Voltage Dependence of Proton Pumping by Bacteriorhodopsin Mutants with Altered Lifetime of the M Intermediate

The light-driven proton pump bacteriorhodopsin (BR) from Halobacterium salinarum is tightly regulated by the [H+] gradient and transmembrane potential. BR exhibits optoelectric properties, since spectral changes during the photocycle are kinetically controlled by voltage, which predestines BR for optical storage or processing devices. BR mutants with prolonged lifetime of the blue-shifted M intermediate would be advantageous, but the optoelectric properties of such mutants are still elusive. Using expression in Xenopus oocytes and two-electrode voltage-clamping, we analyzed photocurrents of BR mutants with kinetically destabilized (F171C, F219L) or stabilized (D96N, D96G) M intermediate in response to green light (to probe H+ pumping) and blue laser flashes (to probe accumulation/decay of M). These mutants have divergent M lifetimes. As for BR-WT, this strictly correlates with the voltage dependence of H+ pumping. BR-F171C and BR-F219L showed photocurrents similar to BR-WT. Yet, BR-F171C showed a weaker voltage dependence of proton pumping. For both mutants, blue laser flashes applied during and after green-light illumination showed reduced M accumulation and shorter M lifetime. In contrast, BR-D96G and BR-D96N exhibited small photocurrents, with nonlinear current-voltage curves, which increased strongly in the presence of azide. Blue laser flashes showed heavy M accumulation and prolonged M lifetime, which accounts for the strongly reduced H+ pumping rate. Hyperpolarizing potentials augmented these effects. The combination of M-stabilizing and -destabilizing mutations in BR-D96G/F171C/F219L (BR-tri) shows that disruption of the primary proton donor Asp-96 is fatal for BR as a proton pump. Mechanistically, M destabilizing mutations cannot compensate for the disruption of Asp-96. Accordingly, BR-tri and BR-D96G photocurrents were similar. However, BR-tri showed negative blue laser flash-induced currents even without actinic green light, indicating that Schiff base deprotonation in BR-tri exists in the dark, in line with previous spectroscopic investigations. Thus, M-stabilizing mutations, including the triple mutation, drastically interfere with electrochemical H+ gradient generation.


Introduction
Bacteriorhodopsin (BR) from Halobacterium salinarum acts as a light-driven proton pump. It generates an electrochemical gradient for protons, which then is utilized for secondary active transport processes and ATP synthesis. The BR transport mechanism is closely coupled to the photocycle, in which, after absorption of a photon, the retinal isomerizes from all-trans to 13-cis, and via the spectroscopically distinct intermediates J, K and L the blue-shifted M intermediate is formed. The key steps in the BR photocycle for the directed, 'vectorial' transport process are the de-and reprotonation of the Schiff base (i.e. the formation of the M 1 intermediate, the transition from M 1 to the intermediate M 2 , and decay of M 2 ). During the formation of M 1 , the Schiff base (SB) releases a proton to Asp-85, and its pK a shifts from .13 to ,4. Still in M 1 , the SB changes its pK a back to .13, and accessibility changes from extracellular to intracellular during the M 1 RM 2 transition. Subsequently, the retinal SB is reprotonated by Asp-96 from the intracellular side, whereby M 2 relaxes back to the ground state BR via the photocycle intermediates N and O. The MRBR decay contains the rate-limiting and main electrogenic steps at depolarizing voltages [1].
Substantial for understanding the pump mechanism of BR was the determination of the protein's structure in the ground state BR and several photocycle intermediates [2,3,4,5,6,7,8,9,10,11,12,13]. Two experimental approaches were used, first, shock-freezing of crystals formed from BR wild-type protein immediately after illumination with actinic light [9,11], second, crystallization of BR mutants, which accumulate certain photocycle intermediates. Mutants BR-D96N and BR-D96G, due to their drastically slowed M decay in spectroscopic experiments, were used for structural analysis of the M intermediate [7,8,10,11], and mutants BR-F171C and BR-F219L were chosen for the structure of the N intermediate, [4,5,12].
The three-dimensional structure of BR-WT reveals seven ahelices (termed A-G) forming a transmembrane pathway for protons, which is subdivided into a cytoplasmic (CP) and an extracellular (EC) part by the chromophore retinal. The retinal is bound via a SB to Lys-216 in helix G ( Figure 1). BR mutant structures showed that during the first half of the photocycle (BRRM) only minor structural changes occur, while the relaxation from the state M 1 (with SB accessibility still towards EC) through M 2 (with SB accessibility towards CP) back to BR is accompanied by distinct, albeit small, structural rearrangements. In detail, after excitation of the retinal, the CP part of the channel still assumes a closed conformation, but deprotonation of the SB and deprotonation of Asp-96 trigger movements of helices B, G and F, which open the CP part of the proton pathway [5,7,8,9,11,12]. These studies revealed that well-concerted conformational rearrangements take place during the BR photocycle.
The mutant BR-D96G/F171C/F219L (BR-tri) which combines several of the above mutations was shown to exhibit structural features of the M intermediate already in the dark [10,11]. In contrast to BR-WT and other structurally investigated BR mutants, BR-tri showed no detectable structural changes during the reaction cycle, although its proton pump activity was reported to reach 66% of BR-WT [14]. These findings have challenged the functional importance of structural changes for the efficiency of proton pumping by BR.
Besides extensive research that is currently devoted to the application of other retinylidene proteins such as channel-or halorhodopsins in optogenetics, there is a long-standing interest from biotechnology in the archetypical BR and its sequence variants, since BR is endowed with outstanding robustness against harsh natural or experimental conditions and can be considered to be the most thoroughly studied membrane transporter protein in biophysical research. The photochromic and optoelectric properties made it particularly interesting for biomolecular optoelectronic high-density or high-speed data processing and storage concepts, such as optical memories [15], holographic storage devices [16,17], document security applications [18] or ultrafast optical switch logics [19,20,21]. For most of these applications, BR mutants with extended lifetimes of photointermediates with large photochromic shifts are desired that enable safe optical switching between quasi-stable spectral species. Promising variants would be BR mutants such as L93A stabilizing the red-shifted O intermediate [22], the aforementioned Asp-96 substitutions stabilizing the blue-shifted M intermediate, or F171C and F219L destabilizing M in favor of N, which are all well characterized by crystallography. However, despite the wealth of spectroscopic and structural data, very little information about the crucial function of these mutants is available, which is light-driven H + transport across cellular membranes. Of particular interest is the study of the transmembrane potential's influence on the photocycle kinetics, which can only be achieved by heterologous expression in cell lines that facilitate electrophysiology. To close this gap, we expressed mutants BR-D96N, BR-D96G, BR-F171C, BR-F219L, and BR-D96G/F171C/F219L in Xenopus oocytes and studied the voltage dependence of transient and stationary photocurrents using the two-electrode voltage-clamp technique (TEVC). These experiments reveal the influence of mutations and negative membrane potentials on proton pumping. In combination with an analysis of the lifetime of M intermediates in the respective mutants allowed us to derive general conclusions about the regulation of proton pumping by the lifetime of M, as exemplified previously for wildtype BR [1,23,24].

Results
Upon expression of wild-type and mutant BR constructs in Xenopus oocytes, transient and stationary photocurrents were measured using the two-electrode voltage-clamp technique. Figure 1 shows the location of the mutated amino acids within the BR structure and illustrates, that all mutated residues are located within the intracellular part (with respect to the retinal) of the proton transport pathway. As observed for BR-WT ( Fig. 2A), all investigated mutants respond to green light (l .495 nm) with positive transient and stationary photocurrents consistent with outwardly directed H + translocation (

Stationary Photocurrents of BR-WT
Photocurrents of BR-WT in response to green light exhibit three phases, which can be attributed to distinct processes of the photocycle (given in brackets). Upon switching-on illumination, a positive transient current is observed (BRRM, on-response), which is followed by a positive stationary current (turnover cycling; ''BRRBR'', or more correctly ''MRM''), and after light switch-off the photocurrent decays to zero (MRBR decay, off-response) ( Fig. 2A). As shown previously [1,24], the off-response consists of two exponential components reflecting two time constants during the MRBR decay. The fast time constant (t f , 1-2 ms) is   prepared with PyMol 1.0 software. The retinal chromophore (magenta) is covalently linked via a Schiff base to Lys-216 in helix G (orange), which -together with the primary proton acceptor (Asp-85) and proton donor group (Asp-96) -is depicted in ball-and-chain representation (oxygen atoms: red, carbon atoms: green, nitrogen atoms: blue). Also shown are Phe-171 and Phe-219, which, together with Asp-96, were mutated herein. doi:10.1371/journal.pone.0073338.g001 apparently voltage-independent and can not be resolved due to the limited time resolution of about 1-2 ms in TEVC experiments. The slow time constant t s is approximately linearly dependent on voltage. It represents the rate-limiting step of the BR pump cycle (,33 ms at 0 mV and pH 7.5 [1]) and therefore determines the stationary current. Figure 2A depicts photocurrents upon stationary illumination at 0 mV and -100 mV. The peak amplitudes are less voltage-dependent than the stationary currents, which can be attributed to a lower voltage dependence of the initial photocycle events (retinal isomerisation followed by transfer of the SB proton to Asp-85 and H + release from the extracellular H + release complex). The current-voltage (I-V) curve for the stationary photocurrents of BR-WT is linear between 2120 mV and +60 mV (Fig. 2B). Unfortunately, the occurrence of timedependent baseline drifts at voltages exceeding +60 mV or 2120 mV precluded the determination of current amplitudes with satisfactory accuracy. Assuming that inversion of the BR photocycle and photocurrent reversal cannot occur, current amplitudes should decrease to zero in a non-linear, but monotonous fashion at more negative potentials. Conversely, at depolarizing potentials, non-linearity due to saturation of photocurrent amplitudes is expected from an enzyme kinetic point of view [25,26].
Notably, non-linear decrease of stationary photocurrents at negative potentials was readily observed for M stabilizing mutants (Fig. 4B,G and Fig. 5B). The voltage dependence from slopes of the stationary I-V curves of the various BR constructs can be compared by linear extrapolation of the normalized current amplitudes (Fig. 2B) to the 'potential of zero current' (V I = 0 ), which is about -220 mV for BR-WT. Of note, V I = 0 should not be confused with the term 'reversal potential' that is used to characterize passively conducting ion channels. The more negative V I = 0 is for a particular mutant, the weaker is the voltage dependence of proton pumping.

Pre-steady State Photocurrents of BR-WT
BR in the M state absorbs blue light, whereupon the retinal reisomerizes and reprotonation of the SB occurs from the extracellular side from Asp-85, thereby short-circuiting the photocycle, which is known as blue light quenching. Therefore, additional illumination with blue laser flashes (l = 396 nm, 10 ns pulse duration) results in transient quenching of the stationary photocurrent, showing up as inwardly directed transient currents ( Fig. 2C,D,G). Figure 2E,F shows such blue laser flash-induced currents (#1 and #12 from Fig. 2C) on an expanded time scale. After the blue flash-induced inward current peak, a small positive transient current is observed, which is due to the fact that those BR molecules, which had just been shot back to the ground state by the blue laser flash, are re-excited by the green background light.
The blue laser flash-induced transient currents are not fully resolved in time due to the system time constant of TEVC experiments, which limits the time constants for the fast rise and decay of the negative transients to ,1.4 ms (Fig. 2E,F). Since these time constants are invariant from flash to flash, the peak amplitude is proportional to the amount of charge translocated during the transient current. In turn, the charge is a relative measure for the amount of BR molecules, which momentarily populate the M intermediate, since during blue light quenching in each BR molecule a proton is translocated across a fraction of the transmembrane field by traveling the distance between Asp-85 and the SB. Thus, the negative peaks of the blue laser flashinduced currents during illumination with green light indicate the amount of BR molecules in the M state.
The blue laser flash-induced negative transient currents in Fig. 2C show the formation and decay of M for wild-type BR at 0 mV. The graph represents an overlay of 12 traces with different time intervals n?Dt (25 ms to 300 ms) between the start or the end of stationary illumination and the blue laser flash. Blue laser flashes applied during illumination resulted in large negative peak currents, which did not depend on Dt during stationary illumination (see also The transient current preceding the stationary current reflecting the reaction BRRM [1], is similar to BR-WT. The stationary current amplitude is determined by the reactions M 1 RM 2 RBR, as long as BRRM is much faster. Changes in the kinetics of these reactions would strongly influence the I peak /I stationary ratio. However, the I peak /I stationary value of BR-F171C is comparable to that of BR-WT indicating a similarly fast photocycle. The decay phase (off-response) of mutant BR-F171C exhibits two exponential components with time constants similar to BR-WT (see Fig. 6) indicating that the rate-limiting step of the transport cycle is not significantly altered. These time constants agree well with the time constants of 3 ms and 8 ms determined at pH 7 by time-resolved FT-IR spectroscopy for the de-and reprotonation of Asp-96, respectively [27,28]. In notable contrast to BR-WT, the fast component has a larger amplitude implying that the major fraction of BR-F171C molecules, which decay to the ground state in the dark reaction does not generate large charge movement. This observation on the N-accumulating BR-F171C mutant is in line with the interpretation that the electrogenicity of the NRORBR partial reaction sequence is smaller than that of the preceding MRN transition. Blue laser flashes applied to BR- F171C during green illumination did not result in negative transient currents at 0 mV (Fig. 3C), and at 2100 mV negative transients were only small (Fig. 3D) indicating that the accumulation of M intermediates during illumination is drastically reduced. This property of BR-F171C (indicating that SB reprotonation proceeds faster to bring about M decay in favor of N) is remarkable considering that Phe-171 resides at the largest distance to the SB from all amino acids studied herein. (2): BR-F219L Figure 3E shows photocurrents of mutant BR-F219L at 0 mV and -100 mV upon illumination with green light. Stationary photocurrent amplitudes reach 30-50% of BR-WT currents, whereas I peak /I stationary values are comparable (Fig. 6). The slope if the stationary I-V curve and V I = 0 (-228 mV), is almost identical to BR-WT (Fig. 3F). The slow time constant from the stationary photocurrent's off-response is similar to BR-WT, but with smaller relative amplitude, even though it is slightly larger than that of BR-F171C (Fig. 6). Blue laser flashes induce smaller negative peak currents during green-light illumination at 0 mV (Fig. 3G) indicating reduced accumulation of M, the effect is weaker compared to BR-F171C. Blue flash-induced peak currents of mutant BR-F219L during green illumination increase strongly at 2100 mV (Fig. 3H), and after green illumination the negative peaks decrease very slowly indicating that for BR-F219L negative potentials even more profoundly affect M decay than for BR-WT.

M-accumulating Mutants (1): BR-D96N
Photocurrents of mutant BR-D96N at 0 mV and -100 mV are at least 20 times smaller compared to BR-WT (Fig. 4A), whereas transient peaks (on-response) still reach ,25% of the BR-WT amplitudes. The stationary photocurrents vanish almost completely at -100 mV, and the I peak /I stationary ratio is strongly altered ( Table 1), indicating a drastically slowed photocycle. This is confirmed by a similarly increased time constant of the photocurrent's off-response (Fig. 6). The decay phase is less well defined due to the small amplitudes, and only single-exponential fits can be performed yielding time constants of ,200 ms at 0 mV (Fig. 6) indicating that the rate-limiting step of the entire pumping process is dramatically slowed. Moreover, the I-V curve of BR-D96N (Fig. 4B) is clearly non-linear and steeper between 0 mV and +60 mV compared to BR-WT. The asymptotic decrease of the I-V curve to zero demonstrates that the vectoriality cannot be inverted simply by applying sufficiently large negative voltages.
Blue laser flash experiments for mutant BR-D96N elicited larger negative transient currents already at 0 mV (Fig. 4D) compared to both, the stationary currents and the blue laser flash-induced transient currents of BR-WT. After the end of green light illumination, the M decay occurred with a much slower time constant at 0 mV (Fig. 4D) compared to BR-WT (,300 ms vs. ,100 ms), and was even slower (.1000 ms) at -100 mV (Fig. 4E), similar to previous observations [1,24].

M-accumulating Mutants (2): BR-D96G
Stationary green light-induced photocurrents of mutant BR-D96G were comparable to BR-D96N (compare Fig. 4F to 4A), with a similar I-V curve that also decayed asymptotically to zero at negative potentials (Fig. 4G). Blue laser flash experiments again showed large accumulation of M during illumination with green light. The decrease of the blue flash-induced peak currents after illumination with green light at 0 mV was even slower than for BR-D96N ( Fig. 4J; ,1200 ms), but did not increase as profoundly at -100 mV (Fig. 4K).

Azide Effects on BR-D96N and BR-D96G
As expected from previous studies [29], azide (20 mM) greatly restores normal proton pump activity in mutants BR-D96N (Fig. 4C) and BR-D96G (Fig. 4H), which can be seen from the strong increase of stationary photocurrents and the decreased time constants of the photocurrents' off-responses (Fig. 2G). Blue laser flashes also revealed a faster M decay in the presence of 20 mM azide (data not shown). Thus, also in the Xenopus oocyte system, the drastically reduced rate of proton pumping by mutants BR-D96N and BR-D96G can be recovered by the addition of azide, with the notable difference that the stationary photocurrent amplitudes of BR-D96G in the presence of azide at -100 mV are more profoundly decreased than for BR-D96N.

M-accumulating Mutants (3): BR-D96G/F171C/F219L
Photocurrents observed for the mutant BR-D96G/F171C/ F219L (BR-tri) in response to continuous illumination with green light resemble those of mutants BR-D96G and BR-D96N. Following a comparatively large transient current, stationary currents are very small (Fig. 5A) and exhibit a slow M decay (from off-response). Similar to BR-D96N and BR-D96G, the I-V curve is steeper than for BR WT in the range between 0 mV and +60 mV and asymptotically decays to zero at negative potentials (Fig. 5B). Some enhancement of the stationary photocurrent occurs upon addition of 20 mM azide (Fig. 5C), but by far not as strong as for the two Asp-96 mutants. Blue laser flash experiments revealed a high degree of M accumulation at 0 mV (Fig. 5D), similar to BR-D96G and BR-D96N, but only small increases of the negative peak currents were observed at -100 mV (Fig. 5G), indicating reduced voltage sensitivity of M accumulation in BR-tri.
The feature that distinguishes BR-tri from all single mutants studied in this work is that negative, inwardly directed transient currents in response to blue laser flashes are observed even after long times in the dark (Fig. 5E). This is in contrast to BR-WT, for which blue laser flashes without green background illumination exclusively induced positive transient currents (Fig. 5F), as expected for single-turnover forward transport caused by residual absorption of blue light.

Discussion
Photocurrents of several structurally and functionally important BR mutants were investigated in Xenopus oocytes under membrane potential control. Mutants BR-D96G and BR-D96N had already been studied in black lipid membrane experiments or with purple membranes oriented in polyacrylamide gels [30,31]), in which no transmembrane potential control is possible. All mutants generated positive photocurrents indicating outwardly directed proton pumping, but large differences in photocurrent amplitudes, voltage dependence and kinetics were observed.

Classification of the BR Mutants by Voltage Dependence of Photocurrents and M Lifetime
The first group of mutants is represented by BR-F171C and BR-F219L, which destabilize M in favor of the N intermediate. The photocurrents of these mutants are approximately of the same size as BR-WT currents, but the voltage sensitivity within the +60 mV to 2120 mV range studied here is reduced for BR-    Fig. 6) indicating that the rate-limiting step of the transport cycle is not significantly altered. However, in contrast to BR-WT, the fast component is much larger than the slow one showing that a major fraction of BR-F171C molecules decaying to the ground state after the end of steady-state illumination does not generate large charge movement. Since BR-F171C in known to accumulate N, these observations suggest that the electrogenicity of the NRORBR partial reaction sequence is small compared to that of the preceding MRN transition. The differential amount of M found for mutants BR-F171C and BR-F219L under identical illumination conditions indirectly suggests a differential degree of N accumulation, which is notable since both mutants were used to determine the 3D structure of the N intermediate. The results from our electrophysiological experiments can be reconciled with findings from X-ray crystallographic studies. In a study using BR-F171C it was claimed that all protein molecules could be driven into the N state [4,5], whereas in another study on BR-F219L [12], only 33% of the pump molecules were found to be present in the N state upon illumination, while the rest remained in the parental state [32]. The incomplete enrichment of N in BR-F219L was attributed to the large conformational changes suggesting that only one pump molecule per trimer in a purple membrane array could undergo the required large conformational change to avoid steric clashes. The BR-F219L molecules in the plasma membrane of oocytes (see Table 1:10 10 molecules per cell, i.e. 3300 molecules per mm 2 for a spherical cell with 1.5 mm diameter) do not achieve such a dense packing comparable to the situation in native purple membranes. Thus, it seems unlikely that the lower population of the N state in mutant BR-F219L is a consequence of steric hindrance. Rather, the reduced accumulation of N should be regarded as a characteristic kinetic property of this mutant H + pump.
The second group consists of mutants BR-D96N, BR-D96N and BR-tri (BR-D96G/F171C/F219L), which all are known to accumulate M strongly during illumination [8,30,31]. All these mutants consistently show largely reduced stationary photocurrents (at least by a factor of 20). This is accompanied by similarly increased decay time constants of the photocurrent's off-response at an extracellular pH of 7 (to ,250 ms, Table 1), which according to previous work represents the rate-limiting step of the photocycle [1]. BR-D96N, BR-D96G, and BR-tri show a large M accumulation during illumination as indicated by large negative transient currents upon blue laser flashes. This is paralleled by a drastically increased lifetime of M after illumination compared to BR-WT, reaching from ,300 ms for BR-D96N over ,800 ms for BR-D96G to .1000 ms for BR-tri (pH 7). These values are in good agreement with published data from spectroscopic measurements (BR-D96N, BR-D96G .500 ms; [30]). Regarding the nature of the M intermediate(s) probed by blue laser flashes, one can refer to a structural study by Luecke et al. [8], which attributed conformational changes in BR-D96G to an early M intermediate such as M 1 , while the structural changes in BR-D96N were proposed to represent formation of a later intermediate M N , which is similar to M 2 . The major difference between M 1 and M 2 is the direction of SB accessibility. Similar to the effect of mutations on the M 1 /M 2 distribution, the orientation of the SB can be changed by the electrical field, as shown for BR-WT [24]. The exclusively negative amplitudes of blue laser flash-induced transient currents show that SB reprotonation upon blur light only occurs from the extracellular side (Asp-85), independent from the previous SB orientation (either promoted by mutation or by the electrical field). Most likely, after the absorption of a blue photon in the M state, the retinal isomerizes back to all-trans with SB accessibility to the EC part of the proton translocation pathway.
It is important to note, that the observed small photocurrents are not due to a greatly reduced expression level. The number of BR molecules in the plasma membrane of oocytes can be estimated as follows: The slow phase of the stationary current's off-response (,27 ms at 0 mV for BR-WT) constitutes the ratelimiting step for the whole photocycle, and stationary turnover cannot proceed faster than the corresponding rate constant (37 s -1 ). Therefore, a lower limit for the number N of BR molecules which gives rise to the observed stationary current can be calculated by the following equation: where I stat is the stationary current amplitude and t 2 the ratelimiting time constant. The resulting N values are given in Table 1 for all studied constructs and show that the expression levels only vary by a factor of ,2-3. This variation cannot account for the drastically reduced stationary currents observed for M-stabilizing mutants (by more than a factor 30). M-accumulating mutants are simply weak proton pumps, and this should also impair the ability to generate an electrochemical H + gradient within a cellular context. At negative voltages, the I-V curves of the M-accumulating mutants are non-linear and approach the zero current line asymptotically. This in accord with the notion that the vectoriality of the light-driven H + pump BR cannot be inverted simply by applying sufficiently negative voltages. The limited voltage range accessible by TEVC so far precluded the observation of such a behavior of BR-WT.

The Role of the M Intermediates for the Voltage Sensitivity of Proton Pumping
The understanding of the voltage sensitivity of proton pumping in BR is an essential aspect of the molecular mechanism. Our results show that the life-time of the M intermediates determines the current-voltage behavior of BR and, therefore, serves in the regulation of proton pumping. Mutants BR-D96N, BR-D96G and BR-tri achieve M accumulation by a drastically increased M lifetime. Mutants BR-F171C and BR-F219L accumulate N by a faster M 2 decay and not by a slowed N decay, since the time constants for the photocurrent's off-response are not significantly changed compared to BR-WT. It has been shown that the voltage dependence of stationary proton pumping by BR is regulated by two effects, which are related to the life-time of the M intermediate. The main electrogenic event occurs during the MRNRORBR branch of the photocycle, which is reflected by the pronounced voltage dependence of the slow decay constant (t 2 ) of the stationary current's off-response. The voltage dependence of t 2 is stronger than that of the stationary I-V curve [1], and at strongly negative membrane potentials an additional, much slower M decay time constant becomes apparent. This has led us to propose a branching -most likely from M 1 -into a futile reaction pathway involving a still M-like intermediate, which also allows for reprotonation of M from the extracellular side [1,24]. The extent to which this branching occurs, depends on the strongly voltagedependent lifetime of M and explicitly implies reversibility of the M 1 «M 2 reaction [1]. At large negative potentials the M lifetime is drastically increased allowing more BR molecules to enter and accumulate in the futile reaction branch.
The strong effect of the lifetime of M on the voltage dependence of proton pumping is corroborated by the differential voltage sensitivity of the mutants studied here. The slopes of the I-V curves of mutants with undetectable (BR-F171C) or reduced (BR-F219L) M accumulation are shallower, and those of mutants with strong M accumulation (BR-D96G, BR-D96N and BR-tri) are steeper than in BR-WT. Accordingly, the increase of M lifetime probed by blue laser flashes follows the same order (BR-F171C # BR-F219L,BR-WT,BR-tri,BR-D96N,BR-D96G). Thus, the profound correlation between voltage dependence of proton pumping and the lifetime of the M intermediate, as well as the augmentation of stationary pump currents of M accumulating mutants by azide is a characteristic feature of BR. These data are consistent with observations on sensory rhodopsins (SR) when expressed in Xenopus oocytes in the absence of their cognate transducer proteins. The SR proteins also exhibit long-lived M analogues during the photocycle together with a profound voltagedependence and steep I-V curves and show substantial augmentation of stationary proton pumping by azide [33]. Of note, these findings do not contradict results from surface-enhanced infrared difference absorption spectroscopy (SEIDAS) studies on SRII from Halobacterium salinarum (or Natronomonas pharaonis) showing a pronounced inhibition of proton transfer from the retinal SB to the primary proton acceptor Asp-75 (or Asp-73, respectively) [34], which according to later work [35] is strongly dependent on pH, in accordance with a profound effect of the applied electric field on the protonation state of Asp-75. In our photocurrent measurements, the initial proton transfer step corresponds to the positive peak current preceding the stationary photocurrent, which also exhibits a marked, albeit weaker, voltage dependence, which was shown to be similarly dependent on the extracellular pH [1].

Proton Pumping in the Absence of Substantial Conformational Changes -the BR-Tri Mutant (BR-D96G/ F171C/F219L)
The most severe changes in photocurrent properties compared to BR-WT are found for BR-tri. The current-voltage behavior of BR-tri is largely governed by the characteristic properties of BR-D96G, which removes the primary proton donor for reprotonation of the SB. Stationary photocurrents are very small (only ,3% of BR WT currents at 0 mV), which severely impairs the ability to pump protons. Again, the small photocurrents are consistent with increased time constants for the rate-limiting step (off-response of the stationary current, .300 ms) and cannot be due to a reduced expression level (see Table 1).
Some features of the triple mutant, however, differ markedly from all other mutants. Firstly, the transient current (on-response) has a different shape compared to BR-WT. Rising and falling phase are slower, which can be interpreted as a slower initial phase of the photocycle (usually the formation of M). This is in line with the slower M formation, as determined by spectroscopy (1 ms, vs. 50 ms for BR-WT) in a study of Tittor and colleagues [14]. How can this behavior be explained? In Figure 7, the distances between the SB and the carboxyl oxygens of Asp-85 are depicted according to structural information for BR-WT and BR-tri. The structures provided by Sass et al. for BR-WT (ground state and M) showed that the distance increases during the BRRM transition (Fig. 7A,C). The BR-tri structure by Subramaniam et al. [11] revealed that (1) the distance between Asp-85 and the SB is increased in the unilluminated ground state already, and (2) that the CP channel is opened (Fig. 7B,D). The increased distance between Asp-85 and the SB may consequently slow down SB deprotonation towards Asp-85 and the formation of M in BR-tri. Furthermore, the primary proton donor for SB reprotonation is removed due to the D96G mutation, which most likely also entails a slowed-down SB reprotonation during the photocycle of BR-tri. Both properties, in effect, would make BR-tri a slow and inefficient proton pump.
Unfortunately, the crystal structure does not allow conclusions about the protonation state of the SB in BR-tri. In this respect, the second exclusive property, which distinguishes BR-tri from the other mutants, is remarkable. Blue laser flash illumination in the dark (i.e. irrespective of pre-excitation) leads to negative peak currents (see Fig. 5E). If these negative currents of BR-tri were due to the same process as in BR-WT (i.e. blue light quenching of M and reprotonation of a deprotonated SB from Asp-85) this would indicate that SB deprotonation occurs spontaneously in BR-tri and that a significant part of the BR-tri molecules would assume a state with the deprotonated SB even in the dark. However, vibrational spectroscopy has demonstrated that the SB in BR-tri is protonated in the dark state [14]. Therefore, it is conceivable that the negative transient current is due to a fast and transient proton release towards the ''wrong'' CP-oriented side, as a consequence of the increased distance to Asp-85 and the already opened CP channel. Similar negative transient current components upon blue laser flash illumination have also been reported for the BR homolog proteorhodopsin (PR) from c-proteobacteria, although in the latter case, this is not due to an increased distance between the retinal SB and the primary proton acceptor group (Asp-97 in PR), but due to the fact that the pK a of Asp-97 (PR) is around 7.2, so that about 50% of the prooton acceptor groups are protonated at neutral pH [36]. This futile side reaction may contribute to the low proton pumping efficiency, as does the substantially slowed SB reprotonation through the (already open) CP pathway, as judged from the time constants of the photocurrent's off-response (indicating slowed M decay). However, whereas photocurrents of BR-D96N/G single mutants are substantially accelerated by azide, thereby restoring rather normal H + pump function, BR-tri photocurrents are hardly augmented by azide, which also suggests that the disturbance of structure and dynamics of the H + pump caused by the triple mutation is more severe than just the defect of slowed SB reprotonation through the CP channel.
It is notable that the triple mutant was found to show structural features of the M intermediate already in the (unexcited) ground state and that only minor conformational changes occur upon illumination [14]. This has raised a discussion about the necessity of structural changes for efficient proton translocation. In the same study, it was claimed that the BR-tri mutant reaches 66% of the WT protein's overall transport activity [14] even in the absence of detectable structural changes. However, the turnover number of ,2 s 21 for BR-tri (at pH 6) determined in the same publication [14] compared to the generally accepted turnover number of BR-WT (50-100 s 21 [37], and ,37 s 21 determined in this study), matches well with the data provided in our study, which showed about 30-fold larger photocurrents of BR-WT compared to BRtri. Thus, our results indicate that the proton pumping process of BR-tri under otherwise identical illumination conditions proceeds by a factor of at least 30 slower than in BR-WT, thus making BRtri a profoundly weaker proton pump than the wild-type protein.
In addition, these observations imply that the disruptive effect of the M-stabilizing D96G mutation, which per se results in a slow and inefficient proton pump, can not be compensated for by the Nstabilizing effect of the F171C (plus F219L) mutations.

Conformational Changes as Basis for Vectorial Transport of BR-WT
The properties of the BR mutants studied here suggest that potent vectorial ion transport by BR is intimately related to distinct, albeit comparatively subtle structural changes that need to occur in a temporally well coordinated manner. This mechanism determines correct sidedness of SB de-and reprotonation, and ensures vectoriality of H + transport as well as efficient energy transduction. The available structures show that formation of M increases the distance between Asp-85 and the SB (compare Fig. 7A and 7C), which is comparable to the distance obtained for BR-tri in the ground state ( Fig. 7B and 7D). In the early transport cycle intermediates, H + transfer from the SB to Asp-85 is allowed, but the structural changes during M formation that increase the distance to Asp-85, prevent the proton from returning to the SB. By the same argument, the large distance between the SB and Aps-85 in the ground state of BR-tri accounts for the slowed onresponse of the BR-tri photocurrents. Subsequently, the SB is reprotonated from Asp-96, and the large conformational changes of helices F, G and also helix B further facilitate reprotonation of Asp-96 through the CP channel. The conformational changes, the opening of the CP channel after deprotonation of the SB, and the change in its accessibility from EC to CP provide the appropriate framework for the capability of BR to efficiently pump protons.

Ethics Statement
Surgical removal of ovary tissue from adult Xenopus laevis females followed registered protocols approved by the responsible state authority (Landesamt für Gesundheit und Soziales [LAGeSo] Berlin, Reg. No. O 0308/06) and the local ethics committee (Tierversuchskommission, Fachgruppe IC beim LAGeSo Berlin), in accordance with the German Animal Protection Act (Tierschutzgesetz). Animals were anesthetized by immersion in water containing 0.2% w/v tricaine (MS-222, Sigma, Deisenhofen, Germany) for 5 min, and subsequently placed on ice during surgical treatment. All efforts were made to minimize animal suffering.

cDNA Constructs
The bacterio-opsin (bop) cDNA was amplified from a pGEM-bop construct (kind gift of Dr. Phil G. Wood) using adapter primers adding NruI and XbaI restriction sites 59 and 39 of the coding sequence, and inserting a Kozak sequence [38] immediately 59 of the start codon. The fragment was cut with NruI and XbaI and subcloned into vector pTLN [39]. Mutations were introduced using the QuickChange TM mutagenesis kit (Stratagene, La Jolla, CA) and verified by sequencing (Eurofins MWG Operon, Ebersberg, Germany).

Electrophysiology
BR photocurrents were recorded in the TEVC configuration on Xenopus oocytes with a GeneClamp500 amplifier (Molecular Devices, Union City, CA) or a Turbotec 10 CX amplifier (npi electronics, Tamm, Germany). Data acquisition, shutter triggering and control of transmembrane potential were performed with pClamp 7 or 9 software via a Digidata 1200B or Digidata 1440 interface (Molecular Devices, Union City, CA). Current traces were usually recorded at 5 kHz after filtering to 1 kHz using the amplifiers' filtering circuits. Data traces obtained were usually averages of 3 runs. Since the observed photocurrents are quite small (,100 nA), changes in the leak conductances of the oocytes (1-10 mS) in the range of 1% lead to changes in offset currents of 1-10 nA over the potential range studied here. Therefore, all currents were calculated as the difference between currents evoked by a specific voltage-pulse first with, then without illumination. Only experiments without recognizable drifts in the background conductance were taken for further analysis. All experiments were carried out using a Na-100 bath solution (containing in mM: 100 NaCl, 15 TEA?Cl, 2 CaCl 2 , 10 TRIS, pH 7.5) at room temperature (21uC) 3-5 days after injection. The pipette solutions contained 3 mM KCl. Typical pipette resistances were between 0.5 and 2 MV.

Optical Equipment
To induce photocurrents, cells were illuminated with light from a mercury arc lamp (Carl Zeiss, Göttingen, Germany, equipped with an Osram HBO 100 W bulb), filtered through a heat protection and a short-wavelength cutoff filter (Schott GG495 = ''green light''), and coupled into an optical light-guide. Illumination was controlled by a fast shutter (Uniblitz LS6ZM2, Vincent Associates, Rochester, NY), which provided a response time of ,1 ms. For single flash experiments, light from a XeCl excimer laser-pumped dye laser (10 ns pulse duration, equipped with dye PBBO, 396 nm = ''blue light'') was used. Due to the lower time resolution in oocyte experiments, the peak amplitudes of blue laser flash-induced currents are generally smaller than obtained in HEK293 cells [1].