FUTURE DIRECTIONS FOR RHODOPSIN STRUCTURE AND FUNCTION STUDIES
Paul A. Hargrave
What does rhodopsin do as a photoreceptor protein and how does it do it? The answer to this question will provide the details of rhodopsin's structure and how it acts to carry out the functions of a photoreceptor protein. We know the details in rough outline -- the folding and insertion of the opsin polypeptide chain into the membrane, its binding of 11- cis retinal, the vectorial transport to the outer segment, absorption of a photon by the retinal, the change in conformation of the protein, and the subsequent binding and activation of transducin. But how is this all actually achieved at the molecular level? How does the protein primary structure adopt the appropriate conformation to accomplish these results? To answer this it will be necessary to first know the details of rhodopsin's structure in its inactive state in the dark, and its structure when activated by light, in its signalling state. It would be important to determine the structure of Metarhodopsin I (which contains all-trans retinal, but can't activate transducin) and the difference between this and Metarhodopsin II (R*). These are necessary and achievable goals that will provide the baseline for understanding rhodopsin function. Rhodopsin is more easily obtained in the amounts needed for physical studies than other members of this large class of G protein-coupled receptors. By analogy, insights into the function of rhodopsin should be valuable for understanding the signalling mechanism of the entire class of receptors.
Mutants of rhodopsin, naturally occurring and otherwise, have proven useful in demonstrating what particular regions or amino acids in rhodopsin are most functionally significant (see Khorana 1992; Nathans 1992; Oprian 1992; Dryja 1992; Berson 1993). This important area must be left for consideration elsewhere. In the present commentary I will discuss the prospects for obtaining a high resolution atomic structure for rhodopsin, which I believe will be critical for understanding the function of G-protein-linked receptors generally.
What do we think we know about the structure of rhodopsin?
There has been a virtual explosion of information about rhodopsin during the past twenty years. Original references to the literature are to be found in several recent reviews (Chabre 1985; Findlay 1986; Applebury & Hargrave 1986; Liebman et al. 1987; Chabre & Deterre 1989; Applebury 1991; Hargrave & McDowell 1992; Khorana 1992; Nathans 1992; Hargrave & McDowell 1993). Although many gaps in our knowledge have been filled, we are still a long way from the goals outlined above. We view rhodopsin as an intrinsic membrane protein, with half of its mass embedded in the lipid bilayer and the remaining half approximately equally distributed between the two hydrophilic surfaces facing the cytoplasm and the intradiscal environment. We believe it to be topographically organized so that the polypeptide chain traverses the lipid bilayer seven times, alternately exposing hydrophilic sequences at membrane surfaces and burying hydrophobic sequences in the lipid bilayer (Fig. 1). The amino terminus, to which carbohydrate is attached at two sites, is located in the intradiscal environment; the carboxyl terminus, containing serine and threonine residues that may be phosphorylated by rhodopsin kinase, is exposed to the rod cell cytoplasm. The transmembrane segments are thought to be largely helical in structure, although several of them may be irregular due to presence of a proline in their sequence. Rhodopsin's chromophore, 11-cis retinal, is attached by Schiff base linkage to the side chain of lysine 296, located midway in the membrane-spanning seventh helix. The retinal is enveloped in a pocket formed by the amino acids lining the interior surfaces of rhodopsin's helices.
Further characteristic features emerge when the sequences of many vertebrate rhodopsins are compared. Two highly conserved cysteines are present that are involved in a stabilizing disulfide bond linking helix III with loop e2. One or more cysteines are generally present in the C-terminal sequence following helix VII. These cysteines become palmitoylated and presumably insert into the lipid bilayer forming a fourth cytoplasmic loop region. Additional amino acids are found to be conserved in all rhodopsins studied to date. These conserved residues are good candidates for structurally/functionally essential residues that are required components of what it means to be a rhodopsin. Some of these residues are also conserved in other members of the class of G-protein-coupled receptors (Probst et al. 1992) and may be assumed to be required for features that are common to members of the receptor class, such as membrane translocation or signal transmission.
What do we think we know about the function of rhodopsin?
Within a millisecond of absorbing light and isomerizing its 11-cis retinal to all-trans, rhodopsin forms Metarhodopsin I (MI). This intermediate is not able to bind and activate transducin, but Metarhodopsin II (MII) is (Fig. 2). Deprotonation of the retinylidene-Schiff base accompanies the conversion of MI to MII [reviewed by (Hofmann 1986)]. This deprotonation event is essential for the activation of transducin.
The protonated retinylidene-Schiff base is stabilized by an ion pair with the carboxyl group of glutamic acid 113 (Zhukovsky & Oprian 1989; Sakmar et al. 1989; Nathans 1990). Studies with mutant rhodopsins show that the maintenance of the interaction between Lys296 and Glu113 is essential in keeping rhodopsin inactive. Opsins lacking either of these charged groups constitutively activate transducin (Robinson et al. 1992). Thus it appears, in part, that when MI loses the proton from its Schiff base, this breaks a key ionic interaction that allows it to assume the more open protein structure, MII.
The cytoplasmic surface of rhodopsin and MI is unable to bind and activate transducin. But the conformational change that accompanies the formation of MII causes a rearrangement of surface residues that provides the appropriate surface array for binding transducin. Amino acids in loops i2, i3 and i4 are involved in this protein/protein recognition event [(Knig et al. 1989); reviewed in (Hargrave et al. 1993)]. Further molecular details concerning how rhodopsin achieves all of these transformations, remain to be elucidated.
What will most advance our knowledge of how rhodopsin acts as a photoreceptor?
In order to better understand how rhodopsin acts to transduce the reception of a photon into activation of a G-protein, it will be necessary to have an understanding of rhodopsin structure at the level of atomic resolution. This will require both a knowledge of where rhodopsin's amino acids are located before absorption of a photon, and where rhodopsin's amino acids become relocated to following absorption of a photon, when rhodopsin becomes transformed to MI and then to its signalling state, MII.
All proteins are stabilized by a variety of individual interactions of their amino acids, involving disulfide bonds, hydrogen bonds, ionic bonds and Van der WaalsU forces. In rhodopsin's inactive state these forces contribute to maintain a stable three-dimensional structure in which the protein is constrained from binding and activating transducin. All proteins have a certain range of dynamic movement at any temperature, but this must be quite small for vertebrate rhodopsin since "noise" (signalling in the absence of photon absorption) is extremely low (Baylor 1987).
The juxtaposition of amino acids in rhodopsin's signalling state serves to define the binding site for transducin. The particular three-dimensional array that permits this binding is not present in MI but is formed when rhodopsin adopts the MII conformation. Thus a knowledge of the structure of MII will show what amino acid interactions have been broken in rhodopsin and have been formed to stabilize the receptor signalling state, MII. Such information would provide the boundary conditions and would be essential in beginning to understand how the isomerization of 11-cis retinal leads to the structural reorganization that allows binding and activation of transducin.
What are the prospects for obtaining a high resolution structure for rhodopsin?
There are basically three methods for obtaining atomic level structural data for proteins: 1) X-ray diffraction of three- dimensional crystals (K hlbrandt 1988), 2) electron diffraction, electron microscopy and image processing of two- dimensional crystals (Amos et al. 1982; K hlbrandt 1992), and 3) nuclear magnetic resonance (NMR). Rhodopsin, at 40 kDa, is too large for analysis by currently available NMR techniques and cannot be examined in the membrane by liquid- state NMR. That leaves crystallography.
What are the prospects for getting a high-resolution structure for any membrane protein or protein complex? At present there are atomic resolution structures for more than 1700 proteins, but only 4 classes of membrane proteins are represented: bacteriorhodopsin from Halobacterium halobium (Henderson et al. 1990); the bacterial photosynthetic reaction centers (Deisenhofer & Michel 1989; Feher et al. 1989; Arnoux et al. 1989); bacterial porins (Weiss et al. 1991; Cowan et al. 1992); and light harvesting complex from green plants (K hlbrandt et al., 1994).
What are the chances of getting crystals of rhodopsin?
The chances are excellent! Several investigators have already devoted many years to this task and crystals have been obtained. Both two-dimensional (Corless et al. 1982; Dratz et al. 1985; Demin et al. 1987; Schertler et al. 1993; Schertler & Hargrave, 1994) and three-dimensional crystals have been obtained (Demin et al. 1987; Yurkova et al. 1990; de Grip et al. 1992; E. Dratz, personal communication).
Two-dimensional crystals, presumably of rhodopsin, have been induced to form in the frog disk membrane following extraction of some of the lipid with the detergent Tween 80 (Corless et al. 1982). The molecules in these crystals appear to be 20-25 in width and 70-80 in length and to have a cross-sectional area similar to that of bacteriorhodopsin. When observed, crystals made up ~5-10% of the membrane. Although they were not definitively demonstrated to be composed of rhodopsin, it seems unlikely that they would be composed of minor membrane proteins. 2-D crystals of bovine rhodopsin that diffract to a resolution of about 25 have also been formed using the Tween 80 extraction method (Demin et al. 1987). In a different study, two-dimensional crystals of bovine rhodopsin were induced to form within the disk membrane in the presence of ammonium sulfate and an amphiphilic compound, chlorhexidine (Dratz et al. 1985). Analysis of these crystals by electron microscopy also showed a surface area for rhodopsin similar to that of bacteriorhodopsin, further supporting the presence of seven transmembrane segments in rhodopsin. However, none of the crystal preparations obtained by these investigators were of sufficient quality to give higher resolution data; e.g., data that would allow individual helices of the protein to be resolved or the position and identity of individual amino acids to be defined.
Visualization of the helices of bovine rhodopsin has been obtained by cryoelectron microscopy of two-dimensional crystals formed from purified rhodopsin in phosphatidyl choline (Schertler et al. 1993). Rhodopsin was solubilized in n-octyl tetraoxy-ethylene (C8E4) and crystals were formed as the detergent was removed by dialysis. Analysis of the crystals as frozen hydrated specimens allowed collection of data adequate for calculation of a 9 resolution map. The map shows an elongated arc-shaped feature flanked by four resolved peaks of density. Orientation of the helices is clearly different from that of bacteriorhodopsin. More recently, tubular structures containing rhodopsin crystals have been formed in good quantity by extracting frog disk membranes with Tween 80 (Schertler & Hargrave 1994). Electron micrographs of the frozen hydrated crystals allowed a projection structure to be calculated to 6 resolution. This showed an arrangement of helices similar to that in the map obtained previously for bovine rhodopsin (Schertler et al. 1993). Helices 4, 6 and 7 are nearly perpendicular to the membrane plane, but helix 5 is more tilted or bent than anticipated from the 9 map. The rhodopsin molecule can be described as a bundle of four tilted helices alongside three perpendicular helices that are arranged in a straight line. The next step will be to collect data from tilted specimens to allow calculation of a three-dimensional map so that helix tilt angles can be determined.
Three-dimensional crystals of rhodopsin have been obtained over a pH range from 5.5 to 7.0 using five different detergents and two different precipitants (de Grip et al. 1992). Crystals have also been obtained from octyl polyoxyethylene (o-POE) using ammonium sulfate, sodium phosphate or sodium citrate as precipitants (Yurkova et al. 1990). Unfortunately, the crystals formed to date have been too small, fragile and disordered to allow high-resolution diffraction analysis. This also has been the experience of other investigators who have worked with rhodopsin (E. Dratz, personal communication). The crystals obtained from o-POE were needle shaped with maximum dimension 70 mm x 70 mm x 1000 mm; too small for X-ray analysis (Yurkova et al. 1990). Such observations are certainly not unique to rhodopsin and appear to be experienced with the majority of membrane proteins. It is often difficult enough to obtain ordered crystals from well-behaved soluble, globular proteins, but these difficulties are compounded when dealing with non water-soluble proteins that have to be handled in detergent solutions.
The conditions that will eventually succeed in producing large, stable and highly-diffracting crystals of rhodopsin will probably be unique for rhodopsin and will have to be determined empirically. The particular properties and behavior of rhodopsin as a protein will dictate what conditions will eventually succeed. For that reason, detailed knowledge of rhodopsin's properties as a protein are a prerequisite to such an undertaking. The body of experience obtained from the crystallization of all previous proteins provides only guidelines for procedures to use as the studies with rhodopsin proceed. We will consider the steps in the process of obtaining protein crystals and look for the most productive approaches that might be utilized with respect to rhodopsin.
Preparation of rhodopsin-containing membranes
Cattle retinas are conveniently obtained in quantity from animals slaughtered under conditions minimizing bright-light exposure. From each retina it is often possible to obtain more than 700 mg of rhodopsin. Purified rod cell outer segments that contain membrane-bound rhodopsin are conveniently obtained by a number of protocols that include homogenization, differential centrifugation, and density gradient centrifugation (Papermaster & Dreyer 1974; McDowell 1993). Following hypotonic lysis of the rod cell outer segments and washing at low ionic strength, membranes are prepared that contain nearly 95% of their protein content as rhodopsin plus opsin. The components other than rhodopsin, are opsin, the "rim protein", peripherin/rds, ROM-1, remaining membrane-associated proteins, plasma membrane proteins, and phospholipids [reviewed in (Molday 1989; Hargrave & McDowell 1993)]. These membrane preparations can be solubilized in any of a variety of detergents and submitted to chromatographic steps that lead to purified rhodopsin.
Purification of detergent-solubilized rhodopsin
For purposes of crystallization it seems most desirable that a reproducible well-defined homogeneous preparation of rhodopsin is obtained, free of opsin and other components. There are two main methods in use for the preparative chromatography of rhodopsin; hydroxyapatite chromatography and lectin affinity chromatography.
When rhodopsin is purified by calcium phosphate chromatography using the commercial detergent preparation Emulphogene, it shows a single protein band by electrophoresis and a spectral ratio of 280nm/498nm of 1.75 (Shichi et al. 1969). However, rhodopsin purified by this method contains variable amounts of lipid (Papermaster & Dreyer 1974). When chromatography on hydroxyapatite is carried out using the cationic detergent tridecyltrimethylammonium bromide (TrTAB), the protein binds to the column and can be conveniently washed free of lipid (Hong & Hubbell 1973). Rhodopsin of the same high spectral purity is then eluted using a salt gradient and contains from 0.2 to 0.8 moles of phosphate (phospholipid) per mole of protein. One disadvantage of this method is that TrTAB is not commercially available and must be synthesized by the investigator. However, it is dialyzable, and rhodopsin purified in TrTAB can be conveniently prepared in another detergent by addition of a non-dialyzable detergent to the rhodopsin TrTAB solution followed by dialysis (Hong & Hubbell 1973).
Probably the most widely used method to prepare rhodopsin employs chromatography on concanavalin A Sepharose. Several methods have been described (Litman 1982; de Grip 1982). Rhodopsin, in a variety of detergents, is applied to the affinity matrix in a buffer containing salts of Mg+2, Ca+2 and Mn+2 (to stabilize the bound concanavalin A tetramer). Washing with detergent-buffer removes lipids and contaminating proteins. Rhodopsin is eluted with a-methyl mannoside in detergent-buffer. Dialysis removes the sugar and presumably the heavy metal ions. One must be alert to possible contamination by variable amounts of divalent metal ions, especially in light of a report that zinc becomes tightly associated with rhodopsin (Shuster et al. 1992). Another contaminant that can be introduced by this purification method is concanavalin A itself. It can be removed by passing the purified rhodopsin over a column containing an antibody to concanavalin A (de Grip 1982) or by mannose agarose affinity chromatography. However, a more attractive approach is to prepare rhodopsin in a mild detergent such as octyl or nonyl glucoside that does not cause concanavalin A to be removed from the column (Litman 1982).
Homogeneity of rhodopsin
Purified rhodopsin should be assessed for homogeneity by SDS polyacrylamide gel electrophoresis (SDS-PAGE). The stained protein band on SDS-PAGE is always broader and less sharp than that of most other proteins. This appears to be due to heterogeneity of glycosylation. Vertebrate rhodopsins contain two sites of glycosylation, but not all molecules are identical in their carbohydrate content. About 70% of the oligosaccharides on bovine rhodopsin contain Man3GlcNAc3, 10% Man4GlcNAc3, and 20% Man5GlcNAc3 (Fukuda et al. 1979; Liang et al. 1979; Hargrave et al. 1984). The distribution of these three components between the two sites in rhodopsin is unknown. This heterogeneity could make a difference in the ability of differently glycosylated molecules to pack and interact in crystals and represents a potential source of crystal disorder. One approach to dealing with this source of heterogeneity would be removal of the oligosaccharides by endoglycosidase digestion (Plantner et al. 1991). Peptide-N- glycosidase F is capable of removing both oligosaccharide chains completely, leaving deglycosylated rhodopsin as the sole product. The rhodopsin species with 0, 1 and 2 oligosaccharide chains are easily distinguished by SDS-PAGE (Plantner et al. 1991). If required, reaction mixtures might be further purified by passing through a column of concanavalin A Sepharose, allowing passage of successfully deglycosylated rhodopsin molecules.
When rhodopsin is separated by SDS-PAGE, a series of molecular weight bands is often generated corresponding to dimer, trimer and higher molecular weight oligomers. This generation of a multiplicity of bands can be considered a unique characteristic of the protein, but is also a nuisance when analyzing the protein for homogeneity, since the presence of contaminating proteins can be obscured. Rhodopsin is thought to be a monomer in disk membranes (Cone 1972; Downer 1985). The production of oligomers by SDS-PAGE is an artifact. Oligomers are probably produced after solubilization of rhodopsin by conditions that promote rhodopsin/rhodopsin collisions that alter detergent/lipid interaction and promote interaction of rhodopsin monomers that leads to aggregation. Oligomer formation may be eliminated by incubating rhodopsin solutions or rhodopsin- containing membranes at low protein concentration (<1 mg/mL) in the presence of high concentrations of SDS (5%) and high concentrations of reducing agent, avoiding elevated temperatures (using room temperature or 37oC), and avoiding the use of stacking gels that concentrate rhodopsin during PAGE (Papermaster & Dreyer 1974).
Since cattle are rarely completely dark adapted, we must consider the possibility that rhodopsin prepared from such light-exposed retinas could contain phosphorylated opsin as a contaminant. The various species of phosphorylated rhodopsin can be conveniently detected by isoelectric focusing. Rhodopsin itself has a pI of 6.0, and the different phosphorylated species have progressively more acidic pIs (K hn & McDowell 1977; Aton et al. 1984). It would be important to assess whether a rhodopsin sample, prepared for purposes of crystallization, contained species other than that of the unphosphorylated protein with a pI of 6.0. If such phosphorylated rhodopsins are found, they can be conveniently removed by passing the detergent-solubilized rhodopsin sample over a Fe+3-Chelex column (Andersson & Porath 1986; J. H. McDowell, personal communication) or by anion-exchange chromatography of rhodopsin at neutral pH (W. de Grip, personal communication).
Rhodopsin is also posttranslationally modified by palmitoylation of two adjacent carboxyl-terminal cysteine residues (Ovchinnikov et al. 1988; Papac et al. 1992). Partial or complete loss of palmitate would result in a significant change in the hydrophobicity and probably also the organization of the carboxyl-terminal region of rhodopsin. Disorganization of the carboxyl-terminal 40 amino acids of rhodopsin could result in a major problem for crystallization. It is known that the cysteine thioester linkage is labile to hydroxylamine, alkali, and to reducing agents (O'Brien & Zatz 1984). It has been shown by O'Brien and coworkers that simply storing solubilized rhodopsin at 4oC for 4 days led to a decrease in fatty acid content from 2.26 moles to 0.83 moles/mole rhodopsin (O'Brien et al. 1987). The rate of loss is temperature dependent and is accelerated with increase in temperature. This suggests that in crystallization experiments that take several weeks to complete, that rhodopsin will be heterogeneous with respect to palmitate content and may lose one or both of its palmitates during the course of the experiment. In experiments reported by de Grip, it was found necessary to include 5 mM dithioerythritol in the rhodopsin buffer in order to prevent damage to rhodopsin due to detergent impurities (de Grip et al. 1992). Presence of this reducing agent and the long times for crystallization would be expected to lead to loss of palmitate from its thioester linkage to rhodopsin.
What are the important characteristics for a detergent for the crystallization of rhodopsin?
An ideal detergent would provide an environment for rhodopsin that simulated its membrane environment so well that its properties in detergent solution were the same as those measured in the disk membrane. Rhodopsin would be as thermally stable in this detergent as in the membrane. It would bleach with the same kinetics, form a stable opsin, and the opsin would be fully regenerable to rhodopsin upon binding 11-cis retinal. No such detergent has yet been found. In addition, such an ideal detergent must have other important properties.
An appropriate detergent for protein crystallization must be a defined chemical substance that is available commercially in pure form or that can be synthesized in quantity economically and relatively easily. It must be pure and chemically stable. The detergent should have good solubility at room temperature and at 4oC, and over a reasonable range of pH and ionic strength. Although the ability to efficiently solubilize membrane-bound rhodopsin would be helpful, this is not required, since a detergent can usually be exchanged.
It is generally assumed that detergent molecules form a ring around solubilized membrane proteins, binding to hydrophobic surfaces previously occupied by lipid fatty acyl chains (Fig. 3). This leaves the hydrophilic ends of the protein exposed to the aqueous environment. The detergent should produce small monodisperse micelles and create the smallest stable protein-detergent complex possible (Garavito 1991). Such a small complex might be expected to enhance ionic interactions between hydrophilic protein surfaces that may promote good crystal formation. Too large a micelle may interfere with protein-protein interactions. An unstable micelle leads to nonspecific hydrophobic interactions, aggegation and precipitation.
Measurement of properties of rhodopsin-detergent complexes
Numerous detergents have been used in the study of the properties of rhodopsins (de Grip 1982; Fong et al. 1982; de Grip et al. 1992) and a variety of parameters have been measured to assess the characteristics of the rhodopsin- detergent complexes. Since crystallization trials may require two to four weeks, it is the stability, in particular the long-term stability of the rhodopsin-detergent complex, that is most pertinent to the consideration of its crystallization.
The primary parameter that has been measured to assess the stability of the rhodopsin-detergent complex has been that of thermal stability. In one study the rhodopsin-detergent complexes were heated at various temperatures and the denaturation of rhodopsin followed by measurement of decrease in absorption at 500 nm (de Grip 1982; de Grip et al. 1992). Temperatures were determined at which the half-time of denaturation is 10 min for each detergent (Table I). Among the detergents tested, the one imparting the greatest stability to rhodopsin is b-dodecylmaltose.
Thermal stability of rhodopsin has also been measured by differential scanning calorimetry (Miljanich et al. 1985; Shnyrov & Berman 1988; Khan et al. 1991; McDowell et al. 1992). By programmed heating of rhodopsin in the membrane and in detergent solution, temperatures of denaturation are obtained (Table 2). Such data show that even the mildest detergent tested, digitonin, falls far short of offering the protective environment that is available to rhodopsin in its native membrane.
Long-term stability of the rhodopsin-detergent complex under conditions used in crystallization has been examined by de Grip and coworkers. They examined rhodopsin for maintenance of spectral integrity, structural stability (avoidance of formation of lower molecular weight fragments measured by SDS-PAGE and immunoblotting), and formation of crystals when rhodopsin/detergent/precipitant solutions were kept at 20oC for 2-3 months (de Grip et al. 1992). Some fragmentation of rhodopsin occurred when detergents or precipitants contained oxyethylene units. This suggested that the degradation was caused by peroxidation, which seems quite likely since it was subsequently eliminated by inclusion of a reducing agent. One wonders whether scrupulous purification of the detergents/precipitants to remove peroxidants, the choice of more stable detergents, or inclusion of vitamin E or butyl- hydroxytoluene as an antioxidant, and an argon atmosphere (Farnsworth & Dratz 1976) might prove helpful, and eliminate the need for inclusion of a reducing agent (which may lead to depalmitoylation, as discussed above).
Small amphiphiles are often helpful in crystallization of membrane proteins
Considerable success has been achieved in crystallization of membrane proteins by the addition of 1-5% of small amphiphilic compounds to the protein-detergent complex. Mixed micelles are formed in which the amphiphiles intercolate into positions that would be occupied by larger detergent molecules. They reduce the size of the protein- detergent complex which is thought to maximize the hydrophilic protein-protein interactions needed for good crystal lattice formation (Michel 1991). Of the more than 100 different compounds investigated, threo-1,2,3-heptane triol and benzamidine have been the most successful in trials with bacteriorhodopsin and the light-harvesting complex. However, inclusion of the amphiphiles is not essential since the proteins have been successfully crystallized under other conditions in their absence.
Other investigators have found that amphiphiles improved the solubility of their membrane protein and influenced the type of crystal formed (Allen & Feher 1991; Garavito 1991). Crystals of a reaction center complex that showed a resolution of 10 were improved to 7 when 1,6-hexanediamine was added. In this instance it has been suggested that the diamine acted as a bridge stabilizing the protein micelles in the crystal lattice (Welte & Wacker 1991). Although the addition of small amphiphiles has been applied to the formation of 3-D crystals of rhodopsin, conditions have not yet been found in which they yield an improvement (Demin et al. 1987; de Grip et al. 1992). However, the quality of 2-D rhodopsin crystals in the membrane was improved by the inclusion of the surface-active agent, chlorhexidine (Dratz et al. 1985).
What are the prospects for obtaining two-dimensional crystals of rhodopsin that will yield a high-resolution structure?
Currently available methods are capable of yielding high- resolution structures from well-ordered two-dimensional crystals of membrane proteins. Given the perfect microscope and the perfect membrane specimen, atomic resolution can be achieved. Bacteriorhodopsin is the classic example where a structure has been obtained with a resolution of 3.5 in the plane parallel to the membrane (Henderson et al. 1990). Interpretation of the structure has allowed assignment of 21 amino acids from all 7 helices that contribute directly to the environment of the retinal, allowing a proposal to be made for the location of amino acids that constitute the proton channel. This is the type of information that would be very useful for vertebrate rhodopsin.
There are a wide variety of examples of membrane proteins that have been examined by electron microscopy and image analysis. The two-dimensional crystals studied thus far fall into three categories: 1) proteins that occur naturally in a semi-crystalline array; bacteriorhodopsin in purple membrane, gap junctions from hepatocytes (Unwin & Zampighi 1980); 2) crystalline arrays that can be produced from membranes in vitro; cytochrome oxidase (Vanderkooi 1974), rhodopsin (Corless et al. 1982; Dratz et al. 1985), acetylcholine receptor (Toyoshima & Unwin 1990), H,K-ATPase (Hebert et al. 1992), and 3) proteins crystallized from protein-detergent complexes, or protein/lipid/detergent-complexes, as two- dimensional sheets; bacterial porin (Dorset et al. 1983), cytochrome reductase (Hovmller et al. 1983), cytochrome b/c1 complex (Hovmller et al. 1983), light-harvesting complex (K hlbrandt 1984), photosynthetic reaction center (Miller and Jacobs 1983), and NADH dehydrogenase (Boekema et al. 1986).
Vertebrate rhodopsin exists in a highly fluid membrane and does not naturally form a crystalline array. Because the lamellar part of the rod cell disk membrane is nearly pure rhodopsin in lipid, it is attractive to attempt to induce rhodopsin to form a crystalline array within its native membrane (Corless et al. 1982; Dratz et al. 1985; Demin et al. 1987). However, the level of resolution obtained by that approach has not come close to what is required. Improvements have been made by examining unstained membranes and by use of improved methods for data analysis (Schertler et al. 1993; Schertler & Hargrave 1994). However, substantial improvements in resolution require that the crystals be larger, and that may require that they be formed by a different approach.
The most versatile approach to the formation of 2-D crystals is to reconstitute detergent-solubilized proteins into a lipid environment formed during the slow dialysis of detergent. With this approach there is control over the type of detergent, the amount and type of lipid, and the method and rate of removal of detergent. Rhodopsin's physical and photochemical properties have been studied in a variety of different lipids (Deese et al. 1981; de Grip et al. 1983; Ryba & Marsh 1992; Mitchell et al. 1992). Based upon these and other studies it may be possible to choose a lipid environment that will be more conducive to the formation and stabilization of rhodopsin in a crystalline array.
It may be easier to study all forms of rhodopsin in 2-D crystals
It is not only possible to examine 2-D crystals of membrane proteins at low temperature, it may be desirable. The best electron microscopic image yet obtained from bacteriorhodopsin was from a sample at the temperature of liquid helium (Henderson et al. 1990). Electron microscopic data is frequently taken from frozen or refrigerated samples. Studies of the neutron diffraction of an intermediate in the photocycle of bacteriorhodopsin involved measuring it at -180oC (Dencher et al. 1989).
Low temperatures have been used to selectively isolate the spectral intermediates in bleaching of rhodopsin. Various intermediates can be produced by irradiation at low temperature followed by warming to the temperature at which the intermediate is stable. MI can be produced by warming to a temperature between -40oC and -15oC, and MII can be formed over the range -15oC to 0oC. Thus by photolyzing the rhodopsin 2-D crystal and subjecting it to the appropriate temperature it should be possible to form these structural intermediates. By immediately freezing in liquid ethane and maintaining the crystals at liquid nitrogen temperature, the intermediates should be preserved for examination by electron crystallography. Difference maps between two rhodopsin forms should indicate the areas where the rhodopsin polypeptide chain has undergone a change in conformation in proceeding from rhodopsin to MI and from MI to MII. Such an approach has been successfully applied to detection of structural changes between the ground state and the M intermediate of bacteriorhodopsin (Subramaniam et al. 1993).
The objective: three-dimensional crystals of rhodopsin
X-ray crystallography of three-dimensional crystals has produced the highest resolution structural data for proteins. Only four classes of membrane proteins have thus far yielded crystals of sufficient order to produce reasonably high resolution structures. The successful methods for production of crystals from membrane proteins have been adaptations of methods used for soluble globular proteins. Concentrated solutions of purified membrane proteins in detergent may form crystals in the presence of increasing concentrations of precipitants such as ammonium sulfate or polyethylene glycol. Critical parameters such as the choice of detergent have been discussed above. Here I wish to discuss the experimental problems introduced by the intrinsic nature of rhodopsin itself: its extreme sensitivity to light.
Working with rhodopsin in the dark
To produce and analyze crystals of rhodopsin demands that all aspects of the process be conducted under dim red light (light of >620 nm) or by infrared image converter. This means that all of the crystallization trials, selection and examination of crystals, mounting and X-ray analysis be conducted under dark-room conditions. Such work is not only tedious but has been reported to lead to failure to identify crystals (de Grip et al. 1992). Eventually another problem will arise when the crystals are examined by X-ray analysis. Interaction of the measuring beam with water in the membrane sample causes fluorescence -- light that would be expected to bleach the rhodopsin being examined. Such considerations have led investigators to attempt to produce light- insensitive rhodopsins. The search for light-stable rhodopsin: the non-isomerizable chromophore
Locking retinal into place so that it would not be photosensitive is one approach to handling rhodopsin more conveniently for crystallization and analysis. What is desired is a rhodopsin that will not be subject to a change in structure upon light exposure. Such a rhodopsin might also have greater long-term stability, enhancing its ability to withstand multi-week crystallization times.
Opsin will combine with a number of isomers and derivatives of retinal to yield visual pigments of varying stability (Derguini & Nakanishi 1986). To be optimally useful for crystallization trials, any visual pigment formed by combination of a retinal derivative with opsin must have the following characteristics: 1) minimal perturbation of its native structure, 2) long-term stability in detergent; 3) insensitivity to light, as measured by maintenance of spectral integrity; and 4) insensitivity to attack by hydroxylamine (This is a measure of the native conformation of the protein-retinal complex. Hydroxylamine attacks the retinylidene-Schiff base linkage when it become accessible to the aqueous environment. The retinylidene-Schiff base is unavailable for attack in native rhodopsin.)
Ring-locked derivatives of retinal designed to eliminate photoisomerization about the 11-12 double bond have been investigated as possible chromophores that would yield a rhodopsin insensitive to photoisomerization. Such compounds have been produced with 5-membered- (Ito et al. 1982), 6- membered- (van der Steen et al. 1989; Bhattacharya et al. 1992) 7-membered (Akita et al. 1980), 8- and 9-membered rings (Hu et al. 1994) built around the retinal 11-12 double bond (Fig. 4). Each of these compounds (except the 9-membered one) has been successfully recombined with opsin, and the pigments formed (Rh5, Rh6, Rh7and Rh8 respectively) have been the subject of many interesting studies (Fukada et al. 1984; Zankel et al. 1990; de Grip et al. 1990; Hu et al. 1994). However, pigments Rh5 and Rh7 have been subject to attack by hydroxylamine and are unstable in digitonin solution. By contrast, Rh6 is stable to hydroxylamine in the dark in detergent solution and is nearly as thermally stable as rhodopsin itself. Although it has unusual photostability, it does bleach in detergent solution at a rate 0.6% that of rhodopsin (de Grip et al. 1990; Bhattacharya et al. 1992). The rhodopsin analog has greatly reduced yet measureable activity in stimulating enzymes in the phototransduction pathway and in serving as a substrate for rhodopsin kinase. It is probable that all of the 11-cis ring-locked chromophores photoisomerize to a limited extent about the 7- and 9-double bonds, and that during this photoisomerization the Schiff base becomes transiently accessible for hydrolysis. To further restrict the potential for photoisomerization, a second ring between the retinal carbons 9 and 11 has been introduced (Fig. 4, compound 7). Unfortunately the photosensitity of the rhodopsin pigment formed with this retinal is not further reduced when compared to Rh6 (W. de Grip, personal communication).
Really light-stable rhodopsin?
There have been two different approaches taken to capitalizing on the greatly enhanced light stability of rhodopsin containing the 6-membered ring locked retinal. One approach has been chemical; to further strengthen the association of the retinal to rhodopsin by covalent attachment (van der Steen et al. 1989). The other approach has been biological; to engineer the protein so that it is less susceptible to bleaching (Ridge et al. 1992). A promising compound for locking rhodopsin in a photo- insensitive state is the acid fluoride of the 6-membered ring-locked retinal (van der Steen et al. 1989). It reacts with opsin to yield a blue-shifted rhodopsin analog with 390nm absorbance maximum that is thermally stable, non- bleachable, and is completely inactive in signal transduction (de Grip et al. 1990). It can be heated in detergent at 60oC (under conditions that denature rhodopsin with t1/2 <1min) with complete stability. The rhodopsin analog can be illuminated for 1 h under conditions that bleach rhodopsin with t1/2 =15 sec, without loss of absorbance (van der Steen et al. 1989). Such photo- and thermal stability appears to come not only from a good fit to the retinal binding site and stabilization towards photoisomerization, but from the irreversible amide bond that replaces the normally hydrolyzable Schiff base. However, this rhodopsin analog has been difficult to purify (W. deGrip, personal communication): the acylation reaction is slow and incomplete, and the acylation must be performed on the completely reductively- methylated protein in which all lysines except Lys296 are methylated. This leads to heterogeneity and difficulty in purification. Another disadvantage is that in formation of the retinyl-amide linkage, the protonated Schiff base is destroyed, thus the ion-pairing that is an important part of rhodopsin structure and function cannot be observed. Although crystals of this derivative might be useful in examining many overall features of the rhodopsin molecule, it would differ in important details in the region of the retinal attachment site.
A novel alternative approach to further stabilizing the Rh6 rhodopsin has been to engineer the protein to make it less photosensitive (Ridge et al. 1992). Previous studies had identified a number of amino acids that were located in the retinal binding pocket. Twelve mutant rhodopsins that exhibited spectral differences in their combination with 11- cis retinal in the ground or excited state were recombined with the cyclohexene-locked retinal. One such mutant rhodopsin, Trp265Phe, was completely stable to illumination. The mutant Rh6 rhodopsin was stable to hydroxylamine and was inactive in phototransduction or as a substrate for rhodopsin kinase (Ridge et al. 1992). The lack of photosensitivity of this mutant resulting from a single amino acid mutation points to an important role of Trp265 in the signal transduction mechanism. Mutant Trp265Phe Rh6 would appear to be an ideal candidate for crystallization trials. However, one must not overlook the need to produce sufficient quantities of the mutant rhodopsin in a large-scale expression system and the requirement for the synthetic retinal for regeneration of rhodopsin. Although expensive and laborious, the approach appears quite promising.
Although many of the above approaches to the crystallization of rhodopsin appear promising, additional approaches still need to be considered.
Rhodopsins of different species may have advantages for crystallization
There are many instances in which a protein from one species will fail to yield crystals of good quality and order, but the protein that differs in only a few amino acids, from a closely related species, will crystallize beautifully. For many of the glycolytic enzymes, the protein from yeast or from rabbit muscle has been quite satisfactory, but for others it has been necessary to seek alternative sources such as crayfish or chicken (Campbell et al. 1971). Although the difficulty in obtaining good crystals of bovine rhodopsin is probably related to its characteristics as a membrane protein, it would be well to consider that rhodopsins from pig, horse or sheep (Findlay 1986) could have small differences that are important in successful crystal formation.
The thermal stability of rhodopsin may be an important characteristic in successful crystal formation, as we have mentioned previously. Rhodopsins from organisms that experience high body temperatures could offer this advantage. Examples include the desert iguana lizard Dipsosaurus dorsalis that has the highest known body temperature for a vertebrate, at 47oC (McFall-Ngai & Horwitz 1990), and the Saharan silver ant Cataglyphis bombycina whose body temperature can reach 53.6oC (Wehner et al. 1992). Gaining information about the properties of rhodopsins from these exotic species offers special challenges. However, if study of these rhodopsins enhances our understanding of rhodopsinUs stability, this would be a valuable addition to our knowledge of rhodopsin structure/function relationships.
Invertebrate rhodopsins may have advantages for crystallization. The primary structures of octopus (Ovchinnikov et al. 1988) and squid (Hall et al. 1991) rhodopsins have been determined and many of their biochemical properties have been investigated. The proteins can be obtained in sufficient quantity. Invertebrate rhodopsins are more photostable than vertebrate rhodopsins inasmuch as their chromophore, 11-cis retinal, does not dissociate from its binding site following photoisomerization. A stable metarhodopsin is formed that is then converted back to rhodopsin upon reception of another photon of the correct wavelength. This ability of invertebrate rhodopsins to form a stable metarhodopsin makes them attractive candidates for more conveniently obtaining a structure for the signalling state of the receptor, metarhodopsin. One disadvantage, however, is their much lower thermal stability in detergents.
The octopus or squid rhodopsins may have an additional characteristic that will be of importance in crystal formation. They are larger proteins, 50-51 kDa, due to a large C-terminal extension of about 150 amino acids (Ovchinnikov et al. 1988; Hall et al. 1991). Although the significance of this region for rhodopsin function is not clear, the presence of such a large globular addition to a predominantly membrane-embedded protein, may promote protein- protein association in the crystal lattice and may reduce some of the problems associated with crystallization from detergents.
Can rhodopsin be made to look more like a soluble protein, for purposes of crystallization?
Based upon the membrane proteins that have formed the most well ordered crystals to date it would appear that a membrane protein that has the maximum amount of hydrophilic surface area would have the best chances of ordered crystal formation. One way to increase the amount of hydrophilic protein on rhodopsin's surface is to form a complex with a protein with which it interacts.
Complexes of antibody-Fab fragments with their protein antigens have been successfully crystallized and have yielded high resolution crystal structures (Mariuzza et al. 1987). There are a number of high-affinity monoclonal antibodies of well-defined specificity for rhodopsin that have been described (Molday 1989; Adamus et al. 1991). IgG class antibodies specific for rhodopsin's carboxyl-terminal sequence are easily accessible to their binding site on the opsin molecule either in its rhodopsin form (in the dark) or following light exposure. Such antibodies would be good choices to investigate the potential of this approach. This approach is not trivial since it is necessary to individually determine the proteolysis conditions suitable for production of an Fab fragment from each monoclonal antibody. Homogeneous Fab fragments must be produced that are suitable for crystallization purposes. Yields can be low, requiring many tens of milligrams of purified antibody as starting material. The Fab-rhodopsin complex should be formed and then purified to homogeneity. The complex must be sufficiently stable to remain intact during the purification and crystallization process. Although the validity of this approach has not yet been demonstrated with membrane proteins, the idea is appealing.
The lectins concanavalin A and wheat germ agglutinin bind to rhodopsin's oligosaccharide chains. These lectins have been used for affinity purification of rhodopsin and for labelling rhodopsin for microscopic analysis. They are potential candidates for protein ligands that would impart a larger hydrophilic surface to the rhodopsin molecule. The lectins would bind to the intradiscal surface of rhodopsin and might serve to mask the heterogeneity of rhodopsin's oligosaccharide chains. However, concanavalin A is a tetramer whose tendency for subunit dissociation is likely to complicate purification and crystallization procedures.
Protein engineering is one route that is free from the problem of dissociation of a bound protein ligand. By fusing the gene for rhodopsin with that for a desired protein, the resulting larger fusion protein is a single covalent entity. de Grip is exploring application of this method to rhodopsin (de Grip et al. 1992). The protein for fusion could be chosen from a list of many, such as lysozyme, that are models often used for testing new methods for protein crystallization. In theory the potential of this approach is great, especially if one has good intuition or guidance concerning what fusion protein construct(s) will be most valuable to investigate. At present this approach suffers from the lack of an expression system capable of producing hundreds of milligrams of the required protein. Expression systems for rhodopsin have been described for COS cells (Oprian et al. 1987), human embryonic kidney cells (Nathans et al. 1989) and an insect cell line (Janssen et al. 1991), but the levels of expression still need to be improved. Overexpression of membrane proteins generally has proven to be difficult (Schertler 1992).
An approach to the crystallization of Metarhodopsin II (R*)
Although we earlier suggested that 2-D crystals might be made by freezing out MII using conditions under which it is stable, this technique is unlikely to be practical for 3-D crystals. Crystallization as the complex with transducin may offer a more viable approach to this problem. This route is another variation on the attempt to crystallize rhodopsin in the presence of a bound protein ligand.
R* may be stabilized in its complex with transducin if GTP is absent and if all traces of GDP are removed (K hn 1980). This leads to a complex R*.Te in which the Ta nucleotide site is empty. This complex is stable "almost indefinitely" under conditions of physiological ionic strength (Bornancin et al. 1989). R* continues to bind retinal and remains spectroscopically in the MII state, stabilized by the binding of transducin. However, for purposes of obtaining a crystal structure of this complex it would be necessary for the complex to remain stable in detergent under conditions suitable for purification and long-term crystallization. It is doubtful whether these stringent conditions could be met. The R*.Te complex may be treated with hydroxylamine to form the Re.Te complex in which rhodopsin has lost its retinal and in which transducin is free of nucleotide (Bornancin et al. 1989). Any dissociation of this complex would be irreversible, leading to decay of the products and denaturation of the proteins. Thus the complex would have to maintain its strong association for periods of weeks under conditions suitable for crystallization.
Another approach to obtaining the crystal structure of R*
The activated state of rhodopsin, R*, is that state of rhodopsin that binds to and causes the activation of transducin. It is possible that this state may be adequately simulated by more than one mutant form of rhodopsin.
Disruption of the ion pair between the protonated Schiff base of Lys296 and the carboxyl group of Glu113 leads to formation of R*. When Glu113 is replaced by glutamine, the resulting rhodopsin is constitutively active once 11-cis retinal is removed from its binding pocket. Similarly, when Lys296 is replaced by site-specific mutagenesis, the resulting opsin becomes constitutively active (Robinson et al. 1992). These observations suggest that if one of these mutant opsins were to be crystallized, that the resulting structure might simulate the signalling state of rhodopsin. However, opsins are much less stable in detergents than rhodopsins, and the stability of the detergent complexes of these mutants may prove to be inadequate.
In the above examples, key mutations in single amino acids in rhodopsin were enough to shift the conformation of R to that of R*. It is possible that mutations in other regions of rhodopsin might also result in formation of a constitutive mutant. It is known, for example, that rhodopsin loops i2, i3 and i4 participate in binding of transducin (Knig et al. 1989; reviewed in Hargrave et al. 1993) and that loops i2 and i3 participate in activating it (Franke et al. 1990; Franke et al. 1992). Recently a single amino acid mutation in the carboxyl end of loop i3 of the a1B-adrenergic receptor was found to produce the constitutively active receptor (Kjelsberg et al. 1992). This shows that the adrenergic receptor is delicately poised and rather easily pushed over the energy barrier toward activation. It is likely that the activation of rhodopsin will not be so easily achieved. Nonetheless, our knowledge of the mechanism of activation of rhodopsin is in its early stages and we should be alert to the possibilities of other participating residues whose alteration may readily lead to the formation of R*. Our present interest is in routes that may aid our understanding of rhodopsin structure/function relationships.
Conclusion
An essential basis for understanding how a protein functions is the determination of its primary, secondary and tertiary structures. A structure at the level of atomic resolution provides the framework on which the understanding of function may develop. Rhodopsin is a particularly important and attractive target for study. From rhodopsin we will not only learn about the mechanism of phototransduction but also, by analogy, about the mechanism of action of receptors that couple to G-proteins. There is a large body of information about rhodopsin that will be of value in helping construct and evaluate experimental strategies. At present the strategies leading to formation of 2-D crystals appear most verstaile and most likely to succeed in production of structures at low to moderate levels of resolution. This will be extremely valuable in elucidating general features of rhodopsin's molecular architecture but may not yield the required level of atomic resolution to help in elucidating function.
The techniques of mutagenesis are powerful and may be required in order to tailor a rhodopsin molecule more amenable to stable 3-D crystal formation. Full utilization of these techniques awaits development of a high yielding expression system capable of producing the hundred milligram amounts of material that will be needed for innumerable trials. Whatever methods eventually succeed will triumph from application of a high degree of creativity coupled with detailed knowledge of the system, and an intense concentration of effort and resources over an extended period of time.
I would like to thank numerous colleagues who have read this manuscript in draft form and have made many valuable suggestions, most of which I have adopted. During preparation of this manuscript Dr. Hargrave's research on rhodopsin was funded by grants from the National Eye Institute of the National Institutes of Health (EY06225 and EY06226), a grant from the International Human Frontier Science Program, and an unrestricted departmental award from Research to Prevent Blindness. Dr. Hargrave is Francis N. Bullard Professor of Ophthalmology.
Supported in part by research grants EY06225 and EY06226 from the National Eye Institute of the National Institutes of Health, a grant from the International Human Frontier Science Program, and an unrestricted departmental award from Research to Prevent Blindness, Inc. PAH is Francis N. Bullard Professor of Ophthalmology.
FIGURE LEGENDS
Fig. 1. A topographic model for bovine rhodopsin in the rod cell disk membrane. Rhodopsin's polypeptide chain is shown traversing the lipid bilayer 7 times yielding 7 hydrophobic helical segments (I to VII) that are separated from each other by hydrophilic segments. Loops i1-i4 and the carboxyl- terminal sequence face the cytoplasmic surface; loops e1-e3 and the amino-terminal sequence are sequestered in the disk membrane lumen. Based on (Dratz and Hargrave 1983; Ovchinnikov et al. 1988).
Fig. 2. The rhodopsin cycle. Rhodopsin (R) is activated by light (hn) and forms Metarhodopsin I (MI) which exists in equilibrium with Metarhodopsin II (MII, which is equivalent to R*). R* causes transducin (T) to become activated (to T*) by GDP -> GTP exchange. R* becomes phosphorylated (R*(Pi)7), allowing it to bind arrestin, thereby blocking the ability of R* to continue to activate transducin. Upon loss of all- trans retinal, phosphorylated opsin (O(Pi)7) is dephosphorylated and rebinds 11-cis retinal, regenerating rhodopsin.
Fig. 3. Model for a rhodopsin/detergent complex. (A) Cross- sectional diagram of a rhodopsin molecule encircled by a detergent micelle. Polar (P) surfaces of rhodopsin, including its carbohydrate chains (CHO) interact with the aqueous environment and with detergent head groups (DH). Buried hydrophobic regions interact with hydrophobic detergent tails (DT). (B). Two rhodopsin/detergent complexes are shown forming interactions between their polar surfaces that presumably are required for crystal formation. Larger detergent micelles would make these protein/protein interactions less effective. Figures based on (Michel 1991).
Fig. 4. Structures of different retinals of interest for rhodopsin function and crystallography. 1, 11-cis retinal; 2, all-trans retinal; 3, Ret-5, 5-membered ring-locked retinal; 4, Ret-6, 6-membered ring-locked retinal; 5, Ret-7, 7-membered ring-locked retinal; 6, 6-membered ring-locked retinoyl fluoride; 7, Ret-6 with an additional ring linking carbons 9 and 11.
Adamus, G., Zam, Z. S., Arendt, A., Palczewski, K., McDowell, J. H. & Hargrave, P. A. (1991) Anti-rhodopsin Monoclonal Antibodies of Defined Specificity: Characterization and Application. Vision Research 31: 17-31.
Akita, H., Tanis, S. P., Adams, M., Balogh-Nair, V. & Nakanishi, K. (1980) Nonbleachable rhodopsins retaining the full natural chromophore. Journal of the American Chemical Society 102: 6370-6372.
Allen, J. P. & Feher, G. (1991). Crystallization of reaction centers from Rhodobacter Sphaeroides. In: Crystallization of Membrane Proteins, ed. H. Michel, pp. 137-153. CRC Press.
Amos, L. A., Henderson, R. & Unwin, P. N. T. (1982) Three- dimensional structure determination by electron microscopy of 2-D crystals. Progress in Biophysics and Molecular Biology 39: 183-231.
Andersson, L. & Porath, J. (1986) Isolation of phosphoproteins by immobilized metal (Fe3+) affinity chromatography. Analytical Biochemistry 154: 250-254.
Applebury, M. A. (1991) Molecular determinants of visual pigment function. Current Opinion in Neurobiology 1: 263-269.
Applebury, M. L. & Hargrave, P. A. (1986) Molecular biology of the visual pigments. Vision Research 26: 1881-1895.
Arnoux, B., Ducruix, A., Reiss-Husson, F., Lutz, M., Norris, J., Schiffer, M. & Chang, C. G. (1989) Structure of spheroidene in the photosynthetic reaction center from Y Rhodobacter sphaeroides. FEBS Letters 258: 47-50.
Aton, B. R., Litman, B. J. & Jackson, M. L. (1984) Isolation and Identification of the Phosphorylated Species of Rhodopsin. Biochemistry 23: 1737-1741.
Baylor, D. A. (1987) Photoreceptor signals and vision. Investigative Ophthalmology & Visual Science 28: 34-49.
Bhattacharya, S., Ridge, K. D., Knox, B. E. & Khorana, H. G. (1992) Light-Stable Rhodopsin .1. A Rhodopsin Analog Reconstituted with a Nonisomerizable 11-cis Retinal Derivative. Journal of Biological Chemistry 267: 6763-6769.
Boekema, E. J., Van Heel, M. G. & Van Bruggen, E. F. J. (1986) Preparation of two-dimensional crystals of Complex I and image analysis. Methods in Enzymology 126: 344-353.
Bornancin, F., Pfister, C. & Chabre, M. (1989) The transitory complex between photoexcited rhodopsin and transducin. Reciprocal interaction between the retinal site in rhodopsin and the nucleotide site in transducin. European Journal of Biochemistry 184: 687-698.
Campbell, J. W., Duee, E., Hodgson, G., Mercer, W. D., Stammers, D. K., Wendell, P. L., Muirhead, H. & Watson, H. C. (1971) X-ray diffraction studies on enzymes in the glycolytic pathway. Cold Spring Harbor Symposium on Quantitative Biology 36: 165-170.
Chabre, M. (1985) Molecular mechanism of visual phototransduction in retinal rod cells. Annual Review of Biophysics and Biophysical Chemistry 14: 331-360.
Chabre, M. & Deterre, P. (1989) Molecular mechanism of visual transduction. European Journal of Biochemistry 179: 255-266.
Cone, R. A. (1972) Rotational Diffusion of Rhodopsin in the Visual Receptor Membrane. Nature New Biology 236: 39-43.
Corless, J. M., McCaslin, D. R. & Scott, B. L. (1982) Two- dimensional rhodopsin crystals from disk membranes of frog retinal rod outer segments. Proceedings of the National Academy of Sciences of the United States of America 79: 1116- 1120.
Cowan, S. W., Schirmer, T., Rummel, G., Steiert, M., Ghosh, R., Pauptit, R. A., Jansonius, J. N. & Rosenbusch, J. P. (1992) Crystal structures explain functional properties of two E. coli porins. Nature 358: 727-733.
de Grip, W. J. (1982) Purification of bovine rhodopsin over concanavalin A-Sepharose. Methods in Enzymology 81: 197-207.
de Grip, W. J. (1982) Thermal stability of rhodopsin and opsin in some novel detergents. Methods in Enzymology 81: 256-265.
de Grip, W. J., Olive, J. & Bovee-Geurts, P. H. M. (1983) Reversible modulation of rhodopsin photolysis in pure phosphatidylserine membranes. Biochimica et Biophysica Acta 734: 168-179.
de Grip, W. J., van Oostrum, J., Bovee-Geurts, P. H. M., van der Steen, R., van Amsterdam, L. J. P., Groesbeek, M. & Lugtenburg, J. (1990) 10,20-Methanorhodopsins: (7E,9E,13E)- 10,20,methanorhodopsin and (7E,9Z,13Z)-10,20- methanorhodoopsin. 11-cis-Locked rhodopsin analog pigments with unusual thermal and photo-stability. European Journal of Biochemistry 191: 211-220.
Berson (1993) Retinitis Pigmentosa. Invest. Ophthalmol. & Vis. Sci. 34: 1655-1676.
de Grip, W. J., Van Oostrum, J. & De Caluw, G. L. J. (1992) Studies towards the crystallization of the rod visual pigment rhodopsin. Journal of Crystal Growth 122: 375-384.
Deese, A. J., Dratz, E. A., Dahlquist, F. W. & Paddy, M. R. (1981) Interaction of rhodopsin with two unsaturated phosphatidylcholines: a deuterium nuclear magnetic resonance study. Biochemistry 20: 6420-6427.
Deisenhofer, J. & Michel, H. (1989) The photosynthetic reaction center from the purple bacterium Rhodopseudomonas viridis. Science 245: 1463-1473.
Demin, V. V., Yurkova, E. V., Kuzin, A. P., Barnakov, A. N. & Abdulaev, N. G. (1987). Crystallization of bovine retinal rhodopsin. Retinal Proteins. pp. 519-524. VNU Science Press.
Dencher, N. A., Dresselhaus, D., Zaccai, G. & B ldt, G. (1989) Structural changes in bacteriorhodopsin during proton translocation revealed by neutron diffraction. Proceedings of the National Academy of Sciences of the United States of America 86: 7876-7879.
Derguini, F. & Nakanishi, K. (1986) Synthetic rhodopsin analogs. Photobiochemistry & Photobiophysics 13: 259-283.
Dorset, D. L., Engel, A., H ner, M., Massalski, A. & Rosenbusch, J. P. (1983) Two-dimensional crystal packing of matrix porin. Journal of Molecular Biology 165: 701-710.
Downer, N. W. (1985) Cross-linking of dark-adapted frog photoreceptor disk membranes. Evidence for monomeric rhodopsin. Biophysical Journal 47: 285-293.
Dratz, E. A. & Hargrave, P. A. (1983) The structure of rhodopsin and the rod outer segment disk membrane. Trends in Biochemical Sciences 8: 128-131.
Dratz, E. A., VanBreemen, J. F. L., Kamps, K. M. P., Keegstra, W. & VanBruggen, E. F. J. (1985) Two-dimensional crystallization of bovine rhodopsin. Biochimica et Biophysica Acta 832: 337-342.
Dryja, T. P. (1992) Rhodopsin and autosomal dominant retinitis pigmentosa. Eye 6: 1-10.
Farnsworth, C. & Dratz, E. A. (1976) Oxidative damage of retinal rod outer segment membranes and the role of vitamin E. Biochimica et Biophysica Acta 443: 556-570.
Feher, G., Atten, J. P., Okamura, M. Y. & Rees, D. C. (1989) Structure and function of bacterial photosynthetic reaction centers. Nature 339: 111-116.
Findlay, J. B. C. (1986). The biosynthetic, functional, and evolutionary implications of the structure of rhodopsin. The Molecular Mechanism of Photoreception. pp. 11-30. Springer- Verlag.
Fong, S.-L., Tsin, A. T. C., Bridges, C. D. B. & Liou, G. I. (1982) Detergents for extraction of visual pigments: Types, solubilization, and stability. Methods in Enzymology 81: 133- 140.
Franke, R. R., Knig, B., Sakmar, T. P., Khorana, H. G. & Hofmann, K. P. (1990) Rhodopsin mutants that bind but fail to activate transducin. Science 250: 123-125.
Franke, R. R., Sakmar, T. P., Graham, R. M. & Khorana, H. G. (1992) Structure and Function in Rhodopsin - Studies of the Interaction Between the Rhodopsin Cytoplasmic Domain and Transducin. Journal of Biological Chemistry 267: 14767-14774.
Fukada, Y., Schichida, Y., Yoshizawa, T., Ito, M., Kodama, A. & Tsukida, K. (1984) Studies on structure and function of rhodopsin by use of cyclopentatrienylidene 11-cis-locked- rhodopsin. Biochemistry 23: 5826-5832.
Fukuda, M. N., Papermaster, D. S. & Hargrave, P. A. (1979) Rhodopsin Carbohydrate. Structure of small oligosaccharides attached at two sites near the NH2 terminus. Journal of Biological Chemistry 254: 8201-8207.
Garavito, R. M. (1991). Crystallizing membrane proteins: experiments on different systems. In: Crystallization of Membrane Proteins, ed. H. Michel, pp. 89-105. CRC Press. .
Hall, M. D., Hoon, M. A., Ryba, N. J. P., Pottinger, J. D. D., Keen, J. N., Sabil, H. R. & Findlay, J. B. C. (1991) Molecular cloning and Primary Structure of Squid (Loligo- forbesi) Rhodopsin, a Phospholipase C-directed G-protein- linked Receptor. Biochemistry Journal 274: 35-40.
Hargrave, P. A., Hamm, H. E. & Hofmann, K. P. (1993) Interaction of rhodopsin with the G-protein, transducin. BioEssays 15: 43-50.
Hargrave, P. A. & McDowell, J. H. (1992) Rhodopsin and Phototransduction - A Model System for G-Protein-Linked Receptors. The FASEB Journal 6: 2323-2331.
Hargrave, P. A. & McDowell, J. H. (1993) Rhodopsin and Phototransduction. International Review of Cytology 137B: 49- 97.
Hargrave, P. A., McDowell, J. H., Feldmann, R. J., Atkinson, P. H., Rao, J. K. M. & Argos, P. (1984) Rhodopsin's protein and carbohydrate structure: Selected aspects. Vision Research 24: 1487-1499.
Hebert, H., Xian, Y., Hacksell, I. & M rdh, S. (1992) Two- dimensional crystals of membrane-bound gastric H,K-ATPase. FEBS Letters 299: 159-162.
Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E. & Downing, K. H. (1990) Model for the structure of bacteriorhodopsin based on high-resolution electron cryomicroscopy. Journal of Molecular Biology 213: 899-929.
Hofmann, K. P. (1986) Photoproducts of rhodopsin in the disc membrane. Photochemistry & Photobiophysics 13: 309-327.
Hong, K. & Hubbell, W. L. (1973) Lipid Requirements for Rhodopsin Regenerability. Biochemistry 12: 4517-4523.
Hovmller, S., Slaughter, M., Berriman, J., Karlsson, B., Weiss, H. & Leonard, K. (1983) Structural studies of cytochrome reductase. Improved membrane crystals of the enzyme complex and crystallization of a subcomplex. Journal of Molecular Biology 165: 401-406.
Hu, S., Franklin, P. J., Wang, J., Ruiz Silva, B. E., Derguini, F. & Nakanishi, K. (1994) Unbleachable rhodopsin with an 11-cis-locked eight- membered ring retinal: the visual transduction process. Biochemistry 33: 408-416.
Ito, M., Kodama, A., Tsukida, K., Fukada, Y., Shichida, Y. & Yoshizawa, T. (1982) A novel rhodopsin analogue possessing the cyclopentatrienylidene structure as the 11-cis-locked and the full planar chromophore. Chemical & Pharmaceutical Bulletin 30: 1913-1916.
Janssen, J. J. M., Mulder, W. R., De Caluw, G. L. J., Vlak, J. M. & J., D. G. W. (1991) In vitro expression of bovine opsin using recombinant baculovirus: the role of glutamic acid (134) in opsin biosynthesis and glycosylation. Biochimica et Biophysica Acta 1089: 68-76.
Khan, S. M. A., Bolen, W., Hargrave, P. A., Santoro, M. M. & McDowell, J. H. (1991) Differential scanning calorimetry of bovine rhodopsin in rod outer segment disk membranes. European Journal of Biochemisty 200: 53-59.
Khorana, H. G. (1992) Rhodopsin, Photoreceptor of the Rod Cell - An Emerging Pattern for Structure and Function. Journal of Biological Chemistry 267: 1-4.
Kjelsberg, M. A., Cotecchia, S., Ostrowski, J., Caron, M. G. & Lefkowitz, R. J. (1992) Constitutive Activation of the alpha-1B-Adrenergic Receptor by All Amino Acid Substitutions at a Single Site - Evidence for a Region Which Constrains Receptor Activation. Journal of Biological Chemistry 267: 1430-1433.
Knig, B., Arendt, A., McDowell, J. H., Kahlert, M., Hargrave, P. A. & Hofmann, K. P. (1989) Three cytoplasmic loops of rhodopsin interact with transducin. Proceedings of the National Academy of Sciences of the United States of America 86: 6878-6882.
K hlbrandt, W. (1984) Three-dimensional structure of the light-harvesting chlorophyl a/b-protein complex. Nature 307: 478-480.
K hlbrandt, W. (1988) Three-dimensional crystallization of membrane proteins. Quarterly Reviews of Biophysics 21: 429- 477.
K hlbrandt, W. (1992) Two-dimensional crystallization of membrane proteins. Quarterly Reviews of Biophysics 25: 1-49.
K hlbrandt, W., Wang, D. N. & Fujiyoshi, Y. (1994) Atomic model of plant light-harvesting complex by electron crystallography. Nature 367: 614-621.
K hn, H. (1980) Light-and GTP-regulated interaction of GTPase and other proteins with bovine photoreceptor membranes. Nature 283: 587-589.
K hn, H. & McDowell, J. H. (1977) Isoelectric focusing of phosphorylated cattle rhodopsin. Biophysics of Structure and Mechanism 3: 199-203.
Liang, C.-J., Yamashita, K., Muellenberg, C. G., Shichi, H. & Kobata, A. (1979) Structure of the carbohydrate moieties of bovine rhodopsin. Journal of Biological Chemistry 254: 6414- 6418.
Liebman, P., Parker, K. R. & Dratz, E. A. (1987) The molecular mechanism of visual excitation and its relation to the structure and composition of the rod outer segment. Annual Review of Physiology 49: 765-791.
Litman, B. J. (1982) Purification of rhodopsin by concanavalin A affinity chromatography. Methods in Enzymology 81: 150-153.
Mariuzza, R. A., Phillips, S. E. V. & Poljak, R. J. (1987) The structural basis of antigen-antibody recognition. Annual Review of Biophysics and Biophysical Chemistry 16: 139-159.
McDowell, J. H. (1993). Preparing rod outer segment membranes, regenerating, and determining rhodopsin concentration. In: Photoreceptor Cells, ed. P. A. Hargrave, Academic Press, pp 123-130.
McDowell, J. H., Khan, S. M. A., Bolen, D. W., Santoro, M. M. & Hargrave, P. A. (1992). Structural stability of rhodopsin and opsin studied by differential scanning calorimetry. In: Signal Transduction in Photoreceptor Cells. ed. P. A. Hargrave, K. P. Hofmann & U. B. Kaupp, pp. 31-41. Springer- Verlag.
McFall-Ngai, M. J. & Horwitz, J. (1990) A comparative study of the thermal stability of the vertebrate eye lens: Antarctic ice fish to the desert iguana. Exp. Eye Res. 50: 703-709.
Michel, H. (1991). General and practical aspects of membrane protein crystallization. In: Crystallization of Membrane Proteins, ed. H. Michel, pp. 73-88. CRC Press.
Miljanich, G. P., Brown, M. F., Mabry-Gaud, S., Dratz, E. A. & Sturtevant, J. M. (1985) Thermotropic Behavior of Retinal Rod Membranes and Dispersions of Extracted Phospholipids. The Journal of Membrane Biology 85: 79-86.
Miller, K. R. & Jacobs, J. S. (1983) Two-dimensional crystals formed from photosynthetic reaction centers. Journal of Cell Biology 97: 1266-1270.
Mitchell, D. C., Straume, M. & Litman, B. J. (1992) Role of sn-1-saturated,sn-2-polyunsaaturated phospholipids in control of membrane receptor conformational equilibrium: effects of cholesterol and acyl chain unsaturation on the Metarhodopsin I<>Metarhodopsin II equilibrium. Biochemistry 31: 662-670.
Molday, R. S. (1989) Monoclonal antibodies to rhodopsin and other proteins of rod outer segments. Progress in Retinal Research 8: 173-209.
Nathans, J. (1990) Determinants of visual pigment absorbance: Role of charged amino acids in the putative transmembrane segment. Biochemistry 29: 937-942.
Nathans, J. (1992) Rhodopsin: Structure, Function, and Genetics. Biochemistry 31: 4923-4931.
Nathans, J., Weitz, C. J., Agarwal, N., Nir, I. & Papermaster, D. S. (1989) Production of bovine rhodopsin by mammalian cell lines expressing cloned cDNA: spectrophotometry and subcellular localization. Vision Research 29: 907-914.
O'Brien, P. J., St. Jules, R. S., Reddy, T. S., Bazan, N. G. & Zatz, M. (1987) Acylation of disc membrane rhodopsin may be nonenzymatic. Journal of Biological Chemistry 262: 5210-5215.
O'Brien, P. J. & Zatz, M. (1984) Acylation of Bovine Rhodopsin by [3H]Palmitic Acid. Journal of Biological Chemistry 259: 5054-5057.
Oprian, D. D. (1992) The ligand-binding domain of rhodopsin and other G
protein-linked receptors. J. Bioenerg. & Biomembranes 24: 211-217.
Oprian, D. D., Molday, R. S., Kaufman, R. J. & Khorana, H. G. (1987) Expression of a synthetic bovine rhodopsin gene in monkey kidney cells. Proceedings of the National Academy of Sciences of the United States of America 84: 8874-8878.
Ovchinnikov, Y. A., Abdulaev, N. G. & Bogachuk, A. S. (1988) Two adjacent cysteine residues in the C-terminal cytoplasmic fragment of bovine rhodopsin are palmitylated. FEBS Letters 230: 1-5.
Ovchinnikov, Y. A., Abdulaev, N. G., Zolotarev, A. S., Artamonov, I. D., Bespalov, I. A., Dergachev, A. E. & Tsuda, M. (1988) Octopus rhodopsin: Amino acid sequence deduced from cDNA. FEBS Letters 232: 69-72.
Papac, D. I., Thornburg, K. R., Bullesbach, E. E., Crouch, R. K. & Knapp, D. R. (1992) Palmitylation of a G-Protein Coupled Receptor - Direct Analysis by Tandem Mass Spectrometry. Journal of Biological Chemistry 267: 16889-16894.
Papermaster, D. S. & Dreyer, W. J. (1974) Rhodopsin Content in the Outer Segment Membranes of Bovine and Frog Retinal Rods. Biochemistry 13: 2438-2444.
Plantner, J. J., Le, M.-L. & Kean, E. L. (1991) Enzymatic deglycosylation of bovine rhodopsin. Experimental Eye Research 53: 269-274.
Probst, W. C., Snyder, L. A., Schuster, D. I., Brosius, J. & Sealfon, S. C. (1992) Sequence Alignment of the G-Protein Coupled Receptor Superfamily. DNA and Cell Biology 11: 1-20.
Ridge, K. D., Bhattacharya, S., Nakayama, T. A. & Khorana, H. G. (1992) Light-Stable Rhodopsin .2. An Opsin Mutant (Trp-265 -> Phe) and a Retinal Analog with a Nonisomerizable 11-Cis Configuration Form a Photostable Chromophore. Journal of Biological Chemistry 267: 6770-6775.
Robinson, P. R., Cohen, G. B., Zhukovsky, E. A. & Oprian, D. D. (1992) Constitutively Active Mutants of Rhodopsin. Neuron 9: 719-725.
Ryba, N. J. P. & Marsh, D. (1992) Protein Rotational Diffusion and Lipid/Protein Interactions in Recombinants of Bovine Rhodopsin with Saturated Diacylphosphatidylcholines of Different Chain Lengths Studied by Conventional and Saturation-Transfer Electron Spin Resonance. Biochemistry 31: 7511-7518.
Sakmar, T. P., Franke, R. R. & Khorana, H. G. (1989) Glutamic acid-113 serves as the retinylidine schiff base counterion in bovine rhodopsin. Proceedings of the National Academy of Sciences of the United States of America 86: 8309-8313.
Schertler, G. F. X. (1992) Overproduction of membrane proteins. Current Opinion in Structural Biology 2: 534-544.
Schertler, G. F. X. & Hargrave, P. A. (1994) 6 projection structure of frog rhodopsin. manuscript in preparation.
Schertler, G. F. X., Villa, C. & Henderson, R. (1993) Projection structure of rhodopsin. Nature 362: 770-772.
Shichi, H., Lewis, M. S., Irreverre, F. & Stone, A. L. (1969) Biochemistry of visual pigments. I. Purification and properties of bovine rhodopsin. Journal of Biological Chemistry 244: 529-536.
Shnyrov, V. L. & Berman, A. L. (1988) Calorimetric study of thermal denaturation of vertebrate visual pigments. Biomedica et Biochimica Acta 47: 355-362.
Shuster, T. A., Nagy, A. K., Conly, D. C. & Farber, D. B. (1992) Direct Zinc Binding to Purified Rhodopsin and Disc Membranes. Biochemical Journal 282: 123-128.
Subramaniam, S., Gerstein, M., Osterhelt, D. & Henderson, R. (1993) Electron diffraction analysis of structural changes in the photocycle of bacteriorhodopsin. The EMBO Journal 12:1-8.
Toyoshima, C. & Unwin, N. (1990) Three-dimensional structure of the acetylcholine receptor by cryoelectron microscopy and helical image reconstruction. Journal of Cell Biology 111: 2623-2635.
Unwin, P. N. T. & Zampighi, G. (1980) Structure of the junction between communicating cells. Nature 283: 545-549.
van der Steen, R., Groesbeek, M., van Amsterdam, L. J. P., Lugtenburg, J., van Oostrum, J. & de Grip, W. J. (1989) All E-10,20-methanoretinylopsin, light-stable rhodopsin. Synthesis and spectroscopy of all E-10,20-methano- and all-E- retinoyl fluoride and their reaction with bovine opsin. Recueil des Travaux Chimiques des Pays-Bas 108: 20-27.
Vanderkooi, G. (1974) Organisation of proteins in membranes with special reference to the cytochrome oxidase system. Biochimica et Biophysica Acta 344: 307-345.
Wehner, R., Marsh, A. C. & Wehner, S. (1992) Desert ants on a thermal tightrope. Nature 357: 586-587.
Weiss, M. S., Kreusch, A., Schiltz, E., Nestel, U., Welter, W., Wechesser, J. & Schulz, G. E. (1991) The structure of porin from Rhodobacter capsulatus at 1.8 A resolution. FEBS Letters 280: 379-382.
Welte, W. & Wacker, T. (1991). Protein-detergent micellar solutions for the crystallization of membrane proteins: some general approaches and experiences with the crystallization of pigment-protein complexes from purple bacteria. In: Crystallization of Membrane Proteins, ed. H. Michel, pp. 107-123. CRC Press.
Yurkova, E. V., Demin, V. V. & Abdulaev, N. G. (1990) Crystallization of membrane proteins: bovine rhodopsin. Biomedical Science 1: 585-590.
Zankel, T., Ok, H., Johnson, R., Chang, C. W., Sekiya, N., Naoki, H., Yoshihara, K. & Nakanishi, K. (1990) Bovine rhodopsin with 11-cis-locked retinal chromophore neither activates rhodopsin kinase nor undergoes conformational change upon irradiation. Journal of the American Chemical Society 112: 5387-5388.
Zhukovsky, E. A. & Oprian, D. D. (1989) Effect of carboxylic acid side chains on the absorption maximum of visual pigments. Science 245: 928-930.
This manuscript was prepared for delivery at the conference "Controversies in Neuroscience III: Signal Transduction in the Retina and Brain", at the Robert S. Dow Neurological Sciences Institute and Good Samaritan Hospital & Medical Center, Portland, Oregon, on October 31-November 1, 1992.