|Home | About | Journals | Submit | Contact Us | Français|
In absence of their natural ligand 11-cis retinal, cone opsin GPCRs fail to traffic normally, a condition associated with photoreceptor degeneration and blindness. We created a mouse with a point mutation (F81Y) in cone S-opsin. As expected, cones with this knockin mutation respond to light with maximal sensitivity red-shifted from 360 nm to 420 nm, consistent with an altered interaction between the apoprotein and ligand, 11-cis retinal. However, cones expressing F81Y S-opsin showed an ~3-fold reduced absolute sensitivity that was associated with a corresponding reduction in S-opsin protein expression. The reduced S-opsin expression did not arise from decreased S-opsin mRNA or cone degeneration, but rather from enhanced ER-associated degradation (ERAD) of the nascent protein. Exogenously increased 11-cis retinal restored F81Y S-opsin protein expression to normal levels, suggesting that ligand binding in the ER facilitates proper folding. Immunohistochemistry and electron microscopy of normal retinas showed that Mueller cells, which synthesize a precursor of 11-cis retinal, are closely adjoined to the cone ER, so could deliver the ligand to the site of opsin synthesis. Together, these results suggest that the binding of 11-cis retinal in the ER is important for normal folding during cone opsin biosynthesis.
Opsins are G-protein coupled receptors (GPCRs) that initiate vision via a signaling cascade whose molecular components are highly concentrated in an elaborated cilium, the photoreceptor outer segment. The synthesis of these components creates a large daily burden of transcription, translation, protein trafficking and turnover, with total opsin renewed at a rate of about 10% per day (1967, 1971; Hollyfield, 1979; Jonnal et al., 2010). More than 100 mutations in rhodopsin have been associated with retinal degeneration, typically autosomal dominant retinitis pigmentosa (Hartong et al., 2006), and most of these mutations can be characterized as causing opsin misfolding (Rakoczy et al., 2011). An increasing number of cone opsin mutations are being associated with retinal disease, thanks to the growing application of high resolution fundus imaging, high throughput genotyping and other novel methodologies (Carroll et al., 2009; Gardner et al., 2010). Because cones are essential for daytime vision, including color and high acuity foveal vision, cone-specific disease has an especially high impact on visual function.
Many neurodegenerative diseases like retinal degeneration result from an overload of the endoplasmic reticulum associated degradation (ERAD) pathway (Lin et al., 2007; Kosmaoglou et al., 2008; Matus et al., 2011). Quality control of nascently translated proteins in the ER is an essential function of all cells (Ellgaard and Helenius, 2003), and ERAD helps insure that proteins exported from the ER are correctly folded (Ron and Walter, 2007; Vembar and Brodsky, 2008). When ERAD is insufficient to cope with improperly folded proteins, they can collect in the ER, aggregate and severely disrupt normal cellular function. Because a substantial fraction of proteins are misfolded during biosynthesis (Petaja-Repo et al., 2000; Schubert et al., 2000; Jung et al., 2009), photoreceptors must have an unusually robust ERAD capacity to accompany the high rate of opsin biosynthesis.
We created a knockin mouse with a point mutation (F81Y) in S-opsin, predicted to alter 11-cis retinal interaction with S-opsin apo-protein, and shift the wavelength of maximum sensitivity (λmax) from the wildtype value of 360 nm to 425 nm (Fasick et al., 2002). Initial characterization revealed that while cones expressing the mutant opsin were morphologically normal and responded to light with the predicted shift in λmax, the quantity of F81Y S-opsin expressed was reduced below 50% of the wildtype (WT) level, with a corresponding reduction in light sensitivity. We thus undertook a series of studies to determine the mechanisms underlying the reduction in F81Y S-opsin expression.
We created mice with the F81Y mutation in S-opsin because previous heterologous expression in COS cells had shown this mutation to shift the spectral maximum from the ultraviolet (360 nm) to ~420 nm (Fasick et al., 2002), close to that of human S-opsin. (Our numbering of residues is based the mouse S-opsin sequence, accession # NP_031564.1, while that used by Fasick et al., F86Y, was based on the bovine rhodopsin sequence.) The strategy for introducing the F81Y point mutation into mouse Opn1sw is detailed in Daniele et al. (2011). To create Opn1swF81Y/F81Y mice, homozygous targeted mice with the Neomycin resistance cassette inserted between exons 3 and 4 (Opn1swNeo/Neo) were bred against mice expressing a transgene for FLP1 recombinase expressed under the human ACTB promoter (Jackson Laboratory, Bar Harbor, Maine, stock no. 003800). Excision of the Neo selection cassette was confirmed from mouse genomic DNA using PCR. The PCR primer pairs (forward: 5’- AGT ATC GAA GCG AGT ACT ACA CCT GG -3’; reverse: 5’- GCCTTT TGT GTC GTA GCA GACTCT TG -3’) detected Opn1swF81Y and Opn1swWT. The primer pair (forward: same as previous; reverse: 5’- CCC GGT AGA ATT CGA GAA GTT CCT AT -3’) detected targeted Opn1swNeo and Opn1swF81Y. Together these PCR primers were sufficient for discriminating Opn1swWT, Opn1swNeo, and Opn1swF81Y alleles. Further crosses of mice with confirmed excision of the Neo selection cassette (Opn1swF81Y/WT) allowed for selection of FLP transgene-negative Opn1swF81Y/WT mice. The FLP transgene was detected by PCR with primers recommended by Jackson Labs (forward 5’-CAC TGA TAT TGT AAG TAG TTT GC-3’; reverse 5’-CTA GTG CGA AGT AGT GAT CAG G-3’).
Total RNA was extracted from whole mouse eyes in Eppendorf tubes with plastic homogenizers using the Tri Reagent® (Molecular Research Center, Inc. Cincinnati, OH). After DNAase treatment (DNA-free™ kit, Applied Biosystems, Austin, TX), cDNA was generated from oligo-dT primed total RNA (5 µg in 15 µl for each reaction; SuperScript® First-Strand Synthesis System for RT-PCR, Invitrogen, Carlsbad, CA). Taq-Man® real time qRT-PCR was performed on serially diluted samples of cDNA from Opn1swF81Y/F81Y and WT littermate eyes with a TaqMan 7500 Real-Time PCR System (Applied Biosystems, Carlsbad, CA) using the exon-spanning probes for mouse visual opsins and β-actin, with exon boundary and catalogue number as follows: rhodopsin (1–2) Mm00520345-m1; M-opsin (1–2) Mm01193546-m1; S-opsin (1–2) Mm01135619_g1; S-opsin (4–5) Mm00432058_m1; β-actin, Mm00607939-s1. Sample dilutions were matched across the two genotypes such that the cDNA used for the qRT-PCR came from equal quantities of total RNA, and we report the cDNA samples in terms of the quantity of ocular RNA from which they were derived. The 96-well plates used for the qRT-PCR reactions were loaded so that each dilution was replicated approximately inversely proportional to the equivalent RNA quantity in order to increase the reliability of estimates of cycle threshold (CT) for target genes in samples with greater dilution.
Cycle threshold data were analyzed with the method of (Pfaffl, 2001), which takes into consideration the efficiency of amplification for the primers. Efficiency (E) is defined as the negative antilog of the slope of a plot of CT vs. logB (sample dilution), i.e., E = B−1/slope. Thus, if sample dilution is expressed in base 2 units (B=2), a slope of −1 corresponds to perfect efficiency, E = 2, i.e., a 2-fold decrease in CT for each 2-fold increase in cDNA loaded. Suboptimum primer efficiency (in the range of cDNA used) gives rise to slopes greater than −1. With these definitions, the estimate of the ratio of expression of a target vs. a reference gene is given by
where ΔCTtarget is the difference in cycle threshold between WT and Opn1swF81Y/F81Y littermate samples for the target gene (S-opsin) for a given dilution, and ΔCTref is the cycle threshold difference for the reference gene (“ref”) between the littermate samples. The expression ratio of S-opsin in WT vs. F81Y mice was calculated with Eq 1 for each of the cDNA dilutions and averaged over the dilultions to obtain an overall estimate of the expression ratio.
A number of different primary antibodies were employed in this investigation for immunoblotting, immunoprecipitation (IP) and immunohistochemistry (IHC). These include three immunopurified polyclonal antibodies raised against mouse S-opsin peptides: against a CT-terminal peptide (“rAb-SopsCT”; amino-acids 317–333, CRKPMADESDVSGSQKT; Yenzym Antibodies, San Francisco; Daniele et al., 2011); against N-terminal peptides -- in rabbit ("rAb-SopsNT", MSGEDDFYLFQ; Zhu et al., 2003) and in goat (“gAb-SopsNT”; sc-14363, Santa-Cruz Biotechnology, Santa Cruz, CA). Additional antibodies used were rabbit anti-M-opsin (“rAb-Mops”; MAb65696-100, Abcam, Cambridge, MA), rabbit anti-GNAT2 (Sc-390; Santa Cruz Biotechnology), mouse anti-rhodopsin (“4D2”, Hicks and Molday, 1986), rabbit anti-β-actin (Ab 34737, Abcam), mouse anti-VCP (Ab 11433, AbCam), rabbit anti-EDEM1 (E8159, Sigma-Aldrich), rabbit anti-PDE6α’ antibody (gift from Tiansen Li), mouse anti-CRALBP (gift of J. Saari; Nawrot et al., 2004), rabbit anti-IRBP (gift of John Nickerson), mouse anti-KDEL ER retention sequence (Ab12223, lot GR14623-2; Abcam, San Francisco, CA). Alexa 555- and Alexa 568-conjugated goat anti-rabbit, and Alexa 488 goat anti-mouse were used to detect primary immunosera in IHC (Invitrogen Life Technologies, Carlsbad, CA). Goat anti-rabbit IgG coupled to IR Dye680 (92632221, LI-COR, Lincoln, NE), donkey anti-goat and anti-mouse IgGs coupled to IR Dye800 (LI-COR 92632214 and 92632212) were used as secondary antibodies for immunoblotting.
Dark adapted retinas were dissected and retinal lysates prepared as previously described (Daniele et al., 2011). Briefly, isolated retinas were processed in 150 µl extraction buffer (20 mM Bis-Tris propane buffer pH 7.5, 10 mM dodecyl-β-Maltoside and 5 mM NH2OH) supplemented with complete mini protease inhibitor cocktail tablets (Roche), homogenized by sonication for 10 s and clarified by centrifugation 5 min at 21000g. The rhodopsin concentration of retinal lysates was determined with bleaching difference spectroscopy as previously described (Lyubarsky et al., 2004). Quantification of rhodopsin provides a sensitive means equating retinal protein mass across genotypes: calibration showed that there is 4.2 pmol rhodopsin for each µg total retinal protein, and calculations show the total quantity of cone opsin in the adult WT mouse retina to be ~10 pmol, 1/70th that of rhodopsin (Daniele et al., 2011). Lysates were brought to a final concentration of 3 pmol rhodopsin/µl in Novex® Tris-Glycine SDS Sample Buffer (2X, Invitrogen Co. Carlsbad, California 92008, # LC2676), and incremental dilutions of the stock solution were loaded for quantitative immunoblotting in Novex® 4–12% Tris-Glycine pre-cast gels (Invitrogen). The quantity of the retinal lysate (expressed in rhodopsin mass) loaded per lane in gels typically ranged in 4 steps from 7.5 to 60 pmol. For immunoprecipitation (IP), lysates from animals of different genotypes were normalized for rhodopsin content and brought to a final volume of 200 µl. IP was performed according to manufacturer’s protocol using Dynabeads® Protein G co-IP kit (Invitrogen Co. Carlsbad, CA). Samples containing 150 µl of dynabeads, 3 µg of S-opsin C-term antibody and lysate were incubated for 40 min at room temperature. The supernatant was then removed and the beads washed three times in washing buffer. Elution of the bound proteins was achieved by adding 20 µl of elution buffer and 20 µl of 2X SDS sample buffer with β-mercaptoethanol. Eluted samples were subjected to SDS-PAGE and immunoblotting. Immunopositive bands were visualized with either enhanced chemiluminescence (ECL) (Thermo Scientific – Fisher) or a LI-COR® Odyssey Infra-red imaging system (LI-COR Lincoln, NE), and quantified with software provided by the manufacturer and cross-checked with customized Matlab scripts. (For display purposes, but not analysis, images of entire PVDF membranes were inverted and their contrast reduced.) For glycosylation assays, 30 µg of protein as measured with the Bradford reagent assay (Fermentas, Vancouver, BC) was treated for 1h at room temperature in a reaction solution containing SDS Sample Buffer (2X), G5 or G7 reaction buffers and 2500 U of Endo H or PNGase F according to the manufacturer’s protocol (New England Labs, Ipswich, MA).
Mice were sacrificed and eyes enucleated and immersion fixed in 4% paraformaldehyde in 1X phosphate buffered saline (PBS) for 30–40 min at room temperature. The anterior segment was cut away with fine scissors and the lens removed. After 3 exchanges of PBS, eyecups were embedded in 4% agarose gel pre-heated to 70C (low melting, GIBCO/Invitrogen Life Technologies, Carlsbad, CA 92008, Cat. No. 18300-012) and oriented as described so that the sectioning plane bisected the dorso-ventral midline (Wagner et al., 2000; Daniele et al., 2011). 150 µm sections were made using a vibrating blade microtome (Leica Microsystems, model VT 1000S, Wetzlar, 35578 Germany). Vibratome sections were incubated with primary immunosera at 4C overnight after ~ 1 hr incubation at room temperature with normal sera (Jackson Immunoresearch, goat, donkey). Secondary immunosera were applied for 1.5–2 hrs at room temperature after 3 washes in PBS. All immunosera were diluted in a PBS buffer containing BSA (0.5%) and Triton X-100 (0.5%). Slide-mounted sections with dual immunofluorescence labeling were visualized with a Zeiss LSM 510 confocal microscope (Zeiss Microscope Imaging, Inc., Thornwood, NY). Images were imported into Huygens Software (Scientific Volume Imaging, Hilversum, Netherlands) and deconvolved with a theoretical point spread function. Deconvolved images were exported as RGB TIFs and adjustments to brightness and contrast made with ImageJ software (Rasband WS, ImageJ, US NIH, http://rsb.info.nih.gov/ij/, 1997–2009).
Mouse eyes were enucleated and pre-fixed for retinal flatmounting in 4% paraformaldehyde in 1X PBS for 1h at room temperature. Cuts were made to mark the orientation (as indicated above) of the retina within the eyecup before the retina was separated from the sclera and fixed in 4% paraformaldehyde in 1X PBS overnight at 4C. Retinal flatmounts were labeled with an AlexaFluor 488 conjugate of peanut agglutinin (PNA, Invitrogen/Molecular Probes) and anti mouse S-opsin (rAb-SopsCT) diluted in PBS buffer with 0.5% BSA and 0.1% Triton X-100. After washing with PBS, retinas were incubated with a goat anti-rabbit Alexa 555 secondary antibody for 1h at room temperature. Finally, the isolated retinas were flattened, mounted on a microscope slide with a coverslip, and imaged with a Nikon Eclipse TE2000-U microscope equipped for epifluorescence using a 10X objective. Cone spatial densities were determined from images with a feature detection algorithm using ImageJ software (Rasband WS, ImageJ, U. S. National Institutes of Health, Bethesda, MD, USA, http://rsb.info.nih.gov/ij/, 1997–2009). The ImageJ output data were further refined with customized Matlab™ image analysis software (Daniele et al., 2011).
For quantitative analysis of S-opsin immunofluorescence, vibratome sections of WT and Opn1swF81Y/F81Y littermates were made and singly labeled in parallel with S-opsin primary antibodies and with an Alexa Fluor 568-conjugated secondary. Z-stacks were collected with a photon-counting two-photon laser scanning microscope operating in its linear range (Calvert et al. 2007). Image acquisition, laser intensity and scan parameters were controlled by a custom LabVIEW based interface (Coleman Technologies Inc.). The titanium-sapphire laser (Chameleon Ultra II, Coherent Inc, Santa Clara, CA) was tuned to 800 nm for excitation of the Alexa Fluor 568 fluorochrome. Custom Matlab (Mathworks, Natick, MA 01760) scripts (Calvert et al., 2007; Nikonov et al., 2008) were used for segmenting complete cone outer segments or oriented “slabs” from the 3D Z-stacks, and for extraction of voxel intensity values (in photon counts) from the 3D image stacks.
Mice were euthanized by cardiac perfusion following asphyxiation with CO2. Fixatives consisted of 2.5% Glutaraldehyde, 2% paraformaldehyde in 0.1 M Na Cacodylate, pH 7.4 for conventional EM and 0.5% Glutaraldehyde, 4% paraformaldehyde in Sorenson buffer for immuno-EM processing. Eyes were dissected into eyecups, divided into wedges centered at the ventral midline and subjected to overnight fixation. Further processing for conventional EM was performed as described previously (Daniele et al., 2010). Immuno-EM of mouse retinas was performed as described by (Erickson, 1987). Images of the sections were acquired with a transmission electron microscope (CM 120, Phillips Biotwin Lens, F.E.I. Company, Hillsboro, OR, U.S.A.) coupled to a digital camera (Gatan MegaScan, model 794/20, digital camera (2K X 2K), Pleasanton, CA) at the Diagnostic and Research Electron Microscopy Laboratory at UC Davis.
Light responses of mouse cone photoreceptors were recorded with published methods (Nikonov et al., 2006). In brief, the cell body region of one to several cells in the most distal portion of the outer nuclear layer of a retinal slice was drawn into a suction pipette, and the membrane current measured with a current-to-voltage converter (a form of loose-patch recording). Rod currents were suppressed with a 500 nm rod-saturating background, and the responses of cones to brief, calibrated flashes of various wavelengths from 340 nm to 630 nm measured. Tissue dissection was performed under IR-illumination. A cautery (Bovie®, Aaron Medical, St. Petersburg, FL) was used to mark the eyecup so that retina orientation could be maintained. To obtain recordings from ventral cones, wedges of retina (~1/8 area of retina) centered at the ventral midline ~ 1.5 mm from the optic nerve to the edge of the retina were cut out of the eye cup and chopped into small pieces (~200 µm on a side) with a fine blade.
To measure its extinction coefficient, F81Y S-opsin was expressed heterologously and purified. Duplicate plates were transfected with cDNA for WT and F81Y-S-opsin (both with C-terminus modified for immunopurification with 1D4 monoclonal antibody) using 15 µg DNA per 15 cm plate using procedures described previously (Babu et al., 2001). Cells were cultured for 55 hrs at 37C in media containing 10% newborn calf serum. Culture was done in incubators kept mostly in the dark, but with occasional brief light exposure when the incubator door was opened. Cells were collected and washed in PBS + glucose at 4C extensively. Under dim red light (Kodak #2 filter), 5 µM 11-cis retinal was then added to the cells, followed by overnight incubation at 4C; this procedure effectively stops translation while allowing the 11-cis retinal to combine with competent opsin to form bleachable visual pigment. Cells were collected, washed with PBS + glucose at 4C, and then solubilized with 1% dodecyl maltoside. The cone opsin was purified using a 1D4 Sepharose column, with elution buffer containing 0.1% dodecyl maltoside at 4C. The extinction coefficient of mouse F81Y S-opsin was measured with bleaching by acid denaturation (Babu et al., 2001), and found to be within a few percent of that of WT mouse S-opsin, εmax = 41,760 (Vought et al., 1999) (data not shown).
Littermate pairs of WT and Opn1swF81Y/F81Y mice were injected every other day for a 10 day period with 20 µg 11-cis retinal in 100 µl vehicle, drawn from a ~40 mM stock solution in 100% ultrapure EtOH purged with argon. (11-cis retinal was provided by Dr. Rosalie Crouch under the auspices of the National Eye Institute.) The injection vehicle consisted of 10% BSA in 0.9% sterile NaCl, and the EtOH with 11-cis retinal constituted <2% by volume. For each littermate pair of mice injected with 11-cis retinal, a yoked control littermate pair of WT and Opn1swF81Y/F81Y mice was sham-injected with vehicle containing EtOH alone. During the course of injections mice were housed in a vivarium with a 12 hr light:dark cycle with light intensities of 50–100 lux. Mice were dark adapted overnight prior to sacrifice for quantification of their cone opsins with Western blotting (as described above).
The gene-targeting strategy used to create the F81Y S-opsin knockin (Opn1swF81Y/F81Y) mouse inserted a Neomycin resistance cassette between the third and fourth exons (Fig. 1A). Successful targeting was confirmed with Southern blotting of ES cells and sequencing of genomic DNA (Fig. 1B). Surprisingly, initial Western blotting suggested a more than 2-fold reduction in the level of expression of the Opn1swF81Y allele relative to the wild-type allele (Fig. 1C). In contrast, there was no reduction in the expression of M-opsin, and cone-specific G-protein (Gtαc) and phosphodiesterase (PDE6c), were only modestly reduced by 28% and 26% (n = 2 littermate pairs), respectively.
To determine if the decrease in expression of the Opn1swF81Y allele arose from altered transcription efficiency, we performed real-time quantitative reverse transcriptase PCR (qRT-PCR) analysis of mRNA from retinas of WT and Opn1swF81Y/F81Y littermate pairs, using rhodopsin and β-actin as reference genes (Fig. 1D). Six replications of the experiment with mRNA from the separately analyzed eyes of two littermate pairs yielded an average upward shift in the logarithmic threshold cycle (CT) of +0.24 ± 0.09 (mean ± s.e.m., p = 0.02) log2 units for Opn1swF81Y vs. WT S-opsin mRNA, corresponding to a 15 ± 5% decrease in Opn1swF81Y mRNA. Application of the method of Pfaffl et al. (2001), which takes into consideration the efficiencies of different PCR primer pairs, yielded the identical estimate of Opn1swF81Y mRNA reduction, because the efficiencies of amplification of the target and reference genes were within 4% of ideal (Methods). In the same samples, the M-opsin mRNA in Opn1swF81Y/F81Y mice was negligibly different from WT (6% ± 6%), and there was no detectable change in rhodopsin mRNA.
To determine precisely the reduction in expression of the protein product of the Opn1swF81Y gene we performed quantitative comparisons of S-opsin immunoblots of Opn1swF81Y/F81Y and WT littermate pairs. In each experiment 4 levels of retinal lysate from an Opn1swF81Y/F81Y and WT pair were blotted with anti-S-opsin antibodies, and the expression ratio estimated as the ratio of the slopes of the blot strength vs. protein load data for the two genotypes (METHODS; for an example, see Fig. 6). Thirty-four such experiments from 12 littermate pairs revealed the average quantity of S-opsin in Opn1swF81Y/F81Y retinas to be reduced 2.6 ± 0.3 fold that of WT controls (mean ± s.e.m.; p < 0.001, for t-test comparison against an expression ratio of unity). In summary, there was a far greater reduction in the Opn1swF81Y protein than in its mRNA, suggesting post-transcriptional downregulation of the mutant gene product.
Mouse rods that underexpress rhodopsin have altered outer segment structure and signaling (Calvert et al., 2001; Liang et al., 2004), raising the question of whether the large reduction in the expression of F81Y S-opsin alters the anatomical structure and physiological signaling of Opn1swF81Y/F81Y cones. To assess the functionality of F81Y cone outer segments, we recorded flash responses from single cones using the suction electrode method (Nikonov et al., 2006). Light responses of cones of the ventral retina of Opn1swF81Y/F81Y mice were very similar to those of WT mice (Fig. 2A, C, Table 1), indicating that the outer segments of Opn1swF81Y/F81Y cones are fully functional. As expected (Fasick et al., 2002), the spectral sensitivity of the Opn1swF81Y/F81Y cones was red-shifted 60 nm, having a λmax of 420 nm as compared to the WT value of 360 nm. However, the absolute sensitivity of Opn1swF81Y/F81Y cones at λmax was ~ 3-fold lower than that of WT cones (p < 10−5; Fig. 2E, Table 1), consistent with the 2.6-fold reduction in F81Y S-opsin protein expression.
The reduction in S-opsin expression and parallel loss of sensitivity of cones of the Opn1swF81Y/F81Y ventral retina reveals that F81Y S-opsin does not traffic to the outer segment in normal quantities, and raises the question of whether the number, size or other features of the cone outer segments are normal in mice expressing the mutant S-opsin. To address this question we analyzed retinal flat mounts and sections immunostained with PNA and S-opsin antibodies. The density of S-opsin expressing cones in the Opn1swF81Y/F81Y retina was normal (Fig. 3). In sections of eyes of littermates selected for adherence of the pigment epithelium, we measured the lengths of Opn1swF81Y/F81Y and WT ventral retinal cone outer segments, and found them to be indistinguishable (data not shown): 12.1 ± 0.02 µm (mean ± s.e.m., 114 cones, 2 mice) vs. 12.4 ± 0.02 (103 cones, 3 mice), respectively. We conclude that despite the reduction in S-opsin expression, the morphology of ventral cones of Opn1swF81Y/F81Y mice appears normal in light microscopy.
The global reduction in S-opsin expression in Opn1swF81Y/F81Y mice measured by Western blotting and reflected in the reduced light sensitivity of cones should also be manifest as a reduction of S-opsin immunofluorescence. To test this prediction, we quantified S-opsin immunolabeled sections of ventral retina of littermate pairs with a photon-counting, two-photon imaging system (Fig. 4). For one littermate pair, the average absolute S-opsin immunofluorescence of Opn1swF81Y/F81Y cone outer segments (COS) was reduced on average 1.6-fold (Fig. 4E; p < 0.00002 for test of no difference); for a second littermate pair the average reduction was 2.2 fold (p < 0.00003).
The apparently normal outer segment morphology of the Opn1swF81Y/F81Y cones suggests that opsin and membrane delivery to the outer segment may be to some extent uncoupled in cones. This contrasts with the situation in rods, whose outer segment length, diameter and disc structure depend on rhodopsin expression level (Liang et al., 2004). This possible difference between rods and cones led us to investigate the subcellular localization of opsin in different compartments of the cones, and in particular in the cell bodies and inner segments, where translation, ER and Golgi sorting, and membrane delivery to the basal discs of the outer segment take place. S-opsin immunofluorescence was readily observed in the cone inner segment, cell body, myoid region, along the axon and in the synaptic pedicles of both WT and Opn1swF81Y/F81Y ventral cones (Fig. 4A, C). We quantified the axial distribution of S-opsin immunofluorescence in oriented digitally excised “slabs” of retina: to a first approximation, there was a proportional reduction of S-opsin in each subcellular region of Opn1swF81Y/F81Y cones (Fig. 4F), and in both genotypes in excess of 25% of the S-opsin immunogenicity was not in the outer segment.
The decrease in F81Y S-opsin expression, combined with the absence of a comparable decrease in its mRNA (Fig. 1D), suggests that Opn1swF81Y mRNA is less efficiently translated than the mRNA of WT Opn1sw, or that newly synthesized F81Y S-opsin protein is rapidly degraded after translation. One well characterized pathway that governs protein removal early in the secretory pathway is ER-associated degradation (ERAD, Vembar and Brodsky, 2008). ER degradation enhancing alpha-mannosidase-like 1 (EDEM1) is a lectin chaperone that accelerates ERAD by selectively targeting misfolded proteins to the site of retrotranslocation (Hosokawa et al., 2001), and has been shown to play a role in targeting immaturely glycosylated P23H rod opsin for ERAD (Kosmaoglou et al., 2009). Another important component of the ERAD retrotranslocation machinery is the transitional AAA-ATPase VCP/P97, which provides the energy required to extract poly-ubiquitinated substrates through the ER membrane (Ye et al., 2001), and has been shown to play a role in the degradation of heterologously expressed P23H rod opsin (Griciuc et al., 2010).
To test the hypothesis that the ERAD pathway is involved in the reduction of F81Y S-opsin expression, we performed immunoprecipitation assays of WT and Opn1swF81Y/F81Y retinas with an S-opsin antibody to determine whether EDEM1 and VCP are differentially complexed with S-opsin (Fig. 5). VCP and EDEM1 were readily detected in the immunoprecipitates from both WT and Opn1swF81Y/F81Y retinas, but were absent from those of S-opsin knockout mice, establishing the S-opsin specificity of the antibody. Comparing the blot densities of VCP and EDEM1 relative to that of S-opsin in the immunoprecipitates, we found VCP to be enhanced 3.7 ± 0.7 fold (n = 5; p < 0.01), while EDEM1 was enhanced 2.6 ± 0.7 fold (n = 4; p = 0.05). These results implicate ERAD in the reduced expression of F81Y S-opsin, and as a consequence suggest that newly translated F81Y S-opsin has a stronger tendency to misfold than WT S-opsin.
Because the F81Y point mutation clearly alters the binding pocket for 11-cis retinal ligand (thus affecting spectral sensitivity), we considered the hypothesis that interaction of the nascently translated S-opsin with 11-cis retinal binding might affect the folding of F81Y S-opsin. We tested this hypothesis by systemically increasing 11-cis retinal with intraperitoneal injections in Opn1swF81Y/F81Y and WT mice. Littermate pairs were injected with 20 µg (70 nmol) of 11-cis retinal on alternate days for 10 days, a time sufficient for complete renewal of their outer segments (Young, 1967, 1971; Hollyfield, 1979; Jonnal et al., 2010); sham-injected, age-matched littermate pairs exposed to the identical light rearing conditions served as controls. Quantitative analysis of Western blots for S-opsin (Fig. 6) from 10 sets of littermate pairs revealed the ratio of expression of S-opsin in Opn1swF81Y/F81Y vs. WT for 11-cis retinal-injected mice to be 0.96 ± 0.06 (mean ± s.e.m.), while for the sham-injected mice the corresponding ratio was 0.52, corresponding to a 1.9 ± 0.05 –fold reduction in F81Y S-opsin expression, as expected from previous results. The difference between the S-opsin expression ratios of sham- and 11-cis retinal injected littermate pairs was highly significant (p < 0.001, t-test for unit ratio). Thus, exogenously elevated 11-cis retinal caused the expression of F81Y S-opsin to increase to the WT level, effectively rescuing the Opn1swF81Y/F81Y phenotype of reduced S-opsin expression.
During the course of the 11-cis retinal injection experiments just reported, we also performed immunoblot comparisons of the quantity of S-opsin extracted from WT mice injected with 11-cis retinal vs. WT mice that were sham-injected. These comparisons suggested that the chromophore injections might be increasing the level of WT S-opsin, and so we pursued this effect in a series of experiments with age-matched WT mice. In 16 immunoblot comparisons (as in Fig. 6) of retinal lysates of 9 pairs of WT mice, the ratio of S-opsin in 11-cis retinal-injected vs. sham-injected was 1.22 ± 0.09 (mean ± s.e.m.; p = 0.023 for 1-tailed t-test). Thus, exogenous 11-cis retinal increased WT S-opsin expression, and so we conclude that during biosynthesis WT S-opsin, as well as the mutant F81Y-S-opsin, is sensitive to the ambient level of chromophore.
We also compared the quantities of rhodopsin and M-opsin of mice injected with 11-cis retinal vs. mice that were sham-injected. The average quantity of rhodopsin extracted was 430 ± 25 pmol/eye and 429 ± 18 pmol/eye (mean ± sem, n = 24 pairs) for sham-injected and 11-cis retinal injected mice, respectively. The average slope ratio of M-opsin signals in Western blots (as in Fig. 6) for sham- vs. 11-cis retinal injected mice was 1.00 ± 0.06 (mean ± sem, n = 5 pairs). In both cases, the statistical power was adequate to have reliably detected a 15% change at p = 0.05. Thus, increased opsin production with exogenous 11-cis retinal injections in WT eyes appears uniquely associated with S-opsin.
To understand more deeply the organization of the retinal mechanisms that govern cone opsin biosynthesis and their relationship to native sources of 11-cis chromophore, we examined the cell body and inner segment regions of cones with high resolution confocal immunohistochemistry, and with electron microscopy (Fig. 7), taking advantage of the ability of S-opsin antibodies to localize S-opsin in these subcellular regions (Fig. 4). The cell body region contains substantial S-opsin and a high density of rough ER (Fig. 7D, J–M), and every cone cell body had a “cap” region, where rough ER and non-outer segment S-opsin were localized (Fig. 7A – C). Remarkably, Mueller cell processes tightly apposed the ER-containing region of the cones, as revealed by immunohistochemical staining for the cellular 11-cis retinoid binding protein, CRALBP (Fig. 7B, E, H) (Bunt-Milam and Saari, 1983; Nawrot et al., 2004), and well known morphological features of Mueller cells, including villous processes that project into the inner segment layer (Fig. 7J, L). Mueller cells in some species have been established to be capable of synthesizing 11-cis retinol (Das et al., 1992), and CRALBP serves as a high-affinity cellular carrier for both 11-cis retinol and 11-cis retinal (Futterman et al., 1977; Saari et al., 1982; Saari et al., 2001). In addition, interphotoreceptor retinoid binding protein (IRBP) is clearly present near the cell body and inner segment region of cones (Fig. 7F). Because IRPB is secreted by photoreceptors into the interstitial space, some IRBP is no doubt also present in the cone cytoplasm (Fig. 7F). Overall, then, these results show that the ER-containing regions of the cone, where its opsin is translated and proofread, are tightly apposed to a source of 11-cis retinol, an immediate precursor of 11-cis retinal, and that 11-cis retinoid-specific carriers are present in the Mueller cell cytoplasm and in the nearby interstitial spaces.
The results reported here show that in the absence of a normal interaction of nascent S-opsin with 11-cis retinal, ER quality control mechanisms reduce the quantity of the protein synthesized. Specific support for the hypothesis that F81Y S-opsin expression is reduced by ERAD is provided by the increased immunopreciptation by F81Y S-opsin of ER-degradation enhancing α-mannosidase-like 1 (EDEM1) and of VCP/p97 (Fig. 5, Fig. 8B). Overall production of F81Y S-opsin is augmented by 11-cis retinal (Fig. 6), whose precursor is synthesized by the adjacent Mueller cell (Figs. 7, ,8A)8A) (Wang and Kefalov, 2009, 2010).
Our results imply that interaction of 11-cis retinal with nascent opsin can contribute to ER quality control by altering the branching ratio between Golgi exit and ERAD (Fig. 8B). The first moment in the life of a nascent S-opsin when 11-cis retinal could interact with its normal binding site is when Lys291 on the 7th transmembrane helix is exposed to the ER membrane during exit from the translocon. More likely, 11-cis retinal binds to the opsin during the lectin cycle: such binding would occur precisely during the critical phase of proofreading following the completion of translation, when the folding occurs that creates the chromophore binding pocket. In the absence of exogenous 11-cis retinal F81Y S-opsin appears more likely to fold improperly and undergo excess mannose trimming (red arrow), removing the opsin from the lectin cycle, binding EDEM1 (Kosmaoglou et al., 2009), leading to its progression to retrotranslocation assisted by VCP/P97, ubiquitination and destruction by the proteasome. Since exogenous 11-cis retinal increases the expression level of F81Y S-opsin to that of correspondingly injected WT mice (Fig. 6), the chromophore effectively biases the branching ratio of the key reactions in ER quality control (red/green arrows in Fig. 8B). In these terms, a useful summary of our experiments can be made in terms of the effects of S-opsin genotype (WT vs. F81Y S-opsin) and 11-cis retinal level on the branching ratio (Table 2).
Normal production of 11-cis retinal is required for the health and survival of rod and cone photoreceptors, as well as their signaling of light capture. Cones of mice with genetic deletions of Rpe65 or Lrat – enzymes expressed in the retinal pigment epithelium cells essential for normal production of 11-cis retinal from retinoid precursors (Jin et al., 2005; Moiseyev et al., 2005; Redmond et al., 2005) – undergo rapid degeneration, with rod degeneration progressing more slowly (Rohrer et al., 2005; Znoiko et al., 2005; Fan et al., 2008; Zhang et al., 2008). Cone outer segment structure and function in Rpe65−/− and Lrat−/− mice can to some extent be restored in young mice by exogenous delivery of 11-cis retinal (Rohrer et al., 2005; Zhang et al., 2008). The cone degeneration that occurs in the absence of normal 11-cis retinal production was initially proposed to be the result of mistrafficking of cone opsins (Rohrer et al., 2005; Zhang et al., 2008). The results presented here are consistent with more recent studies that suggest that 11-cis retinal plays a role in cone opsin synthesis and maturation, likely acting in the ER (Sato et al., 2010; Zhang et al., 2011). Zhang et al. (2001) have hypothesized that S-opsin is distinctly susceptible to forming aggregates during biosynthesis in the absence of chromophore, leading to activation of ER stress pathways, and ultimately resulting in cone photoreceptor degeneration. Remarkably, Opn1swF81Y/F81Y cones show no obvious signs of ER stress, or of S-opsin aggregation, and do not degenerate, and apparently have the capacity to eliminate by ERAD a substantial fraction (at least 50% WT levels) of improperly folded opsin. This suggests that, in the complete absence of 11-cis retinal synthesis in Rpe65−/− and Lrat−/− mice, the fraction of opsin that misfolds may be so large that it strains or exceeds the cell’s ERAD capacity (red branch, Fig 8B).
The neural retina contains a biochemical pathway that can supply 11-cis retinal for the regeneration of cone pigment that has been bleached by light (Wang et al., 2009; Wang and Kefalov, 2009, 2010). Central to this pathway are the Mueller cells, which generate 11-cis retinol from retinoid precursors (Das et al., 1992), and strongly express the 11-cis retinoid-specific binding protein CRALBP (Bunt-Milam and Saari, 1983; Nawrot et al., 2004), which aids in the delivery of 11-cis retinoid to the cones. Genetic deletion of CRALBP slows both rhodopsin and cone pigment regeneration (Saari et al., 2001). The interstitial retinoid binding protein, IRBP, which is secreted by rods and cones, has also been shown to be necessary for normal regeneration of cone pigments (Jin et al., 2009; Parker et al., 2011). Our results suggest a new and important role for the Mueller cell supply to the photoreceptor inner segments and cell bodies, namely, the delivery of chromophore to nascent opsin during the proofreading cycle (Fig. 8).
GPCRs naturally activated by lipophilic ligands provide an even broader example of ligand-sensitivity of GPCR biosynthesis (Nakamura et al., 2010). Such ligands include prostaglandins, leukotrienes, platelet activating factor, lysophosphatidic acid and endocannabinoids. The GPCR literature provides many examples in which co-culture with cell permeant natural ligands or synthetic homologue ligands increases the heterologous expression level of mutant GPCRs. These include the vasopressin 2 receptor (Morello et al., 2000), opioid receptors (Chen et al., 2006; Leskela et al., 2007), melanocortin receptor (Rene et al., 2010) and mutant rhodopsin (Noorwez et al., 2004; Krebs et al., 2010) The authors of several of these studies have hypothesized that the ligands bind in the ER to unstable folding intermediates, stabilizing them and permitting ER exit. A full understanding of the factors governing the critical branching ratio between ER exit and ERAD (Fig. 8) and ERAD capacity in the native tissues and cell types in which GPCRs are expressed is central to therapeutic intervention strategies for hereditary diseases arising from mutant GPCRs (Wiseman et al., 2007).
We are grateful to Marie Burns for helpful comments, to Eric Pierce for help in creating the knockin mouse, to Cheryl Craft, John Nickerson and Jack Saari for providing antibodies, to Paul Fitzgerald and Brad Shibata for assistance with electron microscopy, and to Bob Birge for assistance with spectral analysis. Supported by NIH EY02660.