Breaking the Carboxyl Rule

Background: Lysine rather than a carboxylic residue is in place of the internal proton donor in the E. sibiricum proton pump. Results: H+ uptake precedes reprotonation of the retinal Schiff base. K96A mutation slows it by >100-fold. Conclusion: Lysine 96 facilitates proton delivery to the Schiff base. Significance: This is the first example where lysine mediates proton transfer to the retinal Schiff base. A lysine instead of the usual carboxyl group is in place of the internal proton donor to the retinal Schiff base in the light-driven proton pump of Exiguobacterium sibiricum (ESR). The involvement of this lysine in proton transfer is indicated by the finding that its substitution with alanine or other residues slows reprotonation of the Schiff base (decay of the M intermediate) by more than 2 orders of magnitude. In these mutants, the rate constant of the M decay linearly decreases with a decrease in proton concentration, as expected if reprotonation is limited by the uptake of a proton from the bulk. In wild type ESR, M decay is biphasic, and the rate constants are nearly pH-independent between pH 6 and 9. Proton uptake occurs after M formation but before M decay, which is especially evident in D2O and at high pH. Proton uptake is biphasic; the amplitude of the fast phase decreases with a pKa of 8.5 ± 0.3, which reflects the pKa of the donor during proton uptake. Similarly, the fraction of the faster component of M decay decreases and the slower one increases, with a pKa of 8.1 ± 0.2. The data therefore suggest that the reprotonation of the Schiff base in ESR is preceded by transient protonation of an initially unprotonated donor, which is probably the ϵ-amino group of Lys-96 or a water molecule in its vicinity, and it facilitates proton delivery from the bulk to the reaction center of the protein.


A lysine instead of the usual carboxyl group is in place of the internal proton donor to the retinal Schiff base in the lightdriven proton pump of Exiguobacterium sibiricum (ESR). The involvement of this lysine in proton transfer is indicated by the finding that its substitution with alanine or other residues slows reprotonation of the Schiff base (decay of the M intermediate) by more than 2 orders of magnitude. In these mutants, the rate constant of the M decay linearly decreases with a decrease in proton concentration, as expected if reprotonation is limited by
the uptake of a proton from the bulk. In wild type ESR, M decay is biphasic, and the rate constants are nearly pH-independent between pH 6 and 9. Proton uptake occurs after M formation but before M decay, which is especially evident in D 2 O and at high pH. Proton uptake is biphasic; the amplitude of the fast phase decreases with a pK a of 8.5 ؎ 0.3, which reflects the pK a of the donor during proton uptake. Similarly, the fraction of the faster component of M decay decreases and the slower one increases, with a pK a of 8.1 ؎ 0.2. The data therefore suggest that the reprotonation of the Schiff base in ESR is preceded by transient protonation of an initially unprotonated donor, which is probably the ⑀-amino group of Lys-96 or a water molecule in its vicinity, and it facilitates proton delivery from the bulk to the reaction center of the protein.
ESR 4 is a retinal protein from Exiguobacterium sibiricum that functions as a light-driven proton pump (1, 2). E. sibiricum 255-15, a Gram-positive bacterium that carries the ESR gene, was isolated from a 43.6-m depth of a 2-3-million-year-old permafrost core (3,4). ESR shows homology to other proton pumps: archaeal bacteriorhodopsin (5-7), eubacterial proteorhodopsin (8), and xanthorhodopsin (9 -11). Its amino acid sequence (1) and a mutational study (2) indicate that, as in the other pumps, in ESR the key component of the counterion to the Schiff base and proton acceptor is the conserved Asp-85, homologous to Asp-85 of bacteriorhodopsin, which is involved in proton transfer from the Schiff base to the extracellular surface (12). As in xanthorhodopsin (13) and proteorhodopsin (14,15), a histidine residue (His-57) interacts closely with Asp-85 (2). An unusual and intriguing feature of ESR is that in the cytoplasmic domain, Lys-96 is in place of the internal proton donor, which is Asp-96 in bacteriorhodopsin (16 -20) and a glutamic acid in the eubacterial pumps, proteorhodopsins, xanthorhodopsin, and others (13,21). Lysine at the internal donor site is present also in retinal protein gene sequences from the thermophilic bacterium Exiguobacterium sp. AT1b isolated from a hot spring at Yellowstone National Park, and in a number of marine Euryarchaeota from Puget Sound (22), but the proteins have not yet been studied.
The ESR gene was expressed in Escherichia coli, and upon incubation with all-trans-retinal, the protein formed a pigment with an absorption maximum at 532 nm (1). Proteoliposomes containing ESR, and E. coli cells with ESR expressed, produce robust light-induced acidification of the bulk in a broad pH range from 4.5 to 9 (2), confirming capability for light-induced proton transport. Despite the absence of the usual carboxylic residue as internal proton donor, the turnover of the ESR pho-tocycle is comparable with those in proteorhodopsin (21) and xanthorhodopsin (9). The photocycle of ESR includes mostly red-shifted intermediate(s) but also formation of an M intermediate above neutral pH (2), indicating transient light-induced deprotonation of the Schiff base that in bacteriorhodopsin and proteorhodopsin is the crucial step in proton transport. The rate and yield of M formation were found to be pH-dependent, sensitive to the environment of the protein, and to substitution of His-57 (2). In the commonly used detergent DDM, the M intermediate accumulated only at high pH, with an apparent pK a of 9. In liposomes and the lipid-like detergent, lyso-PG, the M intermediate accumulated with a lower pK a at 6.5 (2). Lightinduced proton uptake from the bulk was followed by proton release in the photocycle of ESR (1, 2), a sequence that is the reverse of that in bacteriorhodopsin but common to eubacterial proton pumps, proteorhodopsin (21), xanthorhodopsin (13), and gloeobacter rhodopsin (23).
This work focuses on elucidating the role of Lys-96 in the reprotonation of the Schiff base during the photocycle. In bacteriorhodopsin and homologous proton pumps, formation of the M intermediate with a deprotonated Schiff base (24) is the first proton transfer step, when a proton is lost from the retinal Schiff base to Asp-85 in the extracellular domain (12,25). The decay of M is associated in turn with reprotonation of the Schiff base from the cytoplasmic surface (26); it can be followed spectroscopically by observing the reversal of the absorbance increase at 410 nm.
We draw upon earlier findings with bacteriorhodopsin, which showed that reprotonation of the Schiff base from the bulk is a two-step process and involves Asp-96 as an internal proton donor (17). Deprotonation of the initially protonated Asp-96 was detected by FTIR during the decay of the M intermediate (16,18,27), which proceeds in a pH-independent manner, but the subsequent reprotonation of the carboxylate during proton uptake from the bulk is pH-dependent (28,29) and is responsible for slowing the N to O transition and overall photocycle turnover at high pH (29 -31). These changes in protonation of the internal carboxyl residue are caused by a transient decrease of its pK a value during the photocycle from above 11 to ϳ 7.5 (28,30,32,33). Furthermore, earlier studies on bacteriorhodopsin (17,19,34), proteorhodopsin (21), and gloeobacter rhodopsin (23) had shown that replacement of the carboxylic residue acting as internal proton donor to the Schiff base with nonionizable residues greatly slows (typically by 2 orders of magnitude at neutral pH) reprotonation of the Schiff base, as reflected in deceleration of the decay of the M intermediate. In the most extensively studied D96N mutant of bacteriorhodopsin, the decay of M intermediate is strongly pH-dependent, linearly decelerating with the decrease of proton concentration (17). Lack of an internal proton donor could be compensated for by addition of sodium azide, which apparently shuttles protons from the bulk to the Schiff base (17,(35)(36)(37).
As with Asp-96 in bacteriorhodopsin, elucidation of the role of Lys-96 in ESR involved studies of mutants in which Lys-96 was substituted with other residues (K96A, K96Q, K96I, K96H, and K96R), examination of the pH dependence of M decay, and correlating it with the kinetics of proton uptake in the wild type protein. The data indicate that the presence of Lys-96 in wild type ESR greatly accelerates reprotonation of the Schiff base (by more than 100-fold). Measurements of the kinetics of lightinduced proton concentration changes in the bulk revealed that proton uptake precedes reprotonation of the Schiff base in wild type, especially evident in D 2 O and at high pH, which suggests that there is a protonatable site in the protein that gains a proton from the bulk after formation of M and passes it with a short delay to the Schiff base in the M to N transition. That reprotonation of the Schiff base is slowed manyfold by mutations of Lys-96 suggests that Lys-96, and possibly interacting water molecules, is the site that mediates proton transfer from the bulk to the Schiff base.

EXPERIMENTAL PROCEDURES
ESR was expressed in E. coli and purified as described previously (1), except lipid-like detergent 16:0 lyso-PG (Avanti Polar Lipids, Inc.) was used for solubilization instead of DDM. The important advantage of this detergent over DDM is that accumulation and decay of the M intermediate can be followed in a broader pH range (above 6), as when the protein is reconstituted in liposomes (2). The protein was solubilized in 0.5% lyso-PG, purified on a nickel column using 250 -300 mM imidazole for elution, and then dialyzed in two steps to lower the concentration of imidazole. Lyso-PG is less efficient in solubilizing ESR than DDM, so the alternative method used in experiments involved solubilization of the ESR with DDM and subsequent replacement of DDM with lyso-PG by washing and elution with buffer containing the latter from a nickel column. Both methods produced samples with similar properties. The samples typically contained 0.05% lyso-PG, 100 mM NaCl, and one or several buffers (citric acid, MES, MOPS, Bicine, CHES, and CAPS). Replacement of H 2 O with D 2 O (99% from Aldrich) was achieved using Microcon centrifugal filter devices (Ultracel YM-30). The effect of D 2 O on kinetic constants developed with time and reached a maximum after 3 days of incubation in D 2 O.
Spectral and kinetic measurements were performed as described recently (2). Absorption spectra were measured on a Shimadzu UV1700 spectrophotometer. The kinetics of laserinduced absorption changes in the 2-s to 100-s time domain were obtained using a single wavelength time-resolved system. The second harmonic of a 7-ns Nd-YAG laser at 532 nm was used for photoexcitation. The energy of flashes was less than 1 mJ/cm 2 , far below the threshold that causes irreversible bleaching of the protein. The fit of the kinetic traces was done using the FitExp program (38). Kinetics of light-induced proton uptake and release during the ESR photocycle was followed using the pH-sensitive dye pyranine in 0.05% lyso-PG in H 2 O at pH between 7.2 and 7.8 and in D 2 O at pD 7.6 -7.8. For the higher pH range (8 -9.5), phenol red and thymol blue were used. The pK a values of these dyes were measured to be 7.9 and 9.6 in 0.05% lyso-PG, respectively. Unlike pyranine, the latter two dyes absorb at the wavelength of photoexcitation, 532 nm, and therefore correction for attenuation of the laser intensity by the dye was necessary. It was done by calculating the fraction of light absorbed by ESR at 532 nm before and after addition of the dye, or by determining the attenuation factor from the decrease of laser-induced absorption changes after addition of the dye at a wavelength where dye does not undergo pH-induced changes (480 nm for phenol red and 500 nm for thymol blue). Typically, the attenuation was 5-10%. The time course of light-induced proton changes was obtained as a difference between the traces at 560 nm for phenol red and 595 nm for thymol blue from a sample with the dye added minus that without the dye. Alternatively, the difference between a trace with a dye minus a trace with 4 -5 mM buffer added at the same pH was used. The advantage of the latter method was that it did not require correction for attenuation, only a small correction for dilution, which was 0.6% (5-6 l of 0.5 M CHES or 1 M HEPES to 900 l sample was added). Cuvettes with a 4-mm pathway for laser excitation and 10 mm for measuring beams were used. Typical concentration of ESR was 10 M; the fraction cycling was 10% (i.e. 1 M of the chromophore), and the concentration of the pH-sensitive dyes was 20 -50 M. All spectroscopy was at 23°C.
Light-induced proton pumping by the K96A mutant in suspensions of E. coli cells and proteoliposomes was examined in same way as described previously for wild type ESR (2).

RESULTS
In the sections below, we first demonstrate that replacing Lys-96 with alanine and other residues dramatically slows reprotonation of the Schiff base and that many features of the K96A mutant are similar to the D96N mutant of bacteriorhodopsin. Evidence is then presented for a two-step mechanism for proton transfer from the bulk to the retinal Schiff base that involves an internal proton donor, which is the side chain of Lys-96 and/or interacting water molecules.
Effect of Substituting Lys-96 with Alanine and Other Residues on Schiff Base Reprotonation- Fig. 1 depicts the kinetics of light-induced absorption changes in the wild type ESR and the K96A mutant at pH 7.4 at three characteristic wavelengths, 410, 510 and 590 nm (Fig. 1, A-C, respectively). The light-induced absorption increase at 410 nm is from accumulation of the M intermediate upon Schiff base deprotonation and transfer of a proton to the counterion and proton acceptor Asp-85. M rise is in the 2-s to 2-ms time domain, similar in the mutant and the wild type, except that in the wild type the component with a slow rise ( ϭ ϳ1.6 ms) is apparently attenuated by the fast M decay. The dramatic difference between the mutant and the wild type is in the rate of M decay, i.e. the reprotonation of the Schiff base. In the wild type, it is biphasic and occurs with time constants of ϳ0.8 -1 ms (80%) and 10 ms (20%). The slower rate constant varies slightly in different preparations in the range of 9 -15 ms. In the mutant, 90% of the M intermediate decays with Ϸ500 ms, which is more than 2 orders of magnitude slower than in the wild type. Thus, in the mutant, M decay becomes the rate-limiting step in the photocycle, so the recov-ery of the initial state, followed at 510 nm, is nearly an order of magnitude slower than in the wild type (500 ms versus 60 ms). Another distinctive difference is that the red-shifted intermediates that are formed after reprotonation of the Schiff base in the wild type (see the rise and decay of the absorbance at 590 nm in 1-100-ms time domain) do not accumulate in the K96A mutant, evidently because the decay of these intermediates, at 10 and 60 ms, respectively (see below), is much more rapid than the rate of their formation. Thus, Lys-96 facilitates reprotonation of the Schiff base in ESR by about 500-fold (500 ms versus 1 ms) at neutral pH and speeds up the overall turnover of the cycle by an order of magnitude. Lysine at location 96 is unique in this regard; a manyfold decrease in the rate of reprotonation of the Schiff base occurs upon substitution of Lys-96 with other residues. Comparative data for the K96A, K96I, K96Q, K96H, and K96R are given in the Table 1. The rate of M decay depends only to a small degree on the particular residue in place of Lys-96. It is 2-fold slower in K96I but 2-fold faster in K96Q, as compared with K96A. Placing of a histidine, which potentially could undergo protonation change, only modestly accelerated M decay compared with K96A at pH 7 and 8, although more so at pH 9. However, M decay was still almost 2 orders of magnitude slower than in the wild type. Arginine was even less effective. These results indicate that the rapid M decay that occurs in wild type ESR is not observed with other residues in place of Lys-96. Attempts to replace Lys-96 with an aspartic acid did not produce an expressed protein.
pH Dependence of the Rate of M Decay in the K96A Mutant-If the protons for reprotonation of the Schiff base in the K96A mutant have to be captured directly from the bulk, one would expect that reprotonation of the Schiff base would become slower at increasing pH values, as in the D96N mutant of bacteriorhodopsin. As shown in Fig. 2A, this is indeed the case. The decay of the M intermediate and the corresponding recovery of the initial state slowed by more than an order of magnitude upon increasing the pH from 6.8 to 8.5. The time constant of M decay was inversely proportional to the concentration of protons (Fig. 2B). The slope was 0.65. In the bacteriorhodopsin mutant D96N, it is 0.75 (17). A slope less than one might originate from the influence of ionizable groups on the protein (19). Similar pH dependences were observed for the K96A, K96Q, and K96I mutants solubilized in DDM (Table 1) and K96A reconstituted into liposomes (59).
Proton Transport by the K96A Mutant in E. coli Cells and Liposomes-To evaluate the effect of K96A mutation on the overall proton pumping capability of ESR, we examined lightinduced pH changes in suspensions of live E. coli cells carrying the K96A mutant and liposomes reconstituted with K96A in comparison with wild type. As shown in Fig. 2C for a cell suspension, the mutant was capable of proton pumping, but lightinduced pH changes produced by the mutant were smaller than those produced by the wild type. The ratio of light-induced pH changes in the mutant versus that in wild type was smaller at higher pH where the slower recovery of the initial state in the mutant (Fig. 1B) resulted in lower cycling turnover and proton transport rate. Similar observations were made on proteoliposomes (not shown).
Effect of Sodium Azide on Reprotonation of the Schiff Base in the K96A Mutant-In the D96N mutant of bacteriorhodopsin, rapid reprotonation of the Schiff base can be recovered by addition of sodium azide (17,19,37), the protonated form of which apparently serves as a mobile proton donor to the Schiff base. At pH 6 in the presence of 2 mM NaN 3 , the Schiff base was reprotonated in the D96N mutant at a rate comparable with that in wild type. We also found that in the K96A mutant of ESR, sodium azide dramatically accelerates the rate of reprotonation of the Schiff base. At pH 7.1, 20 mM NaN 3 accelerates M decay from 350 to 16 ms (Fig. 3A, traces 1 and 2). Faster decay of M results in accumulation of the red-shifted intermediate (Fig.  3B) and faster recovery of the initial pigment. As expected, the rise of the red-shifted states is accelerated to the same degree as the M decay, by about 120-fold at 800 mM azide (Fig. 3B). Even at high concentrations of azide, the rate of the Schiff base reprotonation in K96A was not as rapid as in wild type (Fig. 3C). Unlike in K96A, sodium azide does not produce a large effect on M decay in the wild type. The fast (0.9 ms) component is accelerated only 2-fold at 800 mM sodium azide (Fig. 3C), whereas the slow component accelerated ϳ5-fold. As in bacteriorhodopsin, the effect of sodium azide on M decay in K96A is pH-dependent. NaN 3 becomes more effective in accelerating M decay as the pH is lowered, which is consistent with the protonated form of azide being the source of a proton for the Schiff base (37). To a lesser extent, buffers and especially imidazole are also capable of accelerating M decay in the K96A mutant.
pH Dependence of the Photocycle Steps in Wild Type ESR-Insights to how Lys-96 is involved in proton transfer from the bulk to the Schiff base were gained from examination of the pH dependence of M decay and proton uptake in the wild type photocycle. First, we describe the complex Schiff base reprotonation kinetics. In lyso-PG micelles, between pH 6 and 7, the amount of the M intermediate observed increases with an apparent pK a of ϳ 6.3. M rise is nearly pH-independent, but M decay exhibits three characteristic patterns in three pH ranges, between pH 6 and 7, between 7 and 9, and above 9, as described in the following.
Between pH 6 and 7, the amount of the M intermediate increases but the rate constants of M decay and the rest of the photocycle do not change substantially (Fig. 4A, traces 1-4). The global fit of absorption changes at four wavelengths (410, 510, 550, and 590 nm) indicates that at pH 7.1 (Fig. 5A) the rise of M can be described satisfactory with a sum of three components, with 1 ϭ 3.5 s, 2 ϭ 9.5 s, and 3 ϭ 101 s. The decay of the M intermediate includes two components, with time constants of 4 ϭ 0.8 -1.0 ms and 5 ϭ 9 -15 ms. The fraction of the fast components is 80% and is pH-independent up to pH 7 (Fig. 4, A and D).
Global fit of kinetics at wavelengths between 390 and 700 nm at pH 7.1 shows that the decay of M to the first red-shifted species (N 1 ), with a time constant of 4 ϭ 0.8 ms, is accompanied by an absorbance increase in a broad spectral range from 455 to 700 nm with a maximum at 550 nm (Fig. 5B, spectrum 1). The subsequent rise of absorption at 590 nm, with a time constant 5 ϭ 10 ms, is assigned to the transformation of N 1 to the next intermediate, O (or N 2 ). We prefer to use the term N instead of O, because FTIR data indicate that in this state the configuration of the retinal chromophore is 13-cis (59).The difference spectrum of the N 1 to N 2 transition obtained as a plot of the absorption changes accompanying the 10-ms component, which also involves a conversion of a minor portion of M, is shown in Fig. 5B (spectrum 2). The N 2 intermediate decays to the initial ESR with 6 ϭ 58 ms (spectrum 3 in Fig. 5B).
In the second pH region, between pH 7 and 9, the decay of M becomes pH-dependent. It occurs in such a way that the time constants of the two kinetic phases remain unchanged, but the fraction of the faster component decreases from 80% to near 0, with a pK a of 8.1 Ϯ 0.2, whereas the fraction of the slower, ϳ10 ms, component correspondingly increases (Fig. 4, A, traces  D, curves 1 and 2). This suggests that the pH-dependent transition between fast and slow decaying M is governed by the protonation equilibrium of some residue(s), affecting the fraction but not the apparent rates of reprotonation of the Schiff base. For example, this could be a proton donor that facilitates rapid (1 ms) reprotonation of the Schiff base, and it could become increasingly unavailable as the pH is raised. Thus, an increasing fraction of the Schiff base would be reprotonated more slowly. The rate of Schiff base reprotonation through this slower reaction is not dependent on the bulk concentration of protons, and thus the rate of proton capture, but is limited by the rate of a pH-independent reaction, presumably the decay of N 1 to N 2 , which would shift the Schiff base reprotonation to completion.

-7, and
Above pH 9, the decay of M gradually slows from 9 to 35 ms at pH 10.3, and even slower components (60 and 260 ms) appear (Fig. 4, B and C). This decrease in the rate of M decay causes a decrease in the accumulation of N 1 (Fig. 5C), as seen in the difference spectrum for M decay at pH 9.3, which shows that the N 2 state with absorption maximum at 595 nm is the main product of the reaction. The slowing of M decay linearly decreases with the decrease of proton concentration (the log of time constant linearly increases versus pH, Fig. 4C). Thus, the decay of M at pH above 9 is limited by capturing protons from the bulk (i.e. by proton uptake).
Light-induced Transient Proton Uptake and Release in Wild Type ESR-It was established earlier that in the photocycle of ESR proton uptake occurs before proton release (1, 2). In analogy with bacteriorhodopsin, because ESR is a proton pump in the cytoplasmic to extracellular direction, the proton uptake will be on the cytoplasmic side. Therefore, an important question is whether proton uptake precedes, coincides, or follows reprotonation of the Schiff base. The time course of proton uptake and release was determined from changes in the absorption of the pH-sensitive dye pyranine after photoexcitation at pH 7.2 as shown in Fig. 6A (trace 1). An increase in absorbance at 460 nm corresponds to proton uptake. Uptake, with a time constant of ϳ0.9 ms (0.86 Ϯ 0.05 ms), is nearly the same (or slightly more rapid) as the major component in the decay of the M intermediate (Fig. 6, trace 2) and thus in the rise of N 1 (trace 3), at ϳ1 ms (0.96 Ϯ 0.04 ms). Proton release occurs with time constant 64 Ϯ 3 ms and correlates with the decay of the redshifted intermediate and the recovery of the initial state. The data show therefore that the proton is taken up during the M to N 1 transition. This is different from bacteriorhodopsin where proton uptake occurs during the next reaction, the N to O transition with a time constant 2-3-fold slower than reprotonation of the Schiff base during the M to N transition (39 -43).
The very small difference in the rates of proton uptake and M decay in ESR makes it difficult to decide the exact sequence of events that lead to Schiff base reprotonation. The slightly faster proton uptake seen in experiments pointed to the possibility that uptake from the bulk precedes reprotonation of the Schiff base. More evidence in favor of this was obtained in experiments with replacement of H 2 O to D 2 O.

D 2 O Kinetic Isotope Effect on the Photocycle, Kinetic Separation of Proton Uptake and Protonation of the Schiff Base-Pre-
vious studies on bacteriorhodopsin showed that different steps of the photocycle are slowed to different degrees in D 2 O. The deuterium kinetic isotope effect, which is characterized by the ratio of the time constant in D 2 O to that in H 2 O, was found to vary between 1.1 and 13.1 for different photocycle transitions in bacteriorhodopsin and its mutants (44) . Fig. 6, B and C, shows that the kinetics of M rise and M decay are both slowed in D 2 O. The time constants obtained from global fit of absorption changes at four characteristic wavelengths in a sample in D 2 O at pD 7.6 are given in Table 2. The first three time constants describe the kinetics of M rise, and the next two describe its decay, and the last constant characterizes recovery of the initial state. In D 2 O, the major component of M decay ( 4 ) slowed 3-fold, from 1.0 Ϯ 0.07 ms in H 2 O to 3.0 Ϯ 0.03 ms. Remarkably, the kinetics of proton uptake was affected by D 2 O to a lesser degree. It slowed only 2.3-fold, from 860 Ϯ 50 s to 2.0 Ϯ 0.1 ms (Fig. 6B). The smaller deuterium effect on proton uptake than on the M decay is evident from comparison of the kinetic traces in Fig. 6C where the shift between traces 1 and 2 is greater than between 3 and 4. In D 2 O, the major component of proton uptake ( ϭ 2 ms) clearly precedes the faster component of M decay (reprotonation of the Schiff base) and formation of the intermediate N 1 ( ϭ 3 ms), as shown in Fig. 6B (traces 1 and 3). This critical result (a delay between proton uptake and M decay) indicates that delivery of a proton from the bulk to the Schiff base occurs in two distinguishable steps. First, a proton is taken up by a group in the protein. Second, after a small delay, it is transferred to the Schiff base. It is reasonable to suggest that the group that is protonated during proton uptake fulfills the criteria for a proton donor to the Schiff base. It is unprotonated in the initial state and acquires proton only in M. Once protonated, however, it becomes the internal source of the proton that is passed to the Schiff base.
pH Dependence of Proton Uptake, Estimation of the pK a of the Donor-Further insight into the mechanism of proton uptake and the function of a donor was obtained from the proton uptake kinetics at high pH, using phenol red for pH 8 and thymol blue for pH 9 -9.5 as pH-sensitive dyes. Fig. 6D shows proton uptake and release traces at pH 7.2, 8.0, and pH 9.1 as measured with pyranine, phenol red, and thymol blue, respectively. The kinetics at pH 8.0 is similar to that at pH 7.2 (Fig. 6D, traces  1 and 2), although the uptake is somewhat faster (760 Ϯ 40 s versus 860 Ϯ 50 s). At pH 9.1 (Fig. 6D, trace 3), the uptake is clearly biphasic; the maximum in the dye absorption change from proton uptake is shifted from 4 ms at pH 7.2 to ϳ20 ms. The initial fast phase of uptake occurs with a time constant in the range of 650 Ϯ 150 s, which is slightly faster than at pH 7.2. These data show that the time constant of the faster proton uptake component does not become slower at high pH, but its fraction decreases, similar to the fraction of the faster M decay component. The lack of slowing of the faster rate constant at pH 9.1 suggests that the rate of proton uptake is limited not by diffusion of a proton from the bulk to a group accepting a proton, but by some conformational change resulting in either a change of the pK a value of the group or a change of access that allows its protonation.
The slower component of uptake is clearly seen at pH 9.1. The fit indicates that its amplitude is attenuated by the proton release that follows (presumably at the extracellular surface), and it accounts for about 77% of the proton uptake at this pH. The time constant of the slower component is in the range of 15-18 ms. If one assumes that the fraction of the fast component in proton uptake is determined by the pK a of the group receiving a proton at this time of the cycle (during the M to N transition), one can estimate from the magnitude of this fraction (ϳ23% at pH 9.1, see Fig. 6D) that the pK a should be 8.5 Ϯ

DISCUSSION
Our observations strongly suggest that Lys-96 is involved in the reprotonation of the Schiff base in the photocycle of ESR.
The key evidence is the dramatic, more than 100-fold, decrease in the rate of M decay caused by the replacements of Lys-96 with any of a variety of residues ( Fig. 1 and Table 1), the pH independence of the rate constants of M decay in wild type ESR at pH below 9 (Fig. 4C) but not in the K96A mutant (Fig. 2B), and most importantly, the kinetics of proton uptake in the wild type that indicates that protonation of a protein group precedes reprotonation of the Schiff base (Table 2 and Fig. 6). ESR is the first example where a lysine, rather than a carboxylic acid, i.e. Asp-96 in bacteriorhodopsin and a glutamic acid in the numerous proteorhodopsins and xanthorhodopsins, plays such a role in a light-driven proton pump.
The data we present here decide among three alternative scenarios for how reprotonation of the retinal Schiff base might function in ESR as follows. 1) H ϩ uptake and protonation of an internal donor occurs first, followed by its deprotonation through transfer of a proton to the Schiff base (proton uptake precedes the Schiff base reprotonation). 2) In the opposite sequence of events, a donor is initially protonated, gives up its proton to the Schiff base, and then is reprotonated from the bulk (proton uptake coincides with, or follows reprotonation of the Schiff base as in bacteriorhodopsin). 3) There is no internal proton donor. The proton transfer to the Schiff base from the bulk occurs in a single continuous chain of events through a hydrogen bonding network (proton uptake and Schiff base reprotonation coincide under all conditions). The kinetics of  1 and 2, respectively) and on the kinetics of pyranine absorption changes (traces 3 and 4, respectively). The kinetic components are shown in the Table 2. Traces in D 2 O are shown as dotted lines. D, kinetics of proton uptake and release at pH 7.2, 8.0, and 9.1, and the kinetics of the Schiff base reprotonation followed at 590 nm in H 2 O: trace 1, 460 nm, pyranine at pH 7.2; trace 2, 560 nm, phenol red at pH 8.0; trace 3, 595 nm, thymol blue at pH 9.1; trace 4, 590 nm at pH 9.1 in the absence of thymol blue. Traces are normalized. Actual amplitudes were 2.6, 3.6, 3.5, and 26 mOD for the traces 1-4, respectively. The fit of the trace 3 with three components yielded the following time constants: 1uptake ϭ 650 s, 23% for fast proton uptake; 2uptake ϭ 18 ms, 77% for the second phase of uptake, and release ϭ 45 ms for proton release. proton uptake are inconsistent with the second and third alternatives but support the first alternative involving transient protonation of a donor (Lys-96 or an aqueous network in its vicinity) before reprotonation of the Schiff base, as discussed below. The finding that proton uptake precedes reprotonation of the Schiff base (Fig. 6) indicates that a protein group gains a proton after formation of M but before reprotonation of the Schiff base occurs. In H 2 O, both processes occur within ϳ1 ms after photoexcitation. In D 2 O, a distinct kinetic separation of uptake and Schiff base reprotonation becomes noticeable. The difference in the time constants of the two events ( Fig. 6 and Table 2) is direct evidence for the existence of a group that receives a proton (deuteron) from the bulk and then donates it to the Schiff base, serving as an internal proton donor in the M to N transition.
A mechanistically important observation is that the fast time constant for proton uptake (0.6 -0.9 ms) in the wild type does not become slower with increasing pH, measured up to pH 9.1 (Fig. 6D). It indicates that the limiting step for proton uptake is not the capture of a proton from the bulk but a pH-independent process, e.g. a conformational change that enables protonation of the donor group. This would explain the pH independence of the fast component of M decay in the wild type. A decrease in the fraction of this component at pH above 7.5 could be caused by the smaller fraction of protonated donor at pH near and above the pK a of the donor during proton uptake (see below).
Role of Lys-96 in Reprotonation of the Schiff Base-Fast reprotonation of the Schiff base seen in the wild type is abolished by mutations of Lys-96 (Table 1). This indicates that Lys-96 is critically involved in delivery of a proton to the Schiff base. There are two alternatives. First, the ⑀-NH 2 of Lys-96 is the donor, initially unprotonated but becoming transiently protonated after M is formed. Following transfer of its proton to the Schiff base, its initial unprotonated state is restored. Second, the protonation state of Lys-96 does not change, but the lysine side chain supports a hydrogen-bonded network that can contain an additional proton in a similar way as the extracellular proton release network (45) is supported by Glu-194 and Glu-204 in bacteriorhodopsin (42,43).
The involvement of Lys-96 in proton transfer thus raises the crucial question whether Lys-96 could be reversibly protonated during the proton exchange with the bulk and with the Schiff base. In its hydrophobic environment, Asp-96 in the initial state of bacteriorhodopsin has a pK a of Ͼ11.5 (28,46), elevated from its value of ϳ 4 in water. This extremely high pK a is from the environment of Asp-96, the hydrophobic "barrel" that surrounds it (47). It favors a neutral, nonionized-state. Many residues that constitute the barrel in bacteriorhodopsin are conserved in ESR and will provide a similar environment for the Lys-96. For this reason, the Lys-96 is expected to be neutral (i.e. unprotonated) in the initial state, and its pK a value could be much lower than in water. The pK a values of buried lysine residues in proteins were found to be very sensitive to their environment. The pK a value can be as low as 5.3 and also as high as 10.4, near its value in water (48,49). In the latter case, the pK a of 10.4 was explained by a hydration of Lys by a single water molecule (50). Thus, in the initial state of ESR, buried Lys-96 could well be unprotonated above pH 6, where M accumulates and its decay can be investigated. Transition to the M intermediate causes conformational changes in the cytoplasmic region of bacteriorhodopsin, associated with the appearance of additional water molecules in the cytoplasmic domain (51,52), which are required for proton transfer to the Schiff base (53). Hydrogen bonding with mobile water molecules constitutes a possible way for modulating the pK a of buried Lys-96 during the ESR photocycle also. What should be the pK a value of Lys-96 in the M state if it is to be the proton donor? The proton kinetics at high pH indicates that the fraction of fast proton uptake decreases with a pK a of ϳ8.5, which is likely to be the pK a value of the donor in the M to N 1 transition. Such a rise in the pK a for the ⑀-NH 2 of Lys-96 during the photocycle is reasonable and would correspond to the observed drop of ϳ4 pH units in the pK a of Asp-96 in bacteriorhodopsin at a time when proton uptake takes place in the N to O transition (28 -30).
What determines the two distinct time constants of ϳ1 and 10 -15 ms seen in the kinetics of proton uptake and M decay? A likely possibility is that fast proton uptake is induced by a conformational change following M formation that causes "opening" of a channel from the bulk to the donor. The degree of protonation of the donor is determined by its pK a . The degree of protonation of the Schiff base, in turn, would depend on the fraction of the protonated donor and the relative proton affinities of the Schiff base and the donor; these factors would determine fractions of M and N in the M Ͻ ϭ Ͼ N 1 equilibrium. After transition of N 1 to N 2 , more M can be transformed to N 1 .
A Model for Proton Transfers in the ESR Photocycle-The connection of proton uptake and release with reactions and intermediates of the photocycle can be summarized by the scheme in Fig. 7 in which the steps of proton transfer in ESR and BR are compared. Deprotonation of the Schiff base in the L to M 1 step is followed by the M 1 to M 2 transition during which group D, presumably Lys-96 and interacting water molecule(s), accepts a proton from the bulk. This is likely to involve an opening of the cytoplasmic "half-channel" that provides a pass for a proton from the bulk to the donor and causes an increase in the pK a of the donor so that it accepts a proton at pH as high as 8.5. The M 2 state is short lived because the proton gained is quickly transferred to the Schiff base in the M 2 to N 1 transition that occurs with an ϳ1-ms time constant. The M 2 Ͻ ϭ Ͼ N 1 transition is apparently a reversible reaction. The pH-independent time constant of the fast ϳ1-ms component of M decay is explained by the rate of internal proton exchange between the donor and the Schiff base in the M 2 Ͻ ϭ Ͼ N 1 transition. The decrease in the fraction of Schiff base that reprotonates with this fast phase with increasing pH (pK a 8.1 Ϯ 0.2) is apparently caused either by the decreased fraction of donor molecules that are protonated at pH above 8, or, if transient equilibrium is established between the Schiff base and the bulk, by the pK a value of the Schiff base itself (8.1 in N 1 ). In the latter case, the Schiff base pK a becomes the limiting factor that determines the fraction of M that can undergo rapid transition to N 1 . The estimate based on the amplitude of the fast component of proton uptake yields a value of 8.5 Ϯ 0.3 for the pK a of the donor. The transition to the N 2 state presumably results in elevating the pK a of the Schiff base by at least 1 unit, which then leads to almost complete M decay with a 10 -15-ms time constant at a pH between 8 and 9. At a still higher pH values, the decreasing fraction of protonated donor becomes rate-limiting, and it explains the pH-dependent slowing of the photocycle at pH Ͼ 9. Proton release occurs at the end of the photocycle presumably when the chromophore is reisomerized to all-trans, Asp-85 deprotonates, and its salt bridge with protonated Schiff base is re-established. Proton release is likely to occur through a hydrogen-bonded pathway, yet unidentified. Fig. 7 reveals several similarities but also profound differences. The similarity is that both Asp-96 and Lys-96 appear to be neutral in the ground state and undergo transient change of ionization during the photocycle. Asp-96 is initially protonated to very high pH values of Ͼ11.4 (28,54), but Lys-96 is most likely initially unprotonated, with a pK a value possibly below 6. This would explain the key differences in the sequence of events after photoexcitation. In bacteriorhodopsin, protonation of the Schiff base by Asp-96 during the M to N transition is distinctively more rapid than the proton uptake (39,42,46,55,56) leading to reprotonation of Asp-96. Asp-96 appears to be inaccessible to the bulk at the time of the proton exchange between Asp-96 and the Schiff base (33), because it does not lose the proton to the bulk up to very high pH values. It becomes more accessible during proton uptake, however, and its apparent pK a decreases to pK a ϳ7.5 (28 -30). In contrast, in ESR, protonation of Lys-96 must occur before it can act as the donor. Its pK a was initially low, but during the photocycle, it appears to rise to 8.5 Ϯ 0.3. Above pH 9, Schiff base reprotonation becomes pH-dependent (Fig. 4C) indicating that a decreasing fraction of the donor is protonated as the pH is increased.

Comparison of the Mechanism of the Schiff Base Reprotonation in ESR with That in Bacteriorhodopsin-
At pH Ͼ 9 the proton donor of ESR does not function as efficiently as at lower pH as a source of a proton, presumably because it remains mostly unprotonated in M; M decay in wild type ESR is much more rapid than in the K96A mutant even at the high pH. The observed more than 2 orders of magnitude faster rate in the wild type must reflect the benefit of the presence of an ionizable group in the proton transfer pathway between the protein surface and the buried Schiff base. In such a situation, with only a small fraction of the donor protonated, the rate of M decay becomes pH-dependent at pH Ͼ 9, proton capture becoming rate-limiting, and it assumes the linear pH dependence (Fig. 4C, curves 2 and 3) similar to that in the K96A mutant (Fig. 2B). A similar situation apparently occurs in other pumps with the carboxylic proton donor at high pH when it is only partially protonated and hence proton uptake by the donor should occur first, followed by proton transfer to the Schiff base. Two examples for this are xanthorhodopsin (57) and the fungal rhodopsins (58) where a decrease in the fraction of M that decays fast was observed at high pH.
In conclusion, we provide evidence that Lys-96 is a key part of a mechanism for efficient proton conduction from the bulk to the Schiff base in ESR. Lys-96 probably acts as the group that undergoes transient protonation-deprotonation during the M to N transition with the pK a of 8.5 Ϯ 0.3. Alternatively, it could be a part of a complex with water molecules that undergo protonation changes. Further studies in the infrared could possibly discriminate between these two possibilities. The functioning of the donor in ESR is different from that of Asp-96 in bacteriorhodopsin in that in the latter it undergoes transient deprotonation, whereas in ESR it undergoes transient protonation during the photocycle. In both cases the donor in the ground state is neutral, and the ionized state (positive in ESR and negative in BR) appears only transiently in the photocycle. Since the pioneering studies of the mechanism of proton transport in bacteriorhodopsin (16 -18) and later in proteorhodopsin (21), the presence of two buried carboxylic groups, an internal proton acceptor and an internal donor for the retinal Schiff base, has been considered one of the key identifying features of a lightdriven retinal-based proton pump that distinguish them from sensory rhodopsins. The finding that in ESR Lys-96 in place of a carboxylic residue (Asp or Glu) facilitates proton delivery from the bulk to the Schiff base suggests that there are exceptions to this rule. The initial light reactions BR 3 K, ESR 3 K, and the transition of K to L are omitted. In both proteins, proton transport involves deprotonation and reprotonation of the retinal Schiff base (Sb) a proton acceptor (A), and a proton donor (D). In BR, additionally, a proton release complex (PRC) has been identified. In BR, A is Asp-85 and D is Asp-96. In ESR, these groups are Asp-85 and Lys-96 or associated water molecule(s). In BR, the M 1 to M 2 transition involves proton release from the PRC to the bulk, but in ESR, this step involves proton uptake and protonation of D. Therefore, in BR proton release occurs before proton uptake, but in ESR uptake is before release. The key difference between the two proteins is in the way the Schiff base is reprotonated. In BR it receives a proton from the internal donor, which is initially protonated, and reprotonation of the donor occurs during the N 1 to N 2 transition. In ESR the donor is initially unprotonated, and becomes transiently protonated in the M 1 to M 2 transition, before it can donate a proton to the Schiff base in the M 2 to N 1 transition.