Received 13 Sep 1996 | Accepted 4 Nov | Published 29 Dec 1996
A Molecular Vision Research Article
Philip L. Yeagle,1,* James L. Alderfer,2 and Arlene D. Albert1,3
1Department of Biochemistry, University at Buffalo School of Medicine and Biomedical Sciences, Buffalo, NY 14214
2Department of Biophysics, Roswell Park Cancer Institute, Buffalo, NY 14263
3Department of Ophthalmology, University at Buffalo School of Medicine and Biomedical Sciences, Buffalo, NY 14214
*Address correspondance to Philip L. Yeagle <firstname.lastname@example.org>.
Purpose. High resolution structural information is lacking for any member of the class of G-protein receptors. This dearth of structural information extends to virtually all integral membrane proteins. As part of an alternative approach to examining integral membrane protein structure, we are determining the structures of the extramembraneous domains of the G-protein receptor, rhodopsin.
Methods. The carboxyl terminal domain of bovine rhodopsin was synthesized, containing the last 43 amino acids of the protein sequence (rhoIVe). This sequence included the entire putative fourth cytoplasmic loop as well as a significant portion of helix seven, the transmembrane helix of this receptor to which the carboxyl terminal is attached. The solution structure of rhoIVe was determined by multidimensional 1H nuclear magnetic resonance.
Results. The structure contained a portion of alpha-helix corresponding to the top of transmembrane helix seven of the receptor. This allowed unambiguous docking of the carboxyl terminal domain to a model of the transmembrane domain. Helix seven is longer than suggested by hydropathy analysis. The structure also revealed the fourth cytoplasmic loop. The palmitoylation sites of rhodopsin are located near the deduced membrane surface. However, palmitoylation is not required for formation of this loop.
Conclusions. The carboxyl terminal of rhodopsin forms a structural domain whose structure can be determined separately from the rest of the protein. This structure reveals the fourth cytoplasmic loop that had been suggested to exist based on the presence of palmitoylation sites in the carboxyl terminal domain. Determination of the structure of all of the cytoplasmic domains of rhodopsin in a manner that allows docking to the structure of the transmembrane domain should permit construction of the entire surface of rhodopsin that interacts with the G-protein, transducin. Additionally, the rhodopsin phosphorylation sites and mutations associated with certain autosomal dominant forms of retinitis pigmentosa can now be located in the three dimensional structure of the carboxyl terminal domain.
Rhodopsin in the rod outer segment (ROS) disk membrane mediates the response of the rod photoreceptor cell to light. Rhodopsin is a member of the G-protein receptor family. Upon absorption of a photon of light, rhodopsin can proceed through a photocycle to the metarhodopsin II state and activate the G protein, transducin. Based on circular dichroism (CD) measurements (1), primary sequence (4), and recent projection structures (12, 14), a bundle of seven transmembrane helices has been suggested as part of the structure for bovine rhodopsin.
No high resolution structure is currently available for rhodopsin. Very few high resolution structures are available for any transmembrane proteins. Membrane proteins do not generally crystallize in a form suitable for X-ray crystallography. Membrane proteins also are not generally soluble due to their hydrophobic transmembrane domains. Therefore, multidimensional high resolution nuclear magnetic resonance (NMR) techniques are not useful for determination of intact integral membrane protein structure. Alternative approaches to membrane protein structure are required.
Recently we reported the structure of soluble peptides that represented two cytoplasmic domains of the receptor: an abbreviated carboxyl terminal domain (16) and the third cytoplasmic loop (17). These cytoplasmic domains each exhibited biological activity, and thus likely maintained crucial elements of native structure in the soluble peptides. In each case, compact globular structures were obtained using multidimensional 1H NMR. Our goal is to determine the structures of all the extramembraneous domains of bovine rhodopsin and dock them with the developing structure of the transmembrane domain (12, 14).
In this report, the structure of a peptide consisting of the carboxyl terminal 43 amino acids of rhodopsin is determined. This peptide is ten amino acids longer than the previous structure determination and reveals considerably greater structural information. The structure contained a portion of alpha-helix corresponding to the top of the seventh transmembrane helix of the receptor, allowing unambiguous docking of the carboxyl terminal domain to a model of the transmembrane domain. Furthermore, the palmitoylation sites could be identified and the suggestion of a fourth cytoplasmic loop could be confirmed.
Determination of the structure of all of the cytoplasmic domains of rhodopsin in a manner that allows docking to the structure of the transmembrane domain should permit construction of the entire surface of rhodopsin that interacts with the G-protein, transducin.
MATERIALS AND METHODS
The polypeptide constituting 43 amino acid residues of the carboxyl terminal domain of bovine rhodopsin, rhoIVe, was synthesized in the Biopolymer Facility of Roswell Park Cancer Institute, using 9-fluorenylmethoxy-carbonyl (FMOC) chemistry. It was purified by high performance liquid chromatography (HPLC), the amino acid composition analyzed, and the primary sequence determined. Figure 1 shows the sequence of rhoIVe within the primary structure of rhodopsin (4).
Circular dichroism spectroscopy
Circular dichroism (CD) spectra were obtained on a JASCO J-600 (JASCO, Japan) spectropolarimeter at room temperature, with a path length of 0.1 cm, band width of 1.0 nm, time constant of 2.0 seconds, a wavelength range of 260 to 185 nm, using quartz cuvettes. The spectrum for the buffer was subtracted from all the other spectra.
All NMR spectra were accumulated in 10 mM phosphate buffer at pH 5.9 on a Bruker AMX-600 spectrometer (Bruker, Bellerica, MA) at 10o C. Standard pulse sequences and phase cycling were employed to record: in H2O(10% D2O), double quantum filtered (DQF) correlated spectroscopy (COSY) and nuclear Overhauser effect correlated spectroscopy (NOESY) (400 ms mixing time) (6). All spectra were accumulated in a phase sensitive manner using time-proportional phase incrementation for quadrature detection in F1. Chemical shifts were referenced to a trace amount of internal methanol and expressed relative to (2,2,3,3,d4)-trimethylsilylproprionate (TSP).
The sequence-specific assignment of the 1H NMR spectrum was carried out using standard methods. Assigned nuclear Overhauser effect (NOE) cross peaks were segmented using a statistical segmentation function and characterized as strong, medium, and weak corresponding to upper bounds distance range constraints of 2.7, 3.5 and 5.0 Å, respectively. Lower bounds between nonbonded atoms were set to the sum of their van der Waals radii (approximately 1.8Å). Pseudoatom corrections were added to interproton distance restraints where necessary (15). Distance geometry calculations were carried out using the program DIANA (3) within the SYBYL 6.2 package (Tripos Software Inc., St. Louis, MO). First generation DIANA structures, 150 in total, were optimized from 1 to step 43 with the inclusion of three redundant dihedral angle constraint (REDAC) cycles. Energy refinement calculations (restrained minimizations/dynamics) were carried out on the best distance geometry structures using the SYBYL program implementing the Kollman all-atom force field. Statistics on structures were obtained from the X-plor program. These calculations were performed on a Silicon Graphics 4D/440 computer (Silicon Graphics, Inc., Mountain View, CA).
Imaging of the resulting structures was performed on a Power Macintosh (Apple Computer, Cupertino, CA) with MacImdad (Molecular Applications Group, Palo Alto, CA). Sculpt (Interactive Simulations, San Diego, CA) was used to build the model of the transmembrane domain on a Power Macintosh, based on the figures in the published reports (14). Helices representing the sequences of the seven transmembrane helices of rhodopsin were built with the Biopolymer module in Sybyl. These were combined into one file (pdb format) and placed in Sculpt for Macintosh, on a PowerMac 8500/120. The resulting minimized structure was moved to MacImdad on the same computer, and the carboxyl terminal was docked to the model of the transmembrane domain by superposition of the overlapping portions of the seventh helix.
Circular dichroism spectra of rhoIVe were obtained. Analysis of the CD spectra suggested the presence of about 13% alpha-helix, 29% anti-parallel ß-sheet, and 18% turn. This corresponds to about 5-6 residues in alpha-helix, about 12-13 in anti-parallel ß-sheet, and about 7-8 in turns. As seen below, this is in good agreement with the structure determined by NMR.
NMR structure determination
The structure of rhoIVe was determined by homonuclear 1H NMR. The temperature of these experiments was 10o C, in an attempt to stabilize a single structure for the peptide. A total of 403 unique constraints were obtained: 233 intra-residue constraints, 90 sequential constraints, and 80 long-range constraints. The line widths were too broad to adequately analyze coupling constants. At the temperature and protein concentration (10mg/ml) of these experiments, the solution of rhoIVe tended to form a gel, which may have been the source of the broadened resonances. Figure 2 shows the number of constraints per residue. As can be seen, constraints are available throughout the sequence. The number of available constraints were nonetheless modest, thus limiting the effective resolution of the structure obtained.
The sequence specific assignment resonance was carried out following standard procedures. The results are shown in Table 1.
Analysis of the connectivity in the set of usable constraints is presented in Figure 3. The connectivity suggests the presence of alpha-helix in the amino terminal region of this peptide.
Using these constraints, DIANA was used to generate compatible structures, and the best structures were minimized. An overlay of the best six structures appears in Figure 4. Table 2 shows the statistics for this ensemble of structures, as well as the energies. The average structure appears in Figure 5. The Ramachandran plot shows clusters in the regions of alpha-helix and anti-parallel ß-sheet, as well as dihedral angles outside these regions.
A model of the transmembrane domain of rhodopsin was also constructed. Using the overlap between helix 7 of the transmembrane domain and the helix on the N terminus of rhoIVe, the structure of the carboxyl terminal domain was docked with the transmembrane domain.
The resulting structure of the carboxyl terminal of bovine rhodopsin reveals several important structural features. First, the amino terminal region of this 43mer forms an alpha-helix. The amino-terminal seven to eight amino acids contribute to this helix, in agreement with the CD analysis. This alpha-helix is logically continuous with helix 7 of the transmembrane domain of rhodopsin. These data suggest that helix 7 is longer than originally suggested by the hydropathy plots of the primary sequence of bovine rhodopsin (4). It also defines the connection between the carboxyl terminal domain and the hydrophobic transmembrane domain of bovine rhodopsin (Figure 5).
Second, the structure is organized roughly into three subdomains. The first is the alpha-helix just described. The second and third subdomains form lobes on opposite sides of the molecule. The first lobe contains the two cysteines that are palmitoylated in the native membrane structure. These two cysteines are located at the bottom of the lobe, near the putative location of the bilayer surface. Thus the structure revealed here locates the cysteines in a suitable location for acylation by palmitoyl CoA, as is observed for rhodopsin in the ROS disk membrane (10). In addition, because of this location of the cysteines, a loop is formed between the top of helix 7 and the palmitoylation site. This could be referred to as the fourth cytoplasmic loop. Palmitoylation is not required for the formation of this loop. This may explain why removal of these palmitoylation sites does not inhibit activation of transducin by rhodopsin (5).
The third subdomain of this structure is partly composed of a ß-sheet, as reported previously (16), and in agreement with the CD analysis, as well as prior fourier transform infrared spectroscopy (FTIR) analysis (11). It is an anti-parallel ß-sheet. The two strands of the sheet are connected by a ß-turn. The third subdomain represents the most exposed portion of the carboxyl terminal. This region contains the phosphorylation sites most readily modified by rhodopsin kinase (7-9). The exposure of this region likely makes the phosphorylation sites readily available to the kinase. Whether phosphorylation alters the conformation of this region is being investigated separately.
At the true carboxyl terminus are found the sites of mutations for retinitis pigmentosa that map to the carboxyl terminal domain (2, 13). These are found in the last four amino acids of the primary sequence of rhodopsin. They are located in, or on the edge of, the small anti-parallel ß-sheet that characterizes this section of the protein. Whether these mutations lead to structural changes in the carboxyl terminal domain of rhodopsin will be explored in the future. Also to be explored is whether complexation of this domain with the other cytoplasmic domains of rhodopsin has any effect on conformation.
In summary, this work has revealed the complete structure of the carboxyl terminal domain of bovine rhodopsin. The appearance of a portion of helix 7 allows the docking of this structure to the transmembrane of rhodopsin. The prediction of a fourth cytoplasmic loop is borne out by the structure. The acylation sites are found near the putative location of the membrane surface, as expected. It is anticipated that equivalent structural detail can be obtained for the cytoplasmic loops of rhodopsin, and then all the cytoplasmic domains of rhodopsin can be docked to the transmembrane domain. The result will be a high resolution structure of the surface of rhodopsin that interacts directly with the G-protein, transducin.
We thank Dr. R. Wollman for computational assistance. This work was supported by National Institutes of Health Grant EY03328 and in part by CA16056.
REFERENCES1. Albert, AD and Litman, BJ, Independent structural domains in the membrane protein bovine rhodopsin, Biochemistry 17 (1978) 3893-3900.
2. Dryja, TP, Hahn, LB, Cowley, GS, McGee, TL and Berson, EL. Mutation spectrum of the rhodopsin gene among patients with autosomal dominant retinitis pigmentosa, Proc. Natl. Acad. Sci. USA 88 (1991) 9370-9374.
3. Guntert, P, Braunk, W and Wuthrich, K, Efficient computation of three-dimensional protein structures in solution from NMR data using the program DIANA and the supporting programs CALIBA, HABAS, and GLOMSA, Mol. Biol. 217 (1991) 517-530.
4. Hargrave, PA, McDowell, JH, Curtis, DR, Wang, JK, Juszczak, E, Fong, SL, Rao, JKM and Argos, P, The structure of bovine rhodopsin, Biophys. Struct. Mech. 9 (1983) 235-244.
5. Karnik, SS, Ridge, KD, Bhattacharya, S and Khorana, HG, Palmitoylation of bovine opsin and its cysteine mutants in COS cells, Proc. Natl. Acad. Sci. 90 (1993) 40-44.
6. Kumar, A, Ernst, RR and Wuthrich, K. A two-dimensional nuclear Overhauser enhancement (2D NOE) experiment for the elucidation of complete proton-proton cross-relaxation networks in biological macromolecules. Biochem. Biophys. Res. Commun. 95 (1980)
7. McDowell, JH, Nawrocki, JP and Hargrave, PA, Phosphorylation sites in bovine rhodopsin, Biochemistry 32 (1993) 4968-4974.
8. Ohguro, H, Hooser, JPV, Milam, AH and Palczewski, K, Rhodopsin Phosphorylation and Dephosphorylation in Vivo, J. Biol. Chem 270 (1995) 14259-14262.
9. Ohguro, H, Palczewski, K, Ericsson, LH, Walsh, KA and Johnson, RS, Sequential phosphorylation of rhodopsin at multiple sites, Biochemistry 32 (1993) 5718-5724.
10. Pepperberg, DR, Morrison, DF and O'Brien, PJ, Depalmitoylation of rhodopsin with hydroxylamine, Methods in Enzymology 250 (1995) 348-361.
11. Pistorius, AM and deGrip, WJ, Rhodopsin's secondary structure revisited: assignment of structural elements, Biochem. Biophys. Res. Commun. 198 (1994) 1040-1045.
12. Schertler, GRX, Villa, C and Henderson, R, Projection structure of rhodopsin, Nature 362 (1993) 770-772.
13. Sung, CH, Davenport, CM and Nathans, J, Rhodopsin mutations responsible for autosomal dominant retinitis pigmentosa, J. Biological Chemistry 268 (1993) 26645-26649.
14. Unger, VM and Schertler, GFX, Low resolution structure of bovine rhodopsin determined by electron cryo-microscopy, Biophys. J. 68 (1995) 1776-1786.
15. Wuthrich, K, Billeter, M and Braun, WJ. Pseudo-structures for the 20 common amino acids for use in studies of protein conformations by measurements of intramolecular proton-proton distance constraints with nuclear magnetic resonance. J. Mol. Bio. 169 (1983) 949-961.
16. Yeagle, PL, Alderfer, JL and Albert, AD, Structure of the carboxyl terminal domain of bovine rhodopsin, Nature Struct. Biol. 2 (1995) 832-834.
17. Yeagle, PL, Alderfer, JL and Albert, AD, Structure of the third cytoplasmic loop of bovine rhodopsin, Biochemistry 34 (1995) 14621-14625.