Vibrational Spectroscopy of Bacteriorhodopsin Mutants EVIDENCE THAT ASP-96 DEPROTONATES DURING THE M + N TRANSITION*

The role of Asp-96 in the bacteriorhodopsin (bR) photocycle has been investigated by time-resolved and static low-temperature Fourier transform infrared dif- ference spectroscopy. Bands in the time-resolved difference spectra of bR were assigned by obtaining anal- ogous time-resolved spectra from the site-directed mutants Asp-96 + Ala and Asp-96 + Glu. As concluded previously (Braiman, M. S., Mogi, T., Marti, T., Stern, L. J., Khorana, H. G., and Rothschild, K. J. (1988) Biochemistry 27, 8516-8520) Asp-96 is predomi- nantly in a in intermediate. Upon formation of the N intermediate, deprotonation of Asp-96 occurs. This is consistent with its postulated role as a key residue in the reprotonation pathway leading from the cytoplasm to the Schiff base. A broad band centered at 1400 cm", which increases in inten- sity upon N formation is assigned to the Asp-96 symmetric COO- vibration. The Asp-96 + Ala mutation also causes a delay in the Asp-212 protonation which normally occurs during the L + M transition. It is concluded

' The abbreviations used are: bR, bacteriorhodopsin; h-bR, native bacteriorhodopsin from H. halobium; e-bR, bacteriorhodopsin produced from the expression of a synthetic wild-type gene in Escherichia coli; FTIR, Fourier transform infrared; TR-FTIR, time-resolved FTIR. hR mutants are designated by the wild-type amino acid residue (standard one-letter code) and its position number followed by the substituted amino acid residue. Thus, D96E signifies the mutant in which the aspartic acid a t position 96 has been replaced by glutamic acid.
dation. photocycle, the chromophore and protein undergo structural changes that lead to the net transport of one proton from the cytoplasmic medium to the exterior of the cell. A useful method for studying these structural changes is Fourier transform infrared (FTIR) difference spectroscopy (2). This technique is sufficiently sensitive to detect individual chromophore and protein groups involved in the bR photocycle (3-6). By using FTIR difference spectroscopy in conjunction with isotope labeling and site-directed mutagenesis methods, it is possible to assign vibrational bands to particular residues in bR (7)(8)(9)(10)(11)(12).
In one of the first applications of this approach ( 7 ) , sitedirected mutants of the four membrane-embedded Asp residues in the bR structure (Asp-85, Asp-96, Asp-115, Asp-212) were examined with low temperature FTIR difference spectroscopy. Difference bands that appeared in the 1700-1800cm" region of the spectrum during the bR + K, L, and M steps of the photocycle were each assigned to one of these residues (7). A structural model for the bR active site was developed on the basis of these data and spectroscopic studies of other bR mutants (7,9). According to this model, Asp-85 is protonated by the chromophore Schiff base during M formation, resulting in the release of a proton into the extracellular medium, whereas Asp-96 is involved in the reprotonation of the Schiff base and uptake of a proton from the cytoplasmic medium. However, due to the distance between Asp-96 and the Schiff base it was suggested that Asp-212 is also part of this pathway and acts as the proximal proton donor to the Schiff base during the M + N transition ( 7 ) . Although there is both recent structural (13) and spectroscopic evidence to support the postulated roles of Asp-85 in proton release and of Asp-96 in proton uptake and Schiff base reprotonation (7,10,(14)(15)(16)(17)(18)(19), there is still uncertainty as to whether other residues (such as Asp-212) are involved directly in the proton transport pathway.
In this paper, we utilize stroboscopic time-resolved FTIR (TR-FTIR) (20) and steady-state low-temperature FTIR to explore the role of Asp-96 in the photocycle. We find that Asp-96 undergoes a deprotonation during the M + N transition. Our data also indicate that the protonation of Asp-212 during the M + N transition is delayed when Asp-96 is replaced by either Ala or Glu.

MATERIALS AND METHODS
Spectroscopic Measurements-Time-resolved FTIR spectra were recorded a t 297 K on a Nicolet Analytical Instruments (Madison, WI) 60SX spectrometer using a stroboscopic method described elsewhere (20).
Low-temperature static FTIR difference measurements were made as reported previously on rehydrated films formed by air-drying a sample suspension on an AgCl window (21, 22). Sample Preparation-The construction, expression, and purification of bacteriorhodopsin mutants carrying single substitutions has been reported previously (14,23). Apoproteins were regenerated with retinal and reconstituted in vesicles with polar lipids from H. halobium, using a 1ipid:protein weight ratio of 1:l (24). Fig. 1, shows FTIR difference spectra of the bR + M transition at 250 K of e-bR and the mutants D96A, D96E; and D96N. In agreement with earlier measurements (7, lo), the negative/positive bands at 1742/1748 cm" in e-bR are absent in these mutants, confirming their assignment to the carboxylic C=O stretch of Asp-96. The appearance of both a positive and negative band can be explained if Asp-96 is protonated in both bR and M, but is found in different environments in the two states (7,10). Since at 170 K, where the L to M transition is blocked, the negative 1742-cm" band is more intense than the 1748cm" positive band, it was postulated (7) that a deprotonation occurs during L formation, which is followed by a reprotonation during the L + M transition. Alternatively, Gerwert et al. (10,25) postulate that Asp-96 undergoes only a change in environment during the K + L transition followed by a deprotonation later in the photocycle.

Asp-96 Is Protonated i n the M Intermediate-
Bands due to the symmetric and antisymmetric stretch modes of carboxylate (COO-) groups are expected in the 1300-1600-cm" region (26). Any positive difference band in this region of the wild-type e-bR spectrum that is absent in the D96A spectrum might be due to a deprotonated form of Asp-96 present in the M intermediate. Such a band was suggested previously to be at 1378 cm" (10) in the bR + M difference spectrum at 273 K and was assigned tentatively to the symmetric COO-vibration of Asp-96. This band is also reported to increase in intensity during N formation and therefore might reflect deprotonation of Asp-96 during this step of the photocycle (25). However, the D96A mutation does not significantly affect the intensity near 1378 cm" in the bR + M difference spectrum obtained a t 250 K (Fig. 1). Furthermore, this band does not change significantly in intensity during N formation at room temperature ( Fig. 5; see also "Discussion"). Thus, it seems unlikely that this assignment is correct.
Figs. 2-4 display TR-FTIR difference spectra recorded at room temperature for h-bR, D96E, and D96A, respectively. In each figure, the top difference spectrum corresponds to one set of conditions where M has formed but not appreciably decayed (bR + M difference spectrum), and the other spectra correspond to conditions where M has decayed and appreciable concentrations of the N intermediate have accumulated (bR + (M + N) difference spectra). The time ranges chosen were based on parallel measurements made using kinetic visible spectroscopy' and by comparison with kinetic rates determined from specific bands in the TR-FTIR difference spectra.
The time-resolved bR + M difference spectra of h-bR and the two mutants D96E and D96A in Figs. 2-4 are very similar to the corresponding low-temperature bR + M difference spectra. For example, negative bands due to the bR chromophore appear at 1527 cm" (C=C stretch), 1254, 1201, and 1169 cm" (C-C stretches), and positive bands due to the M chromophore are observed near 1568 cm" (C=C stretch) and 1182 cm" (C-C stretch). The absence of appreciable concentrations of the N intermediate can be deduced from the relative weakness of positive bands which are characteristic of N near 1535 and 1186 cm" (20,27). In addition, the 1761cm" band assigned to the protonation of Asp-85 has not yet shifted to 1755 cm" as it is known to do upon N formation (20).
As in the case of the low-temperature bR + M difference spectra, there is no evidence for significant levels of Asp-96 deprotonation in the bR + M spectra at room temperature.
In the 1300-1450-cm" region, there is again no positive band that could be assigned to a COO-vibration, since such a band should be reduced in intensity upon replacement of Asp-96 by Ala. Deprotonation of Asp-96 during Formation of the N Intermediate-In contrast to room temperature bR "-$ M difference spectra, we find evidence in bR + N difference spectra for deprotonation of Asp-96. The N intermediate is expected to accumulate in wild-type and in the D96E mutant in a time range characteristic of the "fast" component of M decay (17,28). This decay occurs slower in the D96E mutant than in wild-type (16). Figs. 2 and 3 (middle traces) show difference spectra of h-bR and of the D96E mutant in the appropriate time ranges (4-5 ms and 10-40 ms, respectively). Formation of the N intermediate in both samples can be deduced from the shift of the 1761-cm" band to 1755 cm" (20), as well as from an increase in intensity near 1186 cm" (C-C stretch mode in N), a reduction in intensity of the negative 1530cm" band, and a positive shoulder which becomes intense near 1535 cm" (C=C ethylenic stretch in N). In contrast, D96A (Fig. 4), which has a much slower M decay (240 ms), does not exhibit bands characteristic of N formation even at longer times, presumably due to the fast decay of N relative to M (17).
Concomitant with the appearance of N, we find that the broad band centered near 1400 cm" gains intensity in both h-bR (Fig. 2) and the D96E mutant (Fig. 3). In the case of D96E, we show a set of time-resolved difference spectra (Fig.  5 ) which have been scaled in order to compensate for the partial decay of the sample back to the bR state. The 1400cm" and 1186-cm" bands are both found to increase in intensity relative to other bands in the region such as that at 1374 cm". As seen in the inset of Fig. 5, the time course for increase in intensity of the 1400-cm" band parallels that of the 1186-cm" band, indicating that it can also be assigned to the N intermediate.
The  Ftc. 5. Scaled time-resolved FTIR difference spectra of D96E. The earliest spectrum corresponding to 2.4 ms after photolysis pulse is the curve with lowest intensity a t 1186 and 1394 cm". Successive spectra at increasing time intervals of 4.2 ms appear as curves with monotonically increasing intensity a t 1186 and 1394 cm-l. Each spectrum was scaled and base line-corrected globally over the 1970-970-cm" region using the average of all the time-resolved spectra as a reference. This procedure compensates for the normal decay of band intensity throughout the spectrum due to the completion of the photocycle and return of bR back to the bRsm ground state. Note that the growth of the 1186-and 1400-cm" bands relative to other peaks such as at 1374 cm" is apparent in these spectra. Inset, plot of intensity as a function of time with 0.69-ms resolution of the 1186-cm" (+) and 1394-cm" ( X ) bands for spectra of D96E globally scaled and base line corrected as described above. A best tit to a single expotential gave a time constant of 12 ms for the increase in intensity of the 1186-cm" band and 11 ms for the 1394-cm" band.
Asp-96 Deprotonates during the M "+ N Transition of bR (data not shown). Furthermore, a 1400-cm" positive band observed in low-temperature FTIR studies of bR was found to be sensitive to [13C]Asp labeling (29, 30). It should be also noted that both Asp-85 and Asp-212 are unprotonated in light adapted bR (7) and undergo protonation reactions during the photocycle. Thus, no other positive bands due to a COO-are expected to appear. A more conclusive assignment of this band would be possible if it disappeared in the bR + N spectrum of the D96A mutant. However, under normal conditions this mutant does not form an N intermediate, as indicated with transient visible absorption measurements (17).
Assuming that Asp-96 deprotonates upon N formation, an increase in the amplitude of the negative 1742-cm" band should also occur and be closely correlated with the appearance of the 1400-cm-' band. A small increase in negative intensity at 1742 cm" is observed in our time resolved measurements of bR at 297 K (Fig. 2) as well as at 290 K in the first few milliseconds (20). A slight increase in negative intensity of this band has also been reported to occur on the 25ms time scale at 278 K (25). The reason this negative band is small compared for example with the 1762-cm" band, is most likely due to two factors. First, the 1760-cm" band shifts down to 1755 cm" upon N formation and causes a partial cancellation of the negative 1742-cm" band. An additional cancellation occurs due to the positive 1738-cm" band, which is associated with Asp-212 protonation during M formation (7,12) and is expected to be present in the bR + (M + N) difference spectrum. This latter band can be seen clearly in difference spectra of D96E and D96A, where the removal of the negative 1742-cm" band causes an increase in the apparent amplitude of the 1739-cm" band ( Figs. 3 and 4).
The D96E spectra provide confirmation that these overlapping bands explain the low intensity of the 1742-cm" negative band. In this mutant, the 1742-cm" band downshifts to near 1720 cm" (Fig. l), as observed previously at low temperature (7). The absence of the interfering positive peaks in this region allows the observation of a clear intensification of this band concomitant with an increase in the N concentration (Fig. 3). Thus, although the amplitude changes of the 1742em" COOH band assigned to Asp-96 are partially obscured because of band overlap, amplitude changes of the 1720-cm" COOH band of Glu-96 clearly indicate that this group undergoes a deprotonation during N formation.
We also find that the 1400 and 1742-cm" bands are markedly increased in intensity in the difference spectra of h-bR recorded at pH 9 ( Fig. 2) (20). This is again consistent with the assignment of these bands to Asp-96 deprotonation a t N. Under these conditions, the decay of N is slowed (28, 31) and its concentration is increased on a time scale of approximately 10 ms. Thus, the difference spectrum obtained at this pH represents predominantly the bR + N transition. In agreement, the N intermediate marker bands at 1535 and 1185 cm" are particularly intense.
Effects of Asp-96 Substitutions on Asp-212 Protonation Changes-An interesting feature of the low temperature bR + M difference spectrum of the D96A mutant, and to a lesser degree the D96E and D96N mutants (Fig. l ) , is the reduction in intensity of the positive band at 1738 cm-l. This band has been assigned to the protonation of Asp-212 (7,12). Since the negative band at 1742 cm" is no longer present in the three Asp-96 mutants, one would expect the 1738-cm" band to appear more intense but the opposite is observed. A similar effect is observed in the time-resolved bR + M spectra of D96A and D96E (Figs. 3 and 4). In the case of D96A, the 1738-cm" band is almost absent in the early bR + M differ-ence spectrum (Fig. 4, top) gaining intensity only after 55 ms (Fig. 4, bottom). Thus, substitution of Asp-96 appears to cause a delay in the protonation of Asp-212.

DISCUSSION
The present study helps clarify the role of Asp-96 during the bR photocycle. In agreement with an earlier FTIR study (7), we conclude that Asp-96 is predominantly in a protonated state at the M stage of the photocycle. During L formation, a partial deprotonation of Asp-96 (7,29) or alternatively a change in environment of Asp-96 (10) occurs. However, this appears to be only a transient event, with Asp-96 reprotonating by M (20).
Upon formation of the N intermediate we now find evidence for deprotonation of Asp-96. This deprotonation was not deduced from earlier low-temperature data (7), since it is difficult to trap appreciable amounts of N under these conditions. Rather, this conclusion is based on TR-FTIR spectra a t room temperature in which a positive band near 1400 cm" can be assigned to the carboxylate stretch mode of Asp-96 or Glu-96. In addition, a negative Asp-96 COOH band at 1742 cm" and a more intense Glu-96 COOH band near 1720 cm" can be assigned in the respective bR + N difference spectra.
We also find evidence that the Asp-96 + Ala mutation blocks or delays the protonation of Asp-212. This effect can be understood if Asp-212 is part of the proton pathway between Asp-96 and the Schiff base and serves as the proton donor during the M -+ N transition as suggested previously (7,12). In this case, elimination of Asp-96 would be expected to inhibit the protonation reaction of Asp-212, thereby delaying the Schiff base reprotonation. An alternative explanation is that Asp-96 is coupled in a more indirect manner to the protonation of Asp-212, perhaps through structural changes in the C and F helices.
A number of questions still remain about the pathway and sequence of proton movements from the cytoplasmic medium to the Schiff base. The simplest model is a direct movement of the proton from Asp-96 to the Schiff base during N formation (16,17). In this case, it will be important to determine the pechanism by which the proton bridges the approximately 10-A gap (13) as well as the driving force for this movement. Alternatively, if Asp-212 is also involved as an intermediary group (7,12), then the proton reaching the Schiff base during N formation must have a more complicated physical and temporal path.