Retinoids assist the cellular folding of the autosomal dominant retinitis pigmentosa opsin mutant P23H.

The clinically common mutant opsin P23H, associated with autosomal dominant retinitis pigmentosa, yields low levels of rhodopsin when retinal is added following induction of the protein in stably transfected HEK-293 cells. We previously showed that P23H rhodopsin levels could be increased by providing a 7-membered ring, locked analog of 11-cis-retinal during expression of P23H opsin in vivo. Here we demonstrate that the mutant opsin is effectively rescued by 9- or 11-cis-retinal, the native chromophore. When retinal was added during expression, P23H rhodopsin levels were 5-fold (9-cis) and 6-fold (11-cis) higher than when retinal was added after opsin was expressed and cells were harvested. Levels of P23H opsin were increased approximately 3.5-fold with both compounds, but wild-type protein levels were only slightly increased. Addition of retinal during induction promoted the Golgi-specific glycosylation of P23H opsin and transport of the protein to the cell surface. P23H rhodopsins containing 9- or 11-cis-retinal had blue-shifted absorption maxima and altered photo-bleaching properties compared with the corresponding wild-type proteins. Significantly, P23H rhodopsins were more thermally unstable than the wild-type proteins and more rapidly bleached by hydroxylamine in the dark. We suggest that P23H opsin is similarly unstable and that retinal binds and stabilizes the protein early in its biogenesis to promote its cellular folding and trafficking. The implications of this study for treating retinitis pigmentosa and other protein conformational disorders are discussed.

Rhodopsin, the major pigment of vertebrate rod cells, is an integral membrane glycoprotein (ϳ40 kDa) that functions as a G protein-coupled receptor in detecting dim light (1). Rhodopsin biogenesis involves the membrane insertion of opsin, a 348-amino acid polypeptide that must fold to form seven transmembrane ␣-helices and assemble with the chromophore, 11cis-retinal. The chromophore is covalently attached to opsin by a protonated Schiff base linked to Lys 296 in the seventh ␣-helix and interacts extensively with amino acid residues located in transmembrane ␣-helices (2). It also contacts residues at the base of the "retinal plug," a short four-stranded ␤-sheet assembled from the N-terminal segment and the second extracytoplasmic loop of opsin (2). Given these intimate contacts, key questions are whether retinal can bind opsin at the early stages of its biogenesis and whether such binding increases the yield of the protein.
Retinal binding to opsin may be impaired in some forms of autosomal dominant retinitis pigmentosa (ADRP), 1 a group of hereditary disorders that leads to rod photoreceptor death and subsequently to a severe loss of peripheral and night vision (3,4). Approximately 150 mutations in opsin are implicated in this disease (5,6). Based on heterologous expression studies (7)(8)(9), mutant opsins fall into three classes: Class I mutants bind retinal properly and accumulate at the cell surface; Class II mutants fail to bind retinal, are improperly glycosylated, and accumulate intracellularly; and Class III mutants bind retinal slightly, have less severe glycosylation defects than Class II mutants, and are found both on the cell surface and intracellularly. Class II and III mutants may have defects in the membrane insertion of opsin, opsin folding, retinal binding, or in the stability of the pigment once retinal has bound.
A potential strategy to treat ADRP caused by Class II and III rhodopsin mutants is to use retinal as a pharmacological chaperone (10), a ligand that corrects the misfolding of a protein by binding specifically to non-native or native forms. Pharmacological chaperones appear to act mainly in the endoplasmic reticulum (ER), releasing the proteins they target from the ER quality control machinery and increasing their flux to other cellular compartments (10). Because opsin stability in vitro is increased by the binding of 11-cis-retinal and other retinoids (11)(12)(13), these compounds may act as pharmacological chaperones and increase the yield of Class II and III rhodopsin mutants. In support of this idea, the level of T17M rhodopsin (a Class III mutant (8)) regenerated from cell membranes increased 10-fold when the protein was expressed in the presence of 11-cis-retinal (14). Cell surface expression of the Class III opsin, P23H, was increased in COS-7 cells when 9-cis-retinal was provided (15). Similarly, the addition of a 7-membered ring, locked, non-photoisomerizable form of 11-cis-retinal to P23H opsin during its expression promoted trafficking of the mutant protein and increased the yield of the pigment relative to samples in which retinal was added after cell harvesting (16). Although these studies suggest that they can act as pharmacological chaperones for mutant opsins, further experiments are needed to establish the precise function of retinoids.
In this study, we show that the stimulation of P23H pigment production by added retinoids is not limited to the locked retinal used in our earlier work (16). We demonstrate that yields of the protein can also be increased by 9-or 11-cis-retinal, the native chromophore. As observed with the locked retinal, the mutant pigment was formed efficiently only when retinals were added during opsin expression. We show that 9-and 11-cisretinal promote trafficking of P23H opsin through the secretory pathway to the cell membrane, indicating that retinals bind the protein at an early stage. Finally, we show for the first time that P23H rhodopsin has a decreased thermal stability relative to wild-type (WT) rhodopsin, providing a partial explanation for the defect in its production.

EXPERIMENTAL PROCEDURES
Cell Culture, Regeneration, and Purification of Rhodopsin-Cell lines and growth conditions were described previously (16). All steps were carried out under dim red light. Retinals in 100% ethanol were added to a final concentration of 50 M 2 h after growing cultures were induced with tetracycline (0.5 g/ml) or after cells were harvested 48 h after induction. Cells were harvested and lysed in PBS (10 mM sodium phosphate buffer, 137 mM NaCl), pH 7.4, containing 1% n-dodecyl-␤maltoside (DM) (Anatrace), and rhodopsin was purified by immunoaffinity chromatography as described previously (16). Briefly, rhodopsin in cell lysates was bound to monoclonal antibody 1D4, which recognizes the rhodopsin C terminus, and was coupled to Sepharose 4B (Amersham Biosciences). The protein was eluted with a synthetic peptide corresponding to the last 18 amino acid residues of the rhodopsin C terminus in 10 mM sodium phosphate buffer, pH 6.0, or PBS, both containing 0.1% DM. UV-visible spectra of eluted pigments were recorded on a Tidas II spectrophotometer (World Precision Instruments). Visible absorption maxima ( max ) were determined from the peak value of smoothed spectra. The variation in max of two to four independently purified samples was less than Ϯ1 nm.
Retinoid Analysis-All procedures were carried out under dim red light (Ͼ660 nm). Retinoids were analyzed following slight modifications of the procedures of Groenendijk et al. (17) and Smith and Goldsmith (18). Freshly purified rhodopsin was dried under a stream of nitrogen at room temperature. A 1:1 mixture of methanol and 1 M hydroxylamine, pH 6.5, was added to each sample, and the samples were mixed by vortexing. Then methanol was added to a concentration of 70%, and the samples were mixed again in a 1.7-ml centrifuge tube. Water and dichloromethane were added to yield a 1:1:1 ratio of water:methanol: dichloromethane. The extracts were mixed briefly by vortexing and then centrifuged, and the lower organic phase was collected. The procedure of adding dichloromethane, vortexing, centrifugation, and collection of the lower phase was repeated twice. The collected organic phases were combined and dried under a stream of nitrogen gas. Samples were stored in the dark at Ϫ80°C if they were not analyzed immediately. The residue was dissolved in 1% isopropyl alcohol in hexane, and an aliquot was immediately injected into the HPLC sample and resolved according to the procedure of Smith and Goldsmith (18), with detection at 357 nm.
SDS Gel Electrophoresis and Immunoblotting-Purified rhodopsin samples were electrophoresed on 10% SDS polyacrylamide gels and transferred onto Immobilon-NC (Millipore) nitrocellulose membranes. Membranes were incubated at room temperature for 1 h with blocking buffer (Licor), then for 1 h with monoclonal 1D4 antibody, then three times for 5 min each in PBST (PBS containing 0.1% Triton X-100, pH 7.4), and finally for 1 h with IRDye800-conjugated goat anti-mouse secondary antibody (Rockland, Inc.). Finally the membranes were again washed three times with PBST and scanned in an Odyssey infrared scanner (Licor).
For quantifying total opsin, the total cell protein concentration in cell lysates was determined by the DC protein assay (Bio-Rad), with bovine serum albumin as a standard. Equivalent amounts (10.0 g) of total protein from each sample were electrophoresed, immunoblotted, and scanned as described above. Total opsin was quantified with the Odyssey software.
Peptide N-Glycosidase F (PNGase F) Treatment and ConA Blotting-Rhodopsin samples (ϳ0.1 g) eluted from 1D4 Sepharose beads were incubated with 2.5 milliunits of recombinant PNGase F (Prozyme) for 2 h at room temperature in 20 mM Tris-HCl, 5 mM EDTA, 0.5% w/v SDS, 1% v/v Nonidet P-40, 0.1 mM phenylmethylsulfonyl fluoride, pH 8.0. In an attempt to improve digestion, the following variations of this buffer were explored: a Nonidet P-40 concentration of 2.5 or 5% v/v; a 2-fold increase in both SDS and Nonidet P-40 concentrations; digestion of the reaction with a second aliquot of enzyme in an equal volume of buffer; incubation at 37°C with an identical buffer measuring pH 7.5 at room temperature; addition of 50 mM ␤-mercaptoethanol or dithiothreitol; a preincubation period of up to 5 h in a buffer lacking Nonidet P-40 followed by adjustment of Nonidet P-40 to 1% v/v and addition of enzyme; and the manufacturer's recommended buffers. Untreated samples were identical except that water was added instead of PNGase F. Samples were electrophoresed on a 10% SDS-polyacrylamide gel and immunoblotted with 1D4 antibody as described above or blotted with 0.001% ConA-biotin (Sigma) in blocking buffer (Licor) followed by a streptavidin-IRdye800 conjugate (Rockland) at a concentration of 0.2 g/ml in PBST.
Immunocytochemistry-Cells were grown on glass coverslips to 50% confluence, and opsin expression was induced by tetracycline. Cells were treated with different retinoids after 2 h of induction. After 24 h of treatment, cells were fixed for 5 min with ice-cold methanol, treated for 5 min with ice-cold acetone, and washed three times with PBS. Cells were blocked with 10% normal goat serum (Sigma) in PBS for 1 h at room temperature. Cover slips were washed with PBST and were incubated for 1 h at room temperature with purified 1D4 antibody diluted 1:1000 with PBST. Following incubation, coverslips were rinsed in PBST and incubated with rhodamine (TRITC)-conjugated goat antimouse IgG (Jackson ImmunoResearch Laboratories) and DAPI (Molecular Probes). Cells were rinsed in PBST, mounted in 50 l of anti-fade reagent (Molecular Probes) to retard photobleaching, and analyzed with a DMIRE2 Leica microscope.
Characterization of Rhodopsins-Photobleaching, hydroxylamine sensitivity, and thermal stability studies were conducted using a Cary 50 UV-visible spectrophotometer (Varian) and analyzed using Igor Pro software (Wavemetrics). Spectra were corrected for slight base-line displacements by subtracting the average absorbance from 625-650 nm, a frequency range where rhodopsin does not absorb significantly, from the entire spectrum. Photobleaching was performed in 0.1% DM in 10 mM sodium phosphate, pH 6.0, at room temperature using a 150watt fiber-optic light source (Dolan-Jenner) with a Ͼ495 nm filter. Single samples were illuminated for three 10-s intervals, 30 s, and 1 min, with scanning after each period. Hydroxylamine sensitivity was determined as described previously (16), except that 1 M hydroxylamine, pH 6.0, was added to purified rhodopsin samples to a final concentration of 40 mM hydroxylamine. Spectra were multiplied by 1.04 to correct for the dilution because of hydroxylamine addition. Thermal stability was measured by obtaining full spectra every 2 min at 37°C in 0.1% DM in PBS, pH 7.4. The absorbance value at the absorption maximum was determined, corrected for base-line displacements, and normalized to the absorbance at the initial time point. Values from two independent experiments were averaged and fitted using Igor Pro with a monoexponential function for P23H samples and a linear function for WT samples.

RESULTS
Rationale-We showed previously that 11-cis-7-ring but not 11-cis-6-ring or 11-cis-9-demethyl-7-ring retinal increased the yield of P23H rhodopsin if administered during opsin synthesis (16). To test whether simpler retinoids, particularly the native chromophore, are capable of increasing the yield of P23H rhodopsin, we investigated the effects of 9-and 11-cis-retinal. The cellular expression, pigment characteristics, and cellular localization are reported here; full functional characterization of the proteins will be reported elsewhere.
Yield and Pigment Characteristics of WT Rhodopsin-Control experiments were conducted first with WT opsin to compare the yield of rhodopsin when retinals were added during opsin expression with the yield of rhodopsin when retinals were added after opsin had been expressed and cells harvested (postharvest). Forty-eight hours after tetracycline induction, HEK-293 cells expressing WT opsin were harvested and incubated with both 9-and 11-cis-retinals. The rhodopsin formed was purified by immunoaffinity chromatography under conditions that selectively release folded opsin (19), and the first eluted fraction was examined by UV-visible spectroscopy and HPLC. The pigment obtained in post-harvest regeneration of WT opsin with 9-cis-retinal has a shifted absorption spectrum with max ϭ 487 nm (Fig. 1A, dashed line), whereas the pigment obtained with 11-cis-retinal has the characteristic max ϭ 500 nm of native rhodopsin (Fig. 1B, dashed line).
When retinals were added under dim red light to the media of cells actively synthesizing WT opsin, the same spectral characteristics of the purified proteins were observed as in the post-harvest regeneration (Fig. 1, A and B, solid lines). The addition of 9-cis-retinal to the growth medium produced a pigment with max ϭ 488 nm, whereas 11-cis-retinal produced a pigment with max ϭ 500 nm. On acid denaturation (Fig. 1, A and B, insets) both proteins yielded a pigment with a max near 440 nm, indicating that retinal is linked as a Schiff base in the native protein (20). These pigments contained 9-and 11-cisretinal, respectively, as the principal chromophores as shown by HPLC (Fig. 1, E and F).
Immunoaffinity purification indicated that the cellular yield of the WT protein was similar whether retinal was added during expression or post-harvest but was ϳ30% higher with 9than with 11-cis-retinal (compare Fig. 1, A and B). To test whether these observations were representative of the total opsin in the cell, the opsin content of detergent-solubilized cell lysates was determined by quantitative immunoblotting. WT opsin yields increased only slightly (1.3-and 1.2-fold for 9-and 11-cis-retinal, respectively) when retinal was added during expression (Table I), results similar to the immunoaffinity purification. The level of WT opsin obtained in experiments with 9-cis-retinal was slightly less than that obtained with 11-cisretinal (ϳ0.8-fold) and independent of whether the cofactor was added during expression or post-harvest. Thus, the effects of 9and 11-cis-retinal on WT opsin and rhodopsin levels are relatively small.
Yield and Pigment Characteristics of P23H Rhodopsin-Parallel experiments were conducted with cells expressing P23H opsin. Cells were first harvested 48 h after induction with tetracycline and regenerated post-harvest with 9-or 11-cisretinal, and the pigments were purified by immunoaffinity chromatography. The amount of pigment obtained with 9-or 11-cis-retinal (Fig. 1, C and D, dashed lines) was less than 10% of that obtained post-harvest with the WT protein.
Strikingly, when the individual retinals were added to cells during expression of P23H opsin, a significant amount of pigment was produced (Fig. 1, C and D, solid lines). The increase in P23H rhodopsin yield was similar for both retinals, reaching levels that were 5-fold (9-cis) and 6-fold (11-cis) higher than the corresponding post-harvest samples (compare Fig. 1, C and D,  dashed lines). Although the relative increase was similar for both retinals, the absolute yield of P23H rhodopsin was higher with 9-cis-retinal, typically twice that obtained with 11-cisretinal (compare Fig. 1, C and D, solid lines). The pigment generated in the presence of 9-cis-retinal has a max ϭ 479 nm, whereas the pigment generated in the presence of 11-cis-retinal has a max ϭ 492 nm. The 8-nm blue shift in max with either 9-or 11-cis-retinal compared with WT rhodopsin suggests that the structure of P23H rhodopsin is slightly altered. The retinal in both P23H pigments is linked as a Schiff base (Fig. 1, C and D, insets). As in the case of the WT pigments, 11-cis-retinal is the major chromophore bound to P23H rhodopsin when 11-cis-retinal is added to the growth medium, whereas 9-cis-retinal predominates in pigments purified from cells grown in the presence of 9-cis-retinal (Fig. 1, G and H).
Analysis of the total yields of P23H opsin confirmed the results obtained by immunoaffinity purification. When 9-and 11-cis-retinals were added during expression, the P23H opsin level was increased by 3.5-and 3.4-fold, respectively, compared with the level when retinals were added post-harvest (Table I). Consistent with the increased yield of immunoaffinity purified P23H rhodopsin, P23H opsin levels were ϳ50 -70% higher with 9-cis-than with 11-cis-retinal (Table I).
Glycosylation State of Rhodopsin-To determine whether added retinals act early in rhodopsin biogenesis, we examined the glycosylation state of the purified protein. P23H rhodopsin purified from cells and exposed to 9-or 11-cis-retinal during expression acquires a high molecular mass smear above the major band at 36 kDa ( Fig. 2A, lanes 2 and 4), similar to the WT protein (Fig. 2, A and B, lanes 1 and 3). The 36-kDa form (Fig.  2C, band A) and the high molecular mass smear in all samples correspond to glycosylated protein because they are converted by PNGase F to 33 kDa (Fig. 2C, band B) and 30 kDa species (Fig. 2C, band C). Band C (Fig. 2C) is not detected by ConA blotting (Fig. 2D) and therefore corresponds to deglycosylated opsin; band B (Fig. 2C) is more abundant in P23H samples. It is unlikely to be the result of incomplete PNGase F digestion because incubation for 72 h, a second addition of enzyme, or other changes in incubation conditions (see under "Experimental Procedures") did not significantly alter the digestion pattern. Band B is glycosylated (Fig. 2D), but the glycan is apparently inaccessible or modified and is resistant to PNGase F. Although complete conversion to a deglycosylated form is not observed with PNGase F, the smear is clearly absent. The high molecular mass smear results from glycan modification by the Golgi apparatus (8). These results indicate that when P23H expression is induced in the presence of retinoids, an increased amount of the protein reaches the Golgi and is processed sim-ilarly to the WT protein. Thus, when they are added, 9-and 11-cis-retinals must interact with P23H opsin early during its biogenesis in either the ER or the Golgi apparatus.
In P23H rhodopsin samples a distinct 28-kDa immunoreactive band is observed (Fig. 2C, band D). This species is not shifted by PNGase F, suggesting that it does not contain an N-linked carbohydrate. Because this species is recognized by the 1D4 antibody, has an apparent molecular mass lower than that of deglycosylated opsin, and is not detected by N-terminalspecific antibody R2-15 (data not shown), it appears to be a degradation product formed by proteolysis near the N terminus.
Localization of Opsin-To further test whether added retinals act at an early stage of rhodopsin biogenesis, we examined the distribution of WT and P23H opsin in cells exposed to 9-or 11-cis-retinal during expression of the protein. Immunofluorescence microscopy with 1D4 antibody was used to detect opsin. Control experiments with uninduced WT and P23H cells lack an opsin-specific signal; only the DAPI staining of nuclei is detected (Fig. 3, A and E, blue fluorescence). The WT protein reaches the cell surface in the absence of added retinal (Fig. 3B) or in the presence of either 9-or 11-cis-retinal (Fig. 3, C and D) as observed by the presence of yellow-orange fluorescence on the cell periphery. In contrast, only a little P23H opsin is found at the cell surface in the absence of retinal, and distinct intracellular inclusions are observed (Fig. 3F). The inclusions appear to be aggresomes, 2 similar to those observed previously in COS-7 and HEK-293 cells (15,21). Strikingly, there are fewer inclusions and a substantial increase of P23H opsin at the cell surface with 9-or 11-cis-retinal (Fig. 3, G and H). Thus, adding either of the retinals to cells during opsin expression appears to promote the trafficking of the protein to the cell surface.
Characterization of Purified Rhodopsins-The increased yield of P23H rhodopsin in our experiments provides an opportunity to assess the physical and functional properties of the protein in detail. We therefore examined the photosensitivity, hydroxylamine sensitivity, and thermal bleaching of rhodopsin purified from cells to which retinals were added during opsin expression. Illumination with Ͼ495 nm light resulted in decreased absorbance in the visible region, indicating that all of the rhodopsin pigments are photoactive (Fig. 4, A-D). A 10-s illumination converted the 9-and 11-cis WT rhodopsin to 380-nm species (Fig. 4, A and B), consistent with the formation of metarhodopsin II (22). Further illumination produced minor changes in the spectra of these proteins. In sharp contrast, a 10-s illumination of 9-and 11-cis P23H rhodopsin resulted in broad spectra (Fig. 4, C and D). Continued illumination of these pigments decreased absorbance near 480 nm with a slight increase at 380 nm. A similar but slower conversion was observed when the pigments were illuminated for 10 s and monitored in the dark (data not shown). The broad spectra may include species similar to the WT photointermediates metarhodopsin I (Meta I, max ϭ 480 nm), and II (Meta II, max ϭ 380 nm) and III (Meta III, max ϭ 465 nm), which are in thermal equilibrium with Meta I (23,24). A possible explanation for the broad spectra is that this equilibrium is shifted in P23H rhodopsin, resulting in higher levels of Meta I and III at the expense of Meta II. These results indicate that P23H rhodopsins have altered photo-bleaching properties.
Incubation of the detergent-purified samples with hydroxylamine, pH 6.0, revealed that 9-and 11-cis-retinal WT proteins were only slightly sensitive to this treatment (Fig. 4E, data for 9-cis not shown). However, detergent-purified 9-and 11-cis P23H pigments were both vulnerable to hydroxylamine attack 2 S. Kaushal, unpublished results. a The level of WT opsin obtained by adding retinal post-harvest was set to 1.00, and all other samples were normalized to this value. Values shown represent the average and standard deviation from four independent experiments.
b Ratio of the relative opsin level obtained from adding retinal during expression to that obtained post-harvest. (Fig. 4, F and G), consistent with lower protein stability. The rate of reaction was more rapid than thermal bleaching under identical pH and salt conditions (data not shown). Strikingly, the P23H pigments were insensitive to hydroxylamine when treated in HEK-293 cells prior to purification (Fig. 4H), suggesting that the lipid bilayer stabilizes P23H rhodopsin.
Finally, thermal stability studies at elevated temperatures in the dark demonstrate that 9-and 11-cis P23H rhodopsin were bleached more rapidly than the corresponding WT proteins (Fig. 5). The bleaching spectra of each P23H rhodopsin overlapped at a single isosbestic point, consistent with a twostate conversion from rhodopsin to opsin and retinal (data not shown). Analysis of these data at the absorption maximum revealed that 9-cis P23H rhodopsin was bleached with a t1 ⁄2 Ϸ 5.0 min (Fig. 5, squares), whereas the 11-cis P23H pigment was bleached with a t1 ⁄2 Ϸ 5.8 min (Fig. 5, circles). No bleaching was detected in the WT pigments after 20 min (Fig. 5, triangles and  inverted triangles). Thus, P23H rhodopsin containing 9-or 11-cis-retinal is less thermally stable than WT rhodopsin. DISCUSSION We have demonstrated that 9-and 11-cis-retinals act as pharmacological chaperones to increase the yield of correctly folded P23H rhodopsin in HEK-293 cells. When these retinals were provided to cells post-harvest, the yield of pigment was very low, in agreement with earlier results (8,16,25). In contrast, when the retinals were added during opsin expression, P23H rhodopsin levels were increased 5-6-fold. Opsin levels were increased by ϳ3.5-fold under these conditions; the lower relative increase shows that a greater fraction of the protein folds correctly when retinal is added during expression. The higher yield of P23H rhodopsin was the result of more efficient trafficking through the secretory pathway as evidenced by the increase in high molecular weight glycosylation of the purified protein and by the localization of P23H protein at the cell surface. The purified 9-and 11-cis P23H rhodopsins have a major spectral peak in the visible region and contain the expected retinal as the major chromophore. Thus, 9-and 11cis-retinal are specific ligands that act early during biogenesis to increase the cellular yield of P23H opsin and rhodopsin, consistent with a role of these compounds as pharmacological chaperones.
Our results indicate that P23H rhodopsin has a structural alteration that decreases the stability of the protein. In contrast to earlier spectral studies (26), we observed an 8-nm shift in the visible absorption maximum with both 9-and 11-cisretinal, suggesting that the structure of the protein near retinal is different from that in WT rhodopsin. The unusual response of the mutant proteins to light provides further evidence for an altered retinal environment. Most strikingly, 9-and 11-cis P23H rhodopsins have decreased thermal stability and increased hydroxylamine sensitivity, as observed with the locked analog of 11-cis-retinal (16). Pro 23 is located on the intradiscal face of rhodopsin in close proximity to residues that form the retinal plug, and its substitution with His may disrupt this structure and alter the retinal environment, resulting in thermal instability.
To explain retinal rescue, we propose that the thermal stability of P23H opsin is decreased, similarly to P23H rhodopsin. Retinals may bind to native P23H opsin thereby stabilizing it, or they may bind non-native folding intermediates, acting as a scaffold for subsequent folding, and increase the rate at which these intermediates are converted to the native form. Such binding would prevent the protein from being recognized by the ER quality control machinery as misfolded, allowing it to escape degradation and promoting its transport to the Golgi apparatus. Although purified P23H rhodopsin in detergent is highly unstable, the protein accumulates because it is relatively stable in cell membranes.
Because our experiments examine steady-state levels of opsin rather than newly synthesized protein, we are unable to rule out the possibility that 9-and 11-cis-retinal might also act at a stage of biogenesis other than the secretory pathway. For example, retinals may influence the turnover and degradation of opsin at the cell surface, which has not been examined in HEK-293 cells. In support of this idea, when retinals are added to cells post-harvest, the level of P23H opsin is ϳ50% higher with 9-cis-retinal compared with 11-cis-retinal. Under these conditions, no new synthesis of opsin is expected, suggesting that the increase is because of decreased turnover of previously synthesized protein. Thus, reduced turnover may partially account for the increase in P23H rhodopsin levels when retinals are added during expression. Retinals may also increase P23H opsin yields by increasing opsin gene transcription or opsin translation. However, WT opsin levels were not increased significantly under any of the conditions examined, arguing against this possibility.
Thermal instability may be a common property of Class II and III mutant opsins. Previous studies (7,8,15,25,27,28) of these proteins led to the conclusion that the mutant opsins were misfolded based on low pigment yield, intracellular accumulation, altered glycosylation, or non-native disulfide bonds but did not provide direct evidence for defects in protein thermal stability or folding kinetics. However, several mutant rhodopsins associated with ADRP were recently shown to exhibit decreased thermal stability (16,29,30); our studies of P23H rhodopsin provide another example. If thermal instability is common among ADRP mutant opsins, it may be possible to correlate the degree of instability with progress of the disease and to develop therapeutic approaches using retinoids as pharmacological chaperones.
The rescue of heterologously expressed mutant opsin by retinals raises important questions about rhodopsin biogenesis in the mammalian rod cell. First, at what stage of biogenesis does retinal bind newly synthesized opsin? Opsin biosynthesis is initiated in the inner segment of the rod, which contains the ER and Golgi apparatus (31), but the protein ultimately resides in the outer segment discs of the rod. Early autoradiographic and biochemical analyses (32,33) in mice suggested that retinal binds to opsin only in the outer segment. However, subsequent experiments revealed that rhodopsin is present in the rough ER of rod cells, indicating that retinal binds to opsin in the inner segment (34). Our observation that retinals bind early during rhodopsin biogenesis is consistent with this view but does not resolve the contradiction between the earlier findings. Second, can other ADRP opsin mutants be rescued by retinals or other pharmacological chaperones? This intriguing possibility was first raised by Berson and co-workers (14), who demonstrated rescue of mutant T17M in mice with oral doses of vitamin A palmitate. Our results indicate P23H opsin is an excellent candidate for in vivo rescue.
ADRP caused by P23H opsin may be classified as a protein conformational disease (PCD). PCDs are mostly neurodegenerative disorders in which the misfolding of a specific protein generates non-native intracellular or extracellular aggregates correlated with a progressive loss of cell function and cell death (35)(36)(37). Other misfolded proteins, which are associated with diseases that are not formally classified as PCDs, have been found to aggregate in pericentriolar structures termed aggresomes (38 -40). Aggregate formation may therefore be a common attribute of disease-causing misfolded proteins. Because P23H opsin is conformationally unstable and forms aggresomes 2 (15,21), it may be considered a cause of PCD. The results presented here along with our earlier work (16) thus support the idea that an effective therapeutic strategy for PCDs may be to correct the misfolding of the underlying protein (36,41).