Rhodopsin Structure, Function, and Topography The Friedenwald Lecture

The increase in knowledge of rhodopsin structure and function has been nothing short of phenomenal over the past 30 years. Although some of rhodopsin’s properties had been investigated by Kühne as early as the 1870s, the characterization of rhodopsin’s molecular properties was still quite rudimentary into the late 1960s. ¹ Currently, there is information concerning the complete amino acid sequence of as many as 100 vertebrate visual pigments, a wealth of information about the function of rhodopsin in health and disease and a three-dimensional structure of rhodopsin with a resolution of 5 Å. Moreover, the knowledge of rhodopsin structure and function has led to an appreciation for how the entire class of G-protein–linked receptors may function (for reviews see Refs. ² and ³ ). 

Rhodopsin in Rod Cell Function

The vertebrate rod cell consists of an outer segment that contains a stack of rhodopsin-containing disc membranes connected to an inner segment via a ciliary process (Fig. 1) . The inner segment contains the metabolic machinery of the rod cell. Rhodopsin is synthesized in the endoplasmic reticulum and passes to the Golgi membranes where it becomes glycosylated. Rhodopsin-containing vesicles move from the Golgi to the outer segment where they fuse with the outer segment plasma membrane. Morphologically separate disc membranes are formed by evagination and pinching off of the plasma membrane. The regions of rhodopsin that were facing the outside of the cell are now facing the inside surface of the disc membrane. The regions of rhodopsin that were facing the outer segment cytoplasm remain facing the cytoplasm when the plasma membrane forms disc membranes. Rhodopsin comprises more than 90% of the protein content of these disc membranes. It took years of work by many laboratories to develop a working model of rhodopsin (Fig. 1) as a bundle of transmembrane helices that encompass a binding pocket for the light-sensitive 11-cis-retinal. When light strikes a molecule of rhodopsin in the rod cell, a photon is absorbed, causing isomerization of rhodopsin’s 11-cis-retinal to the all-trans isomer. This causes a change in the conformation of rhodopsin that stimulates transducin, setting off a biochemical amplification cascade that results in a drop in cGMP concentration. That in turn leads to hyperpolarization of the plasma membrane and the signaling of second-order neurons. 

The molecules involved in these processes are shown in Figure 2 . When rhodopsin (R) becomes photoactivated (R*), the G-protein transducin (T) can now bind and become activated (T*). This begins the phototransduction cascade. R* would continue to activate T for a long time unless there were some mechanism to shut it off. In vivo that mechanism depends on phosphorylation. Rhodopsin kinase phosphorylates specific serines and threonines in rhodopsin’s carboxyl-terminal sequence. This phosphorylation reduces the ability of transducin to bind, but does not totally eliminate it. However, now arrestin can bind to the phosphorylated, photoactivated rhodopsin (R*P), and this totally prevents further activation of transducin. This complex eventually decays as the all-trans-retinal dissociates, and the phosphates are removed by a phosphatase. The all-trans-retinal is regenerated in a series of enzymatic steps to 11-cis-retinal, and the retinal rebinds to the protein opsin, regenerating rhodopsin. 

Rhodopsin’s Amino Acid Sequence

Basic to an understanding of the function of any protein is its amino acid sequence—its primary structure. Such information provides the groundwork for understanding how the protein carries out its cellular functions. Currently, the complete amino acid sequence of a protein is generally determined by inference from its nucleic acid sequence. When the sequence analysis of rhodopsin was undertaken in the 1970s, one needed to apply chemical and enzymatic methods to the protein. Automated amino acid sequencing could be applied directly to an intact protein to determine part of its N-terminal sequence. However, this was not an option for bovine rhodopsin because its N-terminal amino acid was blocked. When rhodopsin was enzymatically digested to a mixture of peptides, a peptide with a blocked N terminus was identified and purified, and its sequence was determined. ⁴ The only previously reported sequence of rhodopsin, a 9-amino acid glycopeptide, was contained within its sequence. ⁵ However, the larger N-terminal glycopeptide was found to contain two sites of oligosaccharide attachment, at Asn² and Asn¹⁵. ⁴  

Because the varieties of oligosaccharides attached to proteins are assumed to have functional importance, it was of interest to characterize the nature of these oligosaccharides. In collaboration with David Papermaster and Michiko Fukuda from the University of Connecticut, the two sites were found to contain predominantly the uniquely small GlcNAc₃Man₃, with smaller amounts of chains containing Man₄ and Man₅. ⁶ Similar results were obtained by others. ⁷ Only recently has a functional consequence of rhodopsin glycosylation become apparent. Among those cases of retinitis pigmentosa in which rhodopsin mutations are implicated as causing the disease process, threonine residues Thr⁴ or Thr¹⁷ are mutated. These specific mutations result in a failure of rhodopsin’s immediately preceding Asn residue to become glycosylated. When a Thr¹⁷ mutant rhodopsin was expressed by Kaushal et al. ⁸ and its properties were studied, it was found that the mutant could fold to an apparently correct ground-state structure. However, its metarhodopsin II photoproduct was unstable and had an abnormally short lifetime, rendering it defective in signal transduction. Therefore, glycosylation of Asn¹⁵ is required to have a rhodopsin that can function properly in signal transduction. 

The carboxyl-terminal amino acid of rhodopsin was found to be alanine. ⁹ From tryptic and cyanogen bromide digests of rhodopsin, the C-terminal peptide (terminating in alanine) was identified and purified, and its sequence was determined. ^{9 10} This sequence, Asp-Asp-Glu-Ala-Ser-Thr-Thr-Val-Ser-Lys-Thr-Glu-Thr-Ser-Gln-Val-Ala-Pro-Ala, contains many hydrophilic amino acids and is particularly rich in serine and threonine. 

Within the same year, three laboratories independently discovered that rhodopsin became phosphorylated in a light-dependent reaction. Because phosphorylation of proteins is frequently associated with metabolic control processes, it was of interest to further characterize the site and characteristics of this reaction. When rhodopsin was phosphorylated with endogenous rod cell kinase(s) using[ ³²P]ATP, almost all the phosphate could be found in a single peptide, which was identified as the Ser/Thr-rich C-terminal peptide. ¹⁰ It was evident that the sites of phosphorylation in bovine rhodopsin did not resemble those types of protein sequences that were know to become phosphorylated by well-known kinases such as protein kinase A. It became clear that the phosphorylation was being carried out by a new and different kinase, rhodopsin kinase, ¹¹ and that this kinase had its own distinct sequence recognition. ¹² The kinetics of the in vitro reaction shows a steady progression to high levels of phosphorylation, ¹³ although the in vivo reaction may be more limited in its extent. There seems to be general agreement that the early and preferred sites of phosphorylation are Ser³⁴³ and Ser³³⁸. ^{14 15 16}  

Rhodopsin’s Topography

In parallel with the efforts to determine the sequence of rhodopsin and characterize its functional sites were efforts to determine its topography within the rod cell membrane. Transmembrane proteins like rhodopsin expose different parts of their sequence on opposite membrane surfaces. It is these surface-exposed regions that have the potential for functional interactions with other proteins and with small molecules. 

Following earlier work, ¹⁷ rhodopsin in intact disc membranes was digested with the proteolytic enzyme thermolysin, as a probe for these exposed regions. Rhodopsin was converted to a slightly lower molecular weight, in a rapid reaction. ⁹ This was accompanied by release of two peptides from its sequence, Val-Ser-Lys-Thr-Glu-Thr-Ser-Gln and Val-Ala-Pro-Ala, peptides readily recognized to be from rhodopsin’s newly determined carboxyl-terminal sequence. ⁹ This provided clear evidence that the carboxyl-terminal region of rhodopsin was highly exposed on the cytoplasmic disc membrane surface. Additional digestion of rhodopsin by thermolysin converted rhodopsin to two lower molecular weight fragments, F1 and F2. Fragment F1 contained carbohydrate and therefore derived from the amino-terminal region of rhodopsin. When the digestion was performed on rhodopsin that had been phosphorylated with[ ³²P]ATP, fragment F2 contained ³²P, demonstrating that fragment F2 originated from rhodopsin’s carboxyl-terminal region. ⁹  

To obtain larger quantities of both fragments for use in structural studies, the thermolysin digestion of membrane-bound rhodopsin was performed on a preparative scale. The digested rhodopsin was solubilized in mild detergent and chromatographically purified as the noncovalent F1–F2 complex. ¹⁰ Separation of the fragments was accomplished by dissolving in concentrated formic acid and separating by size on a gel filtration column in formic acid/ethanol. The hydrophobic nature of these large polypeptides, derived from an intrinsic membrane protein, required the use of such drastic conditions. 

Automated sequencing of the F2 fragment, as well as peptides derived from it, was successful in completing the amino acid sequence of the carboxyl-terminal one third of the sequence of bovine rhodopsin, 108 amino acids in length. ¹⁸ The sequence showed alternating segments of hydrophilic and hydrophobic amino acids with the two predominantly hydrophobic stretches of 24- and 27-amino acids in length being separated by a short hydrophilic sequence. The lengths of the hydrophobic segments were the correct distance to traverse the hydrophobic portion of the membrane lipid bilayer. This suggested a model for the F2 fragment in which its N-terminal and C-terminal were located on the cytoplasmic surface of the disc membrane (because they were formed by proteolytic digestion at the membrane surface), whereas the two hydrophobic sequences would traverse the lipid bilayer and expose their hydrophilic connecting segment on the opposing membrane surface. 

Evidence was accumulating that rhodopsin’s N-terminal sequence was exposed to the inside surface of the disc membrane (and the external surface of the rod outer segment plasma membrane). Chemical evidence was provided by the reaction of a radioactive nitrene precursor, nitroazidophenyl taurine (NAP-taurine). When this membrane impermeant reagent was reacted with rhodopsin in intact disc membranes, peptides from rhodopsin’s carboxyl terminus became radioactively labeled, but rhodopsin’s N-terminal peptide was poorly labeled. ¹⁹ When the same reaction was performed on reconstituted membranes in which rhodopsin’s orientation was randomized, the relative modification of the N terminus was greatly enhanced, suggesting that it had been previously unavailable because of its exposure on the opposite membrane surface. 

Additional support came from immunocytochemical studies. Grazyna Adamus in our laboratory prepared anti-rhodopsin mouse monoclonal antibodies that were specific to N-terminal sequences in rhodopsin ²⁰ as well as antisera against rhodopsin’s synthetic N-terminal peptide. Experiments by Polans and Papermaster, ²¹ using these and other reagents, demonstrated that rhodopsin’s amino-terminal sequence was exposed on the outside surface of rod cell outer segments. 

Completion of the amino acid sequence of bovine rhodopsin, ^{22 23} in conjunction with the topographic information summarized above, made it possible to develop a topographic model for the organization of rhodopsin within the lipid bilayer (Fig. 3) . Such a model proposed that rhodopsin’s amino terminal segment was exposed on the inside surface of the disc membrane (or outer surface of the plasma membrane) and that the carboxyl-terminal sequence was exposed on the cytoplasmic surface. The polypeptide chain traversed the lipid bilayer seven times, embedding hydrophobic sequences in the lipid side chain environment and exposing hydrophilic linking regions to the membrane surface aqueous environments. The information derived from topographic studies placed constraints on construction of the model and made predictions about the orientation of sequence segments for which evidence was still to be obtained. 

The model (Fig. 3) predicted that rhodopsin’s loop e2 (amino acid sequence 177–202) should be exposed on the outside surface of the rod cell outer segment. An antibody raised against proteins from turtle retina was subsequently identified as an anti-rhodopsin antibody, and peptide competition demonstrated that its specificity was against the e2 loop. This antibody labeled the outside surface of intact frog rod outer segments, and its binding was abolished in the presence of the e2 synthetic peptide, demonstrating that rhodopsin indeed exposed its e2 loop to the extracellular environment. ²⁴  

Rhodopsin Function

Proteins interact with other molecules—substrates, inhibitors, ions, nucleic acids, carbohydrates, lipids, and other proteins—as ways in which they carry out their various functions. By the time rhodopsin reaches its membrane location in the outer segment, it has already interacted with dozens of proteins involved in such activities as its biosynthesis, acetylation, glycosylation, and transport. It has bound two molecules of palmitate and one molecule of 11-cis retinal and has become embedded in a membrane lipid bilayer. Rhodopsin is now ready to carry out its role in phototransduction. 

After photoactivation, rhodopsin exerts its effect on rod cell biochemistry by interacting with rod cell proteins through its cytoplasmic surface. As described previously (Fig. 2) , photoactivated rhodopsin (R*) binds and activates the G-protein, transducin, initiating the phototransduction cascade. R* is then phosphorylated by rhodopsin kinase, and the resulting R*P is bound by arrestin, preventing further activation of transducin. Thus, rhodopsin interacts with three major proteins—transducin, rhodopsin kinase, and arrestin—and is then dephosphorylated by a phosphatase, as part of its functional cycle. 

Rhodopsin Interaction with Transducin

The nature of the interaction with transducin was first explored using thermolysin-digested rhodopsin. Removal of part of rhodopsin’s carboxyl-terminal sequence has no effect on the binding of transducin, indicating that this sequence in the molecule is not involved in transducin binding. ²⁵ However, proteolytic digestion of the third cytoplasmic loop abolishes binding of transducin, suggesting involvement of that portion of the rhodopsin sequence in transducin binding. 

Confirmation of the role of the third cytoplasmic loop in transducin binding was provided by a totally independent set of experiments. In collaboration with the laboratory of K. Peter Hofmann, Humboldt University, Berlin, the sites of interaction of photoactivated rhodopsin (metarhodopsin II) and transducin were probed using a spectroscopic method. This method depends on the ability to spectroscopically measure the amounts of metarhodopsins I and II in their equilibrium mixture. Transducin binds selectively to metarhodopsin II, shifting the equilibrium and creating more metarhodopsin II (and if transducin dissociates, this extra amount of metarhodopsin II is decreased as the metarhodopsin I and II equilibrium is reestablished). When a synthetic peptide comprising the sequence of rhodopsin’s third cytoplasmic loop was added to such a transducin/metarhodopsin II complex, it displaced metarhodopsin II, as monitored spectroscopically. ²⁶ This provided additional evidence that the third cytoplasmic loop was a site of binding for transducin on the metarhodopsin II surface. Similar peptide competition studies showed that the second and fourth cytoplasmic loops (but not the first loop or the C-terminal sequence) were also sites of binding for transducin. These and related studies have been reviewed. ²⁷  

Rhodopsin Kinase

Initial studies on rhodopsin phosphorylation were performed in vitro using rod outer segments. The properties of the kinase(s) phosphorylating rhodopsin remained to be determined. Rhodopsin kinase became recognized as the kinase that catalyzed light-dependent rhodopsin phosphorylation, and it was found to be a representative of a new class of kinases—those phosphorylating G-protein–linked receptors (reviewed in Ref. ²⁸ ). Purification of rhodopsin kinase assisted in determining its properties, and the discovery of conditions for its stabilization and storage aided in its study. ¹¹ One of the ways to determine kinase specificity is to test synthetic peptides as substrates, and this method was used for rhodopsin kinase. Rhodopsin kinase can phosphorylate synthetic peptides from rhodopsin’s carboxyl-terminal sequence, but with K _m values that are approximately three orders of magnitude higher than that for rhodopsin. ^{12 29} This behavior is in contrast to that of a kinase such as protein kinase A, which phosphorylates synthetic peptides containing its site of phosphorylation nearly as well as its entire protein substrate. In addition, some peptides from rhodopsin’s cytoplasmic surface (particularly the third cytoplasmic loop) inhibit the phosphorylation reaction. ²⁹ These experiments are done by performing the phosphorylation of membrane-bound, freshly photolyzed rhodopsin in the presence of a small amount of synthetic peptide representing sequences from the cytoplasmic surface of rhodopsin. The amount of peptide is too small to become phosphorylated, if it were a substrate, but sufficient to compete for binding sites on rhodopsin by the kinase. Taken together these findings suggest that more than one part of rhodopsin’s surface is involved in the interaction between rhodopsin and rhodopsin kinase. Studies from other laboratories have further implicated cytoplasmic loop i3 ³⁰ as a binding site for rhodopsin kinase, because rhodopsin kinase phosphorylates target serines in rhodopsin’s carboxyl-terminal sequence. 

Rhodopsin Dephosphorylation

Proteins that become phosphorylated for purposes of metabolic control become dephosphorylated by the action of phosphatases. Kühn, in an elegant study, ³¹ demonstrated both the phosphorylation and dephosphorylation of rhodopsin in vivo in frogs. Because there is great specificity in the phosphorylation of rhodopsin (rhodopsin kinase phosphorylates only freshly photolyzed rhodopsin and no other substrate), it is of interest to determine the specificity of the dephosphorylation reaction. There is clearly some specificity in this reaction because of the finding that it is catalyzed by the catalytic subunit of protein phosphatase 2A but not by protein phosphatases 1, 2B, or 2C. ^{32 33} However, this is a phosphatase that can act on other protein substrates. The basis of its ability to dephosphorylate phosphorylated rhodopsin is not known but may derive from a specificity residing in its noncatalytic subunits. Protein phosphatase 2A represents one more in the growing list of proteins that interact with rhodopsin, whose presence and activity are necessary for rhodopsin function. The action of protein phosphatase 2A on photoactivated, phosphorylated rhodopsin is blocked in the presence of arrestin. ³³  

Arrestin

Although arrestin has had a separate life in the vision immunology literature as S-antigen, it first came to notice in the vision biochemistry literature as a MW_app 48-kDa protein that bound reversibly to photolyzed, phosphorylated rhodopsin. ³⁴ This stimulated an effort to determine the role and mechanism of action of arrestin in rod cell biochemistry. In the search for other molecules that might bind to arrestin as part of its mechanism of action, there were many conflicting reports concerning arrestin’s biochemical properties. Using a highly purified preparation of arrestin, it was clear that it did not bind Ca²⁺ or the nucleotides ATP or GTP ³⁵ and thus was not modulated by fluxes of these important rod cell compounds. The specificity of arrestin’s binding was localized to its carboxyl-terminal sequence. Native arrestin binds only to photolyzed phosphorylated rhodopsin, but arrestin that is proteolytically truncated at its carboxyl terminus binds to phosphorylated rhodopsin independent of rhodopsin’s exposure to light. ³⁶ This binding is tighter than the binding of native arrestin. Thus, arrestin’s acidic C-terminal region appears to act as a modulator of the binding interaction. 

In an effort to learn about the sites of binding of arrestin to the surface of rhodopsin, synthetic peptides from rhodopsin’s surface were used as competitors in binding tests. These experiments did not clearly identify binding regions, although they had done so in similar experiments with transducin. However, they did lead to identifying the mechanism of arrestin’s binding to rhodopsin. Because binding of arrestin had been shown to occur to photolyzed, phosphorylated rhodopsin but not to photolyzed, unphosphorylated rhodopsin, the phosphorylated region of rhodopsin was thereby implicated as a site of binding. The surprise was in the way that this binding interaction functioned. This was demonstrated by use of a synthetic peptide representing the fully phosphorylated C-terminal sequence of rhodopsin, amino acids 331 to 348. In the presence of this phosphopeptide, arrestin is stimulated to bind to photolyzed, unphosphorylated rhodopsin. ³⁷ Because rhodopsin itself does not need to be phosphorylated, the phosphopeptide is able in some manner to substitute for the phosphorylated region of the protein. Based on a general knowledge of protein function, it was reasonably hypothesized that arrestin was binding the phosphopeptide and then undergoing a change in conformation that enabled it to bind to the (unphosphorylated) surface of photolyzed rhodopsin. 

This led to several experiments, the results of which supported this hypothesis. A K₅₀ of 36 μM was determined for the binding of the phosphopeptide to arrestin. Limited trypsin digestion yielded different arrestin fragments in the presence and absence of phosphopeptide, suggesting that arrestin in the presence of phosphopeptide assumed a different conformation. ³⁷ The ligand-bound form must have presented a different availability of peptide bonds for trypsin to cleave. A change in the conformation of arrestin in the presence of phosphopeptide is also supported by changes in the chemical reactivity of arrestin’s three sulfhydryl groups toward dithionitrobenzoic acid (DTNB). In the presence of phosphopeptide, the rate of reaction of one mole of sulfhydryl group is greatly enhanced, and the rate of another sulfhydryl is reduced. ³⁸ Here the ligand-bound form of the protein leads to a different microenvironment surrounding two of the sulfhydryl groups, affecting their rate of reaction with DTNB. Taken together, these data suggest that the binding of phosphopeptide to arrestin has the same effect as the binding of photolyzed phosphorylated rhodopsin, making it possible for arrestin to then bind to the rhodopsin surface. In a functional assay, the phosphopeptide substitutes for phosphorylated rhodopsin. In rod outer segments, phosphodiesterase activity can be quenched by addition of phosphopeptide in a manner similar to the quenching achieved by light-stimulated rhodopsin phosphorylation. ³⁸  

Although arrestin initially binds to rhodopsin’s phosphorylated C-terminal region, the carboxyl-terminal is not the final site of its binding on rhodopsin’s surface. This can be demonstrated by the use of phosphopeptide to induce binding of arrestin to photolyzed rhodopsin whose 19-amino acid carboxyl-terminal has been proteolytically removed. ³⁷ Studies in the laboratories of Weiss ³⁹ and Benovic ⁴⁰ have demonstrated that arrestin, after its “activation” by binding to the phosphorylated region, then binds to amino acids in rhodopsin’s cytoplasmic loops i1, i2, and i3. 

Rhodopsin’s Three-Dimensional Structure

Basic to an understanding of how a protein carries out its functions is the knowledge of the protein’s three-dimensional structure. Such structural information provides the necessary framework for determining mechanisms of protein action. For rhodopsin it would not only be important to know its structure in its ground state, before photoactivation, but also to know the conformation that it assumes in metarhodopsin II, its signaling state, in which it is able to bind and activate transducin. Knowledge of the signaling state of rhodopsin will also have important implications in helping to understand how the entire class of G-protein–coupled receptors may function. 

The three-dimensional structures of thousands of proteins have been solved by crystallizing the proteins and determining their structures by x-ray diffraction. Only a handful of membrane proteins have had their structures determined in this way because of technical difficulties. Membrane proteins must first be solubilized in detergents, and it has proven difficult to obtain sufficiently large and well-ordered crystals from the detergent complexes of most membrane proteins. The prospects for application of these methods to rhodopsin have been reviewed. ⁴¹ Incremental progress is currently being made in several laboratories in the areas of sample preparation, additives and cryoprotectants for crystallization, production of larger crystals, use of microfocus beam synchrotron radiation, and improved methods of data analysis. 

Fortunately it has been possible to obtain high-quality structural information from membrane proteins that can form well-organized two-dimensional crystals. Frog rhodopsin can be induced to form two-dimensional crystals in frog disc membranes after extraction with weak polyoxyethylene sorbitan detergents. ⁴² By optimizing conditions (detergent type, detergent/protein ratio, pH, buffer composition, and time of extraction) it has been possible to improve the reproducibility, yield, and quality of these crystals. ⁴³  

Samples of membrane preparations were stained and examined by electron microscopy to select preparations containing crystalline arrays (Fig. 4) . Promising membrane preparations were frozen on grids for electron microscopy at liquid nitrogen temperatures. Electron micrographs were inspected by optical diffraction, and the diffraction data were collected and processed for many dozens of crystalline membrane samples. By these methods it was possible to determine that rhodopsin molecules have planar dimensions of 28 × 39 Å and are approximately 64 Å in height. ⁴⁴ Contour cross sections of the electron density map show the position of electron densities representing rhodopsin’s transmembrane helices (Fig. 5A ). For the first time it was possible to resolve each of the seven helices, providing clear biophysical evidence supporting the seven-helix model. ⁴⁵  

By comparing individual slices of the map taken at different distances within the membrane, it is possible to note differences between the helices and in their relationships. Four of the helices (4, 6, and 7) are found at nearly the same position at different levels within the membrane; that is, they are essentially perpendicular to the plane of the membrane. In contrast, the other helices (and helix 3 in particular) are tilted. Near the cytoplasmic surface, helices are packed closely together. Near the intradiscal surface, helices are more spread apart, creating a larger cross-sectional area. A space is created between helices 3 through 7 that is the presumptive binding site for rhodopsin’s light-sensitive chromophore, 11-cis-retinal. From the individual contour cross sections, a solid model has been constructed (Fig. 5B) . From this and subsequent studies with bovine rhodopsin (summarized in Ref. ⁴⁴ ), a clearer picture of rhodopsin’s three-dimensional structure is beginning to emerge. 

Conclusions

Considerable knowledge has been obtained during the past 30 years concerning the structure and function of rhodopsin and its orientation within the lipid bilayer. This has laid the groundwork for understanding how rhodopsin functions as a receptor of photons, how G-protein–coupled receptors may function generally, and how mutations in rhodopsin lead to a malfunctioning protein in diseases such as retinitis pigmentosa. A higher resolution, three-dimensional structure of both rhodopsin and of photoactivated rhodopsin is needed to more precisely understand the molecular basis of receptor function. 

 Supported in part by National Institutes of Health Grants EY06225, EY06226, EY08571 and an unrestricted departmental award from Research to Prevent Blindness. PAH is a Senior Scientific Investigator of Research to Prevent Blindness.

 Submitted for publication July 11, 2000; accepted August 2, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

A vertebrate rod cell, disc membrane, and model for rhodopsin in the membrane. Rod cell outer segments consist of stacks of disc membranes containing the photoreceptor protein rhodopsin. The disc membrane shown in expanded cross section shows the oval shapes of densely packed rhodopsin molecules that traverse the lipid bilayer. To the front right of the figure is a drawing of the polypeptide chain of rhodopsin as it may exist in the lipid bilayer in three dimensions. The N-terminal exposes two attached oligosaccharide chains at the intradiscal surface. The cutaway view shows retinal linked to the seventh helix, in a pocket formed by the inside surface of the helical bundle. Reproduced with permission from Hargrave PA, Hamm HE, Hofmann KP. Interaction of rhodopsin with the G-protein, transducin. Bioessays. 1993;15(1):43–50 © 1993 Wiley-Liss, Inc., a subsidiary of John Wiley& Sons, Inc.

Figure 2.

 

The rhodopsin cycle. Rhodopsin (R) is activated by light, hv, (to R*). R* activates the G-protein, transducin (T). R* then becomes phosphorylated by rhodopsin kinase, which allows it to become inactivated by binding arrestin. The complex R*P-arrestin loses all-trans-retinal and then arrestin, and the phosphorylated opsin (OP) becomes dephosphorylated by the action of protein phosphatase 2A (PrP2A). All-trans-retinal is transformed, through a series of steps, to 11-cis-retinal, which subsequently rebinds opsin, forming rhodopsin. Redrawn from and reproduced with permission from Hargrave PA, McDowell JH. Rhodopsin and phototransduction. Int Rev Cytol. 1992;137B:49–97 © 1992 Academic Press, Inc.

Figure 3.

 

Topographic model for bovine rhodopsin in the membrane lipid bilayer. Rhodopsin’s polypeptide chain traverses the lipid bilayer seven times as the helix segments I through VII. Hydrophilic loop sequences are exposed to the cytoplasmic surface (i1–i4) and to the intradiscal or extracellular surface (e1–e3). The amino-terminal sequence is acetylated at Met¹ and contains oligosaccharide chains at Asn² and Asn¹⁵. Cysteines 110 and 187 are linked by a disulfide bridge, and Cys³²² and Cys³²³ are palmitoylated. Retinal is linked via a Schiff base to Lys²⁹⁶ in helix VII. Reproduced with permission from Hargrave PA, McDowell JH. Rhodopsin and phototransduction. Int Rev Cytol. 1992;137B:49–97 © 1992 Academic Press, Inc.

Figure 4.

 

Micrograph of frog disc membranes after mild detergent extraction. Frog disc membranes were extracted with Tween 80 or a Tween 80/20 mixture and stained with uranyl acetate. ⁴³ Membranes containing crystals of rhodopsin in the p2 crystal form are shown in (A) and (C), and rhodopsin crystals in the p22₁2₁ form are shown in (B) and (D). Reproduced with permission from Schertler GF, Hargrave PA. Projection structure of frog rhodopsin in two crystal forms. PNAS. 1995;92(25):11578–11582© 1995 National Academy of Sciences, USA.

Figure 5.

 

(A) Structure of frog rhodopsin. The seven helices of the frog rhodopsin structure. A slice through the center of the density map (B) is shown. Peaks can be seen for each of the seven transmembrane helices. The grid spacing is 10 Å. Reproduced with permission from Unger VM, Hargrave PA, Baldwin JM, Schertler GF. Arrangement of rhodopsin transmembrane alpha-helices. Nature. 1997;389(6647): 203–206 © 1997 Macmillan Magazines Limited. (B) A solid model of the frog rhodopsin projection map showing a view of helices 1 to 4. The model was constructed from 33 contour sections 2 Å apart. Top: rhodopsin’s cytoplasmic side; bottom: its intradiscal (extracellular) side.

The author thanks, in particular, J. Hugh McDowell, Anatol Arendt, Grazyna Adamus, Kris Palczewski, Shao-Ling Fong, Jaime Puig, and our collaborators the late Hermann Kühn, David Papermaster, K. Peter Hofmann, Gebhard Schertler, W. Clay Smith, and Vijay Sarthy, among the many colleagues who have contributed to these studies.