Retinitis Pigmentosa Mutants Provide Insight into the Role of the N-terminal Cap in Rhodopsin Folding, Structure, and Function*

Background: Retinitis pigmentosa mutations in the N terminus of rhodopsin lead to misfolding. Results: Pigments obtained by pharmacological rescue remain defective, whereas some of those obtained by disulfide bond-mediated repair regain function. Conclusion: The N-terminal cap has multiple roles in rhodopsin folding, stability, and signal transduction. Significance: In combination, pharmacological rescue and protein engineering provide insight into rhodopsin stability, function, and disease mechanisms. Autosomal dominant retinitis pigmentosa (ADRP) mutants (T4K, N15S, T17M, V20G, P23A/H/L, and Q28H) in the N-terminal cap of rhodopsin misfold when expressed in mammalian cells. To gain insight into the causes of misfolding and to define the contributions of specific residues to receptor stability and function, we evaluated the responses of these mutants to 11-cis-retinal pharmacological chaperone rescue or disulfide bond-mediated repair. Pharmacological rescue restored folding in all mutants, but the purified mutant pigments in all cases were thermo-unstable and exhibited abnormal photobleaching, metarhodopsin II decay, and G protein activation. As a complementary approach, we superimposed this panel of ADRP mutants onto a rhodopsin background containing a juxtaposed cysteine pair (N2C/D282C) that forms a disulfide bond. This approach restored folding in T4K, N15S, V20G, P23A, and Q28H but not T17M, P23H, or P23L. ADRP mutant pigments obtained by disulfide bond repair exhibited enhanced stability, and some also displayed markedly improved photobleaching and signal transduction properties. Our major conclusion is that the N-terminal cap stabilizes opsin during biosynthesis and contributes to the dark-state stability of rhodopsin. Comparison of these two restorative approaches revealed that the correct position of the cap relative to the extracellular loops is also required for optimal photochemistry and efficient G protein activation.

Autosomal dominant retinitis pigmentosa (ADRP) mutants (T4K, N15S, T17M, V20G, P23A/H/L, and Q28H) in the N-terminal cap of rhodopsin misfold when expressed in mammalian cells. To gain insight into the causes of misfolding and to define the contributions of specific residues to receptor stability and function, we evaluated the responses of these mutants to 11-cisretinal pharmacological chaperone rescue or disulfide bondmediated repair. Pharmacological rescue restored folding in all mutants, but the purified mutant pigments in all cases were thermo-unstable and exhibited abnormal photobleaching, metarhodopsin II decay, and G protein activation. As a complementary approach, we superimposed this panel of ADRP mutants onto a rhodopsin background containing a juxtaposed cysteine pair (N2C/D282C) that forms a disulfide bond. This approach restored folding in T4K, N15S, V20G, P23A, and Q28H but not T17M, P23H, or P23L. ADRP mutant pigments obtained by disulfide bond repair exhibited enhanced stability, and some also displayed markedly improved photobleaching and signal transduction properties. Our major conclusion is that the N-terminal cap stabilizes opsin during biosynthesis and contributes to the dark-state stability of rhodopsin. Comparison of these two restorative approaches revealed that the correct position of the cap relative to the extracellular loops is also required for optimal photochemistry and efficient G protein activation.
Autosomal dominant retinitis pigmentosa (ADRP) 2 is an inherited degenerative disorder of the retina that leads to pro-gressive and irreversible vision loss (1). More than 100 ADRP mutations map to RHO, the gene encoding the visual receptor rhodopsin located in rod photoreceptor cells (2). Most of these mutations result in rod opsin misfolding, but the molecular details of pathogenesis remain elusive, and there is no effective treatment. To gain insight into the underlying causes of receptor misfolding and malfunction, we performed a sideby-side comparison using two complementary experimental approaches. The first, pharmacological rescue, is an established procedure for rescuing folding in mutant G protein-coupled receptors (GPCRs), including ADRP rhodopsin mutants (3). However, it is our seldom used second tool, disulfide bondmediated repair, that provided us with novel insight into mechanisms of disease and receptor function. Our first target, a cluster of ADRP mutations (T4K, N15S, T17M, V20G, P23A/H/L, and Q28H) located in the N-terminal tail of rhodopsin ( Fig. 1), includes P23H, which accounts for the largest fraction of ADRP mutations in the United States (4).
As predicted by mutagenesis and biochemical studies (5,6), crystal structures of rhodopsin reveal a high degree of order in the extracellular (intradiscal) domain (7,8), a feature absent in most other GPCRs (9,10). The tight association between the N terminus and the extracellular domain was also inferred from binding studies using B6-30N, a monoclonal antibody to rhodopsin N-terminal residues Gly-3-Ser-14 (11). These ideas were integrated into the general proposition that folding of the extracellular and seven-transmembrane (7TM) domains is tightly coupled (12), as are conformational changes in these domains during photoactivation and G protein activation (13,14).
During biosynthesis, rod opsin inserts into and folds within the endoplasmic reticulum (ER) of rod cell inner segments. Misfolded forms are withheld and targeted for elimination (see Fig. 6) by the ER-associated protein degradation machinery (15). Correctly folded rod opsin is transported to the rod outer segment (ROS) and becomes densely packed into ROS disc membranes (16). In non-rod cell recombinant expression systems, correctly folded rod opsin is transported instead to the cell surface. As part of the visual cycle in rod cells, 11-cis-retinal, synthesized by retinal pigment epithelial cells, is delivered to ROS and combines with rod opsin to form rhodopsin. However, it is not known when or where de novo synthesized rod opsin first encounters and combines with 11-cis-retinal, an event that can influence folding, especially for ADRP mutants susceptible to 11-cis-retinal-mediated pharmacological chaperone rescue (17,18).
ADRP rhodopsin mutants were previously classified according to their expression profiles in transfected HEK-293S or COS-1 cells (19,20). Type I rod opsin mutants fold normally and reach the cell surface, whereas Type II mutants have reduced expression levels, traffic abnormally, and form pigment inefficiently (19). The absence of 11-cis-retinal during rod opsin folding in these and many similar studies has led us and others to question the relevance of these folding outcomes because, in rod cells, 11-cis-retinal is naturally abundant and may circulate to the ER located in rod cell inner segments. Pharmacological rescue was previously shown to restore pigment formation in ADRP mutant T17M (21), as well as T4R and P23A/H/L (22). Here, we have extended these analyses to all known human ADRP rhodopsin mutations in the N terminus (T4K, T17M, N15S, V20G, P23A/H/L, and Q28H), and for the first time, we have assessed the active-state properties of these rescued mutant pigments.
As the mutant pigments resulting from pharmacological rescue were found to be defective, we used an alternative strategy to repair the defects in the N-terminal cap. Our hypothesis was that mutations within the N terminus of rhodopsin destabilize the entire protein by decoupling key interactions of the cap with FIGURE 1. Location of ADRP rhodopsin mutants examined in this study. A, secondary structure model of rhodopsin highlighting ADRP mutation positions (red circles) in the N terminus (NT; black circles). Polypeptide segments for EL-1 (blue circles), EL-2 (green circles), and EL-3 (purple circles) are shown. Disulfide bonds (dashed black lines) formed either naturally (Cys-110 -Cys-187) or between an engineered cysteine pair (Cys-2-Cys-282) are indicated. Transmembrane segments are indicated by gray rounded rectangles. The model shows only the top half of rhodopsin. B, location of the panel of ADRP mutants from A and the packing of EL-1-3 in a three-dimensional crystal structure model of rhodopsin (Protein Data Bank ID 1U19). The location of the natural disulfide bond (yellow) between Cys-110 and Cys-187 is indicated. The positions of Asn-2 and Asp-282 (the mutation of which to cysteine results in an additional disulfide bond between them) are indicated. C, packing of Pro-23 and Gln-28 onto EL-1 and EL-2. The residues in EL-1-3 that are sites of ADRP are shown as van der Waals spheres. These residues surround the critical Cys-110 -Cys-187 disulfide bridge. The points of contact between Pro-23 and Gln-28 are Pro-180 and Gln-184 on EL-2. Glu-184 is adjacent to Cys-185, which can form an incorrect disulfide bond with Cys-110. Gly-106, Gly-109, and Thr-289 are sites of ADRP mutations that interact with EL-2 in the region of the Cys-110 -Cys-187 disulfide bond. the extracellular loops. Stabilization of the transmembrane helices and connecting loops with bound ligand can overcome these defects, as we have shown here. However, to test our hypothesis directly, we introduced a disulfide bond in an attempt to tether the damaged cap to the 7TM domain. This was achieved by superimposing this panel of ADRP mutants onto a rhodopsin background containing a juxtaposed cysteine pair (N2C/D282C) known to form a disulfide bond (Cys-2-Cys-282) that increases the thermostability of rod opsin (23). For a subset of these ADRP mutants, the N2C/D282C background restored folding of rod opsin, and the resulting pigments had much improved thermostability, photobleaching behavior, and activation of the G protein transducin (G T ). These findings highlight how the extracellular domain of rhodopsin has evolved, under selection pressure constraints requiring high sensitivity vision in dim light, to contribute to receptor stability and signaling efficiency.

EXPERIMENTAL PROCEDURES
Materials and Buffers-Bovine retinas were purchased from J. A. Lawson Co. (Lincoln, NE). 11-cis-Retinal was provided by Rosalie K. Crouch (Storm Eye Institute, Medical University of South Carolina). n-Dodecyl ␤-D-maltoside (DDM) was purchased from Anatrace (Maumee, OH). Bis-tris propane, BES, GTP␥S, 9-cis-retinal, and hydroxylamine hydrochloride were purchased from Sigma. PMSF was purchased from Fluka. Disposable polystyrene columns (2-ml bed volume) were purchased from Thermo Scientific. Restriction endonucleases were purchased either from New England Biolabs or Thermo Scientific. The nonapeptide corresponding to the nine amino acid residues (TETSQVAPA) at the C terminus of rhodopsin was purchased from Peptide Protein Research. Sepharose 4B beads were purchased from Amersham Biosciences. The HEK-293S cell line was provided by J. Nathans (The John Hopkins School of Medicine), as was pRSV-TAg (SV40 large T-antigen expression plasmid). DMEM, DMEM/Ham's F-12 medium, FBS, trypsin, L-glutamine, penicillin, and streptomycin were purchased from PAA Laboratories Ltd.
Preparation of ROS Membranes-Rhodopsin in ROS membranes was prepared from frozen bovine retinas by sucrose density gradient centrifugation (24).
Construction of Opsin Mutants-Site-directed mutagenesis using complementary primer pairs was performed with a synthetic bovine opsin gene in the expression plasmid pMT4 (25) as the target template. All mutants were also superimposed onto the N2C/D282C rhodopsin background (23) by KpnI-PstI (restriction endonuclease sites) fragment exchange. Mutations were verified by DNA sequencing of both strands of the entire opsin gene.
Cell Culture and Transfection-HEK-293S cells were maintained in DMEM/Ham's F-12 medium supplemented with heat-treated (55°C, 30 min) 10% (v/v) FBS, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. HEK-293S cells were grown in a humidified incubator under a 5% CO 2 atmosphere at 37°C. Calcium phosphate precipitationmediated transfection was performed as follows using a modification to a procedure developed by Chen and Okayama (26) and O'Mahoney and Adams (27). On the day before transfection, confluent cell monolayers of HEK-293S cells were trypsinized and seeded at a density of 2-2.5 ϫ 10 6 cells in 22.5 ml of DMEM (supplemented as described above) in a 15-cm tissue culture dish. 24 h after cells were seeded, a transfection mixture (2.5 ml) containing plasmid DNA and Buffers I and J was added to the pre-confluent cell monolayer. To prepare the transfection mixture, plasmid pMT4 containing opsin genes (54 g) and pRSV-TAg (6 g) was diluted in sterile water (1.125 ml). Buffer J (125 l) was then added slowly with continuous mixing. Finally, Buffer I was added over 1 min with Vortex mixing, and the mixture was immediately added dispersed over the cell monolayers. The cells were then kept for 19 h in a humidified incubator set at 1% CO 2 and 37°C. After this incubation period, the spent medium was removed, and the cells were carefully washed once with unsupplemented DMEM (20 ml). The cells were subsequently fed with DMEM (25 ml) containing supplements and kept for a further 48 h in a humidified incubator set at 5% CO 2 and 37°C prior to harvest.
Treatment of Transfected Cells with Pharmacological Chaperones-HPLC-purified retinal isomers were added to transfected cells using a slight modification to the method described previously (17). Briefly, 11-cis-retinal or 9-cis-retinal, prepared as 100 mM solutions in cell culture-grade dimethyl sulfoxide, was dispensed over transfected HEK-293S cells using safelight illumination conditions (Kodak No. 2 filter). These retinoids were added to cells in two 10 M installments, 24 and 48 h after removal of the transfection mixture, to give a final retinoid concentration of 20 M. The culture dishes were wrapped with a single layer of aluminum foil to maintain dark conditions during incubation. Cells were harvested under safelight illumination 24 h after the final addition of retinoids.
Generation of Rhodopsin Pigment and Detergent Solubilization of Cells-Transfected HEK-293S cells obtained from a single 15-cm tissue culture dish were washed twice with 10 ml of ice-cold Buffer A prior to resuspension in 0.9 ml of chilled Buffer B. Pigment formation and subsequent procedures were carried out in a dark room illuminated by Kodak No. 2 safelight filters. Cell suspensions were treated with 11-cis-retinal or 9-cis-retinal (5 M) added in two installments over 3 h at 4°C with end-over-end nutator mixing. Whole cells were solubilized with 1% (w/v) DDM for 1 h at 4°C with end-over-end mixing. The solubilized cell extracts containing pigment were centrifuged at 14,000 rpm for 30 min at 4°C to remove insoluble material prior to purification of pigments. A fraction of the supernatant was kept for examination by UV-visible difference spectroscopy and SDS-PAGE for an assessment of rhodopsin yield.
Rho-1D4 Immunoaffinity Purification of Rhodopsin-Rhodopsin was purified from solubilized bovine ROS membranes or from detergent-solubilized transfected HEK-293S cell extracts using Rho-1D4-Sepharose 4B beads with modifications to the methods described previously (25,28,29). All purification procedures and manipulations were carried out at 4°C. 50 l of Rho-1D4-Sepharose beads with a binding capacity of ϳ1 g of rhodopsin/l of settled beads was added to the cell lysate and mixed end-over-end at 4°C for 2 h to allow binding. The suspension was packed into a 2-ml disposable column and washed with 40 ml of Buffer D (high salt), followed by 20 ml of salt-free Buffer E. Correctly folded rhodopsin was purified by elution with salt-free Buffer H (0.5 ml) for up to three elutions. Once elution of rhodopsin pigment was complete, rod opsin was recovered from the column using Buffer G (high salt). Pigments were stored in the dark at 4°C.
UV-visible Absorption Spectroscopy of Rhodopsin and Photobleaching-UV-visible spectra of pigments purified in Buffer H were recorded at 20°C using a PerkinElmer Life Sciences 35 UV-visible spectrophotometer equipped with waterjacketed cell holders. Spectra were acquired between 250 and 650 nm using a 2-nm bandwidth, a response time of 1 s, and a scan speed of 480 nm/min. The molar extinction coefficient value used to calculate rhodopsin yields was 40,600 M Ϫ1 cm Ϫ1 at 500 nm (30). Purified rhodopsin samples were photobleached at 20°C by direct illumination in a cuvette using a Schott KL 1500 compact fiber optic light guide fitted with a Ͼ495-nm long-pass filter. Rhodopsin samples were manually bleached for 30 s, followed immediately by UV-visible spectroscopy. This process was repeated four times. Finally, the protonated Schiff base was trapped by acidification of the sample (500 l) by the addition of 2 N H 2 SO 4 (2 l).
Thermal Stability of Detergent-solubilized Rhodopsin Mutants-Purified rhodopsin pigments in Buffer H were monitored by UV-visible spectroscopy. Samples were kept at either 37 or 55°C for 2 min in quartz cuvettes to allow for temperature equilibration. UV-visible spectra were recorded at 2-min intervals for 30 min and afterward at 20-min intervals for at least another 24 h, except as stated otherwise. Normalized data were best fit using the single-or double-component exponential decay function of SigmaPlot 12.0, and these values were used to calculate the half-life (t1 ⁄ 2 ) of the mutant rhodopsin pigments. Pigments that decayed rapidly at 55°C were monitored at 5-s intervals using the time drive function on the spectrometer.
Rhodopsin Sensitivity to Hydroxylamine-Schiff base hydrolysis of rhodopsin by hydroxylamine (22) was measured in Buffer F; the final concentration of hydroxylamine hydrochloride (titrated to pH 6.0 using NaOH) was 50 mM. UV-visible absorbance spectra were collected at the same time intervals described above.
Measurement of the Rate of Metarhodopsin II Decay in Rhodopsin-Metarhodopsin II (MII) decay was monitored at 20°C as described previously (31) using instead a PerkinElmer Life Sciences LS 50 fluorometer. Excitation of rhodopsin tryptophans (0.25 M in Buffer F) was at 280 nm (2.5-nm slit width), and emission was monitored at 330 nm (15-nm slit width). Rho-dopsin was converted to MII by photobleaching the sample for 30 s with light (Ͼ495 nm). Fluorescence was recorded at 30-s intervals with a 2-s integration time for 120 min. Data were normalized and best fit to either a single-or double-component rise-to-plateau exponential function (SigmaPlot 12.0). Derived rate constants were used to calculate the half-life of MII decay.
G T Activation Assay-Freshly prepared (1-2 days old), similarly aged rhodopsin mutants with no measurable loss of pigment during storage were used for this assay. G T was purified from ROS membranes according to the method of Kühn (32), except that the final supernatant containing G T was concentrated using a 60-ml Jumbosep TM centrifugal concentration device fitted with a 10-kDa cutoff filter (Pall Corp.) before dialysis against the storage buffer (Buffer L). The activation of G T by rhodopsin MII was measured by monitoring the increase in intrinsic fluorescence of G T that accompanies the exchange of GDP for GTP (33). G T activation was measured at 20°C in Buffer K using the PerkinElmer Life Sciences LS 50 fluorometer as described above. The fluorometer was equipped with an electronic stirrer (Model 300, Rank Brothers Ltd.) set to mix the reaction cuvette contents at 240 rpm. Excitation of G T was at 280 nm, and emission was recorded at 340 nm. The interval for data collection was 3 s, and the reaction was followed for 1 h. The reaction mixture contained 250 nM G T in 1 ml of Buffer K. Rhodopsin (20 nM) was added and allowed to equilibrate for 1 min before bleaching the sample for 30 s with light (Ͼ495 nm). GTP␥S (5 M) was then added to initiate the reaction. The relative initial rate of G T activation was determined from the slope of the increase in relative fluorescence over the first 60 s following the addition of GTP␥S. The G T activation rates for ADRP mutants were expressed as a percentage of WT rhodopsin initial activation rates. To reduce the Cys-2-Cys-282 disulfide bond, rhodopsin pigments were incubated with DTT (50 mM final concentration) on ice for 1 h prior to commencement of the G T activation assay.
Packing Analysis-The packing analysis of the N-terminal domain of rhodopsin was carried out using the method of occluded surfaces (34). The occluded surface method calculates packing values at the level of individual atoms, amino acids, or entire proteins. Packing values range from 0.0 to 1.0, corresponding to totally exposed and totally occluded environments. Hexagonally packed spheres have a maximum packing value of 0.8 due to the void space that exists where the spheres are not in direct contact (35). In the occluded surface calculation, van der Waals surfaces are drawn around each atom in the protein, and normals are constructed that extend outward until they reach another surface or a length of 2.8 Å, the diameter of a water molecule. The cutoff of 2.8 Å between amino acid surfaces accounts for the possibility that water can occupy that space, and therefore, the corresponding surface is defined as being non-occluded. The definition of the occluded surface packing value takes into account the normalized occluded (or buried) surface area weighted by the distance to the occluding neighbors.

RESULTS
Pharmacological Chaperone Rescue of Misfolded ADRP Rhodopsin Mutants-HEK-293S cells transfected with ADRP rhodopsin mutants were incubated in the presence or absence of NOVEMBER 22, 2013 • VOLUME 288 • NUMBER 47

JOURNAL OF BIOLOGICAL CHEMISTRY 33915
11-cis-retinal during expression. Harvested cells were treated with 11-cis-retinal, and rhodopsin pigments were purified from solubilized cell extracts by Rho-1D4-Sepharose immunoaffinity chromatography (see "Experimental Procedures"). Preferential elution of correctly folded rhodopsin was achieved using salt-free Buffer H. Eluted rhodopsin fractions were examined by UV-visible absorbance spectroscopy, and the results are presented in Fig. 2A and Table 1. These N-terminal ADRP mutants made very small or undetectable amounts of pigment. These very low yields are attributed to receptor misfolding and inefficient formation of correctly folded opsin capable of combining with 11-cis-retinal to form pigment. However, when 11-cis-retinal was supplied to transfected cells during opsin biosynthesis, all of these ADRP mutants made WT-like rhodopsin pigment, but the expression levels varied ( Fig. 2A and Table 1). The yields of ADRP rhodopsin mutants T4K and P23A reached WT levels, whereas all other mutants were recovered at levels ϳ50% that of WT rhodopsin. WT rhodopsin expression levels were unaffected by the presence of 11-cis-retinal during opsin expression. For comparison, we also performed pharmacological rescue experiments using the 9-cis-retinal isomer; similar results were obtained, except expression levels were lower, and the purified pigments had blue-shifted absorbance maxima (data not shown). Our results for T17M and P23A/H/L are in full agreement with previous 11-cis-retinal pharmacological rescue experiments (17,18,21,22,36). We now included T4K, N15S, V20G, and Q28H to demonstrate that all known ADRP mutants in the N-terminal tail are susceptible to 11-cis-retinal pharmacological rescue.
Restoration of ADRP Rhodopsin Mutants in the N2C/D282C Background-The N2C/D282C mutation was previously shown to confer thermal stability to purified rod opsin apoprotein a Average rhodopsin yield from a 15-cm tissue culture dish containing transiently transfected HEK-293S cells grown in the absence ( n ) or presence ( r ) of 11-cis-retinal. b Absorbance maxima ( max ) in the visible region. c A 280 /A max ratio of pigments obtained from low salt elution (A 280 /A max ϭ 1. 6 for pure rhodopsin). d NA, not applicable. (23) and the pigment of rhodopsin mutant N15D (37). Our hypothesis was that correct folding in this panel of ADRP rod opsin mutants would be repaired if we could re-establish the stability of the 7TM domain by improving association of the defective cap with extracellular loops (ELs) 1-3. To test this idea, these ADRP mutants were subcloned into the N2C/ D282C background. Transient transfection experiments were performed as described under "Experimental Procedures," but 11-cis-retinal was not added until after the transfected cells were harvested. Pigments were purified and analyzed by UVvisible absorbance spectroscopy as described above, and the results are presented in Fig. 2B and Table 1. The rhodopsin double mutant N2C/D282C and the single mutants N2C and D282C all formed pigments at levels comparable to WT rhodopsin, findings that are in agreement with Xie et al. (23). WT levels of pigment were also obtained from ADRP mutant P23A in the N2C/D282C background. Several other ADRP mutants also formed rhodopsin pigment in this N2C/D282C background, but at lower levels compared with WT. These mutants were T4K (90% of WT), N15S (80% of WT), and both Q28H and V20G (both ϳ50% of WT). Pigment formation was negligible or absent for ADRP mutants T17M, P23H, and P23L in the N2C/ D282C background. The relative expression levels of these mutants restored by the two separate approaches are shown in Fig. 2C. The T17M, P23H, and P23L mutants in the N2C/ D282C background formed some pigment when 11-cis-retinal was present during expression; however, these levels of expression were lower than those observed in the WT background under 11-cis-retinal rescue conditions (Table 1).
Thermal Stability of Restored ADRP Rhodopsin Mutant Pigments-Thermal stability measurements were performed using the purified pigments obtained either by 11-cis-retinal pharmacological rescue or by N2C/D282C repair. The decrease in absorbance at 500 nm at 37 or 55°C as a function of time was monitored, and the data collected were used to determine halflife values (see "Experimental Procedures"). When kept at 37°C, WT rhodopsin purified from HEK-293S cells has a halflife of ϳ58 h. This value is comparable to the thermal decay profile at 37°C of WT rhodopsin prepared from COS-1 cells described in a previous study (38). All ADRP mutant pigments obtained by 11-cis-retinal pharmacological rescue were much less stable at 37°C compared with WT rhodopsin (Fig. 3A and Table 2). In addition, the thermal decay profiles of these mutant pigments had two kinetic components as revealed by curve fitting of the data. The most stable ADRP mutant at 37°C was P23A (t1 ⁄ 2 ϭ 14 h for component I), whereas the most unstable was Q28H (t1 ⁄ 2 ϭ 19 min for component I). The observed order of stability for these mutant pigments in the WT rhodopsin background at 37°C was as follows: To investigate the consequences of tethering the defective cap mutants to the extracellular surface, we measured the thermostability of the purified ADRP mutant pigments in the N2C/ D282C rhodopsin background. At 37°C, rhodopsin N2C/ D282C had a t1 ⁄ 2 of Ͼ4 days (115.5 h), whereas WT rhodopsin had a t1 ⁄ 2 of ϳ3 days (58 h). The increased thermal stability of rhodopsin N2C/D282C is attributable to the additional disulfide bond (Cys-2-Cys-282) in the extracellular domain that forms between the two engineered cysteine residues (Fig. 1A). The ADRP mutant pigments obtained using the N2C/D282C background all displayed higher thermal stabilities than those lacking the additional N2C/D282C mutations obtained by rescue with 11-cis-retinal. For example, the P23A mutant had a half-life of ϳ14 h at 37°C, whereas the corresponding P23A (N2C/D282C) mutant had a half-life of 58 h. P23A in the N2C/ D282C background was also treated with DTT before monitoring pigment stability. In this case, the thermal stabilization effect of the N2C/D282C background was reversed, and the pigment subsequently regained a thermal decay profile similar to that observed in the WT background ( Fig. 3 and Table 2). We also conducted thermal stability measurements at 55°C, and the relative stabilities of the mutants at that temperature were similar to those at 37°C (Fig. 3B and Table 2).
Susceptibility of ADRP Rhodopsin Mutants to Hydroxylamine-To examine the physical accessibility of the retinal Schiff base-Lys-296 linkage to the bulk solvent, hydroxylamine was used as a probe. The loss of absorbance at 500 nm in the presence of 50 mM hydroxylamine at 37°C was examined by UVvisible absorbance spectroscopy (see "Experimental Procedures"). WT rhodopsin in the dark showed some resistance to attack by hydroxylamine (t1 ⁄ 2 ϭ 11.5 h) compared with all of the purified ADRP mutant pigments, which were more susceptible ( Fig. 3C and Table 3). The increase in susceptibility to hydroxylamine attack for ADRP rhodopsin mutants in the WT rhodopsin background is in agreement with results from previous studies (22). We now report that those ADRP mutants repaired in the N2C/D282C background were more resistant to hydroxylamine than those in the WT background. For example, P23A in the WT rhodopsin background (t1 ⁄ 2 ϭ 55.9 min) was 6-fold more sensitive than P23A in the N2C/D282C background (t1 ⁄ 2 ϭ 330.1 min). Similarly, the V20G mutant in the WT background (t1 ⁄ 2 ϭ 10.6 min) acquired an increased resistance to hydroxylamine (t1 ⁄ 2 ϭ 80.6 min) in the N2C/D282C background. In addition, mutant pigments obtained by pharmacological rescue in the WT background exhibited double-component decay kinetics in the presence of hydroxylamine compared with singlecomponent decay kinetics exhibited by mutants repaired by the N2C/D282C background.
Photobleaching Properties of Purified Rhodopsin Mutant Pigments-UV-visible absorbance spectroscopy was used to examine the photobleaching properties of immunoaffinity-purified mutant pigments (see "Experimental Procedures"). Spectra were recorded before and after illumination (Ͼ495-nm light) as described under "Experimental Procedures." Purified WT rhodopsin (visible max ϭ 500 nm) was readily photoconverted to the MII 380-nm form (39,40). Further illumination of the sample for up to 2 min resulted in no further spectral change. Subsequent acidification of the sample moved the absorption maximum from 380 to 440 nm, which corresponds to the formation of a protonated Schiff base (41). The rhodopsin N2C/D282C mutant displayed photobleaching profiles similar to WT rhodopsin (Fig. 4), as did the single N2C and D282C pigments (data not shown). When 11-cis-retinal-rescued ADRP mutant pigments were illuminated, conversion to the 380-nm species was incomplete, and a small but detectable shoulder (380 -480 nm) with a peak at ϳ450 nm was evident (Fig. 4). Further 30-s pulses of illumination were required for complete conversion to the 380-nm form. The degree of resistance to photobleaching varied among the ADRP mutants. Some mutants (T17M, P23A, P23L, P23H, and Q28H) possessed a higher content of a photobleaching-resistant form, whereas others (T4K, V20G, and N15S) photobleached more readily. Upon acidification, several of the photobleached ADRP mutant pigments did not convert fully to the 440-nm species. This property is best explained by the fast decay of MII and release of all-trans-retinal that occur in these mutants prior to sample acidification. A summary of the photobleaching properties of the ADRP rhodopsin mutants is reported in Table 4. These data are consistent with those described in a recent study (22), as well as in earlier reports (20,42).
We next examined the photobleaching properties of purified ADRP mutant pigments in the N2C/D282C background. Mutants T4K, N15S, V20G, and P23A in the N2C/D282C back-ground now displayed WT-like photobleaching profiles (Fig. 4). By contrast, when these ADRP mutants were superimposed onto the single N2C or D282C (control) background, their photobleaching profiles remained abnormal (data not shown).
MII Decay Rates of Rhodopsin Mutants-The rate of rhodopsin MII decay was measured by following the kinetics of alltrans-retinal release. The assay exploits the fluorescence of native tryptophan residues in proximity to retinal that is quenched when retinal is present (see "Experimental Procedures") (31). The results for selected mutants are shown in Fig.  5A, and all data are compiled in Table 4. The t1 ⁄ 2 of MII decay for WT rhodopsin purified from HEK-293S cells was ϳ15.8 min, a value similar to that reported for rhodopsin purified from ROS or transfected COS-1 cells (13,43). In our experiments, the t1 ⁄ 2 of MII decay for ROS was 16.5 min. The t1 ⁄ 2 of MII decay for the N2C/D282C mutant was 14.7 min, a value comparable to that for WT rhodopsin purified from HEK-293S cells. The t1 ⁄ 2 values FIGURE 3. Purified ADRP mutant pigments exhibit thermal instability and increased susceptibility to attack by hydroxylamine. Mutant pigments were obtained by 11-cis-retinal rescue (red squares) in the WT background or by using the N2C/D282C background (blue circles). Purified pigments in Buffer H were transferred into a prewarmed cuvette at either 37°C (A) or 55°C (B) and allowed to equilibrate for 2 min. Samples were examined by UV-visible absorbance spectroscopy (650 to 250 nm), with scans collected at 2-min intervals for 20 min and then at 20-min intervals for at least an additional 24 h. The pigment remaining (percentage) at each time point (first time point value, 100%) was determined and used to construct a scatter plot that was curve-fitted to the singleor double-component exponential decay function of SigmaPlot 12.0 The effect of DTT (50 mM final concentration) on pigments prepared in the WT (brown triangles) or N2C/282C (inverted green triangles) background is indicated. C, pigment stability was recorded as described for A at 37°C but in the presence of hydroxylamine. The susceptibility of single-cysteine rhodopsin mutants N2C (purple diamonds) and D282C (cyan hexagons) to hydroxylamine is also shown. This figure highlights how these ADRP mutant pigments (red squares) acquire improved thermal stability and are more resistant to attack by hydroxylamine when they are in the N2C/D282C rhodopsin background (blue circles). Furthermore, the improved thermostability we observed for the P23A (N2C/D282C) pigment (A and B) was reversed by pretreatment of this sample with DTT (inverted green triangles), which is known to reduce the Cys-2-Cys-282 disulfide bond.
of MII decay for the single N2C and D282C mutants were 15.5 and 18.1 min, respectively.
The MII decay kinetics of the 11-cis-retinal-rescued human ADRP mutants in the N terminus have not been examined previously. They were found to be complex: curve fitting of the data revealed the presence of two MII decay kinetic components (slow and fast). For example, the P23A MII decay profile has a slow component (t1 ⁄ 2 ϭ 18.3 min, 64.8% contribution) and a fast component (t1 ⁄ 2 ϭ 2.3 min, 35.2% contribution). Similarly, the P23H MII decay profile had a slow component (t1 ⁄ 2 ϭ 13.1 min, 63.5% contribution) and a fast component (t1 ⁄ 2 ϭ 1.11 min, 36.5% contribution). A summary of the MII decay profiles is compiled in Table 4. All of the MII decay rates for ADRP mutants modified by N2C/D282C repair were faster compared with WT rhodopsin. However, unlike the double-component decay kinetics observed for ADRP mutants in the WT background, they now exhibited single-exponential decay kinetics. The MII decay profiles of V20G (12.4 min), T4K (12.1 min), and N15S (9.8 min) in the N2C/D282C background were fit to a single-component exponential decay function.
To further prove that a disulfide bond between Cys-2 and Cys-282 is responsible for the single-component MII decay kinetics acquired by the ADRP mutants in the N2C/D282C rhodopsin background, we performed the same experiments using pigments pretreated with DTT. This reducing agent had no effect on the MII decay kinetics of WT opsin purified from ROS (t1 ⁄ 2 ϭ 16.5 min) or from transfected HEK-293S cells (t1 ⁄ 2 ϭ 15.7 min). Similarly, DTT had only a small effect on the rate of MII decay of rhodopsin N2C/D282C (t1 ⁄ 2 ϭ 14.5 min) or the single mutant N2C or D282C (t1 ⁄ 2 ϭ 15.5 and 18.0 min, respectively). By contrast, DTT had a pronounced effect on the rate of MII decay of ADRP mutants rescued in the N2C/D282C rhodopsin background. The MII decay profiles of ADRP mutants in the N2C/D282C background treated with DTT displayed Values are the means Ϯ S.D. from three independent experiments. Data were collected at time intervals as described under "Experimental Procedures," except at 37°C ( e ), which were recorded every 6 min. Pigments were obtained by the addition of 11-cis-retinal after transient transfection ( n ) or during opsin biosynthesis ( r ).   7%). Activation of G T by ADRP Rhodopsin Mutants-To examine the signal transduction properties of the N-terminal ADRP mutants, we examined their ability to activate G T using a fluorescence-based assay (33) as described under "Experimental Procedures." This assay measures the increase in relative fluorescence of Trp-207 in the G T ␣-subunit that occurs during MII-catalyzed exchange of GDP for GTP␥S upon activation (33,44,45). The results for selected mutants are shown in Fig. 5  (B and C), and all data are compiled in Table 4. The values of G T activation are expressed relative to that of WT MII (100%). Upon light activation, the MII species of all detergent-purified ADRP mutants showed a markedly reduced level of G T activation compared with WT MII. For example, the relative G T activation rates for T4K, N15S, and P23A mutant MII species were 25, 8.9, and 15.3% of WT MII, respectively (Table 4). Only the V20G mutant MII species retained a significant amount of G Tactivating ability (50.8% of WT MII). The rate of G T activation by N2C/D282C rhodopsin MII was 79.8% of the WT MII level. This finding is in agreement with the value reported by Oprian and co-workers (23), who used a filter binding-based G T activation assay. We found that the MII forms of purified ADRP mutants restored in the N2C/D282C rhodopsin background had improved levels of G T activation. ADRP mutants T4K and N15S in the N2C/D282C rhodopsin background exhibited G T activation rates that were 78.2 and 73.4% of WT MII, respectively, cf. 25 and 9% in the absence of the disulfide bond formed by N2C and D282C. These values are comparable to the value of N2C/D282C MII (79.8% of WT). The activation of G T by ADRP rhodopsin mutants in the single N2C or D282C rhodopsin background was essentially the same as that observed in the WT rhodopsin background. We next measured G T activation by the ADRP N2C/D282C mutant in the presence of DTT. As shown in Fig. 5B and Table 4, DTT had no effect on G T activation by WT MII. The N2C/D282C rhodopsin G T activation level was reduced by 10% in the presence of DTT, whereas the activation rates with the single N2C or D282C mutant treated with DTT were ϳ5% lower. In those ADRP mutants in which G T activation was improved in the N2C/D282C background, DTT typically reversed this effect. In the case of Q28H, signal transduction properties were not noticeably improved by the N2C/D282C background, and DTT had no effect.

DISCUSSION
GPCRs are a large superfamily of cell surface receptors that respond to diverse ligands (46). The extracellular domain of GPCRs usually mediates the recognition and entry of the specific ligands that regulate signaling activity. However, visual pigments, such as rhodopsin, are unique GPCRs that have evolved into binary molecular on-off switches that contain a light-sensitive covalently bound retinylidene ligand. Unlike most other GPCRs, the extracellular surface of the opsins is not the site of ligand entry; instead, retinal enters (11-cis) and exits (all-trans) via the 7TM domain (47). As a consequence, we propose that the extracellular domain has been free to evolve into a compact structure that contributes to the extraordinary thermostability and very low dark activity of rhodopsin. The N terminus is an integral component of this domain (Fig. 1), and to investigate its specific role, we performed a detailed analysis of ADRP mutations in this segment.
The fact that 11-cis-retinal ligand can function as a pharmacological chaperone for all known ADRP mutations in the N terminus of rhodopsin suggests that the N-terminal cap is not absolutely required for coordinating assembly of the 7TM bundle and formation of the ligand-binding pocket (5,6,48). Instead, we propose that the cap packs onto the extracellular domain after assembly of the helical bundle and then contributes to the general stability of the ligand-free opsin state, as well as ligand-bound rhodopsin pigment. The differences we observed in the absolute expression levels of the different mutants under pharmacological rescue conditions may be attributable to the severity of the mutational defect, the efficiency of pharmacological rescue, the efficiency of trafficking to the cell surface, and/or the stability of the resultant rescued pigment.
All of the N-terminal ADRP mutants obtained by pharmacological rescue with 11-cis-retinal formed pigments with WT rhodopsin-like dark-state spectral properties, showing that the 7TM helices adopt the correct topology and that EL-2 attains the correct fold. However, the rapid thermal decay of these mutant pigments indicates that a functional cap is required for structural stability. The degree of thermal instability was dependent upon the position of the mutation and the amino acid change and can be understood, in part, by how the folded N-terminal cap packs onto the extracellular surface of the Half-life of MII decay of purified pigments obtained as described in the legend to Fig. 5. Half-life values were derived by curve fitting data to the single-or double-component rise-to-plateau exponential function using SigmaPlot 12.0. In all cases, R 2 Ͼ 0.99). c-e When two (MII) decay components were identified, the t1 ⁄ 2 of the additional component ( d ) and the proportion of each component as a percentage ( c and e ) are indicated. f Initial rates of transducin activation were derived using fluorescence increase measured over the first 60 s after GTP␥S addition. a.u., arbitrary units. g Initial rate of G T activation relative to WT rhodopsin (100%). h ND, not determined; NA, not applicable. i Pigments pretreated with DTT. receptor ( Fig. 1 and Table 5). The crystal structure of the rhodopsin extracellular domain reveals the position of the N terminus above EL-2, sandwiched between EL-1 and EL-3. The ADRP sites are situated at the core of this structure (Fig. 1), with Pro-23 and Gln-28 mediating key interactions with EL-2. The three most unstable pigments, Q28H, P23L, and P23H, have two of the highest packing values in the N terminus, consistent with their position within the core of the N-terminal cap ( Table  5) and their tight interactions with both EL-1 and EL-2. Previously, mutations designed to impair the Arg-177-Asp-190 salt bridge in the EL-2 lid also gave rise to unstable rhodopsin pigments (49); however, expression levels were normal, as was susceptibility to attack by hydroxylamine and G T activation.  Table 4. Each graph is a representative result collected using three independently purified samples. B, G T activation profiles of selected ADRP mutants. The mutant pigments used were prepared as described for A, and G T assays were performed using a fluorescence assay as described under "Experimental Procedures." Red lines, WT background; brown lines, WT plus DTT; blue lines, N2C/D282C background; green lines, N2C/D282C plus DTT. The traces shown are representative of three independent experiments. a.u., arbitrary units. C, comparison of relative initial rates of G T activation by light-activated ADRP mutants prepared using 11-cis-retinal rescue in the WT background (red bars) or the N2C/D282C rhodopsin background without (blue bars) or with (green bars) DTT treatment. The initial rate of G T activation by WT rhodopsin was taken as 100%. The results shown are the means Ϯ S.D. from three independent experiments and are reported in Table 4. All mutant pigments obtained by pharmacological rescue displayed impaired photobleaching (Fig. 4), as reported previously (22). Furthermore, the light-activated mutant pigments had abnormal MII decay profiles with two kinetic components. Similar MII decay profiles have been described previously for other rhodopsin mutants, including those that cause ADRP (50,51). Although the molecular basis of the double-component decay profile is not known, it could be due to incomplete photoconversion of the mutant pigments to MII (Fig. 4). These observations point to several additional roles for the N-termi- Correctly folded opsin is transported to the cell surface, and rhodopsin pigment is formed by treatment of harvested intact cells with 11-cis-retinal. B, in N-terminal ADRP mutants (point mutation indicated by the red star), the cap fails to engage fully with the extracellular domain, leading to destabilization of the transmembrane bundle. Only a small fraction of folded mutant rod opsin evades the quality control machinery of the ER to reach the cell surface; the majority of the mutant rod opsin is retained and targeted for ER-associated protein degradation (ERAD). C, pharmacological rescue. 11-cis-Retinal added to cells reaches the ER and stabilizes the correct rod opsin conformation by occupying the ligand-binding pocket, thus negating the need for the N-terminal cap. ADRP opsin mutants containing 11-cis-retinal are released from the ER and traffic to the cell surface. Harvested cells are treated with 11-cis-retinal once more prior to pigment purification. Pigments formed are unstable in detergent and have altered functional properties because the cap defects remain. D, certain ADRP mutants can be repaired by formation of a disulfide bond (Cys-2-Cys-282 (2C/282C)) in the N2C/D282C opsin background. The defective cap is fixed in place by the disulfide tether (dashed circle), thus conferring to the 7TM domain sufficient stability for opsin release from the ER. Pigment formation is achieved by treatment of transfected cells with 11-cis-retinal after harvest. Pigment function is partially restored because the defective cap is held in place by the Cys-2-Cys-282 disulfide bond. Reduction of the Cys-2-Cys-282 bond by DTT treatment breaks this tether, resulting in conversion back to the mutant phenotype (see Figs. 3 and 5). nal cap: 1) guiding correct photoisomerization of 11-cis-retinal, 2) contributing to associated conformational changes in the 7TM bundle, and 3) mediating MII stability. These observations might explain why G T activation is impaired in all of the N-terminal cap mutants. We suggest that point mutations affect the normal function of the cap by changing its interactions with local structures formed by EL-1-3. The involvement of the extracellular domain in photoactivation and signal transduction has received previous attention (52,53). Conformational changes in the extracellular domain upon light activation are also supported by cysteine accessibility studies (29) and NMR experiments (54). Another possibility is that the cap stabilizes EL-2 to ensure that 11-cis-retinal is isomerized specifically to all-trans-retinal to drive efficiently the subsequent steps of receptor activation, but this idea remains to be tested.
Results from studies using animal cell models show that pharmacological rescue of some N-terminal ADRP mutants occurs in vivo, especially when animals are reared in the dark (55). This may explain why mutants in the N terminus give rise to relatively moderate ADRP (22). However, we found these mutant pigments to be unstable and defective, properties that might give rise to cytotoxic dominant-negative effects if these damaged pigments reach the ROS. Images of deformed discs and ROS in P23H transgenic animal models (56) may be explained by such an outcome.
Engineering in a disulfide bridge (Cys-2-Cys-282) led to restoration of rod opsin folding in a subset of mutants, T4K, N15S, V20G, P23A, and Q28H, a finding that substantiates our suggestion that the N-terminal cap has a role in stabilizing the 7TM bundle and in maintaining the correct conformation of the binding pocket in the rod opsin form. Surprisingly, P23H, P23L, and T17M did not fold correctly in this background, a result that validates our dual experimental strategy because it enabled us to discriminate between the mutants. One possibility why these three mutants failed to respond is that they severely deform the cap such that Cys-2 and Cys-282 do not come into close enough proximity for disulfide bond formation. Another possibility is that the Cys-2-Cys-282 bond does form, but the cap is still unable to attain a conformation needed to productively interact with EL-1-3 and to stabilize the 7TM bundle.
To obtain pigment for P23H and P23L in the N2C/D282C background, we needed to perform simultaneous pharmacological rescue with 11-cis-retinal; however, pigment formation was even lower than in the WT rhodopsin background (Table  1) and may be explained by the non-bonded Cys-2 and Cys-282 residues acting as additional point mutations that exacerbate misfolding. An alternative explanation is that the Cys-2-Cys-282 bond does form, but the major fraction of the protein stabilized is a misfolded species. Interestingly, despite the small yield, these pigments were more thermostable than those in the WT background and had improved photobleaching properties, properties consistent with the formation of the Cys-2-Cys-282 bond.
The purified T4K, N15S, V20G, P23A and Q28H mutant pigments obtained using the N2C/D282C rhodopsin background had much improved thermostability, WT-like photobleaching properties, and single-component MII decay kinetics. Some (T4K, N15S, V20G, and P23A) activated G T to an extent comparable to the control pigment (WT N2C/D282C), but G T activation by repaired Q28H did not improve. Taken together, these observations indicate that the Cys-2-Cys-282 bond tethers the defective caps in place, and although this is sufficient for improved stability and photoactivation, the fit is not always optimal. In certain experiments, the improved function was reversed by treating the pigments with DTT to reduce the Cys-2-Cys-282 disulfide bond (Figs. 3 and 5 and Tables 2  and 4), providing further evidence that the Cys-2-Cys-282 disulfide bond is responsible for the repaired phenotype. Although disulfide bond-mediated repair has no obvious use as a therapeutic tool, our findings indicate that regions other than the orthosteric ligand-binding site (57), such as the N terminus, should be considered as future drug targets for stabilization of ADRP rhodopsin mutants.
A direct comparison of the two approaches, pharmacological rescue and N2C/D282C repair, highlights the role of the N-terminal cap in rhodopsin structure and function and sheds light on likely defects arising from ADRP mutations in this region. In both cases, we propose that repair/rescue is likely to occur relatively late in the protein-folding process. This is because rescue requires a native-like binding pocket, whereas repair requires that folding has reached a stage such that the two cysteine residues, separated by 280 amino residues in the rod opsin polypeptide chain, acquire close proximity. However, we cannot rule out the possibility that disulfide bond formation between Cys-2 and Cys-282 happens early in folding due to flexibility of the N-terminal segment prior to adopting its compact folded-state structure. We thus propose that these N-terminal ADRP mutations do not prevent assembly of the 7TM bundle, but rather lower the stability of the final folded rod opsin structure.
Our observations allowed us to construct a model to help explain rhodopsin misfolding and restoration of function (Fig.  6) and to draw the following conclusions. 1) A functional cap is required to stabilize the 7TM bundle of rod opsin (Fig. 6A). 2) Defects in the cap result in instability of the assembled 7TM bundle and retention of unstable rod opsin by the ER prior to ER-associated protein degradation (Fig. 6B). 3) Instability in these ADRP mutants can be compensated by stabilizing the 7TM bundle by occupancy of the retinal ligand pocket during opsin biosynthesis (Fig. 6C) or, in certain cases, repaired by securing the defective lid by a Cys-2-Cys-282 disulfide bridge (Fig. 6D). 4) The cap also stabilizes the positions of EL-1-3, which are integral to coordinated movement of the 7TM domain, receptor activation, and signaling. In conclusion, the dual strategy we have used here in attempts to restore folding in ADRP rhodopsin mutants is significant because it has shed light on the multifunctional role of the N terminus in rhodopsin and provides much needed insight into mechanisms by which mutations in this domain might initiate ADRP.