INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Bacteriorhodopsin (BR)¹ is an integral membrane receptor that functions as a light-driven proton pump in the purple membrane of Halobacterium salinarium (1-4). Several features make BR highly attractive for in vitro studies of membrane protein folding and assembly. First, the denatured apoprotein can be spontaneously refolded to the native state with quantitative recovery of secondary structure, chromophore binding, and proton-pumping activity (5, 6). Reconstitution of the native structure has also been accomplished with complementary combinations of proteolytic fragments and synthetic peptides, comprising one or more of the transmembrane regions (7-11). Furthermore, refolding and chromophore binding has been demonstrated for numerous mutants containing amino acid substitutions, deletions, or insertions (12-15). By using time-resolved spectroscopy, transient intermediates in the folding process of native BR have recently been identified (16, 17). Second, the structure of BR has been solved at high resolution, revealing the detailed arrangement of the seven membrane-spanning -helices and of the surface loops (18-20). The retinal chromophore is linked covalently through a protonated Schiff base to Lys-216 (Fig. 1) and is located in a central cavity where it contacts each of the seven helices. The membrane topology of BR is shared by the visual pigment rhodopsin, which in turn is a member of the large and functionally diverse superfamily of receptors coupled to guanine nucleotide-binding proteins (3). Third, detailed insight into intramolecular interactions present in the ground state and photocycle intermediates of BR has been provided over the past years by the combined application of site-directed mutagenesis and various biophysical techniques (21-23). The majority of these methods have relied on the absorption properties of the retinal chromophore, which represents an extremely sensitive probe to monitor the environment and detect structural changes within the protein.

Based on thermodynamic arguments and renaturation experiments on BR, a model for the folding of helical membrane proteins has been proposed, involving two principal steps (24-26). In the first stage, hydrophobic segments insert into the membrane as -helices, which are independently stabilized by hydrophobic interactions with the bilayer and by main chain hydrogen bonds. In the second stage, the preformed helices assemble without major rearrangement to form the native bundle structure. The transmembrane helix association is driven by helix-helix interactions, external constraints provided by surface loops, lipid-packing effects, and occasionally prosthetic groups (26). This model predicts that individual transmembrane -helices represent independent folding domains, since they achieve a defined conformation in the absence of other parts of the molecule.

The large degree of structural autonomy of membrane-spanning polypeptides is supported by experiments, in which native structures have been assembled from fragments that were separately refolded in vitro or synthesized in vivo. Besides the heptahelical membrane proteins BR (7-11, 27), rhodopsin (28, 29), adrenergic receptor (30), and muscarinic acetylcholine receptor (31), such studies also include members of other protein families that form helical bundle structures, like the voltage-gated sodium channel (32), yeast a-factor transporter (33), or lactose permease (34). Specifically, in the case of BR it has been shown that proteolytic fragments comprising the two helices AB² or FG reassociate with the complementary five-helix fragments CG or AE, respectively, to form the native chromophore (7-9, 27). In addition, the BR structure could be regenerated upon substitution of the two-helix fragments by two synthetic peptides corresponding to the individual transmembrane helices (10, 11). These results indicate that the A-B, B-C, E-F, and F-G loops are not required for the assembly process, although they contribute to the stability of BR, as shown for the A-B and B-C loops by thermodynamic measurements (35). In contrast, the roles of the short C-D and D-E surface loops (Fig. 1) in folding and stabilization of the protein have not been evaluated, primarily due to a lack of selective accessibility to proteases and availability of an appropriate expression system.

To characterize the in vitro folding and assembly process of BR with independent structural domains, an efficient procedure for the production of polypeptide fragments in Escherichia coli has been developed in the present work. By introducing the C-terminal tail of BR as a purification tag into each construct, eight fragments containing two (AB, FG), three (AC, EG), four (AD, DG), or five (AE, CG) of the membrane-spanning helices could be prepared in milligram quantities. Upon reconstitution into phospholipid/detergent micelles, all pairs of fragments comprising complementary (AB·CG,² AC·DG, AD·EG, AE·FG) or overlapping (AC·CG, AD·CG, AD·DG, AE·CG, AE·DG, AE·EG) parts of the protein regenerated a native-like chromophore with high efficiency, whereas in the absence of one or more of the transmembrane segments pigment formation was abolished. This shows that the assembly of BR requires each of the seven helices but does not depend on any of the covalent connections provided by the surface loops. Denaturations in SDS and pH titrations revealed a general destabilization of the chromophores formed by fragment complexes, relative to intact BR. Regions that are susceptible to denaturation were identified in helices D and F-G. The overall results indicate that the C-D and E-F surface loops contribute to the specificity of helix interactions in BR.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Materials-- Purified oligonucleotides were purchased from Eurogentec (Seraing, Belgium), and restriction enzymes were obtained from Pharmacia (Freiburg, Germany) or New England Biolabs (Schwalbach, Germany). Sequenase (version 2.0) and -³⁵S-dATP (500 Ci/mmol) were from Amersham (Braunschweig, Germany), and DNA purification kits were from Qiagen (Hilden, Germany). Casamino acids and yeast extract were obtained from Difco (Augsburg, Germany), and kanamycin, all-trans-retinal, DMPC, and soybean lipids (L--phosphatidylcholine, type II-S) were from Sigma (Deisenhofen, Germany). Ampicillin, DNase I, RNase A, lysozyme, and CHAPS were purchased from Boehringer (Mannheim, Germany). Protran nitrocellulose (0.2 µm) was from Schleicher & Schuell (Dassel, Germany), horseradish peroxidase-conjugated rabbit anti-mouse IgG from Dako (Glostrup, Denmark), and DEAE-Trisacryl from BioSepra (Frankfurt, Germany). Reagents used for N-terminal sequencing and amino acid analysis were obtained from Applied Biosystems (Weiterstadt, Germany). All other chemicals were of analytical grade and purchased from Merck (Darmstadt, Germany).

Construction of BO Gene Fragments-- All BO gene fragments were constructed by restriction fragment replacement in the cloning vector pSBO2 that contains a synthetic BO gene (36). Oligonucleotide duplexes encoding amino acid deletions in the wild-type protein sequence (Table I) were synthesized to span the following restriction sites (cf. Fig. 1 of Ref. 36): KpnI-XhoI for fragment AB, BglII-XhoI for AC, BssHII-XhoI for AD, SphI-XhoI for AE, AatII-ApaI for CG, HindIII-NarI for DG, HindIII-PstI for EG, and AatII-SphI for FG. The gene fragments were constructed with a three-component ligation procedure (37), and the sequences of sections containing newly synthesized oligonucleotides were confirmed by direct plasmid sequencing (38). The cloned BO genes were introduced into the expression vector pPL1 as HindIII-EcoRI fragments (36).

Expression of BO Gene Fragments-- E. coli strain W3110 harboring the temperature-sensitive repressor plasmid pcI857 was transformed with expression vectors carrying the BO gene fragments. Cells were grown in a 2-liter culture vessel at 30 °C under aeration in a rich medium initially consisting of 50 mM potassium phosphate, 15 mM ammonium sulfate, 4 mM sodium citrate, 2 mM MgSO₄, 0.2 mM CaCl₂, 1% casamino acids, 0.5% yeast extract, 0.1% glucose, 0.02% tryptophan, with 100 µg/ml ampicillin and 50 µg/ml kanamycin. During cell growth 150 ml of nutrient was added, consisting of 25% glucose, 10% casamino acids, and 5% yeast extract. When the cells reached an A₆₀₀ of 8-12 (corresponding to 2-3 g dry cell weight/liter), the temperature of the culture was rapidly raised to 42 °C to induce production of BO fragments. After 20 min the cells were harvested by centrifugation (15 min, 5,000 × g, 4 °C) and then frozen. Thawed cells were resuspended in phosphate-buffered saline containing 1 mM MgCl₂, 0.1 mM CaCl₂, 50 µg/ml DNase I, 20 µg/ml RNase A, 0.5 mg/ml lysozyme, and 0.2 mg/ml phenylmethylsulfonyl fluoride and were disrupted by ultrasonication. Membranes were collected by centrifugation (2 h, 40,000 × g, 4 °C), lyophilized, and stored at 20 °C.

Purification of BO Polypeptide Fragments-- To purify the expressed BO fragments, the procedure developed by Braiman et al. (39) for eBO was modified as follows. Lyophilized membranes (~5 g) were mixed directly with 40 ml of solvent A, and the solution was homogenized thoroughly (2 min, Ultra-Turrax T25 Tissuemizer, IKA-Labortechnik, Staufen, Germany). Following centrifugation (15 min, 5,000 × g, 4 °C), the clear supernatant was decanted and the pellet reextracted with 40 ml of solvent A. The two solvent extracts were combined, and a phase separation was then induced by the addition of an equal volume of water. Following centrifugation (20 min, 5,000 × g, 4 °C), the aqueous and organic phases were both removed by decantation. The solid interface was mixed with solvent B (~40 ml) until a single liquid phase was obtained. After removal of insoluble material by centrifugation, the phase separation and interface resolubilization steps were repeated. The interface obtained after the third phase separation was redissolved in 25 ml of solvent C containing 30 mM TEA acetate. Following sedimentation of insoluble material, the supernatant was applied to a DEAE-Trisacryl column (1 × 25 cm) equilibrated in solvent C containing 30 mM TEA acetate. The column was washed with 30 mM TEA acetate in solvent C until the absorbance at 280 nm of the eluate reached the base-line value. Elution of BO fragments was performed with a linear gradient of TEA acetate (30-150 mM, using 2 column volumes each of the initial and final concentration) in solvent C. The polypeptide fragments generally eluted from the column at a TEA acetate concentration of about 120 mM. BO fragments were recovered by phase separation of pooled fractions and were redissolved in solvent B. To transfer the proteins into aqueous solutions, an aliquot of 10% (w/v) SDS in water was added. The exact amount of SDS was chosen to give a final SDS/protein ratio of 5:1 (w/w). Following evaporation of the organic solvent in a speedvac concentrator, the BO fragments were redissolved by the addition of water and lyophilized.

Molecular Characterization of BO Fragments-- Purified BO fragments were analyzed by nonreducing SDS-PAGE with a 5% stacking and a 15% resolving gel (40). For Western blot analysis, proteins were separated by SDS-PAGE and transferred onto nitrocellulose by semidry blotting (41). Immunoreactive proteins were detected by using BR114 (42) as primary antibody (dilution 1:2000) and peroxidase-conjugated rabbit anti-mouse IgG as secondary antibody (dilution 1:800). The protein bands were visualized with 4-chloro-1-naphthol.

Regeneration of BR-like Chromophores from Fragments-- Chromophores were regenerated by the addition of all-trans-retinal to two fragments present in equimolar amounts (10-18 µM) in a solution containing 1.5% DMPC, 0.5% CHAPS, 0.2% SDS, and 10 mM sodium phosphate, pH 6.0. The kinetics of chromophore formation were measured at 22 °C in the presence of a 3-fold molar excess of retinal. Under these conditions the regeneration rates were found to be independent of the retinal concentration. The absorbance increases were monitored at the _max of the dark-adapted chromophores, using an Uvicon model 930 or 932 spectrophotometer (Kontron, Eching, Germany). The traces of absorbance versus time were fit to the sum of two exponential processes, as described previously (43). Time constants and amplitudes were calculated using a nonlinear least squares algorithm (Sigmaplot, Jandel Scientific).

Spectral and Functional Characterization of Regenerated Chromophores-- The _max values of regenerated BR chromophores were measured at 4 °C after overnight dark adaptation followed by light adaptation for 5 min using a fiber optic illuminator (model A3200, Dolan-Jenner, Lawrence, MA) equipped with a 475-nm long-pass filter. Extinction coefficients of fragment complexes were determined by acid denaturation in the dark to give a chromophore with _max at 442 nm (44). The ratio of the absorption at the _max to the absorption at 442 nm after acidification to pH 2.2 was compared with that of wild-type eBR. The extinction coefficient of eBR was assumed to be 52,000 M¹ cm¹ (6). Spectrometric titrations of regenerated chromophores were carried out in steps of 0.1-0.3 pH units by adding microliter aliquots of 0.1-2 N H₂SO₄ or 0.1-5 M NaOH. Following complete equilibration (~3 min), pH readings and absorption spectra were recorded for each point. The amount of titrated pigment was determined from difference spectra as described (45). For the transition to a deprotonated Schiff base at alkaline pH, the absorbance increase at 365 nm was measured. Denaturation at acidic pH was assessed based on the formation of a 442-nm absorbing protonated Schiff base devoid of retinal-protein interactions. The absorbance change A was then plotted versus the pH, and the pK_a values as well as the number of protons involved in the transition were obtained from a previously reported three-parameter curve that was fitted to the data points (45).

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Preparation of BO Polypeptide Fragments-- Eight polypeptide fragments comprising two (AB, FG), three (AC, EG), four (AD, DG), or five (AE, CG) of the transmembrane regions of BR were prepared. The strategy used for the production of the fragments was based on previously developed procedures for the expression of wild-type BO and site-specific variants in E. coli (36, 37, 44) and their purification by solvent extraction and ion-exchange chromatography (39). Initially, all of the BO gene fragments were designed to encode the N-terminal eight amino acids and the C-terminal tail (residues 226-248) of BR (Fig. 1 and Table I). The coding sequence at the N terminus of BR is an important determinant of the expression level (47), whereas the negatively charged C-terminal segment is required for binding of the protein to the anion-exchange matrix.³

View larger version (39K): [in this window] [in a new window]   Fig. 1.   Secondary structure of bacteriorhodopsin. The alignment and length of the seven -helical transmembrane segments A-G are based on the structure of BR obtained by x-ray and electron diffraction (18, 19). Lys-216, the site of attachment of retinal, is marked by a circle. The N-terminal sequence, Met-Gln-Ala-Gln-, shown is that encoded by the synthetic BO gene. The N-terminal sequence of native BR from purple membrane is pyro-Glu-Ala-Gln.

Characterization of Purified BO Fragments-- The purity of the fragments was examined by SDS-PAGE and Western blot analysis. Fragments AB, AC, AD, and AE each appeared as one prominent band when visualized by Coomassie Blue staining (Fig. 2A). In addition, a faint band of lower mobility showed immunoreactivity, presumably representing a dimeric form (Fig. 2B). An increased proportion of high molecular weight bands was observed in polyacrylamide gels for fragments CG, DG, EG, and FG. By immunoblotting the majority of these bands could be assigned to oligomeric states of the fragments. Based on scanning gel densitometry the purity of each fragment was estimated to be 95%. In general, the apparent molecular weight of the monomer in nonreducing gels was in reasonable agreement with the value calculated from the sequence (Fig. 2A and Table I). For fragment AB a significant deviation from the expected mobility was noticed, possibly originating from a reduced extent of denaturation.

View larger version (67K): [in this window] [in a new window]   Fig. 2.   SDS-PAGE of purified polypeptide fragments of BR. Fragments were expressed in E. coli, purified by solvent extraction and ion-exchange chromatography, and complexed with SDS. eBO and fragments AB, AC, AD, AE, CG, DG, EG, and FG were analyzed in a 15% polyacrylamide gel. A, protein bands detected by Coomassie Blue staining (15 µg per lane); B, immunoreactive protein bands detected by the monoclonal antibody BR114, following transfer onto nitrocellulose membrane (4 µg per lane). The positions and molecular masses (in kDa) of marker proteins are indicated on the left (A, unstained; B, prestained protein ladder).

Regeneration of BR-like Chromophores from Complementary Fragments-- The refolding and assembly of BR-like structures from the different fragments was analyzed in DMPC/CHAPS/SDS micelles. In the case of native BO from purple membrane or eBO, this lipid-detergent system has previously been shown to effect essentially complete recovery of the native chromophore (6, 17, 45). In contrast to the intact protein, the regeneration efficiency of the four complexes assembled from a pair of complementary fragments was strongly dependent on the DMPC/CHAPS ratio. By using the standard conditions of 1% DMPC, 1% CHAPS previously established for eBO and site-specific variants, regeneration yields of 74, 19, 59, and 23% were obtained for AB·CG, AC·DG, AD·EG, and AE·FG, respectively. Conditions for optimal regeneration of the chromophores were examined in the concentration range of 0.5-2% DMPC and 0.5-2% CHAPS. Especially in the case of AC·DG and AE·FG a substantial improvement of the regeneration yields was observed upon lowering the CHAPS concentration and simultaneously increasing the DMPC concentration. To allow a comparison of the spectral properties of the different complexes a uniform concentration of 1.5% DMPC, 0.5% CHAPS was chosen for all subsequent experiments. At this lipid/detergent ratio the extent of regeneration was <5% of that maximally observed under any conditions tested and amounted to 86, 51, 76, and 74% for AB·CG, AC·DG, AD·EG, and AE·FG, respectively (Table II).

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Time course of chromophore formation for fragment complexes of BR. A, eBR and complexes assembled from complementary fragments; B, complexes assembled from overlapping fragments. Chromophores were regenerated at 22 °C by the addition of excess all-trans-retinal to equimolar amounts of two fragments in DMPC/CHAPS/SDS micelles. The absorbance increases were monitored at the max of the dark-adapted chromophores. Data points were collected at intervals of 0.1 and 1 min (for time points 10 min and >10 min, respectively) until the regeneration was completed to >85%. The traces were rescaled after determination of the final chromophore absorbance. Exponential time constants for the regeneration process are listed in Table III.

View this table: [in this window] [in a new window]   Table III Kinetics of chromophore formation for fragment complexes of BR Chromophores were regenerated at 22 °C by the addition of excess all-trans-retinal to equimolar amounts of two fragments in DMPC/CHAPS/SDS micelles. The absorbance increases were monitored at the max of the dark-adapted chromophores. The absorbance versus time traces (cf. Fig. 3) were fit to the sum of two exponential processes, as described under "Experimental Procedures."

Spectral and Functional Properties of Complexes Assembled from Complementary Fragments-- Absorption spectra of the dark- and light-adapted states of the chromophores formed by pairs of complementary fragments are shown in Fig. 4. Except for AC·DG (Fig. 4B), the fragment complexes displayed a normal pattern of dark-light adaptation, which in the case of eBR leads to a 9-nm red shift and an increased extinction, due to conversion of the 13-cis-/all-trans-chromophore into essentially 100% all-trans-retinal (44). Despite an incomplete light-adaptation reaction, the AC·DG chromophore was found to be stable toward extended periods of illumination. For the AB·CG, AC·DG, and AE·FG complexes, the absorption maxima of both the dark- and light-adapted states were shifted by 3 nm relative to those of the intact protein (Table II). A minor blue shift of 7 nm was observed for the corresponding absorption maxima of the AD·EG complex, which in addition displayed broadened chromophore bands (Table II). Upon reconstitution into soybean lipid vesicles, all of the chromophores regenerated from complementary fragments showed proton pumping activity at steady-state levels amounting to 38-71% of that measured for eBR (Table II).

View larger version (30K): [in this window] [in a new window]   Fig. 4.   Absorption spectra of the dark- and light-adapted states of complexes assembled from complementary fragments of BR. A, AB·CG; B, AC·DG; C, AD·EG; D, AE·FG. The chromophores were regenerated in DMPC/CHAPS/SDS micelles as described under "Experimental Procedures." The spectra were recorded after overnight dark adaptation (continuous lines) and after 5-min light adaptation (broken lines). The corresponding absorption maxima are listed in Table II.

Structural Stability of Individual Fragments and Complexes Assembled from Complementary Fragments-- The refolding capacity of each fragment was evaluated based on an analysis of the regeneration efficiency in the presence of varying concentrations of the complementary fragment. An increase in the regeneration yield was noticed for all of the fragments as the molar ratios were raised above 1:1 (Fig. 5). In the case of the AB-CG and AD-EG fragment pairs chromophore regeneration saturated at a molar ratio of about 2:1, thereby revealing strong interactions between the partners. In contrast, the regeneration curves reached a plateau at a molar ratio above 3:1 for the AC-DG fragment pair (Fig. 5B). Except for the AC/DG titration (closed symbols in Fig. 5B), the maximum regeneration yields attained were in the range of 91-99%. This shows that apart from DG, which displayed a maximum regeneration of 76%, all other fragments are capable of refolding to near completeness. By comparing the two sets of experiments in each panel, it is evident that limiting amounts of the N-terminal fragment (open symbols in Fig. 5 for molar ratios above 1:1 and closed symbols for molar ratios below 1:1) always result in higher regeneration yields than limiting amounts of their C-terminal counterpart.

View larger version (35K): [in this window] [in a new window]   Fig. 5.   Effect of the fragment ratio on the extent of chromophore formation. A, fragments AB and CG; B, fragments AC and DG; C, fragments AD and EG; D, fragments AE and FG. In the two sets of experiments in each panel the concentration of either the N-terminal (open symbols) or C-terminal fragment (closed symbols) was kept at a constant level of 15 µM. The concentration of the complementary fragment was varied in the range of 7.5-75 µM (0.5-5 ×). The chromophores were regenerated in DMPC/CHAPS/SDS micelles as described under "Experimental Procedures." The extent of regeneration was based on the constant fragment concentration of 15 µM and was calculated using the extinction coefficients listed in Table II.

View larger version (36K): [in this window] [in a new window]   Fig. 6.   Effect of fragment preincubation in DMPC/CHAPS/SDS micelles on the extent of chromophore formation. A, fragments AB, CG, and AB·CG; B, fragments AC, DG, and AC·DG; C, fragments AD, EG, and AD·EG; D, fragments AE, FG, and AE·FG. Individual fragments (open symbols for N-terminal fragments and closed symbols for C-terminal fragments) or combined complementary fragments (×) were preincubated in DMPC/CHAPS/SDS micelles for the indicated length of time. All-trans-retinal and, if appropriate, an equimolar amount of the complementary fragment were subsequently added. The concentration of each fragment in the regeneration mixture was 15 µM. The absorbance of the regenerated chromophore was measured after 20 h of incubation. The data points have been normalized with respect to the regeneration extent determined for complementary fragments without a preincubation period (Table II). The ordinate scale in A and C has been expanded to allow better presentation of the data.

View larger version (33K): [in this window] [in a new window]   Fig. 7.   Absorption difference spectra of the denaturation at acidic pH for complexes assembled from complementary fragments of BR. A, AB·CG; B, AC·DG; C, AD·EG. Difference spectra were obtained by subtracting a dark-adapted absorption spectrum recorded near pH 6 (pH 5.77 for AB·CG, pH 5.74 for AC·DG, and pH 5.82 for AD·EG) from spectra recorded at the indicated pH values. The spectra are labeled from top to bottom in order of decreasing chromophore absorbance near 550 nm. Spectrometric titrations were carried out in steps of 0.1-0.3 pH units, as described under "Experimental Procedures."

View larger version (25K): [in this window] [in a new window]   Fig. 8.   Effect of increasing SDS concentrations on the stability of complexes assembled from complementary fragments of BR. Following regeneration in DMPC/CHAPS/SDS micelles varying amounts of SDS were added to eBR and fragment complexes, as described under "Experimental Procedures." The samples were equilibrated in the dark for 20 h at room temperature, and the residual chromophore absorbance was determined.

Characterization of Chromophores Assembled from Overlapping Fragments-- Whereas chromophore formation was not observed for individual fragments or pairs of fragments that lacked one or more of the seven transmembrane segments of BR (i.e. AB·DG, AC·EG, AD·FG, AB·EG, AC·FG, and AB·FG), regeneration of a BR-like pigment was effected by all possible combinations of fragment pairs that contain one (AC·CG, AD·DG, AE·EG), two (AD·CG, AE·DG), or three (AE·CG) redundant helices. Assuming that a single retinal-binding pocket is formed by these pairs of overlapping fragments, the regeneration yields were calculated to be about 80%, except for the AD·DG complex (about 60%; Table II). Time courses of chromophore formation for complexes assembled from overlapping fragments are shown in Fig. 3B. As observed for complementary fragments (Fig. 3A), the regeneration processes were overall slower than that of the intact protein. Particularly slow regeneration kinetics were noticed for the AC·CG, AD·CG, and AE·CG complexes, which displayed a striking correspondence of the time constants and amplitudes for the two kinetic components (Table III). Similar time courses were also observed for regeneration of the AE·DG and AE·EG complexes, as well as for the AD·DG and AD·EG complexes (Fig. 3 and Table III). Absorption spectra of the chromophores formed by pairs of overlapping fragments are presented in Fig. 9. Compared with eBR, the absorption maxima of both the dark- and light-adapted states were generally blue-shifted by 6-9 nm (Table II). A larger blue shift of 26 nm was seen, however, for the corresponding absorption maxima of the AD·DG chromophore, which in addition displayed markedly broadened absorption bands (Table II and Fig. 9A). Upon reconstitution into lipid vesicles, light-dependent proton pumping was observed for all of these fragment complexes at steady-state levels that were reduced to 27-61% of the value determined for eBR (Table II).

View larger version (21K): [in this window] [in a new window]   Fig. 9.   Absorption spectra of the dark-adapted states of complexes assembled from overlapping fragments of BR. A, AC·CG, AD·CG, and AD·DG; B, AE·CG, AE·EG, and AE·DG. The spectra are labeled from top to bottom in order of decreasing chromophore absorbance. The chromophores were regenerated in DMPC/CHAPS/SDS micelles using equimolar fragment concentrations. The chromophore absorbance therefore reflects the extent of regeneration of the different complexes. The corresponding regeneration yields and absorption maxima are listed in Table II.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

In the present study an efficient strategy has been developed for the preparation of sets of complementary polypeptide fragments of the heptahelical membrane protein BR. The procedure is based on previous methods that have been used to produce wild-type BO and site-specific variants in a folding competent state (36, 37, 39, 44). To apply these protocols to the preparation of fragments, the N- and C-terminal segments of the wild-type protein were initially introduced into each of the eight constructs. The N-terminal residues have been shown to contribute to the stability of the expressed BO, possibly by enhancing its integration into the E. coli membrane (47). Expression of the different constructs occurred at levels comparable to that of the intact protein, except for the DG and EG fragments. In the latter cases internal sequences may interfere with the membrane insertion process, thereby causing extensive protein degradation. By positioning an initiator methionine directly in front of the coding sequence of the respective N-terminal helix, the DG and EG fragments could nevertheless be produced at levels similar to the other polypeptides. Sequence analysis of the purified DG and EG proteins revealed altered processing compared with fragments that contain the N-terminal residues of BO, suggesting different membrane orientations or localizations of the polypeptides in E. coli.

The purification of the fragments was facilitated by inserting the negatively charged C-terminal tail of BR (residues 226-248; Fig. 1) into each of the constructs. This segment was chosen for the following reasons. (i) It mediates binding to the anion-exchange matrix.³ (ii) Its C-terminal location prevents the purification of molecules that contain truncations in this region. (iii) It is recognized by the monoclonal antibody BR114 (42), thereby enabling monitoring of the expression and purification of all of the fragments. (iv) It increases the polarity of the generally extremely hydrophobic polypeptides. (v) It does not influence the spectral or functional properties of the protein, as shown by the phenotype of C-terminally truncated BR generated by proteolysis (51) or recombinant expression.³ Thus, removal of this endogenous tag, which in the case of membrane proteins is notoriously difficult to perform (52), can be avoided.

Based on the present data it is anticipated that the purification procedure is generally applicable to the isolation of polypeptides comprising transmembrane regions. The BR fragments display significant differences in their average hydrophobicity (53), yet they could all be purified in acceptable yields. A particular advantage of the method is that secondary structures are preserved in the solvent system (9, 54), thereby allowing the isolation of proteins in a partially folded state. Analyses of the regeneration efficiencies in the presence of excess amounts of the counterpart demonstrate that all of the BR fragments are capable of refolding to >90% (Fig. 5). The lower refolding capacity of 76% observed for DG can be attributed to the existence of a subpopulation amounting to 18% that is proteolytically degraded (Table I). It is expected that a lack of six residues (approximately two turns) at the beginning of helix D will preclude the proper association of complementary fragments or chromophore formation.

Several observations indicate that the C-terminal part of BR contains regions that are prone to denaturation. Chromophore regenerations with varying fragment ratios revealed a generally small but consistent reduction of the refolding efficiency of C-terminal fragments compared with their respective N-terminal counterpart (Fig. 5). In polyacrylamide gels an increased tendency to aggregation was noticed for DG and FG and to a lesser degree for CG and EG (Fig. 2). Denaturation of individual fragments was directly assessed based on the capability to interact with the counterpart and bind retinal following preincubation in DMPC/CHAPS/SDS micelles. Whereas fragments AB, AC, AD, AE, and CG were found to be largely stable, a significant extent of denaturation was observed for the DG, EG, and FG polypeptides (Fig. 6). This indicates that a structural element crucial for proper folding is contained within the last two transmembrane segments of BR, in agreement with a previous study that used a corresponding proteolytic fragment (9). The increased susceptibility to denaturation for DG relative to EG suggests that a structurally labile region is also present within helix D. The structural integrity of helix D is apparently dependent on the preceding loop and the interaction with helix C, as shown by the comparable stabilities of fragments AC and AD (Fig. 6). The DG, EG, and FG polypeptides could be protected to a significant extent from denaturation by incubation with the complementary component in the absence of retinal. This effect was particularly pronounced for the AD·EG complex (Fig. 6C), reflecting strong intermolecular interactions between the counterparts. The association of complementary fragments in DMPC/CHAPS/SDS micelles provides evidence that in the absence of retinal the complexes acquire conformations similar to those they possess in the regenerated state. This conclusion is also supported by corresponding circular dichroism measurements of fragment complexes⁴ (8) and intact BR (6, 55).

The spectral properties of the chromophores regenerated from pairs of complementary fragments revealed close correspondence with those of the wild-type protein (Fig. 4 and Table II). This shows that the retinal-binding pocket, which is sensitive to small structural alterations within the membrane-embedded regions (12, 19), is properly assembled in all of these complexes. In addition these chromophores displayed light-induced proton pumping, indicating that they can undergo the conformational changes of the photocycle. The generally lower proton transport activity of the fragment complexes compared with intact BR can be caused by several factors. Reconstitution experiments indicate that, unlike wild-type BR, the chromophores of fragment complexes react very sensitively to the lipid environment (see "Results"; Refs. 7 and 10). The intermolecular interactions within the ternary complexes could be weakened in the asolectin vesicles compared with the DMPC/CHAPS/SDS micelles, leading to destabilization of the chromophores upon illumination. Alternatively, the essentially uniform orientation observed for intact BR molecules in asolectin vesicles (56, 57) could be altered in the case of fragment complexes, especially if they comprise overlapping sections of the protein.

The present results, along with previous studies that used complexes assembled from proteolytic fragments and synthetic peptides (7-11), demonstrate that the covalent connections in each of the three cytoplasmic and three extracellular surface loops of BR (Fig. 1) are dispensable for a correct association of the helices. This indicates that the secondary or tertiary structures of these loops are not required to specify the position and orientation of the helices to which they are connected. This conclusion has been supported for several of these loops by insertion and deletion mutagenesis, showing that their structural integrity is not a prerequisite for chromophore formation and function (13, 15).

Although the covalent connections in the four surface loops studied are not essential for the folding of BR, they do contribute to the stability of the protein. Incubations in the presence of SDS revealed that denaturation of the AB·CG, AC·DG, AD·EG, and AE·FG complexes occurs at significantly lower SDS concentrations, compared with intact BR (Fig. 8). Furthermore, destabilization of the helical bundles was manifested by the decreased pK_a values of PSB deprotonation at alkaline pH and the increased pK_a values of denaturation at acidic pH observed for all of the fragment complexes (Table IV). The reduced formation of a red-shifted species noticed upon acidification of the AC·DG, AD·EG, and AE·FG chromophores (Fig. 7) identifies minor structural alterations in the assembled complexes affecting the interaction between the PSB at Lys-216 and Asp-85, which becomes protonated in the purple to blue transition (58). Notably, this perturbed interaction between groups located in helix G and C, respectively, was observed for all complexes that contained these two helices in separate fragments, whereas for the AB·CG complex, where these helices are present within a single fragment, a spectral transition analogous to that of intact BR was seen (Fig. 7A). Compared with the other complementary fragment complexes, a significantly larger destabilization was measured for the AC·DG chromophore, as both the pK_a of PSB deprotonation and the pK_a of acid denaturation deviated by >1.2 pH units from the wild-type values (Table IV). This suggests that the contribution of the short C-D loop to the stability of BR may be greater than that of other surface loops. Calorimetric measurements of BR complexes assembled from proteolytic fragments and synthetic peptides have shown that discontinuities in the A-B or B-C loops decrease both the temperature and the enthalpy of denaturation (35). Thus, these interhelical connections stabilize the entire structure to some extent and do not simply maintain helices in close proximity.

A striking observation is that the heptahelical bundle structure of BR can be regenerated by all pairs of fragments containing overlapping transmembrane segments. Thus, the presence of one (AC·CG, AD·DG, AE·EG), two (AD·CG, AE·DG), or three (AE·CG) redundant helices did not prevent the assembly process. For the AE·CG complex this result is consistent with a previous study that used corresponding fragments produced by proteolysis of purple membrane (59). The spectral properties and stabilities of the resulting chromophores showed in general close agreement with those of the complementary fragment complexes (Tables II and IV), indicating that a retinal-binding pocket similar to native BR is formed in each case. Based on a comparative analysis of the spectral characteristics of the complexes, evidence was obtained regarding the origin of the seven helices that assemble the chromophore-binding pocket. For the AC·CG, AD·CG, and AE·CG complexes a precise correspondence of the chromophore properties (Table II), regeneration kinetics (Fig. 3B and Table III), and pK_a values of acid denaturation and PSB deprotonation (Table IV) was noticed, suggesting an identical arrangement of their retinal-binding pocket. Notably, their spectral transitions from purple to blue involved normal amplitude changes, which was observed solely in the case of the AB·CG complementary fragment complex (see above). Apparently the helical bundles surrounding the chromophores of the AC·CG, AD·CG, and AE·CG complexes contain a discontinuity between helices B and C, indicating that helices C, C-D, and C-D-E, respectively, of their N-terminal fragment are displaced. Thus, the retinal-binding pocket of these three complexes is evidently assembled from helices A and B of the respective N-terminal fragment and the C-terminal CG fragment. For the AD·DG complex the kinetics of chromophore formation (Fig. 3 and Table III) and the pH-induced spectral transitions were very similar to those observed for the AD·EG complex. In addition, the stability of the AD·DG chromophore, as determined by pH titration (Table IV), was markedly increased compared with AC·DG, suggesting that helix D of the DG fragment is displaced, and the bundle structure is analogous to the AD·EG complex. The AE·DG and AE·EG complexes displayed close correspondence in their regeneration kinetics (Fig. 3B and Table III), chromophore properties (Table II), and stabilities toward SDS, thereby implying an identical assembly of their retinal-binding pocket. The spectral transitions of the AE·DG and AE·EG chromophores at acidic pH involved an intermediate with _max at 459 nm. This species was also observed upon acidification of the AD·EG chromophore (Fig. 7C), whereas in the case of the AE·FG complex it was not formed. Based on these results is is likely that the AE·DG and AE·EG complexes also adopt a chromophore-binding pocket analogous to the AD·EG complex.

The derived assemblies of the heptahelical binding pocket from overlapping pairs of fragments can be explained by the following arguments. Strong helix-helix interactions, which represent a driving force for transmembrane -helix association (26), are particularly evident for the AB-CG and AD-EG fragment pairs (Figs. 5 and 6). This suggests that the connecting B-C and D-E loops, respectively, make minor contributions to the overall stability of BR, compared with the other surface loops. If in addition the loops are sufficiently large, they can provide the flexibility required to accommodate redundant polypeptide segments. On the other hand, the interactions between the AC-DG and AE-FG fragment pairs are apparently weaker and depend to a significant extent on connectivity between the respective helices. A lack of continuity in the C-D and E-F loops may destabilize the terminal helices, thereby reducing their affinity in the assembly process, relative to the same helix contained within a continuous polypeptide segment. The combined data indicate that a lack of connectivity in the C-D and E-F surface loops reduces the specificity of the helix assembly of BR.

The availability of complementary sets of BR fragments that are capable of regenerating the native structure with high efficiency provides an opportunity to analyze the thermodynamics and kinetics of in vitro folding and assembly of this prototypic membrane receptor using independent structural domains.