 regulates the pKa of the group responsible for the light-driven proton release in bacteriorhodopsin.

In wild-type bacteriorhodopsin light-induced proton release occurs before uptake at neutral pH. In contrast, in mutants in which R82 is replaced by a neutral residue (as in R82A and R82Q), only a small fraction of the protons is released before proton uptake at neutral pH; the major fraction is released after uptake. In R82Q the relative amounts of the two types of proton release, "early" (preceding proton uptake) and "late" (following proton uptake), are pH dependent. The main conclusions are that 1) R82 is not the normal light-driven proton release group; early proton release can be observed in the R82Q mutant at higher pH values, suggesting that the proton release group has not been eliminated. 2) R82 affects the pKa of the proton release group both in the unphotolyzed state of the pigment and during the photocycle. In the wild type (in 150 mM salt) the pKa of this group decreases from approximately 9.5 in the unphotolyzed pigment to approximately 5.8 in the M intermediate, leading to early proton release at neutral pH. In the R82 mutants the respective values of pKa of the proton release group in the unphotolyzed pigment and in M are approximately 8 and 7.5 in R82Q (in 1 M salt) and approximately 8 and 6.5 in R82K (in 150 mM KCl). Thus in R82Q the pKa of the proton release group does not decrease enough in the photocycle to allow early proton release from this group at neutral pH. 3) Early proton release in R82Q can be detected as a photocurrent signal that is kinetically distinct from those photocurrents that are due to proton movements from the Schiff base to D85 during M formation and from D96 to the Schiff base during the M-->N transition. 4) In R82Q, at neutral pH, proton uptake from the medium occurs during the formation of O. The proton is released during the O-->bacteriorhodopsin transition, probably from D85 because the normal proton release group cannot deprotonate at this pH. 5) The time constant of early proton release is increased from 85 microseconds in the wild type to 1 ms in R82Q (in 150 mM salt). This can be directly attributed to the increase in the pKa of the proton release group and also explains the uncoupling of proton release from M formation. 6) In the E204Q mutant only late proton release is observed at both neutral and alkaline pH, consistent with the idea that E204 is the proton release group. The proton release is concurrent with the O-->bacteriorhodopsin transition, as in R82Q at neutral pH.