Primary structure of C-terminal functional sites in ovine rhodopsin

Detecting molecular interactions that stabilize native bovine rhodopsin

Using single-molecule force spectroscopy we probed molecular interactions within native bovine rhodopsin and discovered structural segments of well-defined mechanical stability. Highly conserved residues among G protein-coupled receptors were located at the interior of individual structural segments, suggesting a dual role for these segments in rhodopsin. Firstly, structural segments stabilize secondary structure elements of the native protein, and secondly, they position and hold the highly conserved residues at functionally important environments. Two main classes of force curves were observed. One class corresponded to the unfolding of rhodopsin with the highly conserved Cys110-Cys187 disulfide bond remaining intact and the other class corresponded to the unfolding of the entire rhodopsin polypeptide chain. In the absence of the Cys110-Cys187 bond, the nature of certain molecular interactions within folded rhodopsin was altered. These changes highlight the structural importance of this disulfide bond and may form the basis of dysfunctions associated with its absence.

Rhodopsin and phototransduction

Recent studies on rhodopsin structure and function are reviewed and the properties of vertebrate as well as invertebrate rhodopsin described. Open issues such as the 'red shift' of the absorbance spectra are emphasized in the light of the present model of the retinal-binding pocket. The processes that restore the rhodopsin content in photoreceptors are also presented with a comparison between vertebrate and invertebrate visual systems. The central role of rhodopsin in the phototransduction cascade becomes evident by examining the main reports on light-activated conformational changes of rhodopsin and its interaction with transducin. Shut-off mechanisms are considered by reporting the studies on the sites of rhodopsin phosphorylation and arrestin binding. Furthermore, recent findings on the energetics of phototransduction point out that the ATP needed for photoreception in vertebrates is synthesized in the outer segments where phototransduction events take place.

Site-directed mutagenesis of conserved charged residues in the helical region of the human C5a receptor. Arg2O6 determines high-affinity binding sites of C5a receptor

The human C5a receptor (C5aR) belongs to the family of G-protein-coupled receptors with seven transmembrane helices. This part of the molecule is thought to contain part of the ligand-binding pocket, specifically to bind the C-terminal Arg of human C5a. Guided by sequence similarity and molecular modelling studies, several residues including polar (Asn119, Thr168, Gln259) as well as all conserved charged amino acids in the upper transmembrane region of the C5aR (Asp37, Asp82, Arg175, Arg2O6, Asp282) were exchanged by site-directed mutagenesis. Receptor mutants were transiently expressed in COS cells and analyzed for altered binding behaviour and/or localization at the cell surface by immunofluorescence. For all residues, suitable mutants could be found that exhibited wild-type affinity towards the ligand, providing evidence against a major contribution of these residues to high-affinity ligand binding. Some mutants, however, exhibited a complete (Asp282-->Ala) or partial loss of ligand-binding capacity (Arg175-->Ala, Arg2O6-->Gln) despite adequate expression levels on the cell surface. This phenotype was further analyzed in the [Gln2O6]C5aR mutant: quantitative flow cytometric analysis of epitope-tagged receptor derivatives in 293 cells confirmed an equal level of wild-type and mutant C5aR on the cell surface. Competitive binding curves revealed the presence of only a small population (<10%) of high-affinity sites (Kd approximately 2 nM), which was functionally active at 20 nM in the heterologous Xenopus oocyte expression system after coexpression of G alpha-16. The number of high-affinity sites of wild-type and [Gln2O6]C5aR in 293 cells could be up-regulated by coexpression of Gi alpha-2 and down-regulated by GTP[gamma S]-mediated uncoupling of the G-protein receptor interaction in membrane preparations. These findings are compatible with a model in which the Arg2O6 residue located in the upper third of transmembrane helix V determines high-affinity binding in the human C5aR by affecting the intracellular G-protein coupling.

Evidence that the extracellular N-terminal domain of C5aR contains amino-acid residues crucial for C5a binding

The human C5a anaphylatoxin is a cationic 74 amino-acid long glycopeptide which derives from proteolysis of the fifth component of complement. It interacts with high affinity with a receptor that belongs to the G protein-coupled receptor superfamily. Several studies have previously suggested that multiple contact points between C5a and the receptor are required to achieve high-affinity interaction. However, at the receptor level little is known about the sites of interaction with C5a. We have investigated by in vitro mutagenesis whether the N-terminal extracellular sequence of the C5a receptor, which is rich in aspartic acid residues, could play some role in C5a binding. Conversion of Asp10 into asparagine did not impair the level of expression at the plasma membrane, nor did it alter the affinity for C5a. However, we consistently observed a discrepancy between an apparent high level of surface expression and a weak capacity to bind C5a with high affinity, suggesting that many receptor molecules, although present on the cell surface, might be misfolded and unable to bind C5a. Replacement of Pro9 by an isoleucine had little effect, if any, on either the affinity or the C5a-binding capacity, whereas the conversion of Pro36 into leucine dramatically reduced the expression of high-affinity receptor at the cell surface. N-glycosylation of human C5a receptor was found to be dispensable for the function of the receptor.

Phosphorylation of solubilised dark-adapted rhodopsin. Insights into the activation of rhodopsin kinase

A protocol for the separation of phosphorhodopsin from phospho-opsin has been developed. The method takes advantage of the finding that, while 0.5% N,N-dimethyldodecylamine-N-oxide completely solubilises membrane-embedded phosphorhodopsin, at this concentration of detergent, phospho-opsin is only sparingly soluble. Phosphorhodopsin solubilised in this manner may be freed from contaminant phospho-opsin by chromatography on hydroxyapatite. Using this method, the rhodopsin-kinase-catalysed phosphorylation of photoexcited rhodopsin and native rhodopsin was studied in rod outer-segment membranes at different levels of bleaching. Prior to analysis of the phosphorylation mixture, the phosphorylated form of photoexcited rhodopsin was converted into phospho-opsin by treatment with NH2OH. It was found that, while at a 5% bleach level the amount of phosphorhodopsin produced was 15% that of phospho-opsin, at 60% bleaching the phosphorhodopsin was less than 1% of phospho-opsin. The phosphorylation reaction under different bleaching conditions was also studied in a completely soluble system (using 2% dodecyl maltoside) and the pattern of phosphate incorporation into rhodopsin versus opsin was identical to that in the membrane system. We have previously proposed that rhodopsin kinase normally exists in an inactive form and is only activated following interaction with photoexcited rhodopsin. The present work strengthens this conclusion and also shows that, following activation, the kinase preferentially phosphorylates photoexcited rhodopsin but can also act upon unbleached rhodopsin. Two possible mechanisms for the activation of the kinase are considered. From the distribution of phosphorhodopsin and phospho-opsin at different bleaching levels, the relative rates of the phosphorylation of photoexcited rhodopsin (kR*) and rhodopsin (kR) were calculated. kR*/kR values for the membrane system of 71 +/- 20 and, for the solubilised system, of 80 +/- 19 were obtained. The algebraic equation used to obtain these values highlights the fact that the ratio of the concentrations of the two substrates, photoexcited rhodopsin and rhodopsin, in a sample, determines the final distribution of phosphate between bleached and unbleached rhodopsin. This conclusion may contribute to the understanding of 'high-gain' phosphorylation observed previously.

Topography of opsin within disk and plasma membranes revealed by a rapid-freeze deep-etch technique

Rod outer segments in fresh rat retinas were examined by a rapid-freeze, deep-etch technique to explore how membrane proteins are organized at the macromolecular level. Cross-fractures revealed that intradiscal membranes are adherent to each other except at the rim. When an isolated fresh retina was incubated in a hypotonic solution for a few minutes, the interdiscal space was expanded and the cytoplasmic surface of the disk membrane was found to be covered with protrusions except at the rim. A few particles were scattered among the protrusions and were attached to the cytoplasmic surface. Since the distribution density of the cytoplasmic surface protrusions was similar to that of the P-face particles, which are known to reflect opsins, the protrusions were considered to be portions of opsins extending into the cytoplasm. The intradiscal surfaces in chemically-fixed retinas were rather smooth and were labelled with anti-opsin antibodies and wheat germ agglutinin. The true surfaces of the plasma membrane were found to be similar in fine structure to those of the disk. A model of the macromolecular organization of rod outer segments is proposed on the basis of these observations. The model shows apposed opsins within a disk membrane adhering to one another except at the rim. These opsins, as well as those in the plasma membrane, are minimally exposed to the extracellular surface, but protrude deeply into the cytoplasm.

In vitro mutagenesis and the search for structure-function relationships among G protein-coupled receptors

Images

Molecular cloning and primary structure of squid (Loligo forbesi) rhodopsin, a phospholipase C-directed G-protein-linked receptor

The sequence of squid (Loligo forbesi) rhodopsin was determined by protein and cDNA sequencing. The protein has close similarity to octopus rhodopsin, having an N-terminal region (residues 1-340) which resembles other guanine-nucleotide-binding protein (G-protein)-linked receptors and a repetitive proline-rich C-terminus (residues 340-452). Comparison of the sequence of squid rhodopsin with those of other members of the G-protein-linked receptor superfamily reveals features which we predict to have both structural and functional importance.

Phototransduction: different mechanisms in vertebrates and invertebrates

The photoreceptor cells of invertebrate animals differ from those of vertebrates in morphology and physiology. Our present knowledge of the different structures and transduction mechanisms of the two animal groups is described. In invertebrates, rhodopsin is converted by light into a meta-rhodopsin which is thermally stable and is usually re-isomerized by light. In contrast, photoisomerization in vertebrates leads to dissociation of the chromophore from opsin, and a metabolic process is necessary to regenerate rhodopsin. The electrical signals of visual excitation have opposite character in vertebrates and invertebrates: the vertebrate photoreceptor cell is hyperpolarized because of a decrease in conductance and invertebrate photoreceptors are depolarized owing to an increase in conductance. Single-photon-evoked excitatory events, which are believed to be a result of concerted action (the opening in invertebrates and the closing in vertebrates) of many light-modulated cation channels, are very different in terms of size and time course of photoreceptors for invertebrates and vertebrates. In invertebrates, the single-photon events (bumps) produced under identical conditions vary greatly in delay (latency), time course and size. The multiphoton response to brighter stimuli is several times as long as a response evoked by a single photon. The single-photon response of vertebrates has a standard size, a standard latency and a standard time course, all three parameters showing relatively small variations. Responses to flashes containing several photons have a shape and time scale that are similar to the single-photon-evoked events, varying only by an amplitude scaling factor, but not in latency and time course. In both vertebrate and invertebrate photoreceptors the single-photon-evoked events become smaller (in size) and faster owing to light adaptation. Calcium is mainly involved in these adaptation phenomena. All light adaptation in vertebrates is primarily, or perhaps exclusively, attributable to calcium feedback. In invertebrates, cyclic AMP (cAMP) is apparently another controller of sensitivity in dark adaptation. The interaction of photoexcited rhodopsin with a G-protein is similar in both vertebrate and invertebrate photoreceptors. However, these G-proteins activate different photoreceptor enzymes (phosphodiesterases): phospholipase C in invertebrates and cGMP phosphodiesterase in vertebrates. In the photoreceptors of vertebrates light leads to a rapid hydrolysis of cGMP which results in closing of cation channels. At present, the identity of the internal terminal messenger in invertebrate photoreceptors is still unsolved.(ABSTRACT TRUNCATED AT 400 WORDS)

An analysis of the periodicity of conserved residues in sequence alignments of G-protein coupled receptors. Implications for the three-dimensional structure

Twenty-three sequences from the family of G-protein coupled receptors have been aligned according to the 'historical alignment' procedure of Feng and Doolittle. Fourier transform analysis of this reveals that parts of five of the seven putative membrane-spanning regions exhibit a periodicity of conserved/nonconserved residues which is compatible with the periodicity of the alpha-helix. This would place the conserved residues on one side of the helix, which may face the inside of the proposed seven membered helical bundle.

Assignment of groups responsible for the "opsin shift" and light absorptions of rhodopsin and red, green, and blue iodopsins (cone pigments)

A modified structural model of rhodopsin is presented. Seven (alpha-helical) segments of 24 largely hydrophobic amino acid residues are assembled with exobilayer connecting strands into an aligned set, using the sequences of human red, green, and blue iodopsins (cone pigments) and human and bovine rod rhodopsins. (Aligned set numbering is used in this article). The inner region of the heptahelical hydrophobic domain includes one His-Glu (Asp) ion pair (red, green rod) near the retinylidene moiety in addition to an iminium ion Asp-99 pair. The negative charges posited in the "point-charge model" to cause the shift of the retinylidene iminium ion light absorption to longer wavelengths in the protein ("opsin shift") are Asp-99 (red, green rod), Glu-102 (red, green), and Glu-138 (rod). Blue iodopsin lacks both an ion pair and a counter charge to the iminium ion in the inner region, a fact that explains its absorption relative to rod rhodopsin. The spectroscopic difference between rod rhodopsin and the red/green iodopsins is due to the influence of Glu-102 in the latter. The red-green difference is due to the net effect of seven OH groups around the chromophore, all such groups being found within one helix turn of the retinylidene location. The tryptophan, which rotates as the retinylidene group isomerizes, may be Trp-142 or Trp-177. The geometric change (the rhodopsin "photoswitch") resulting from cis-trans isomerization in the first excited electronic state (S1), ultimately leads to RX (photoactivated rhodopsin, metarhodopsin II) and changes the activity of exobilayer groups, possibly causing dissociation of Lys-83 and Arg-85 from the carboxylate groups at positions 263 and 265.

Homology between the pyrazine-binding protein from nasal mucosa and major urinary proteins

Sequence analysis of the pyrazine-binding protein from bovine olfactory mucosa reveals marked homology with a family of proteins of unknown function found in the urine of the adult male mouse and rat. In view of the dramatic biological responses to odorants transmitted in male rodent urines, it is proposed that these proteins play important roles in some aspects of odor transmission and reception.

Cloning and sequence analysis of the human brain beta-adrenergic receptor. Evolutionary relationship to rodent and avian beta-receptors and porcine muscarinic receptors

Two cDNA clones, lambda-CLFV-108 and lambda-CLFV-119, encoding for the beta-adrenergic receptor, have been isolated from a human brain stem cDNA library. One human genomic clone, LCV-517 (20 kb), was characterized by restriction mapping and partial sequencing. The human brain beta-receptor consists of 413 amino acids with a calculated Mr of 46480. The gene contains three potential glucocorticoid receptor-binding sites. The beta-receptor expressed in human brain was homology with rodent (88%) and avian (52%) beta-receptors and with porcine muscarinic cholinergic receptors (31%), supporting our proposal [(1984) Proc. Natl. Acad. Sci. USA 81, 272 276] that adrenergic and muscarinic cholinergic receptors are structurally related. This represents the first cloning of a neurotransmitter receptor gene from human brain.

The opsin family of proteins

Images

The human erythrocyte anion-transport protein. Further amino acid sequence from the integral membrane domain homologous with the murine protein

Sequences from the human erythrocyte anion-transport protein homologous with residues 417-449 and 794-813 of the murine erythrocyte anion-transport protein have been determined. The former sequence includes the putative transmembrane helix closest to the N-terminus of the protein. The latter sequence traverses almost all of the lipid bilayer and is located towards the C-terminus of the protein. Sites have been identified by alignment with the murine sequence in the integral membrane domain that are accessible to proteolytic enzymes. Sequences from the integral membrane domain of the erythrocyte anion-transport protein are highly conserved.

Modification of ovine opsin with the photosensitive hydrophobic probe 1-azido-4-[125I]iodobenzene. Labelling of the chromophore-attachment domain

The hydrophobic photosensitive probe 1-azido-4-[125I]iodobenzene (AIB) partitioned preferentially into photoreceptor disc membranes and, upon u.v. irradiation, became covalently bound to opsin and phospholipid. The labelling of both protein and phospholipid was linearly related to AIB concentration. The amount of probe incorporated into protein was not significantly different when membranes were irradiated at -100 degrees, 4 degrees or 25 degrees C, but irreversible aggregation of monomeric opsin was dramatically reduced by performing the photolysis at -100 degrees C. Labelling of opsin after irradiation at -100 degrees or 4 degrees was not significantly reduced by the presence of lysine in the aqueous buffer, indicating that significant amounts of reactive species did not enter the aqueous phase. The incorporation into phospholipid, unlike that into opsin, decreased as the temperature of irradiation increased. Some labelling of opsin occurred on incubation with pre-photoactivated AIB, indicating that reaction may also occur with reactive species of longer lifetimes than the nitrene. Proteolysis of labelled opsin with Staphylococcus aureus V8 proteinase yielded two radiolabelled membrane-bound fragments. The location of the modified sites (cysteine, tryptophan, tyrosine, lysine and histidine residues: all nucleophiles) in the smaller fragment was entirely consistent with putative models for the protein derived from other studies.

Differential immunogold-dextran labeling of bovine and frog rod and cone cells using monoclonal antibodies against bovine rhodopsin

Eleven monoclonal antibodies against different segments of bovine rhodopsin were used with immunogold-dextran markers to label Lowicryl thin sections of bovine and frog retinal photoreceptor cells for visualization by transmission electron microscopy. Antibodies against the C-terminus, F1-F2 loop and N-terminus of rhodopsin were all observed to label bovine rod outer segments (ROS) densely, but to label rod inner segments (RIS) only sparsely. Most antibodies bound 200-600 gold particles per micron2 in the ROS, 10-60 gold particles per micron2 in the RIS and 5-20 particles per micron2 on the Lowicryl resin. One antibody against the N-terminus and one antibody against the C-terminus resulted in the binding of over 1000 particles per micron2 in bovine ROS. Cone outer segments (COS) were labeled with only one antibody, rho 3D6, having a specificity for the 1'-4' C-terminus of bovine rhodopsin. Ninety per cent of the COS were observed to be labeled with this antibody. Immunogold-dextran labeling was also used to study the cross-reactivity of these antibodies to rhodopsin in red and green frog ROS and COS. Monoclonal antibodies directed against sites along the F1-F2 loop, and the N-terminus labeled red frog ROS densely, but did not label either green ROS or COS. Three C-terminal specific antibodies against binding sites along the 1'-8' segment labeled both green and red ROS, but a higher extent of labeling was observed on the green ROS. Antibody rho 3D6, which bound to bovine COS, also labeled frog COS. These results indicate that the F1-F2 loop and segments along to the N-terminus and the C-terminus of bovine rhodopsin show a high degree of homology with corresponding regions of frog rhodopsin from red ROS; the C-terminal 1'-8' segment of bovine rhodopsin is closely related to the corresponding segment of frog rhodopsin from green ROS; and the 1'-4' C-terminus of bovine rhodopsin is highly homologous to segments of opsin from most bovine and frog COS. The labeling of frog red ROS in relation to multiple forms of rhodopsin observed by SDS-gel electrophoresis is discussed.

The Drosophila ninaE gene encodes an opsin

The Drosophila ninaE gene was isolated by a multistep protocol on the basis of its homology to bovine opsin cDNA. The gene encodes the major visual pigment protein (opsin) contained in Drosophila photoreceptor cells R1-R6. The coding sequence is interrupted by four short introns. The positions of three introns are conserved with respect to positions in mammalian opsin genes. The nucleotide sequence has intermittent regions of homology to bovine opsin coding sequences. The deduced amino acid sequence reveals significant homology to vertebrate opsins; there is strong conservation of the retinal binding site and two other regions. The predicted protein secondary structure strikingly resembles that of mammalian opsins. We conclude the Drosophila and vertebrate opsin genes are derived from a common ancestor.

Phosphorylation of ovine rhodopsin. Identification of the phosphorylated sites

Light-dependent phosphorylation of sheep opsin was obtained in purified discs to which was added a partially purified preparation of rhodopsin kinase. A maximum ratio of 1.8 mol of phosphate/mol of rhodopsin bleached was obtained. Perturbing the lipid bilayer did not alter the phosphorylation ratio. Dephosphorylation in both segments and discs was only achieved when the supernatant fraction from a retina homogenate was added. Complete dephosphorylation required the presence of the detergent dodecyltrimethylammonium bromide in the incubation medium. Treatment of phosphorylated disc membranes with Staphylococcal aureus V8 proteinase generated two membrane-bound fragments, only one of which (V8-S, Mr 12 000) was labelled, together with a soluble seven-residue peptide that contained [32P]phosphoserine. Peptide sequencing, together with subdigestion procedures, localized the phosphorylation sites to serine residues at positions 334, 338 and 343 in the whole sequence and threonine residues at positions 335 and 336.

Labelling of the cytoplasmic domains of ovine rhodopsin with hydrophilic chemical probes

The disposition of polypeptide chain of ovine rhodopsin in the photoreceptor disc membrane was investigated by using two hydrophilic reagents, 3,5-di-[125I]iodo-4-diazobenzenesulphonate [( 125I]DDISA) and [14C]succinic anhydride. Both reagents were used to modify rhodopsin in intact disc membranes under conditions where no loss of A500 occurred. Reaction of [125I]DDISA with rhodopsin approached completion after 30 min. Binding was saturated at a 75-fold molar excess of reagent, which gave binding ratios of up to 2 mol/mol of rhodopsin. Proteolysis of rhodopsin, using Staphylococcus aureus V8 proteinase, yielded two membrane-bound fragments, both of which contained bound radioactive probe. Subsequent CNBr cleavage of these fragments produced five radiolabelled peptides which corresponded to the C-terminal region and cytoplasmic loops of rhodopsin. Similar studies with [14C]-succinic anhydride also gave binding ratios of up to 2 mol/mol of rhodopsin. Sequencing of the [14C]succinylated peptides identified the location of the reactive sites as lysine residues 66, 67, 141, 245, 248, 311, 325 and 339 in the polypeptide chain. Non-permeability of both probes was demonstrated by the absence of any radioactivity associated with the intradiscal N-terminal glycopeptide. Sonication of membranes in the presence of [125I]DDISA led to the incorporation of label in this peptide.

Sequence variability in the retinal-attachment domain of mammalian rhodopsins

Ovine rhodopsin was regenerated with 11-cis-[15-3H]retinal and cleaved in situ by Staphylococcus aureus V8 proteinase to give two membrane-bound fragments of Mr 27 000 (V8-L) and 12 000 (V8-S). After purification of the proteolysed complex by affinity chromatography with concanavalin A-Sepharose 4B, [3H]retinal was covalently linked to the protein by reduction with borohydride. The purified [3H]-retinyl V8-S fragment was cleaved with CNBr and trifluoroacetic acid, the resulting peptides resolved by gel filtration and the [3H]retinyl peptide sequenced. The protocol developed for the isolation and sequencing of this region of the ovine protein was applied directly, and reproducibly, to bleached and unregenerated porcine and equine opsins. Comparisons of the primary structures of the fragments reveals marked variation in the sequence immediately after the lysine residue shown in the ovine protein to be the attachment point for the aldehyde group of the chromophore. Mutable positions are localized in regions previously predicted as adopting nonregular or distorted conformations and hint at structural arrangements that may provide a better understanding of the spectral and functional properties of the visual pigment.

The structure of mammalian rod opsins

Ovine rhodopsin is organised in disc membranes as a monomer. The determination of its amino acid sequence has permitted the utilisation of structure prediction programmes which indicate the probable disposition of the polypeptide chain in the bilayer. This putative model is consistent with labelling data using the chemical probes, [14C]succinic anhydride, [125I]diazodiido sulphanilic acid and [125I]iodophenyl azide, and with the cleavage points for several proteases. More surprisingly the predicted structure points to the occurrence of breaks/distortions in the transmembrane helical segments. These distorted regions may be of primary functional importance to the protein and at least one is associated with the attachment point of the chromophore. This particular part of the structure is also identified as a "mutational hot spot", for bovine, equine, ovine and porcine opsins exhibit different sequences (but conserved molecular volumes) in the four residues following the retinyllysine. In an otherwise highly conserved protein with no obvious functional differences between the four species, the high substitution rate in this region is unexplained.

Isolation and characterization of the CNBr peptides from the proteolytically derived N-terminal fragment of ovine opsin

Ovine rhodopsin may be cleaved in situ by Staphylococcus aureus V8 proteinase into two membrane-bound fragments designated V8-L (27 000 mol.wt.) and V8-S (12 000 mol.wt.). After purification of the proteolysed complex by affinity chromatography in detergent using concanavalin A immobilized on Sepharose 4B, the two polypeptide fragments may be separated by gel-permeation chromatography on Sephadex LH-60. Digestion of the N-terminal-derived V8-L fragment with CNBr in 70% (v/v) trifluoroacetic acid resulted in a peptide mixture that could be fractionated by procedures involving gel-permeation chromatography in organic and aqueous solvents and the use of differential solubility. The complete or partial sequences of all ten peptides are reported.

Structural studies on membrane-bound bovine rhodopsin

Our previous work has shown that the treatment of bovine rhodopsin with the proteolytic enzyme papain gives rise to a cleaved, but fully functional, complex consisting of three fragments, H, M and L (heavy, medium and light), held together by strong non-covalent forces. By using some of the chemical and physical differences between the three fragments, a protocol for the preparative isolation of each fragment was devised. Purified M-fragment, which had been radiochemically labelled at the retinal-binding site was treated with CNBr and the mixture subjected to a multi-step separation to furnish a retinyl peptide. The sequence analysis of the latter showed that the retinal-binding lysine residue was located at position 296 from the N-terminal of rhodopsin (or residue 53 from the C-terminal). In order to ascertain the position of the cytoplasmic loop which exists between the M- and L-fragments, radiochemically labelled L-fragment was isolated from the cleaved complex. The purified L-fragment was shown to consist of two populations of peptides which were produced by the action of papain on the bonds between Lys-311 and Gln-312 and between Gln-312 and Phe-313.

The structure of bovine rhodopsin

We have isolated 16 peptides from a cyanogen bromide digest of rhodopsin. These cyanogen bromide peptides account for the complete composition of the protein. Methionine-containing peptides from other chemical and enzymatic digests of rhodopsin have allowed us to place the cyanogen bromide peptides in order, yielding the sequence of the protein. We have completed the sequence of most of the cyanogen bromide peptides. This information, in conjunction with that from other laboratories, forms the basis for our prediction of the secondary structure of the protein and how it may be arranged in the disk membrane.

Structural prediction of membrane-bound proteins

A prediction algorithm based on physical characteristics of the twenty amino acids and refined by comparison to the proposed bacteriorhodopsin structure was devised to delineate likely membrane-buried regions in the primary sequences of proteins known to interact with the lipid bilayer. Application of the method to the sequence of the carboxyl terminal one-third of bovine rhodopsin predicted a membrane-buried helical hairpin structure. With the use of lipid-buried segments in bacteriorhodopsin as well as regions predicted by the algorithm in other membrane-bound proteins, a hierarchical ranking of the twenty amino acids in their preferences to be in lipid contact was calculated. A helical wheel analysis of the predicted regions suggests which helical faces are within the protein interior and which are in contact with the lipid bilayer.