Vibrational Spectroscopy of Bacteriorhodopsin Mutants EVIDENCE THAT THR-46 AND THR-89 FORM PART OF A TRANSIENT NETWORK OF HYDROGEN BONDS*

The role of Thr-46 and Thr-89 in the bacterio- rhodopsin photocycle has been investigated by Fourier transform infrared difference spectroscopy and time-resolved visible absorption spectroscopy of site-di- rected mutants. Substitutions of Thr-46 and Thr-89 reveal alterations in the chromophore and protein structure during the photocycle, relative to wild-type bacteriorhodopsin. The mutants T89D and to a lesser extent T89A display red shifts in the visible X,,,of the light-adapted states compared with wild type. During the photocycle, T89A exhibits an increased decay rate of the K intermediate, while a K intermediate is not detected in the photocycle of T89D at room tempera- ture. In the carboxyl stretch region of the Fourier transform infrared difference spectra of T89D, a new band appears as early as K formation which is attrib-uted to the deprotonation of Asp-89. Along with this band, an intensity increase occurs in the band assigned to the protonation of Asp-212. In the mutant T46V, a perturbation in the environment of Asp-96 is detected in the L and M intermediates which corresponds to a drop in its pK,. These data indicate that Thr-89 is located close to the chromophore, exerts steric con- straints-on it during all-trans to 13-cis NJ), interfaced with a 340s monochromator (Spex Industries, Edison, NJ) using procedures which have been previously reported (25). Sample Preparatwn-The construction, expression, and purifica- tion of the bacterioopsin mutants T46V, T89A, and T89D has been previously reported (26). Apoproteins were regenerated with retinal and reconstituted in vesicles with polar lipids from H. halobium, using a lipid/protein weight ratio of 1:1 (30). Samples were suspended at a concentration of approximately 10 p~ bR in a standard buffer con-sisting of 150 mM KC1,30 mM sodium phosphate buffer, and the pH was adjusted with 0.1 M NaOH or 0.1 M HC1. After pelleting (15,000 X g for 15 min), these samples were used for both the low temperature FTIR and time-resolved visible measurements.

Institutes of Health Grant GM28289 and A111479 and ONR Grant N00014-82-K-0189 (to H. G. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
3 To whom correspondence should be addressed.
1) Recipient of a fellowship from the Swiss National Science Foundation.
The abbreviations used are: bR, bacteriorhodopsin; CHAPS, 3-[ (3-cholamidopropyl)dimethylammonio]-l-propanesulfonic acid; FTIR, Fourier transform infrared; WT, wild-type. bR 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, T46V signifies the mutant in which threonine at position 46 has been replaced by valine. found in the purple membrane of Halobacterium halobium (1). The primary photoevent in bR is an all-trans to 13-cis isomerization of the retinylidene chromophore (2). This isomerization triggers protein structural changes which result in the active transport of a proton from the cytoplasmic to the extracellular medium. Since no active biomembrane transport mechanism has yet been elucidated at the molecular level, an understanding of the proton transport mechanism in bR is of general interest.
It is known from studies utilizing site-directed mutagenesis that the membrane embedded Asp residues, Asp-85, Asp-96, and Asp-212, are essential for normal proton tran!port by bR (3)(4)(5)(6)(7)(8)(9)(10)(11). A recent structural map of bR at 3.5-10 A resolution (12) reveals the approximate location of these residues. In agreement with an earlier spectroscopically derived model (?, 13), Asp-85 and Asp-212 are located approximately 3-4 A below the Schiff base, while Asp-96 is located approximately 10 A above the Schiff base and close to the cytoplasmic membrane surface.
On the basis of FTIR spectroscopy, a model of the bR proton transport mechanism was proposed that accounts for protonation changes occurring in Asp-96, Asp-85, and Asp-212 during the photocycle (5,13). In light-adapted bR (bRS7,,), Asp-85 and Asp-212 are unprotonated (5, 14), while Asp-96 is protonated ( 5 , 6). Upon M formation, a proton is transferred from the Schiff base to Asp-85. This event triggers the release of a proton into the extracellular medium, possibly due to the breaking of a salt bridge between Arg-82 and Asp-85. The Schiff base is then reprotonated during N formation, receiving a proto? from Asp-96 ( 5 , [8][9][10][15][16][17][18]. However, because of the 10 A gap between Asp-96 and the Schiff base, it was proposed that this proton transfer is indirect and involves intervening residues including Tyr-185 and Asp-212, whose protonation states were also observed to change during the photocycle (5,19).
In this paper, we have utilized low temperature FTIR difference spectroscopy (20)(21)(22)(23)(24) and time-resolved visible absorption spectroscopy (25) to investigate the effects of sitespecific substitutions of residues Thr-46 and Thr-89. These residues are in a position in the bR structure to participate in a hydrogen-bonding network which spans from Asp-96 to the retinal Schiff base. In a recent study, the mutants T46V, T89A, and T89D were characterized in L-a-dimyristoylphosphatidylcholine/CHAPS/sodium dodecyl sulfate micelles (26). T46V displayed an accelerated decay of the M intermediate, thereby indicating an interaction with Asp-96, the internal proton donor to the Schiff base. The Thr-89 mutants showed significantly increased regeneration rates with 13-cis retinal compared with all-trans retinal, and abnormal darklight adaptation reactions, suggesting that this residue is in direct contact with the retinal chromophore. The present 1615 work now shows that substitutions of Thr-46 and Thr-89 do not alter the structure of the all-trans chromophore in the light-adapted ground state. However, the light-adapted Thr-89 mutants exhibit a red-shifted  visible X   ,  ,  , and altered structure and stability of the K intermediate. Replacement of Thr-89 by Asp results in deprotonation of Asp-89 early in the photocycle and increased protonation of Asp-212. This suggests the existence of a proton pathway between the 2 Asp residues in T89D. In the case of T46V, the stretching frequency of the Asp-96 carboxyl group is perturbed in the L and M intermediates, indicating the formation of a Thr-461 Asp-96 hydrogen bond at this stage of the photocycle. In order to account for these findings, a model for proton transport is described which extends an earlier spectroscopically derived model. It includes as a key feature the existence of a transient network of hydrogen bonds which acts to transport a proton from Asp-96 to the Schiff base during the M to N transition.

MATERIALS AND METHODS
FTIR Measurements-Low temperature static FTIR difference measurements were made as previously reported on rehydrated films formed by air-drying an aqueous suspension of a sample on an AgCl window (27,28). Each sample was sealed in a specially designed cell, light-adapted, and cooled to 80 K (bR-+K), 170 K (bR-L), or 250 K (bR-M) (28,29). All spectra were recorded at 2-cm" resolution using a Nicolet 740 spectrometer equipped with an MCT detector. As described under "Results," in the case of the mutant T89D a 660-nm narrow band interference filter was used in place of the normal 628nm narrow band interference filter for K-bR photoregeneration.
Time-resolued Absorption Spectroscopy-Time-resolved spectra were obtained with a 1420 UV-enhanced optical multichannel analyzer and model 1460 controller (Princeton Applied Research, Princeton, NJ), interfaced with a 340s monochromator (Spex Industries, Edison, NJ) using procedures which have been previously reported (25).
Sample Preparatwn-The construction, expression, and purification of the bacterioopsin mutants T46V, T89A, and T89D has been previously reported (26). Apoproteins were regenerated with retinal and reconstituted in vesicles with polar lipids from H. halobium, using a lipid/protein weight ratio of 1:1 (30). Samples were suspended at a concentration of approximately 10 p~ bR in a standard buffer consisting of 150 mM KC1,30 mM sodium phosphate buffer, and the pH was adjusted with 0.1 M NaOH or 0.1 M HC1. After pelleting (15,000 X g for 15 min), these samples were used for both the low temperature FTIR and time-resolved visible measurements.

RESULTS
Substitutions of Thr-89 Perturb the K Chromophore-Both low temperature FTIR difference spectroscopy and room temperature time-resolved visible absorption spectroscopy indicate that the substitutions T89A and T89D perturb the structure of the K intermediate. Fig. 1 shows the bR-K FTIR difference spectra of wildtype bR (WT) and the mutants T46V, T89A, and T89D recorded at 80 K. Bands in the wild-type bR difference spectrum have been previously assigned to vibrational modes of the chromophore of bR670 (negative bands) and the K630 intermediate (positive bands) by comparison with the corresponding resonance Raman spectra (21-23, 31, 32). The negative chromophore bands in the T46V, T89A, and T89D difference spectra appear at very similar frequencies to the wild-type spectrum, particularly in the conformationally sensitive C-C stretching region from 1150 to 1300 cm". These results indicate that the chromophore structure in the lightadapted state of the T46V, T89D, and T89A mutants is largely the same as in wild-type bR.
In contrast, shifts in the positive bands assigned to the K chromophore of both T89A and T89D indicate that alterations occur in the K chromophore structure relative to wildtype bR. For example, the frequency of the hydrogen-out-of plane mode is shifted from 957 cm" in wild-type bR to 950 cm" in both T89D and T89A. The intensity of the 1184 cm" shoulder is increased in T89D relative to wild-type bR. Both these features are characteristic of the L intermediate chromophore, suggesting that the Thr-89 substitutions remove some of the constraints which normally prevent the KsS0 chromophore from adopting a more relaxed 13-cis configuration (33).
In order to further investigate the properties of the K intermediates of T89D and T89A, we measured the timeresolved visible absorption difference spectra of these mutants at room temperature. Surprisingly, the positive band near 630 nm, which appears in the difference spectrum of wild-type bR ( Fig. 2 A ) recorded during the first 200 ns, is not observed in T89D (Fig. 2C). In contrast, a K intermediate was detected in T89A, although it had a faster decay time of 280 ns compared to 500 ns for wild-type bR (Fig. 2, A and B ) .
The most likely explanation for the above results is that Thr-89 does not play a major role in determining the configuration of the all-trans chromophore in light-adapted bR but comes into direct contact with the chromophore during all-trans+l3-cis isomerization. The proximity of Thr-89 to retinal (12) supports this picture. It is also interesting to note that the negative-positive bands at 1423/1429 cm" assigned to structural changes involving the Xaa-Pro C-N peptide bond (34) are absent in the T89D bR-+K, bR+L, and bR+ M difference spectra. Thus, it is possible that an interaction occurs between the chromophore and Thr-89 during chromophore isomerization, which acts as a trigger for a protein conformational change involving 1 or more proline residues. Several other residues in helix C appear to also be structurally active or play a critical role during the photocycle including Asp-85 (5), Trp-86 (35), Asp-96 (5, 6,17,18), and possibly Leu-93 (36). We also found that the T46V mutation had no significant effects on the structure of the K intermediate at either low temperature, as determined by the similarity of the T46V bR+K difference spectrum with wild-type bR (Fig. 11, or at room temperature, based on the normal time-resolved visible absorption difference spectra (data not shown).
T89A and T89D Have Red-shifted Chromophores-The ethylenic stretch frequency, UC=C, of the retinylidene chromophore, which appears at 1527 cm" in light-adapted bR, is downshifted in the bR-K, bR-L, and bR+M low temperature difference spectra of the T89A and T89D mutants by 2 and 5 cm", respectively (Figs. 1, 3, and 5 ) . Based on an empirical linear correlation which exists between the UC=C and Xmax of bR and its photointermediates (32,37), the X, , , of T89A and T89D should be red-shifted by approximately 8 and 20 nm, respectively, relative to wild-type bR. The timeresolved visible bR-M difference spectrum of T89D confirms that there exists a red-shifted X, , , for this mutant, but only by 10 nm relative to wild-type bR (Fig. 20). No red shift could be detected in the T89A visible difference spectrum, however, this may be due to a reduced production of N2 which would tend to shift this band to lower wavelength. We therefore conclude that T89D and to a lesser extent T89A have red-shifted Xmax values and that the degree of the red shift may increase at lower temperature.
A red-shifted X, , , of T89D could also account for the dramatic reduction in amplitude we observed for the K-bR difference spectrum (data not shown). Apparently, the 628nm light normally used for photoreversal of the K intermediate back to bR is also absorbed strongly by the red-shifted bR state of T89D. In contrast, photoreversal using 660-nm light (Fig. 1) produced a more normal amplitude bR+K difference spectrum relative to wild-type bR.
In a previous study, in which these mutants were characterized in L-a-dimyristoylphosphatidylcholine/CHAPS/sodium dodecyl sulfate micelles (26) shift was noted for the light-adapted form. Furthermore, the X, , , of the light-adapted state of T89A was blue-shifted by 28 nm relative to the wild type. However, extractions of the lightadapted chromophores of these mutants in micelles indicate the presence of significant fractions of inactive cis isomers (26), which are blue-shifted, whereas in reconstituted membranes their proportion is likely to be significantly reduced. This has previously been observed for example in the tryptophan mutants W182F and W189F (38). In addition, difference spectroscopy reveals solely the X , , , of the fraction of sample which is photoactive. Thus, it is possible that the X, , , of the visible absorption spectrum of a mutant can differ from that determined by difference spectroscopy, as was observed here.
Asp-89 Deprotonates Early in the Photocycle-The bR-L and bR-M difference spectra of T89D exhibited alterations in the 1700-1800-cm" carboxyl stretch region relative to the corresponding wild-type difference spectra (Figs. 3-6). Most apparent is the negative band near 1753 cm" which most likely originates from the carboxyl group of Asp-89. Resolution enhancement shows that this band is superimposed on the normal set of bands previously assigned in the wild-type difference spectra to the 4 membrane-embedded Asp residues in bR ( 5 ) . For example, in the T89D bR+L difference spectrum (Figs. 3 and 4), the 1753-cm" band is superimposed on top of the pair of negative/positive bands at 1742/1748 cm" due to Asp-96 and a second set of bands at 1735/1728 cm" assigned to Asp-115 ( 5 ) . The bR+M difference spectrum of T89D (Figs. 5 and 6) displays a positive band near 1761 cm" assigned to the protonation of Asp-85. However, this band is upshifted in frequency and diminished in intensity due to partial cancellation with the strongly negative 1753-cm" band.
A second effect of the T89D mutant in the bR-M difference spectrum (Figs. 5 and 6) is the intensification of a band at 1738 cm", near the frequency assigned to the carboxyl stretch mode of Asp-212 (5, 14). A similar effect may occur also in the bR-L difference spectrum of T89D (Figs. 3 and  4), accounting for the reduction in intensity of the negative of bR bR => L 170K bands near 1741 and 1736 cm-l. The most likely explanation for this data is that Asp-89 undergoes a partial deprotonation during the bR+L and M stages of the photocycle, while Asp-212 undergoes an increased protonation. As discussed later, transfer of a proton from Asp-89 to Asp-212 can account for several of the other properties of T89D, and suggests the existence of a proton transport pathway between the 2 residues.
T46V Perturbs the Environment of Asp-96 in the L and M Intermediates-The positive/negative bands at 1748 and 1742 cm-I in the wild-type bR+L and bR+M difference spectra reflect a change in the environment of the Asp-96 carboxyl group (5, 6, 17). As seen in Figs. 4 and 6, the T46V mutation causes a shift in the positive component from 1748 to 1754 cm" in both the bR+L and bR+M difference spectra. The negative band at 1742 cm" also appears to be broadened and slightly upshifted in frequency. However, spectral deconvolution (data not shown) reveals that this shift is mainly due to the upshift in frequency of the overlapping positive component at 1748 cm". Thus, the T46V substitution mainly affects the environment of the Asp-96 carboxyl group at the L and M stages of the photocycle. As discussed below, these results suggest that Thr-46 forms a hydrogen bond with Asp-96 during the early part of the bR photocycle.

DISCUSSION
In this study, we have combined FTIR difference spectroscopy (20,24,39) and site-directed mutagenesis (3) in order to investigate the effects of the substitutions T46V, T89A, and T89D on structural changes which occur during the photocycle of bR. These mutants were previously studied by kinetic visible absorption spectroscopy in L-a-dimyristoylphosphatidylcholine/CHAPS/sodium dodecyl sulfate mixed micelles (26). T46V exhibited an increased decay rate of the M intermediate while the Thr-89 mutants had abnormal dark-light adaptation reactions and greater than 10-fold increased regeneration rates with 13-cis retinal compared with all-trans retinal.
The present study provides additional information about the roles of Thr-89 and Thr-46 in the bR photocycle as well as about several other key residues in bR, including Asp-96, Tyr-185, and Asp-212. These findings are discussed below.
Interaction between Thr-46 and Asp-96 during the Early Photocycle-The T46V substitution perturbed the environment of Asp-96 at the L and M stages of the photocycle, shifting the carboxyl stretching vibration of Asp-96 from 1748 to 1753 cm". The direction of this shift is consistent with a drop in the pK, of Asp-96 (20,40), as previously postulated in order to explain the increase in the rate of the M+N transition as well as a slow down in the proton uptake observed for the T46V mutant (26).
These results, along with the proximity of Thr-46 and Asp-96 in the electron diffraction-derived bR structure (12), point to the occurrence of an interaction between Asp-96 and Thr-46 during the early photocycle. At low temperature, Asp-96 appears to undergo a change in its environment during L formation (5, 6). However, recent room temperature timeresolved FTIR measurements (16)   \ tants were destabilized with the largest effect observed for T89D. 4) The rate of dark adaptation of T89D was increased severalfold relative to wild-type bR.* 5) The rate of chromophore regeneration with 13-cis retinal in T89D is increased relative to wild-type bR (26). 6) The 0 decay is slowed down in T89D relative to wild-type bR (26).
The above effects could be explained if a network of polarizable hydrogen bonds existed in light-adapted bR which extends from Thr-89 to Asp-212. A direct interaction between these 2 residues is unlikely based on the electron diffractionderived structure of bR (12). However, as shown in Fig. 7 and discussed in the next section, such a network could be formed if one or more water molecules were located between Thr-89 and Tyr-185. Evidence for the existence of a polarizable hydrogen bond between Tyr-185 and Asp-212 has been previously presented (14). Alternatively, a pathway might exist between Thr-89 and Asp-212 which does not involve Tyr-185.
One of the major consequences of such a hydrogen-bonding network would be the existence of a pathway for proton transfer from Thr-89 to Asp-212. Normally, protons in such a network are expected to be localized predominantly on the hydroxyl groups of the high pK, residues Thr-89 and Tyr-185 (44). However, substitution of Thr-89 by Asp, which has a much lower pK,, is expected to result in an increased protonation of Asp-212. A similar effect, although less pronounced, could occur for the mutant T89A, if an additional water molecule substituted for the hydroxyl group of Thr-89 in the network. If Thr-89 were involved in a transient proton-transporting network of hydrogen bonds (see next section), the presence of such a water molecule could further explain why the mutant T89A is able to pump protons with a steady-state activity which is approximately two-thirds of that of the wild type (26). In addition to the apparent deprotonation of Asp-89 and increased protonation of Asp-212 in T89D, the bR+ M difference spectrum of T89D (Fig. 5) shows a reduction in the 1271-cm" band assigned to Tyr-185 deprotonation during formation of the M intermediate (19). Such a decrease would be expected since Asp-89 can now function as the major proton donor for Asp-212, rather than the higher pK, residues which form a network of polarizable hydrogen bonds.
Several other features of the Thr-89 mutants can be explained by an increase in the protonation state of Asp-212 as discussed below.
Intrinsic Red Shift in the X , , , of Thr-89 Mutants-Since Asp-212 is located close to the positively charged Schiff base and is in a position to serve as a partial counterion, increased protonation of Asp-212 relative to wild-type bR would be expected to cause a red shift of the X , , , (45). Such a red shift is observed for T89D and to a lesser extent for T89A. Furthermore, this red shift has been shown to reflect an intrinsic property of the light-adapted chromophore4 and is not due to an increased amount of acid-blue membrane present in these mutants, as has been found in the case of the mutants R82A, R82Q, D85E, and Y185F (25,46,47).
Increased Isomerization Rate in Thr-89 Mutants-Based on the concepts previously proposed by Seltzer (48), Asp-212 is in an ideal position to play an important role in determining the energy barriers for retinal isomerization during the photocycle and light-dark adaptation. In the case of the Thr-89 mutants, partial neutralization of Asp-212 could explain the above listed properties 3-5, all of which involve the ability of the chromophore to undergo isomerization. For example, the rapid decay of the K intermediate in T89D is likely to reflect an enhanced ability of the chromophore to relax into a more planar 13-cis configuration and/or a facilitation of single bond isomerization around the c 1 4 -c 1 5 bond as is postulated to occur during the K or L transition (49). The increased rate of dark adaptation is also consistent with a lowering of the energy barrier for all-trans to 13-cis isomerization.
Slow 0 Decay Rate of T89D"Recent time-resolved FTIR measurements demonstrate that the 0 decay involves deprotonation of Asp-212. 3 The slower decay rate of the 0 intermediate in T89D relative to wild-type bR (26) could thus be caused by an inhibition of Asp-212 deprotonation. As described in the model below, this deprotonation occurs by movement of a positive charge back into the network of hydrogen-bonded high pK, residues, including Thr-89. The T89D mutation would be expected to slow down the deprotonation of Asp-212 due to the introduction of a low proton affinity residue (Asp-89).
A Model for Proton Transport-An extended network of hydrogen bonds formed between hydrophilic residues such as Asp, Glu, Tyr, Thr, and Ser as well as water molecules could provide the basis for proton transport within the interior of membrane proteins (44, 50, 51). In support of such a proton P. Rath transport mechanism, Zundel and co-workers (44, 52) have demonstrated that the polarizable hydrogen bonds formed in such networks allow collective proton motions. Thus far, however, there is no direct evidence for the existence of an extended hydrogen-bonded network in bR.
Some of the "channel lining residues" recently identified by Henderson and co-workers (12), which are located between the Schiff base and the cytoplasmic side of bR, are in good positions to form a hydrogen-bonded network. In an attempt to account for the protonation changes of specific residues deduced from FTIR as well as a variety of other data, we have constructed a model of such a network that is based on the electron diffraction-derived bR structure (12) and an earlier proposed mechanism for proton transport (5).
Specific steps in the proton transfer process are shown in Fig. 7 and described below.
bR570-h the bR570 state (light-adapted bR), Asp-212 and Asp-85 exist in a predominantly ionized form which is stabilized by the protonated Schiff base and Arg-82 as originally proposed (5). As indicated in Fig. 7 by a series of shaded arrows (see figure legend), a network of hydrogen-bonded residues which includes Asp-212, Tyr-185, and Thr-89 allows a small delocalization of the negative charge on Asp-212 into the network. This is equivalent to a movement of a proton toward Asp-212. This delocalization could account for the partial deprotonation observed in Tyr-185 by FTIR (19). However, the extent of ionization of any of the high pK, residues in the chain is still expected to be small, consistent with recent NMR and UV resonance Raman studies which reveal a predominantly protonated state for Tyr-185 (53, 54).
K6so-Upon K formation, the movement of the positively charged Schiff base causes a disruption in the network of hydrogen bonds including the Tyr-185/Asp-212 hydrogen bond. As a consequence the protonation of Tyr-185 increases (19,55,56). The red-shifted X , , , of the K chromophore is most likely due to the movement of the positively charged Schiff base away from its counterions Asp-85 and Asp-212. This charge separation could also account for the downshift in the C=N stretch frequency (32)  While none of the high pK, residues in this network are expected to exist in a predominantly ionized form, as a consequence of this partial protonation the polarizable hydrogen bonds allow some of the negative charge of Asp-212 to be delocalized throughout the network. This is denoted by placing a 6(-) near Tyr-185. Note, that in a complete description, several other conjugate forms would be shown with the 6(-) placed at the other high p% residues in the network. The movement of a proton in this network towards Asp-212 (i.e. partial protonation of this residue) can be understood as due to the transfer of the Schiff base proton to Asp-85 (5). Prior to formation of the M intermediate, the positively charged Schiff base forms ionic interactions with Asp-212, Asp-85, and Arg-82 (5,11). The disruption of these ionic interactions due to Schiff base deprotonation, and the subsequent release of Arg-82 from the active site act as a trigger for diffusion of a proton to the extracellular medium (5) as indicated in Fig. 7. NSs0-The M+N transition involves a net proton transfer from Asp-96 to the Schiff base (8)(9)(10). The deprotonation of Asp-96 along with a protein secondary structural change at this stage of the photocycle was recently detected by timeresolved FTIR difference spectroscopy (16,17). As illustrated in Fig. 7, this proton transfer involves the entire transient network of hydrogen-bonded residues including Asp-212, which acts as the proximal proton donor to the Schiff base. The overall process involves the transfer of a proton from Asp-212 to the Schiff base, reprotonation of Asp-212 and simultaneous donation of a proton from Asp-96 into the hydrogen-bonded network. In effect, this amounts to a net movement of a proton from Asp-96 to the Schiff base. Asp-96 is now deprotonated and no longer interacts with Thr-46, accepting a proton from the cytoplasmic medium during the next photocycle step. Similar to the M intermediate, a partial negative charge is delocalized among the high pK, residues in the chain and Asp-212 is partially protonated.
0640-The N+O transition involves the reprotonation of A~p -9 6 ,~ thus accounting for the observed slow N decay at high pH (9,43). This transition also involves a reisomerization of the chromophore from 13-cis to all-trans. As discussed in the previous section, the partial neutralization of Asp-212 could act to facilitate this isomerization. It is not yet clear, however, how Asp-96 reprotonation influences this process. One possibility is that conformational changes of the protein associated with the N+O transition3 are triggered by Asp-96 reprotonation. The red-shifted Amax of the 0 chromophore is most likely due to neutralization of Asp-85 and partial neutralization of Asp-212, which serve as the counterions of the Schiff base when the chromophore is in the all-trans configuration. 0 Decuy-The entire system resets itself during the 0-bR transition. This involves a movement of the proton localized on Asp-212 back toward the high pK, residues including Tyr-185 and Thr-89. A block in the deprotonation of Asp-212 is expected to slow the 0 decay rate, as is observed for the mutant Y185F.5 The ionic interactions between Asp-85, Asp-212, Arg-82, and the Schiff base are also reestablished during this step of the phot~cycle.~ Many of the features of this model remain to be established. For example, bands in the FTIR difference spectrum due to protonation changes of Thr-46 and Thr-89 have not yet been identified. In addition, the postulated role of Asp-212 as the proximal proton donor to the Schiff base is based on indirect evidence (5,14). Alternative candidates for a proximal proton donor include Thr-89 (58, 59). However, such a model is consistent with Asp-85 serving as the proximal proton acceptor for the Schiff base. In this case as seen in Fig. 7, a small movement of the Schiff base would allow it to switch from a position where it donates a proton to Asp-85 (L+M transition) to a position where it accepts a proton from Asp-212 (M+N transition). In support of the role for Asp-212 as the proximal proton donor, recent time-resolved FTIR measurements reveal that a slowed M decay, as observed in the mutant D96A, is also accompanied by a delay in the protonation of ASP-212 (17). tural changes occurring in the bR photocycle. The information obtained here and in earlier studies utilizing a combination of FTIR, isotope labeling and site-directed mutagenesis (5,13,14,16,17,19) has led us to propose a tentative model for proton transport in bR which can account for a variety of data on bR and its mutants. Several structural aspects of this model, first proposed in 1988, have recently been confirmed by electron diffraction (12). Higher resolution analyses may help determine the possible existence of a hydrogen-bonding network in a position to transport protons from Asp-96 to the Schiff base. However, the movement of protons through such a hydrogen-bonding network cannot easily be observed directly by diffraction techniques. Thus, further progress and tests of this and other models will most likely continue to rely on the application of spectroscopic probes including static and time-resolved FTIR difference spectroscopy.