The Retinylidene Schiff Base Counterion in Bacteriorhodopsin*

Previous studies of bacteriorhodopsin have indicated interactions between Asp-85, Asp-212, Arg-82, and the retinylidene Schiff base. The counterion environ- ment of the Schiff base has now been further investi-gated by using single and double mutants of the above amino acids. Chromophore regeneration from bacter-ioopsin proceeds to a normal extent in the presence of a single aspartate or glutamate residue at position 85 or 212, whereas replacement of both charged amino acids in the mutant Asp-85 abolishes the binding of retinal. This indicates that a carboxylate group at either residue 85 or 212 is required as counterion for formation and for stabiliza- tion of the protonated Schiff base. Measurements of the pK,, of the Schiff base reveal reductions of >3.5 units for neutral single mutants of Asp-85 but only decreases of ~ 1 . 2 units for corresponding substitutions of Asp-212, relative to the wild type. Substitutions of Asp-85 show large red shifts in the absorption spec- trum that are partially reversible upon addition of anions, whereas mutants of Asp-212 display minor red shifts or blue shifts. We conclude, therefore, that Asp- 86 is the retinylidene Schiff base counterion in wild-type bacteriorhodopsin. In the mutant Asp-85 + Asnl complete equilibration (-3 for each The pH-induced absorbance changes were determined from difference spectra as fol- lows. A reference spectrum was recorded at a pH where the was completely converted to the protonated form. This was subtracted from each of the spectra recorded at higher pH values, and the of titrated was determined from the ance increase of the deprotonated SB at (for ebR, (for the case of R82D/D85R the the form was incompletely separated from that to the form. Thus, of titrated was determined from the increase at 382 the isosbestic point of the two species. To obtain pK, the change LA was plotted versus the pH. The following three-parameter curve (25) was then fitted to the points.

Previous studies of bacteriorhodopsin have indicated interactions between Asp-85, Asp-212, Arg-82, and the retinylidene Schiff base. The counterion environment of the Schiff base has now been further investigated by using single and double mutants of the above amino acids. Chromophore regeneration from bacterioopsin proceeds to a normal extent in the presence of a single aspartate or glutamate residue at position 85 or 212, whereas replacement of both charged amino acids in the mutant Asp-85 + Asn/Asp-212 + Asn abolishes the binding of retinal. This indicates that a carboxylate group at either residue 85 or 212 is required as counterion for formation and for stabilization of the protonated Schiff base. Measurements of the pK,, of the Schiff base reveal reductions of >3.5 units for neutral single mutants of Asp-85 but only decreases of ~1 . 2 units for corresponding substitutions of Asp-212, relative to the wild type. Substitutions of Asp-85 show large red shifts in the absorption spectrum that are partially reversible upon addition of anions, whereas mutants of Asp-212 display minor red shifts or blue shifts. We conclude, therefore, that Asp-8 6 is the retinylidene Schiff base counterion in wildtype bacteriorhodopsin. In the mutant Asp-85 + Asnl Asp-212 .-, Asn formation of a protonated Schiff base chromophore is restored in the presence of salts. The spectral properties of the double mutant are similar to those of the acid-purple form of bacteriorhodopsin. Upon addition of salts the folded structure of wild-type and mutant proteins can be stabilized at low pH in lipidldetergent micelles. The data indicate that exogenous anions serve as surrogate counterions to the protonated Schiff base, when the intrinsic counterions have been neutralized by mutation or by protonation. brane of Halobacterium halobium (1). The protein contains seven a-helical transmembrane segments and a molecule of all-trans-retinal, which is covalently linked to Lys-216 as a protonated Schiff base (PSB) (Fig. 1). The transport of protons involves a photochemical cycle that consists of at least five transient intermediates (4,5). Recently, site-specific mutagenesis has identified several amino acid residues that are involved in proton translocation. The interactions between Asp-85, Asp-212, Arg-82, and the protonated Schiff base (cf. Fig. 1) were shown to be critically important for the regulation of function and absorption maximum of bR (6)(7)(8)(9)(10)(11). In the recently proposed structural model of bR the side chains Of Asp-85 and Asp-212 are approximately equidistant (- 4 A) from the PSB (2), and both residues appear to be deprotonated in the ground state (7,12). Moreover, the N --., H bond of the chromophore is oriented toward the extracellular side of the membrane (13,14), allowing an interaction of the PSB with Asp-85 or Asp-212. Thus, either or both aspartate residues are candidates for being counterions to the PSB. In a previous study we showed that replacement of Asp-85 by neutral amino acids leads to a significant reduction in the pK, of the Schiff base to between 7 and 8.2 and suggested that Asp-85 serves as the primary counterion to the PSB (10). Arg-82 is presumably ionized as well in bR and is likely to interact with Asp-85 and/or Asp-212. In addition, there is evidence that Asp-85, Asp-212, and Arg-82 are involved in the proton release from the Schiff base in the early phase of the photocycle (10). Specifically, Asp-85 functions as the proton acceptor during M formation (7,10,(15)(16)(17). Based on action spectra it was proposed that Asp-85 also becomes protonated during the purple to blue transition at low pH (9).
In the present work, we have used single and double substitution mutants of Asp-85, Asp-212, and Arg-82 to investigate the counterion environment of the PSB in greater detail ( Fig. 1). Our results show that a single carboxylate group (Asp or Glu) at position 85 or 212 fulfills the requirement for folding and subsequent formation of a PSB in bR. Measurements of the pK, values of the SB show large decreases for neutral substitutions of Asp-85 (2-4 orders of magnitude), whereas corresponding mutants of Asp-212 reveal SB pK, values similar to that of the wild type. Because of the greater effects on the absorption spectrum and on the SB pK, for neutral replacements of Asp-85 relative to Asp-212, we conclude that Asp-85 and the PSB show a stronger electrostatic interaction and that Asp-85 is the retinylidene Schiff base counterion in the ground state of bR. In the absence of an endogenous counterion, as is the case in the double mutant D85N/D212N: exogenous anions can serve as surrogate coun- terion and allow chromophore regeneration. The spectral properties of the D85N/D212N mutant are similar to those of the so-called acid-purple form of wild-type bR. The nature of the anions (halides, perchlorate, and different carboxylic acids) greatly influences the absorption maxima of the mutants. Large spectral shifts are observed for D85A ( X,,, between 544 and 603 nm), D85E ( X,,, between 560 and 607 nm), and D212N (X, , , between 494 and 562 nm) in the presence of different anions. This indicates that electrostatic interactions between the PSB and its counterions play an important role in regulating the wavelength of absorbance in bacteriorhodopsin.

EXPERIMENTAL PROCEDURES
Preparation of bR Mutants-The preparation of mutants containing the single substitutions R82Q, D85A, D85E, D85N, D212A, D212E, or D212N has been reported (6,8,10). Genes encoding the double mutants R82Q/D212N, D85E/D212E, D85E/D212N, D85N/ D212E, and D85N/D212N were assembled using restriction fragments of previously constructed pSBO2 vectors that carry the corresponding single mutations (18,19). In each case a small PuuI-BglII fragment (encoding the mutation at residue 82 or 85) was ligated to a large BglII-PuuI fragment of pSBO2 (encoding the mutation at residue 212). For the double mutants R82Q/D85N and R82D/D85R, ApaI-BglII restriction fragments containing the desired mutations were synthesized and ligated with the small Puul-ApaI fragment and the large BglII-PuuI fragment of wild-type pSB02. The sequences of regions carrying the mutations were verified by direct plasmid sequencing via the dideoxy method (20). The mutant genes were then introduced into the vector pPLl as HindIII-EcoRI fragments (18) and expressed heterologously in Escherichia coli under the control of the XP,,promoter (€419). All of the mutant apoproteins were extracted from crude E. coli membranes with an organic solvent mixture and purified to apparent homogeneity by DEAE-Trisacryl chromatography (21). The average yield of mutant apoproteins was 40-50 mg/lO g of freeze-dried membranes.
Regeneration and Characterization of bR Mutants-bR-like chromophores were regenerated from the apoproteins (16 p~) in a solution of 1% DMPC, 1% CHAPS, 0.2% SDS, and 1 mM sodium phosphate, p H 6.0, by the addition of all-trans-retinal (19). The kinetics of chromophore formation were measured a t 20 "C in the presence of a >3-fold molar excess of retinal. Under these conditions the regeneration rate is independent of the retinal concentration (22). The absorbance increase at the X , , , of the dark-adapted chromophore was monitored. The effect of p H on the chromophore was studied in DMPC/CHAPS/SDS mixed micelles, and the pH was adjusted by adding microliter aliquots of 0.1-0.5 M NaOH or 0.1-2 N H,SO,. Extinction coefficients were determined by acid denaturat,ion of each mutant in the dark to give a chromophore with X, , , , a t 442 nm (23). The ratio of ahsorption at the A,,, to the absorption at 442 nm after acidification to pH 1.9 was compared with that of wild-type ebR. The extinction coefficient of ebR was assumed to be 52,000 M" cm" (22). The X, , , , values of the chromophores were measured a t 4 "C after dark adaptation followed by light adaptation for 3 min (250-watt projector lamp equipped with a 475-nm long-pass filter). Retinal was extracted from the chromophores after dark adaptation followed by light adaptation for 3 min, as described previously (23,24).
Effects of Exogenous Salts on the Absorption Spectrum of bR Mutants-Solutions of sodium salts of halides, perchlorate, and different carboxylic acids (formate, acetate, tartrate, citrate, chloroacetate, dichloroacetate, and trichloroacetate) were prepared a t 2 X (0.1, 1, or 4 M ) concentrations. The salt solutions were titrated with their respective acids to give a pH of 6.0 upon dilution to 1 X. The mutants were regenerated as described above and the individual 2 X salt solutions were then added. In the case of the mutant D85N/D212N the chromophore was directly regenerated in the presence of different salts. All measurements were taken at a protein concentration of 6.4 p~ in 1% DMPC, 1% CHAPS, 0.2% SDS, and 1 mM sodium phosphate at pH 6.00 * 0.05.
Determination of the pK, of the Schiff Base-The mutant proteins (6.4 PM) were prepared as described above in the absence or presence of 2 M NaCI. Titrations were carried out in steps of 0.1-0.3 pH units, and following complete equilibration (-3 min) pH readings and absorption spectra were recorded for each point. The pH-induced absorbance changes were determined from difference spectra as follows. A reference spectrum was recorded at a pH where the pigment was completely converted to the protonated form. This spectrum was subtracted from each of the spectra recorded at higher pH values, and the amount of titrated pigment was determined from the absorbance increase of the deprotonated SB at 365 nm (for ebR, R82Q, D85E, D212A, D212E, D212N, and R82Q/D212N) or 405 nm (for D85A, D85N, and D85N/D212N). In the case of R82D/D85R and R82Q/D85N the transition to the 405-nm form was incompletely separated from that to the 365-nm form. Thus, the amount of titrated pigment was determined from the absorbance increase at 382 nm, the isosbestic point of the two species. T o obtain the apparent pK, the absorbance change LA was plotted versus the pH. The following three-parameter curve (25) was then fitted to the points.
= LAT/ll + 101"' P" -lJHli I The three parameters were: LAT, the total absorbance change of the deprotonated SB; n, the number of protons involved in the transition; and pK, the midpoint of the titration.

I. Effects of Substitutions of Asp-85 and Asp-212 on Chromophore Regeneration
Substitutions of Asp-85 or Asp-212 Alter the Kinetics of Chromophore Formation-We have reported previously (6) that substitution of Asp-85 or Asp-212 by Asn slows down the rate of chromophore regeneration (tlr2 = 41 and 31 min, respectively; Table I), compared with ebR (tip = 1 min). A similar effect was noticed for the mutant D212E (tIr2 = 38 min), whereas for D85E (tllz = 0.2 min) the rate was increased over the wild type. Striking differences in the kinetics are also observed with double mutants of Asp-85 and Asp-212 ( Fig. 2; Table I). The regeneration process for D85N/D212E ( t 1 / 2 = 190 min) is slower by more than 2 orders of magnitude compared to the wild type. On the other hand D85E/D212N (t1/2 = 0.5 min) and D85E/D212E (tlP2 = 2.8 min) reveal relatively normal regeneration times, suggesting that Glu-85  of the dark-adapted chromophore was recorded. This was replotted after determination of the final absorbance. The corresponding t1p2 values are listed in Table I. can compensate the effect of the Asp-212 substitution in these double mutants. In comparing R82Q/D212N (tllz = 11 min) with R82Q/D85N (tIl2 = 73 min) we note that chromophore formation in the former double mutant is accelerated about 7-fold. An extremely slow regeneration process is observed for the double mutant R82D/D85R (tl12 = 760 min), presumably due to introduction of a positively charged residue near the Schiff base.
Chromophore Regeneration Requires a Carboxylate Group at Position 85 or 212"Studies with the mutants D212N and D85N have shown that a single aspartate residue at either position 85 or 212 is sufficient to form a bR-like chromophore (6). A normal extent of chromophore regeneration (>SO% ;   Table I) is also observed for the double mutants D85E/D212N and D85N/D212E, which contain a single glutamate residue at position 85 or 212, respectively. However, removal of both aspartate residues in the double mutant D85N/D212N prevented chromophore formation (<2%). This demonstrates that formation of a protonated retinylidene Schiff base in bR  Table   I.

ZZ. Effects of Double Mutants of Asp-85, Asp-212, and Arg-82 on the Absorption Spectrum
Properties of the Dark-and Light-adapted Chromophores-It has been reported previously that single mutants of residues Asp-85, Asp-212, and Arg-82 generally show red-shifted chromophores at pH 6 relative to ebR (6,8,9) and exhibit altered dark to light adaptation reactions (Table 1; Ref. 26). Consistent with this observation, all double mutants display in the dark at pH 6 an absorption maximum that is red-shifted to between 556 and 597 nm (Table I). Absorption spectra of the dark-and light-adapted states are shown for R82Q/D85N, R82Q/D212N, and R82D/D85R in Fig. 3, A-C. None of the double mutants displays a normal pattern of dark-light adaptation, which in ebR leads to a 10-nm red shift and an increased extinction, due to conversion of the 13-cis/all-trans chromophore into essentially 100% all-trans-retinal (cf. Fig.  2A of Ref. 23). Instead, light adaptation results in a reduced shift to longer wavelength (Fig. 3, A and B ) or in a shift to shorter wavelength (e.g. in D85E/D212N and D85N/D212E; Table I). The light minus dark difference spectra of these mutants reveal a decrease in the absorbance of the chromophore and formation of blue-shifted species with X , , , between 420 and 460 nm, except for R82Q/D212N. Retinal extraction experiments indicate a relatively normal 13-cis to all-trans ratio in both the dark-and light-adapted states of R82Q/ D212N (Table I) near 558 nm between p H 5 and 10, whereas D85N has a red-shifted chromophore with X , , , near 590 nm in this pH range (9,lO). In Fig. 4 the visible absorption maximum is plotted as a function of p H for the double mutants R82Q/D85N and R82Q/D212N. In comparing the two mutants, it is evident that between pH 4 and 8 the chromophore of R82Q/D85N is red-shifted by approximately 20 nm relative to R82Q/D212N. Both double mutants display shifts toward shorter wavelengths in the acidic and alkaline p H range. The increased spectral red shifts observed upon replacement of Asp-85 indicate that this residue exerts a greater electrostatic effect on the protonated Schiff base imine in bR, compared with Asp-212.

III. Exogenous Anions as Counterions to the Protonated
Schiff Base Properties of the Double Mutant D85NID212N in the Presence of Salts-In the presence of different salts, chromophore regeneration is restored in the double mutant D85N/D212N. Fig. 5 shows that a maximum of 68% regeneration of the chromophore occurs in the presence of 3 M NaCI. The absorption maximum a t 558 nm remains constant in the concentration range tested. Absorption spectra of the dark-and lightadapted forms of D85N/D212N in 2 M NaCl are displayed in   Retinal extraction experiments reveal for the DA chromophore an all-trans to 13-cis ratio of 85:15, which remains unchanged upon illumination (Table I).
The effect of p H on the absorption spectrum of D85N/ D212N in 2 M NaCl is shown in Fig. 6. At low pH the chromophore of this mutant remains purple even at pH 1.5 ( X , , , 554 nm; broken line in Fig. 6) and does not convert to a 442-nm absorbing species (free PSB). At alkaline pH the PSB chromophore reversibly titrates to a species with X , , , at 402 nm, representing a deprotonated SB (see below). A single isosbestic point at 460 nm is observed for this transition. Besides alkalinization, decreasing the NaCl concentration from 2 M to 0.5 M also shifts the equilibrium toward the deprotonated form (Fig. 7). The experiment indicates that in the presence of NaCl protonation of the SB is facilitated in this mutant. This effect is likely due to a specific interaction of chloride anions with the PSB.
The Presence of Anions Affects the Absorption Spectrum of Asp-85, Asp-212, and Arg-82 Mutants-According to the hypothesis that chloride anions interact with the PSB in D85N/ D212N, the absorption spectrum of this and other mutants that alter the counterion environment should be sensitive to the type of anion present. This prediction was tested by measuring the X ,   (Table I1 and Fig. 8F).
The absorption maxima of the mutants R82Q, D85A, D85E, D212A, D212N, and D85N/D212N vary significantly in the presence of halides (Fig. 8 and Table 11). For R82Q and D85E a direct correlation between a red shift in the X , , , and an increase in the ionic radius is observed in the order F-< C1-< Br-< I-. However, the mutants D85A, D212A, D212N, and D85N/D212N do not show such a relationship. For example, in each mutant the X , , , value in the presence of iodide is less than that in the presence of chloride.
The effect of different monovalent cations on the absorption spectrum of the D85A mutant was tested as well. Fig. 9 shows that in the presence of 1 M LiC1, NaC1, KC1, or RbC1, this mutant displays identical X , , , values at 575 or 576 nm. This observation supports the idea that the salt-induced spectral shifts in these mutants are caused by anions.
Anions Stabilize the PSB Chromophore at Low pH-The absorption spectra of wild-type ebR and mutants in DMPC/ CHAPS/SDS mixed micelles display at acidic pH a transition to a species with X , , , at 442 nm (broken lines in Fig. lo), normally occurring between pH 4 and 2 (23,27). This conversion represents the formation of a free PSB due to denaturation of the protein. Fig. 10, A  values are 593,576,573, and 566 nm in NaC104, NaI, NaBr, and NaC1, respectively. These anion-induced blue shifts in the sequence C10; < I-< Br-< C1-are identical to those observed upon formation of the acid-purple state of purple membrane (28). For D85N the blue shifts seen in the presence of these salts are generally larger, and the sequence differs (Fig. 10B). The experiment indicates that at low pH solution anions can stabilize the PSB of the folded protein by electrostatic interaction.   Table 11. The absorbance scale for A-D refers to the top curve only. The scale of the other spectra has been reduced by 4 5 % to allow better presentation.

ZV. Effects of Substitutions of Asp-85, Asp-212, and Arg-82
on the pK, of the Schiff Base The Apparent S B pK, of ebR 1s 11.3"To evaluate the interaction between the Schiff base and its environment, the S B pK, was determined for wild-type ebR and mutants. Spectrometric titration of dark-adapted ebR in DMPC/ CHAPS/SDS mixed micelles results in a reversible shift of the A, , , from 551 nm at p H 6 to 544 nm at p H 10 (Fig. 11). Between pH 10 and 12 the purple chromophore converts to a species with X , , , at 365 nm. This transition proceeds through an isosbestic point at 410 nm and is complete at pH 11.9 (Fig.  11). If the alkaline system at p H 11.6 is rapidly neutralized to p H 6, the purple chromophore is partially recovered and in addition a peak a t 440 nm arises (broken line in Fig. 11). To differentiate between a PSB and unbound retinal in the latter  values for ebR are 566,573,576, and 593 nm and for D85N 559, 560, 548, and 554 nm, respectively, in NaC1, NaBr, NaI, and NaClO+ The absorbance differences represent changes in the chromophore extinction. fraction, several samples were acid denatured at pH 1.9 (23), following alkalinization to between pH 11 and 12. Fig. 11 (inset) shows that the 442-nm absorbance of the free PSB observed at pH 1.9 is identical, starting from samples at pH 6, 11, or 11.3. Above p H 11.3 a minor portion of the chromophore becomes hydrolyzed (-22% at pH 11.9). This indicates that the 365-nm absorbing species formed at alkaline pH represents essentially a deprotonated SB. The high pH, however, induces partial denaturation of the protein, thereby preventing complete recovery of the purple chromophore upon lowering the pH to around neutral. Fig. 12A shows the absorbance increase of the deprotonated SB of ebR as a function of pH. Analysis of the titration data (cf. "Experimental Procedures") yields an apparent SB p& of 11.3 and a stoichiometry of 2.5 protons ( FIG. 11. Formation of a deprotonated Schiff base in wildtype ebR at alkaline pH. Absorption spectra were recorded in steps of 0.1-0.3 pH units (cf. Fig. 12.4). Only spectra at the indicated pH values are shown. Between pH 10 and 12 the PSB converts to a deprotonated SB with X , , , a t 365 nm. This transition involves an isosbestic point at 410 nm. Neutralization of the alkaline system at pH 11.6 results in the partial recovery of the purple chromophore, in addition to the appearance of a 440-nm absorbing species (broken line). This indicates that the chromophore contains a Schiff base linkage. The inset shows that upon denaturation with acid at pH 1.9 the 442-nm absorbance of the free PSB is identical, starting from samples at pH 6, 11, or 11.3. Above p H 11.3 a minor portion of the chromophore becomes hydrolyzed.  Table 111.
Asp-85, Asp-212, and Arg-82. The resulting SB pK, values are listed in Table 111. In R82Q, D85E, D212N (Fig. 12B), and R82Q/D212N formation of a deprotonated SB ( X , , , at 365 nm) occurs with a pK, comparable with the wild type, whereas in D212A (Fig. 12B) and D212E the SB pK, is slightly decreased to 10.1 and 10.5, respectively. The transition to a deprotonated SB in these mutants is cooperative, except in D212A, where it involves a single proton.
Mutants that lack a carboxylate group at residue 85, namely  D85A, D85N, D85N/D212N, R82Q/D85N, and R82D/D85R, display between pH 5 and 10 reversible transition from protonated to deprotonated SB chromophore (cj. Fig. 6 for D85N/ D212N). In each case, the titration proceeds through an isosbestic point and involves a single proton (Table 111) at 365 nm is formed (Fig. 6), analogous to the transition observed in the case of ebR (Fig. 11) and all other mutants (e.g. R82Q, DZlZN). The transition between the two deprotonated SB species is cooperative ( n > 1.5) and proceeds through an isosbestic point at 382 nm (Fig. 6). The pK, of SB deprotonation for mutants that lack a carboxylate group at residue 85 was determined based on the absorbance increase at either 405 or 382 nm (cf. "Experimental Procedures"). Analysis of the titration data for D85N (Fig. 1ZC) and D85A yields SB pK, values of 6.7 and 7.6, respectively (Table 111). Similar values have been reported previously for these mutants under slightly different experimental conditions (10). The SB pK, for D85N/D212N was measured in 0.5 M NaC1, as the chromophore of this mutant becomes unstable at low salt concentrations. The titration data reveal a pK, of 7.1, similar to the D85N single mutant. The SB pK, is also decreased in the double mutants R82D/D85R and R82Q/D85N (Fig. 120); the pK, values determined are 7.5 and 9.4, respectively. The results show that in the absence of a carboxylate group at residue 85 the SB pK, is lowered from 11.3 in ebR to between 6.7 and 9.4 in single and double mutants.
The SB pKc, of Asp-85, Asp-212, and Arg-82 Mutants Is Affected by Salt-The pK, of the SB of wild-type bR decreases with increasing ionic strength, indicating that the Schiff base senses the negative charge at the membrane surface (30,31). The decrease of the SB pK, is in the order of one pH unit at molar salt concentrations and originates from unspecific screening of the negative surface charge by the salt. Superimposed on this effect there may be an increase of the SB pK, in certain mutants, due to specific binding of anions near the PSB, thereby stabilizing the protonated state. In the case of wild-type ebR in DMPC/CHAPS/SDS mixed micelles the apparent SB pK, is essentially identical in the presence and absence of NaCl (Fig. 12A), presumably because deprotonation of the SB at high pH is associated with unfolding of the protein. In the mutants D85E and D212N the SB p% also remains unchanged in the presence of salt (Table 111).
As suggested by the experiment shown in Fig. 7, the pK, of the SB in the D85N/D212N mutant is raised by 0.3 units upon increasing the NaCl concentration from 0.5 to 2 M. A comparable effect of NaCl on the SB pKa is also observed for D85A, D85N (Fig. 12C), D212A, D212E, and R82D/D85R. In each mutant the pK, of the SB is increased by 0.2-0.9 units in 2 M NaCl (Table 111), indicating that the specific effect due to binding of chloride anions dominates the surface charge effect. The higher SB pK, in the presence of salt demonstrates that solution anions interact electrostatically with the PSB of these mutants.
Single and double mutants in which Arg-82 has been neutralized by Gln show a transition to a deprotonated SB that is decreased by 0.4-0.8 pH units in 2 M NaC1. Thus, the SB pK, values of R82Q, R82Q/D212N, and R82Q/D85N (Fig.  120) are lowered to 10.6, 10.9, and 8.7, respectively, in the presence of salt (Table 111). In these three mutants the surface charge effect seems to dominate. Removal of the positively charged Arg-82 could lead to increased surface charge effects compared with other mutants, if Arg-82 is accessible from or located close to the extracellular surface, as suggested (2). Alternatively an increase of the SB pK, is not observed, since a specific chloride-binding site involving Arg-82 has been eliminated in these mutants.

DISCUSSION
In the present study we have used single and double mutants of residues Asp-85, Asp-212, and Arg-82 to examine the interactions between these groups and the retinylidene Schiff base in bR (Fig. 1). It has been reported previously (6,10) that substitution of either Asp-85 or Asp-212 singly by Asn or Ala does not affect the ability of the resulting apoproteins to fold, bind retinal, and form a characteristic bR-like chromophore. The extent of chromophore regeneration is also normal for the double mutants D85N/D212E and D85E/ D212N, in which one of these aspartate residues is neutralized and the other substituted by glutamate. However, removal of both aspartate residues in the double mutant D85N/D212N abolishes chromophore formation. This demonstrates that a single negatively charged carboxylate group (Asp or Glu) a t position 85 or 212 is required for formation of a protonated Schiff base. The experiments further provide evidence that both Asp-85 and Asp-212 are ionized in wild-type bR and have the potential to serve as counterions to the PSB. This result is consistent with Fourier transform infrared spectroscopic studies, which have suggested that Asp-85 and Asp-212 are deprotonated in the ground state of bR (7,12,16,17). The lack of chromophore formation observed for D85N/ D212N indicates that besides the two aspartate residues, no other amino acid in the protein can function as the counterion t o the PSB, including Tyr-185, which is now known not to be located in the immediate vicinity of the Schiff base (2).
The chromophore regeneration studies identify both Asp-85 and Asp-212 as candidates for being counterions to the protonated Schiff base. On the basis of the following arguments we conclude that Asp-85 serves as the PSB counterion in the ground state of bR. The strength of interaction between the Schiff base and its counterion should be reflected in the pK, of the SB. Therefore, we determined the pKa values for single and double mutants of Asp-85 and Asp-212 and compared these with that of ebR. Spectrometric titration of ebR at alkaline p H results in the partially reversible formation of a deprotonated SB with X , , , a t 365 nm (Fig. 11). The midpoint of the conversion from protonated to deprotonated SB chromophore is observed at pH 11.3, and the transition is cooperative (Fig. 12A). Thus, deprotonation of the SB at this high p H is associated with the titration of other residues, thereby causing partial denaturation of the protein. The apparent SB pKa of 11.3 for ebR in DMPC/CHAPS/SDS mixed micelles may be compared with the corresponding pK, of 13.3 for bR in purple membrane (32), which was shown to directly reflect titration of the Schiff base imine (33). Identical experiments with the D212N and D212A mutants reveal that the SB pK, is decreased by 4 . 2 units relative to the wild type upon substitution of the negatively charged Asp-212 (Fig. 123). Single and double mutants that lack a carboxylate group a t residue 85 show a reversible transition from protonated to deprotonated SB with pK, between 6.7 and 9.4. In each case, the unprotonated SB displays a X , , , near 405 nm, thereby reflecting the protein-chromophore interactions at the M state of the photocycle, where Asp-85 is known to be nonionized (protonated) (7,16,17). The titration data show that in D85N and D85A the SB pK, is strikingly reduced to 6.7 and 7.6, respectively, as reported previously (10). In the double mutants D85N/D212N, R82D/D85R, and R82Q/D85N the pK, of the SB is also lowered to between 7.1 and 9.4 (Table  111). In contrast to the cooperative transition of ebR at high pH, which probably involves unfolding of the protein, the reversible transition a t lower pH to the 405-nm state in mutants with a neutral residue at position 85 is noncooperative. Similar noncooperative titrations have been observed in related systems with strongly reduced SB pK, values, such as bR regenerated with a retinal analogue (30), halorhodopsin (44), or octopus metarhodopsin (25). The close correspondence between the SB pK, values of the D212N mutant and ebR, as well as between those of the D85N/D212N and D85N mutants, indicates a minor role for Asp-212 in regulating the pK, of the SB. In contrast, the large decrease in the SB pK, observed upon substitution of Asp-85 implies that the high pK, of wild-type bR is largely due to stabilization of the PSB by the carboxylate group of Asp-85. The carboxylate group of the Schiff base counterion is the source of a negative electrostatic potential. Reducing this potential should cause a red shift in the absorption spectrum, besides decreasing the pK, of the SB (34). Consistent with this prediction, the X values of 560, 540, and 505 nm for D212N, D212A, and D212Q,3 respectively. This observation supports the idea that the carboxylate group of Asp-85 exerts greater electrostatic influence on the PSB in the ground state of bR, compared with Asp-212.
It was of interest to study the effects of the double mutants R82Q/D85N and R82Q/D212N on the absorption maximum and the SB pK,, since Arg-82 has been proposed to form a salt bridge with Asp-85 and/or possibly Asp-212 (7)(8)(9)(10)12). The results show that Asn-85 is also red shifted relative to Asn-212 when combined with a Gln-82 substitution (Fig. 4).
Assuming an interaction between Asp-85 and Arg-82, the X,,, as well as the SB pKa would be expected to have comparable values in both D85N and R82Q/D85N. The fact that R82Q/ D85N shows a blue-shifted absorption maximum and significantly higher SB pK, relative to D85N could suggest an interaction between Arg-82 and Asp-212 in the D85N mutant, which may not occur in wild-type bR. Alternatively, it is possible that in D85N the positively charged guanidinium group of Arg-82 is located close to the PSB and causes lowering of its pK,. Since compensatory effects in the mutants cannot be excluded, it remains unclear whether Arg-82 interacts with Asp-85 and/or Asp-212 in the ground state of bR.
The spectral changes observed for the mutants in the pres-:' T. Marti  from -600 nm to near the wild-type value (Table 11), consistent with a decrease in the distance of charge separation. This indicates that upon replacement of Asp-85, exogenous anions can substitute for the protein counterion, presumably by binding to the PSB and/or Arg-82. An electrostatic interaction of anions with the PSB in D85A and D85N is further supported by the observed increase in the SB pK, by 0.3-0.5 units in the presence of 2 M NaCl (Table 111). The fact that relatively high anion concentrations are required to induce changes in the X , , , as well as in the SB pK, indicates that the PSB in these mutants has a limited accessibility to aqueous solutions. This interpretation is consistent with a previous study that measured the reactivity of the Schiff base of mutants in the dark, using hydroxylamine as a probe. D85A and D85N revealed a 30-fold reduction in the rate of reaction, compared with ebR, whereas D85E showed a 10-fold increase (35). This difference in the accessibility of the PSB could explain why carboxylic acids generally cause a spectral shift in D85E but not in the D85A and D85N mutants. Significant changes in the absorption maximum are also observed for neutral substitutions of Asp-212 in the presence of salts (Table 11). However, the X , , , values are generally blue shifted relative to the wild type, suggesting that the solute anions do not replace the endogenous counterion. It is possible that some of the larger anions that shift the X , , , of these mutants to near 500 nm induce a structural perturbation in the vicinity of the Schiff base. In D212A and D212N large spectral changes occur even at low anion concentrations. An increased accessibility of the Schiff base to water in these mutants is also suggested by the light instability (6,26) and the enhanced reactivity with hydroxylamine (35). For R82Q and D212E minor variations in the X , , , are observed in the various anion solutions, consistent with the limited accessibility of the Schiff base to aqueous agents (35). The mutants R82Q and D85E are known to display a purple to blue transition near neutral pH (8,9), and it is likely that the pK, of this transition is altered in the presence of different salts.
Studies by Blatz et al. (36) have shown that the absorption maxima of protonated Schiff bases of retinal in nonpolar solvents undergo a progressive red shift, as the ionic radius of the halide increases.
The mutants R82Q, D85A, D85E, D212A, D212N, and D85N/D212N all display significant variations of the X , , , in the presence of halides (Fig. 8). However, a red shift in the X , , , with increasing ionic radius in the order F-< C1-< Br-< I-is only observed in the case of R82Q and D85E. There are several possible explanations for the lack of correlation in our experiments. First, aqueous solutions are polar solvents, and hydrogen bonding to the SB imine is not prevented. Second, mutations of the Schiff base environment can affect the ability of anions to interact with the PSB to different extents. Third, the opsin shift in bR results from interactions with the polyene chain, in addition to those involving the PSB (37-39). The above mutations could affect both types of interactions. Fourth, resonance structures with partial positive charges located for example on retinylidene C-15, C-13, or C-5 may become stabilized in the mutants. Based on the magnitude of the anion-induced spectral shifts in these mutants, it is evident that the interactions between the PSB and its counterions play an important role in regulating the wavelength of absorbance in bR.
In the double mutant D85N/D212N formation of a chromophore is restored in the presence of various salts. Thus, in the absence of a protein counterion in this mutant, solution anions can substitute and promote Schiff base protonation. This effect apparently is due to a direct electrostatic interaction between anions and the Schiff base. In support of this idea, the pK, of the SB is increased at higher anion concentrations (Fig. 7), and the absorption maximum of the mutant is sensitive to the type of solution anion present (Table 11).
The spectral properties of the D85N/D212N double mutant in the presence of salt are strikingly similar to those of the so-called acid-purple form of wild-type bR (28,(40)(41)(42). This conclusion is supported by (i) the color, (ii) the effect of anions on the X , , , , (iii) the stability of the purple chromophore at low pH, (iv) the isomer composition of the retinal chromophore, and (v) photocycle measurements.4 The acidpurple state of bR arises from the acid-blue state by lowering the pH below 1 with hydrochloric acid or by selective binding of anions at pH <2 (28,40,42). Based on studies with mutants (9), it appears that Asp-85 is protonated in the blue form of bR (pK, of protonation -3.5 and 2.9 in micelles and purple membrane, respectively; cf. Refs. 23 and 40), and at pH 1 Asp-212 is presumably protonated as well. Indeed, recent Fourier transform infrared measurements suggest that one carboxylate group, possibly Asp-212, becomes protonated during the chloride-induced transition from blue to acid-purple (42). It is therefore likely that in the acid-purple form exogenous anions replace the intrinsic counterion. This proposal is corroborated by the experiments with wild-type ebR and mutant proteins in DMPC/CHAPS/SDS mixed micelles at low pH.
In the presence of various salts unfolding of the protein, normally occurring near pH 3, is not observed (Fig. 10). The variation in the X , , , seen for ebR and the D85N mutant with the same anion at pH 1.9 could reflect differences in the pK, of Asp-212 and in the nature of the side chain at position 85 (protonated aspartate versus asparagine). These studies demonstrate that solution anions function as counterions to the PSB and stabilize the folded protein, when both Asp-85 and Asp-212 are neutralized by mutation and/or by protonation at low pH.
It is instructive to compare the Schiff base counterion environment of bR with that of other retinal-based pigments. In halorhodopsin, which functions as a light-driven chloride pump in Halobacterium halobium, the residues corresponding to Asp-212 and Arg-82 are conserved, whereas Asp-85 is replaced by the neutral Thr (residue 111). This substitution could be responsible for the increased opsin shift ( X , , , at 578 nm) relative to bR and a weakening of the electrostatic interaction between the PSB and its counterion, as suggested by resonance Raman spectroscopy (43). Besides, several properties of halorhodopsin, namely the SB pK, of 7.4, the increased SB pK, in the presence of salts, and the anion-induced blue shift of the X , , , (44), are highly similar to those observed for the D85A and D85N mutations in bR. A close relationship between the two halobacterial pigments is underscored by recent studies which indicate that the acid-purple form of bR is capable of translocating halides (45). Based on this analogy it is plausible that anions interact directly with the PSB also in halorhodopsin. In the case of bovine rhodopsin, Glu-113 in transmembrane helix C has been identified as the Schiff base counterion by site-specific mutagenesis (46,47). Its substitution by neutral amino acid residues causes a large reduction in the pK, of the SB to around 6. In the presence of solute anions formation of a PSB is promoted in these mutants and the X , , , shows anion-dependent shifts (46,  The Schiff Base Countel retinylidene Schiff base counterion in the ground state of bR.
During the photocycle Asp-212 may transiently function as counterion to the chromophore, for example when Asp-85 is protonated. Further studies are needed to identify the Schiff base counterion at the intermediate states of the photocycle and to evaluate the precise roles of Asp-212 and Arg-82 in the proton translocation process.