Wild-type opsin does not aggregate with a misfolded opsin mutant

Abstract

Rhodopsin is the light receptor in photoreceptor cells that plays a central role in phototransduction and photoreceptor cell health. Mutations in rhodopsin are the leading cause of autosomal dominant retinitis pigmentosa (adRP), a retinal degenerative disease. A majority of mutations in rhodopsin cause misfolding and aggregation of the apoprotein opsin. The pathogenesis of adRP caused by misfolded opsin is unclear. It has been proposed that physical interactions between wild-type opsin and misfolded opsin mutants may underlie the autosomal dominant phenotype. To test whether or not wild-type opsin can form a complex with misfolded opsin mutants, we examined the interactions between wild-type opsin and opsin with a G188R mutation, a clinically identified mutation causing adRP. Förster resonance energy transfer (FRET) was utilized to monitor the interactions between fluorescently tagged opsins expressed in live cells. The FRET assay employed was able to discriminate between properly folded opsin oligomers and misfolded opsin aggregates. Wild-type opsin predominantly formed oligomers and only a minor population formed aggregates. Conversely, the G188R opsin mutant predominantly formed aggregates. When wild-type opsin and G188R opsin were coexpressed in cells, properly folded wild-type opsin did not aggregate with G188R opsin and was trafficked normally to the plasma membrane. Thus, the autosomal dominant phenotype in adRP caused by misfolded opsin mutants is not predicted to arise from physical interactions between wild-type opsin and misfolded opsin mutants.

1. Introduction

Rhodopsin is the G-protein coupled receptor (GPCR) that initiates phototransduction in photoreceptor cells of the retina. Rhodopsin consists of the apoprotein opsin covalently bound to the chromophore 11-cis retinal. Dysfunctions in rhodopsin can impair phototransduction or lead to photoreceptor cell death and retinal degeneration. Over 100 mutations in rhodopsin have been discovered in patients with inherited retinal diseases [1]. A majority of these mutations are incapable of binding 11-cis retinal, cause misfolding and aggregation of the apoprotein opsin, and lead to an autosomal dominant form of retinitis pigmentosa (adRP) [2–5], the most common inherited retinal degeneration [6–9]. The complete pathogenesis of adRP caused by misfolded opsin mutants is unclear.

Multiple factors likely contribute to photoreceptor cell death in adRP caused by misfolded opsin mutants. In healthy rod photoreceptor cells, rhodopsin is synthesized in the inner segment and transported to the outer segment. Rhodopsin is predominantly found in the rod outer segment disk membranes (Fig. 1A). In adRP, photoreceptor cell death occurs, at least in part, due to toxic misfolded opsin aggregates in the endoplasmic reticulum of the inner segment [10–12]. This toxicity derives from inhibition of the proteasome and activation of the unfolded protein response [4,5,12–17]. For partially misfolded mutants, photoreceptor cell death may also result from mutants that are trafficked to the rod outer segment and disrupt normal disk membrane structure [18–20].

Fig. 1.

Opsin localization in a rod photoreceptor cell. The localization of wild-type (red) and misfolded mutant (blue) opsins in the outer segment disk membranes or endoplasmic reticulum in the inner segment is illustrated. Localization of opsin under normal (A) and diseased (B and C) states is shown. In diseased states, scenarios where wild-type opsin cannot physically interact with mutant opsin (B) or where wild-type opsin can physically interact with mutant opsin (C) are shown. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

In rod outer segment disk membranes, rhodopsin and opsin have been shown to form oligomers arranged as nanodomains [21–24]. Quaternary structure formation has been suggested to underlie the dominant retinal degeneration phenotype in adRP, where mutant receptors physically interact with wild-type receptors [5,25]. Whether these physical interactions occur and the type of complexes formed must be determined to more fully understand the pathogenesis of adRP and rationally develop therapeutics. The possible retinal degeneration mechanisms and viable therapeutics can be different depending on whether or not mutant and wild-type opsins physically interact [26]. A loss of function mechanism is unlikely since haploinsufficiency of rhodopsin is not predicted to cause retinal degeneration, at least when the level of rhodopsin is reduced by up to half [27–29]. In fact, photoreceptor cells can adapt to reduce rhodopsin expression to maintain constant rhodopsin density in rod outer segment disk membranes, presumably to maintain a constant photon catch capability [30].

If mutant and wild-type opsins do not interact, then the biosynthesis, trafficking and function of the wild-type receptor are predicted to be unaffected. The mechanism of adRP would then be exclusively due to a gain of function mechanism, either because mutant opsin aggregates cause toxicity in the endoplasmic reticulum or rod outer segment disk membranes are disrupted by properly trafficked mutant opsins (Fig. 1B). If mutant and wild-type opsins do interact (Fig. 1C), then the adRP mechanism could include both gain of function and dominant negative mechanisms. Dominant negative mechanisms could include the promotion of misfolding and aggregation of wild-type opsin in the endoplasmic reticulum, thereby increasing the level of toxicity and suppressing rhodopsin expression below 50%, which may be detrimental. Interactions between mutant and wild-type opsins could also disrupt normal quaternary structure and thereby alter normal packing and function of rhodopsin in rod outer segment disk membranes. To better understand the mechanism of adRP, we must discriminate between these scenarios by determining whether or not physical interactions occur between mutant and wild-type opsins and the types of complexes formed.

Studying the interactions between opsin molecules within a cellular context is a challenge. Förster resonance energy transfer (FRET) methods provide a tool to investigate protein–protein interactions within a cellular context [31]. FRET has been used to study the oligomerization of wild-type opsin and aggregation of P23H and G188R misfolding opsin mutants in COS-1 or HEK293 cells [4,32,33]. FRET has also been used to investigate the physical interactions between the P23H misfolding opsin mutant and wild-type opsins in HEK293 cells [25]. Though FRET was detected between P23H and wild-type opsins, indicating a physical interaction between the two receptors, the data was somewhat ambiguous. No negative control was conducted to ensure that the FRET signal derives from physical interactions. Since a small subset of P23H mutant opsins can fold properly [34], it was unclear whether the FRET signal originated from oligomers of properly folded opsin or from aggregates of misfolded opsin.

To overcome the ambiguities present in previous studies, more detailed FRET studies with appropriate controls were conducted on a G188R misfolding opsin mutant in the current study. The G188R opsin mutant, similarly to the more common P23H opsin mutant, misfolds, aggregates, and causes adRP [33–36]. However, the G188R mutation results in a more severe misfolding phenotype compared to the P23H mutation since all of the receptor molecules are misfolded and incapable of binding the chromophore 11-cis retinal [2,34,35]. Studying G188R opsin reduces ambiguity in the interpretation of results since a homogeneous population of misfolded receptor can be examined.

2. Materials and methods

2.1. DNA constructs

DNA vectors were generated for the expression of receptor tagged with either the yellow fluorescent protein (YFP) variant SYFP2 or mTurquoise (mTq). The vectors pmRho-SYFP2-1D4, pmRhoG188R-SYFP2-1D4, and pmTq-C1 were generated as described previously [33, 37]. The following forward and reverse primers were used to amplify the sequence for mTq by PCR using pmTq-C1 as the template: 5′ ACGA TGGGATCCACCGGTCGCCACCATGGTGAGCAAGGGCGAGGA and 5′ CATCGTGCGGCCGCTAAGGCTGGAGCCACCTGGCTGGTCTCCGTCTTGTACAGC TCGTCCATGC. The amplified product contained a BamHI restriction endonuclease site at the 5′ end and added the sequence for a 1D4 epitope (TETSQVAPA) [38] and NotI restriction endonuclease site at the 3′ end after the sequence for mTq. This generated the product mTq-1D4. The sequences for the fluorescent proteins in pmRho-SYFP2-1D4 and pmRhoG188R-SYFP2-1D4 were replaced with the PCR product mTq-1D4 at the BamHI and NotI restriction endonuclease sites to generate the vectors pmRho-mTQ-1D4 and pmRhoG188R-mTq-1D4, respectively.

Untagged opsin constructs were generated by first PCR amplifying the opsin sequences from pmRho-SYFP2-1D4 and pmRhoG188R-SYFP2-1D4 using the following forward and reverse primers: 5′ ACGA TGAAGCTTCGAATTCGCCACCATG and 5′ CATCGTGCGGCCGCTTAGGCTG GAGCCACCTGGCTGGT. These primers added an EcoRI restriction endonuclease site at the 5′ end and a stop codon and NotI restriction endonuclease site at the 3′ end of the opsin sequences. The PCR products replaced the sequence mRho-SYPF2-1D4 in pmRho-SYFP2-1D4 at the EcoRI and NotI restriction endonuclease sites to generate the vectors pmRho and pmRhoG188R.

The vector coding for the m2 muscarinic receptor tagged with mTq containing the 1D4 epitope was constructed as follows. The template containing the sequence for the m2 muscarinic receptor was from the FLAG-m2-pBlueBac4.5 vector generated previously [39]. The sequence for the m2 muscarinic receptor was amplified by PCR from this template using the following forward and reverse primers: 5′ ACGATGAAGCTTAT GAATAACTCAACAAACTCCTCTAA and 5′ CATCGTTCTAGACCTTGTAGCG CCTATGTTC. These primers added HindIII and XbaI restriction endonuclease sites at the 5′ and 3′ end of the m2 muscarinic receptor sequence, respectively. The PCR product was inserted into the vector pFLAG-CMV-3 (Sigma Aldrich, St. Louis, MO) at the HindIII and XbaI restriction endonuclease sites to generate the vector pFLAG-m2-CMV-3, which codes for the m2 muscarinic receptor tagged at the amino terminal end with a cleavable signal sequence and FLAG epitope. This entire sequence was amplified by PCR with the following forward and reverse primers: 5′ ACGATGGAATTCGCCACCATGTCTGCACTTCTGATCCTAG and 5′ CATCGTACCGGTGGCCTTGTAGCGCCTATGTTCT. This generated a PCR product with EcoRI and AgeI restriction endonuclease sites at the 5′ and 3′ ends, respectively. This PCR product was inserted into the vector pSYFP2-1D4-N1 [33] at the EcoRI and AgeI restriction endonuclease sites to generate the vector pFLAG-m2-SYFP2-1D4. The sequence for SYFP2-1D4 was replaced by the sequence for mTq-1D4 as described earlier to generate the vector pFLAG-m2-mTq-1D4, which was used in transfections.

2.2. Transient transfection of HEK293 cells

HEK293T/17 cells (American Type Culture Collection, Manassas, VA) were cultured and transfected in 12 well plates as described previously [33]. Cells used for confocal microscopy were grown on poly-L-lysine treated #1.5 coverslip glass (Thermo Fisher Scientific, Waltham, MA). Cells were transfected with DNA vectors described earlier using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). Coexpression of YFP-and mTq-tagged receptors was achieved by cotransfecting DNA vectors coding for each. Different acceptor to donor ratios was achieved by changing the relative amounts of each DNA vector transfected. For FRET and microscopy experiments involving wild-type and mutant opsins, the total amount of DNA vector transfected was kept constant at 400 ng. For FRET experiments involving the m2 muscarinic receptor, 2500 ng of the vector pFLAG-m2-mTq-1D4 was transfected to achieve comparable expression levels as that of tagged opsins. 50 to 800 ng of the vector pmRho-SYFP2-1D4 was cotransfected with this vector to achieve different acceptor to donor ratios. Cells used for Western blot analysis were transfected with 200 ng of a single vector. Cells were washed once with 1 mL PBS (4.3 mM Na₂HPO₄·7H₂O, 1.4 mM KH₂PO₄, 137 mM NaCl, 2.7 mM KCl, pH 7.3) and examined 24 h post-transfection.

Cells treated with the proteasome inhibitor MG-132 were processed as follows. Cells were cotransfected with 200 ng each of the vectors mRho-mTq-1D4 and mRho-SYFP2-1D4. At 12 h post-transfection, cells were treated with either 1 μL of dimethyl sulfoxide (DMSO) (Sigma Aldrich, St. Louis, MO) or 1 μL of 5 mg/mL MG-132 (Millipore, Temecula, CA) dissolved in DMSO. Cells were washed and harvested 12 h post-treatment (24 h post-transfection) and examined by FRET.

2.3. FRET assay

The FRET assay and detergent treatment were conducted as described previously [33]. FRET assays were conducted on a FluoroMax-4 spectrofluorometer (Horiba Jobin Yvon, Edison, NJ). YFP fluorescence was monitored by exciting samples at 485 nm (5 nm slit width) and collecting the emission spectra with a 10 nm slit width. The FRET efficiency (E) was computed by measuring the dequenching of the donor fluorescence signal (i.e., mTq) [31], which occurred upon treatment with detergent. The mTq fluorescence emission peak at 476 nm was measured in untreated cells, cells treated with DM, and cells treated with SDS. Total E was computed as follows: SDS: E = 1 − (mTq_untreated / mTq_SDS-treated). The DM-sensitive E was computed as follows: E = (mTq_DM-treated − mTq_untreated) / mTq_SDS-treated. The DM-insensitive E was computed as follows: E = 1 − (mTq_DM-treated / mTq_SDS-treated). The acceptor to donor ratio was determined from cells coexpressing YFP- and mTq-tagged receptors. The ratio was calculated by dividing YFP fluorescence emission at 527 nm in untreated cells by mTq fluorescence emission at 476 nm in SDS-treated cells. FRET curves were generated by plotting the FRET efficiency (E) versus the acceptor to donor ratio (A:D). The data were fit by non-linear regression to the following rectangular hyperbolic function using Prism 6 (GraphPad Software, La Jolla, CA): E = (E_max × A:D) / (EC₅₀ + A:D). The fitted values for E_max and EC₅₀ are reported in Table 1.

Table 1.

FRET curve analysis.

Coexpressed receptors	Parameter
	Emax			EC50
	Total	DM-sensitive	DM-insensitive	Total	DM-sensitive	DM-insensitive
WT-YFP + WT-mTq	0.59 ± 0.02	0.54 ± 0.02	0.11 ± 0.02	2.5 ± 0.2	3.0 ± 0.4	1.4 ± 0.8
G188R-YFP + G188R-mTq	0.56 ± 0.02	0.14 ± 0.01	0.44 ± 0.03	2.3 ± 0.3	0.9 ± 0.4	3.7 ± 0.5
WT-YFP + G188R-mTq	0.49 ± 0.04	0.20 ± 0.03	0.28 ± 0.04	4.2 ± 0.8	3.6 ± 1.2	4.8 ± 1.5
G188R-YFP + WT-mTq	0.21 ± 0.02	0.08 ± 0.02	0.13 ± 0.01	1.6 ± 0.5	1.5 ± 1.0	1.7 ± 0.5
WT-YFP + m2-mTq	0.23 ± 0.03	0.14 ± 0.02	0.08 ± 0.02	3.6 ± 1.4	3.9 ± 1.9	2.5 ± 2.0

Data in Figs. 2, 3 and 7 were fit to a rectangular hyperbolic function as described in the Materials and methods. Fitted values for the maximal FRET efficiency (E_max) and EC₅₀ are shown along with the standard errors.

2.4. Confocal microscopy

Cells were prepared on coverslips and confocal microscopy conducted to detect fluorescence from YFP and DAPI as described previously [33]. Confocal microscopy was carried out on a Leica SP8 confocal microscope equipped with a 100×/1.4-NA oil objective (Leica, Buffalo Grove, IL). Fluorescence from mTq was detected by exciting samples at 458 nm using an Argon laser and collecting the emission signal at 465–500 nm. Images were collected sequentially.

2.5. Western blotting

The cell suspension from a single well was pelleted and used as is or was deglycosylated with PNGase F (New England Biolabs, Ipswich, MA). The pellet was resuspended in NuPage LDS sample buffer containing NuPage sample reducing agent (Novex, Carlsbad, CA) and loaded on a 12% SDS-PAGE RunBlue precast gel (Expedeon, San Diego, CA). The Precision Plus Protein Kaleidoscope molecular weight protein standards (Bio-rad Laboratories, Hercules, CA) were also loaded alongside samples on each gel. Samples were separated by SDS-PAGE and transferred onto Immobilon-P PVDF transfer membrane (Millipore, Temecula, CA). Opsin was detected with either the primary antibodies anti-1D4 [38], from de-glycosylated cell suspensions, or anti-4D2 (Millipore, Temecula, CA), from cell suspensions used as is. Primary antibodies were detected with an anti-mouse IgG alkaline phosphatase conjugate (Promega, Madison, WI). Immunolabeled proteins were visualized using the 5-bromo-4-chloro-3-indolyl-phosphate/nitro blue tetrazolium substrate (Promega, Madison, WI).

2.6. Transfection efficiency determination

Coverslips were prepared similarly as for confocal microscopy but with the following change. Cell nuclei were stained with Draq5 (Abcam, Cambridge, MA). Cells were visualized on a Leica DMI 6000B inverted microscope with a 63×/1.4-NA oil objective (Leica, Allendale, NJ). 14 bit images were collected using a Retiga Aqua Blue camera (Q-imaging, Surrey, BC). 10 different images were captured and analyzed per transfection. Image analysis was performed using Metamorph Imaging Software version 7.8.0.0 (Molecular Devices, Downingtown, PA). Draq5 was detected using filter cube number 41008 (ex. 620/60 and em. 700/75) (Chroma Technology Corp, Bellows Falls, VT). mTq was detected using filter cube number 31044v2 (ex. 436/20 and em. 510/40) (Chroma Technology Corp, Bellows Falls, VT). YFP was detected using filter cube number 49003 (ex. 500/20 and em. 535/30) (Chroma Technology Corp, Bellows Falls, VT). The area of a single cell was defined from images of nuclei stained by Draq5. The area of the nucleus was increased by applying a dilation morphology filter twice using a three pixel circular filter each time. This generated a mask that encompassed the nucleus and cytoplasm of a single cell. The dilated nuclear mask was superimposed on images exhibiting mTq or YFP fluorescence. Cells expressing mTq and/or YFP were determined by monitoring the overlap between the fluorescence signal and the dilated nuclear mask.

2.7. Statistical analyses

All graphs were generated and statistical analyses conducted in Prism 6 (GraphPad Software, La Jolla, CA). Comparisons of the mean value for two conditions were conducted by an unpaired two-tailed t-test. The level of significance was taken as p < 0.05.

3. Results

3.1. Characterization of wild-type opsin and G188R opsin by FRET

Cultured cells are a good model system to investigate the misfolding, aggregation, and preciliary targeting of opsin mutants [40–42]. We have previously established an HEK293 cell system expressing wild-type or mutant opsin tagged with fluorescent proteins for FRET studies [33]. The fluorescent tags were shown not to interfere with the normal structure and function of the light receptor. In that study, variants of cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) were used as the donor and acceptor molecules, respectively, for FRET measurements. In the current study, another CFP variant called mTurquoise (mTq) was used as the donor molecule [37].

FRET occurs between a donor and acceptor molecule over distances in the nanometer range and is therefore useful for investigating protein–protein interactions [31]. FRET between tagged wild-type opsins or tagged G188R opsins was monitored in HEK293 cells spectroscopically using methods reported previously [33]. HEK293 cells were cotransfected with equal amounts of DNA coding for either wild-type opsin tagged with YFP or mTq (WT-YFP or WT-mTq) or G188R opsin tagged with YFP or mTq (G188R-YFP or G188R-mTq). Spectra were collected from cells excited at 425 nm, near the excitation maxima for the donor mTq. FRET was observed for both wild-type and G188R opsins, as indicated by the sensitized emission peak from YFP at 527 nm (Fig. 2A and B, black curve). As we showed previously, the FRET signal from wild-type opsin and G188R opsin exhibited different behaviors upon treatment with the detergent n-dodecyl-β-D-maltoside (DM) [33]. Treatment of cells with DM disrupted most of the FRET signal for wild-type opsin whereas it only partially disrupted the FRET signal for G188R opsin (Fig. 2A and B, red curve). This differential effect of DM is independent of the cellular localization of the receptor since wild-type opsin retained in the endoplasmic reticulum behaves the same as that in the plasma membrane [33]. Disruption of FRET resulted in dequenching of the mTq fluorescence signal and loss of sensitized emission from YFP. For G188R opsin, SDS treatment was required to fully disrupt the FRET signal (Fig. 2B, blue curve).

Fig. 2.

FRET between tagged wild-type opsin or tagged G188R opsin. A, B. Emission spectra of cells coexpressing equal levels of YFP- and mTq-tagged receptors. FRET was conducted on cells coexpressing YFP- and mTq-tagged wild-type opsins (A) or YFP- and mTq-tagged G188R opsins (B). Emission spectra were collected sequentially on untreated cells (black), DM-treated cells (red), and SDS-treated cells (blue). C, D. FRET curves for wild-type (C) and G188R (D) opsins. Total (black, circles), DM-sensitive (red, squares) and DM-insensitive (blue, triangles) FRET efficiencies were computed from emission spectra (e.g., A and B), as described in the Materials and methods. FRET efficiencies were computed from cells expressing a range of acceptor (i.e., YFP) to donor (i.e., mTq) ratios (A:D). FRET curves were generated by plotting the computed FRET efficiencies and the measured A:D ratio. Data were fit with a rectangular hyperbolic function. Fitted lines are shown and values obtained from fits are reported in Table 1. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

To obtain more detailed insights about the relative contributions of the FRET signal that was sensitive or insensitive to DM treatment, the FRET efficiency was computed for a range of acceptor to donor ratios. Changes in acceptor and donor expression were achieved by changing the amount of each DNA transfected while keeping the total amount of transfected DNA constant. Increasing the acceptor to donor ratio resulted in increased FRET efficiencies until saturation was achieved (Fig. 2C and D). The data were fit with a rectangular hyperbolic function to determine the maximal FRET efficiency (Table 1). The total FRET signal (black, circles) is shown along with the FRET signal separated out into DM-sensitive (red, squares) and DM-insensitive (blue, triangles) components.

The curves for the total FRET signal were similar for wild-type and G188R opsins (Table 1). As indicated by FRET studies conducted at a single acceptor to donor ratio, the two opsin forms exhibited differences when the FRET curves were separated into DM-sensitive and DM-insensitive components. For wild-type opsin, the maximal FRET efficiencies for the DM-sensitive and DM-insensitive components represented 92% and 19% of the total maximal FRET efficiency, respectively. For G188R opsin, the maximal FRET efficiencies for the DM-sensitive and DM-insensitive components represented 25% and 79% of the total maximal FRET efficiency, respectively. Thus, the total FRET signal predominantly derived from the DM-sensitive component for wild-type opsin whereas it predominantly derived from the DM-insensitive component for G188R opsin.

3.2. Origin of the FRET signal from wild-type and G188R opsins

A FRET signal in and of itself cannot be interpreted as deriving from physical interactions without appropriate controls. FRET can occur from chance encounters of freely diffusing proteins that bring an acceptor and donor molecule within close proximity allowing for energy transfer even in the absence of physical interactions [43–45]. We refer to this type of FRET as non-specific FRET and that arising from physical interactions as specific FRET. Non-specific FRET was defined by generating a control FRET curve from cells coexpressing opsin with an unrelated GPCR, the m2 muscarinic receptor (Fig. 3). This control defines the level of FRET that must be exceeded to indicate physical interactions [43,44,46,47]. The control FRET curve obtained from untreated cells was lower than either the FRET curves obtained for wild-type or G188R opsin (Fig. 3A). Thus, at least some component of the FRET signal for both receptor forms derives from specific FRET.

Fig. 3.

Control FRET curves. FRET curves were generated from cells coexpressing YFP-tagged wild-type opsin and mTq-tagged m2 muscarinic receptor. Total (A), DM-sensitive (B), and DM-insensitive (C) FRET curves are shown. Data were fit with a rectangular hyperbolic function. Fitted lines are shown and values obtained from fits are reported in Table 1. The fitted lines from FRET curves for wild-type (blue dashed curve) and G188R (red dashed curve) opsins are shown for reference. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

FRET curves for the DM-sensitive component indicated that the maximal FRET efficiency for wild-type opsin was greater than that of the control FRET curve (Fig. 3B, Table 1). In contrast, the maximal FRET efficiency for G188R opsin was similar to that of the control FRET curve. Thus, the DM-sensitive FRET component for wild-type opsin is indicative of specific FRET whereas the DM-sensitive FRET component for G188R opsin derives from non-specific FRET. The opposite was observed for the DM-insensitive FRET component. The DM-insensitive maximal FRET efficiency for G188R opsin was greater than that of the control FRET curve whereas the DM-insensitive maximal FRET efficiency for wild-type opsin was similar but slightly greater than that of the control FRET curve (Fig. 3C and Table 1). Thus, the DM-insensitive FRET signal is specific for G188R opsin and predominantly non-specific for wild-type opsin.

The elimination of the FRET signal observed after treatment of cells with detergent can arise from disruption of protein–protein interactions for specific FRET, disruption of the cell membrane for non-specific FRET, or denaturation of the fluorescent proteins by the detergent. For wild-type and G188R opsins tagged with mTq, treatment of cells with DM and SDS did not change the level of fluorescence from mTq (Fig. 4). For receptors tagged with YFP, treatment with DM had no effect on the fluorescent protein. In contrast, treatment with SDS resulted in a 30% reduction in fluorescence, indicating the denaturation of the fluorescent protein. Disruption of the FRET signal upon treatment with DM is not a result of denaturation of either fluorescent protein. Rather, it derives from disruption of protein–protein interactions for wild-type opsin and disruption of the cell membrane for G188R opsin. Native rhodopsin oligomers from rod photoreceptor cells are disrupted by DM treatment [48]. Thus, the protein–protein interactions disrupted by DM treatment observed for wild-type opsin appear to derive from oligomers normally formed by the light receptor.

Fig. 4.

Sensitivity of fluorescence signal to detergent treatment. A. Emission spectra for mTq- and YFP-tagged wild-type opsins. Emission spectra were collected from cells expressing either mTq-tagged (solid curves) or YFP-tagged (dashed curves) wild-type opsin. Spectra were collected on untreated cells (black) and on cells treated with DM (blue) and then SDS (red). B. Changes in fluorescence after treatment with detergent. The change in the fluorescence signal observed after treating cells with DM and then SDS is shown for mTq-tagged (blue, circles) and YFP-tagged (red, squares) wild-type (WT) or G188R opsin. The mean and standard deviation are plotted along with the individual data points (n = 7). No significant difference was observed in the fluorescence signal of mTq-tagged wild-type or G188R opsin after treatment with DM and SDS, as assessed by a t-test (p > 0.05). A significant difference was observed in the fluorescence signal of YFP-tagged wild-type or G188R opsin after treatment with DM and SDS, as assessed by a t-test (p < 0.001). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The inability of DM to disrupt the DM-insensitive specific FRET signal is indicative of aggregate formation [49]. In contrast to the disruption of the DM-sensitive FRET signal by DM treatment, disruption of the DM-insensitive FRET signal by SDS results, at least in part, from the observed denaturation of the acceptor YFP. Thus, SDS does not appear to fully disrupt opsin aggregates. The inability of SDS to fully disrupt opsin aggregates is also indicated by Western blot analysis of wild-type and G188R opsins (Fig. 5). Bands corresponding to the full-length monomeric wild type opsin either untagged or tagged with fluorescent protein were present at about 37 kDa and 60 kDa, respectively. Faint bands corresponding to multimeric forms were also present. Western blots for G188R opsin revealed bands corresponding to dimers and higher order aggregates. Differences in the appearance of bands between Western blots probed with the anti-4D2 antibody and anti-1D4 antibody are due to the deglycosylation of samples by PNGaseF used in blots prepared with the latter antibody. The Western blots are consistent with FRET data indicating that G188R opsin forms aggregates that are not fully disrupted by SDS treatment.

Fig. 5.

Western blot analysis of opsin expressed in HEK293 cells. Western blots of samples prepared from cells expressing either wild-type (WT) opsin (A, C) or G188R opsin (B, D) are shown. In each blot, the samples were loaded in the following order: YFP-tagged opsin (lane 1), mTq-tagged opsin (lane 2), and untagged opsin (lane 3). The molecular mass of protein standards is indicated in kDa. A, B. Opsin was detected with the anti-4D2 antibody, which detects the N-terminal end of the receptor. C, D. Opsin was detected with the anti-1D4 antibody, which detects the C-terminal end of the receptor. Only samples used for blotting with the anti-1D4 antibody were treated with PNGaseF prior to SDS-PAGE.

The FRET data for wild-type opsin indicates that the receptor predominantly forms oligomers. The small increase of the DM-insensitive FRET signal over that of the control raises the possibility that a small fraction of the wild-type receptor may also aggregate (Fig. 3C and Table 1). This small fraction that aggregates may represent misfolded opsin that has not yet been degraded by the quality control in the endoplasmic reticulum. Misfolded opsin is normally degraded by the ubiquitin–proteasome system [4,5,25]. If aggregates formed by wild-type opsin result from misfolded receptor, increasing the level of misfolded wild-type opsin by inhibiting proteasome activity is predicted to increase the level of aggregates. Cells coexpressing tagged wild-type opsin were treated with the proteasome inhibitor MG-132. Inhibition of the proteasome resulted in a 22% increase in the DM-insensitive FRET signal (Fig. 6), which represents aggregates. Thus, the small specific DM-insensitive FRET signal observed in FRET curves for wild-type opsin originates from aggregates formed by a minor fraction of the receptor that misfolds and has not been degraded. In summary, FRET analysis indicates that wild-type opsin is predominantly properly folded forming oligomers, which can be disrupted by DM. A minor fraction of wild-type opsin misfolds and forms aggregates. G188R, on the other hand, forms aggregates that are resistant to DM and cannot be fully disrupted by SDS.

Fig. 6.

Impact of MG-132 treatment on the DM-sensitive FRET signal. FRET was determined from cells coexpressing equal levels of YFP- and mTq-tagged wild-type opsins. FRET was measured on untreated cells and cells treated with DMSO or MG-132 in DMSO. The percentage of the total FRET signal derived from DM-insensitive FRET is shown. Mean values and standard deviations were plotted along with the individual data points (n = 10). No significant difference was observed between untreated and DMSO-treated cells, as assessed by a t-test (p > 0.05). A significant difference was observed between cells treated with DMSO or MG-132, as assessed by a t-test (p < 0.0001).

3.3. Examining interactions between wild-type and G188R opsins by FRET

Patients with adRP will express both wild-type and G188R opsins. Here, we examined whether or not these two opsins can physically interact with each other, as has been proposed for wild-type and P23H opsins [5,25]. Wild-type and G188R opsins were coexpressed in HEK293 cells. Two coexpression scenarios were considered. Wild-type opsin tagged with YFP and G188R opsins tagged with mTq (WT-YFP and G188R-mTq) or wild-type tagged with mTq and G188R opsins tagged with YFP (WT-mTq and G188R-YFP). A range of acceptor to donor ratios was tested and FRET curves generated (Fig. 7). Different FRET curves were generated depending on the combination of florescent proteins the opsins were tagged with. Examining FRET curves for untreated cells, both coexpression scenarios resulted in curves that were lower than that obtained for either wild-type opsin or G188R opsin expressed alone (Fig. 7A, Table 1). The FRET curve for cells coexpressing WT-YFP and G188R-mTq (green, circles) was above the control FRET curve (black dashed curve), thereby indicating that the FRET signal has a specific component. In contrast, the FRET curve for cells coexpressing WT-mTq and G188R-YFP (magenta, squares) was similar to the control FRET curve, thereby indicating that the FRET signal mostly derived from a non-specific component.

Fig. 7.

FRET curves generated from cells coexpressing wild-type and G188R opsins. FRET curves were generated from cells coexpressing WT-YFP and G188R-mTq (green, circles) or G188R-YFP and WT-mTq (magenta, squares). Total (A), DM-sensitive (B), and DM-insensitive (C) FRET curves are shown. Data were fit with a rectangular hyperbolic function. Fitted lines are shown and values obtained from fits are reported in Table 1. The fitted lines from the control FRET curve (black dashed curve) and FRET curves for wild-type (blue dashed curve) and G188R (red dashed curve) opsins are shown for reference. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The DM-sensitive and insensitive components of the FRET signal were examined for more details about the FRET signal from cells coexpressing wild-type and G188R opsins. The FRET curves for the DM-sensitive component for both types of coexpressed cells were similar to or below the FRET curve obtained from cells expressing G188R opsin alone (red dashed curve) or the control FRET curve (Fig. 7B, Table 1). In each case, this FRET component in cells coexpressing the two opsins derives mainly from non-specific FRET. Cells coexpressing WT-YFP and G188R-mTq may have a minor specific DM-sensitive FRET component since the maximal FRET efficiency is slightly higher than that of the control FRET curve. The FRET curve for the DM-insensitive component of cells coexpressing WT-mTq and G188R-YFP was similar to that of cells expressing wild-type opsin alone (dashed blue curve) (Fig. 7C, Table 1). The DM-insensitive FRET curve obtained from cells coexpressing WT-YFP and G188R-mTq was in between the FRET curves obtained from cells expressing either wild-type opsin or G188R opsin alone. Thus, some aggregation occurs between wild-type and G188R opsins, albeit at a low level.

3.4. Origin of FRET curve differences in the two coexpression scenarios

Differences in the FRET curves observed depending on which opsin species is tagged with either mTq or YFP could arise from a variety of sources, which are examined here. If one receptor form expresses only in a subset of cells, this would lower the measured FRET efficiency. High transfection efficiencies were achieved for all transfections conducted in this study and a majority of cells expressed receptor tagged with both fluorescent proteins (Table 2). High transfection efficiencies were observed even for transfections where the receptor tagged with YFP was in excess over that tagged with mTq. Thus, the differences in FRET curves did not arise from differences in expression of the two opsin forms.

Table 2.

Transfection efficiency.

Coexpressed receptors	Transfected DNA ratio	Co-transfection efficiency (%)	Transfection efficiency (%)
WT-YFP + WT-mTq	1:1	96.7 ± 1.7	99.5 ± 0.8
	7:1	89.8 ± 9.4	98.1 ± 1.7
G188R-YFP + G188R-mTq	1:1	86.8 ± 11.4	89.9 ± 5.9
	7:1	79.0 ± 5.8	91.9 ± 10.3
WT-YFP + G188R-mTq	1:1	94 ± 6.6	97.5 ± 3.3
	7:1	88.1 ± 5.3	98.1 ± 14.2
G188R-YFP + WT-mTq	1:1	85.8 ± 10.8	93.8 ± 8.8
	7:1	91.1 ± 10.3	88.5 ± 6.8

Transfection efficiencies were calculated as described in the Materials and methods. Co-transfection efficiencies were calculated from cells expressing both YFP- and mTq-tagged receptors while transfection efficiencies were calculated from cells expressing at least one of the tagged receptors. Transfection efficiencies are reported for two different transfected DNA ratios of YFP- and mTq-tagged receptors. Mean values are reported along with the standard deviations (n = 4).

The presence of truncated opsin species could also decrease the measured FRET signal. A truncated and untagged receptor species could form a complex with tagged opsins, thereby reducing the number of complexes containing a FRET pair. Since misfolded opsins are retained in the endoplasmic reticulum and may become degraded, there is a potential for the presence of an appreciable population of untagged truncated receptor. The integrity of untagged and fluorescent protein-tagged opsins was evident in Western blots. Cell extracts expressing each receptor form individually were examined on Western blots using either an anti-1D4 antibody, which detects an epitope at the C-terminus of rhodopsin and was added at the end of the fluorescent protein sequence, or an anti-4D2 antibody, which detects the N-terminal end of rhodopsin. The former labels non-truncated receptor while the latter would label both truncated and non-truncated receptors. Western blots labeled with either antibody gave similar results (Fig. 5). An appreciable population of truncated mutant receptor is not present in cell extracts and therefore does not contribute to the observed differences in FRET curves.

The last consideration for the difference in FRET curves is biologically relevant. The DM-insensitive FRET curves suggest the presence of a small population of misfolded and aggregated receptors even for wild-type opsin (Fig. 3C, Table 1). If only misfolded wild-type opsin can physically interact with the mutant, then differences in FRET can arise depending on which species is tagged with the two fluorescent proteins. This idea is illustrated with a simulation, where we assume a fraction of wild-type opsin misfolds (10% in this case) and only the misfolded opsins aggregate with the mutant. For simplicity, we assume that aggregates are dimers. Using the binomial distribution, we can predict the fraction of receptors that can form a heterocomplex, and therefore capable of FRET, over a range of acceptor to donor ratios (Fig. 8). When examining the expected number of complexes capable of FRET, the simulation predicts that cells coexpressing WT-YFP and G188R-mTq (green curve) will have a larger fraction of receptor tagged with the donor mTq in heterocomplexes compared to cells coexpressing the two opsin forms with the reverse complement of fluorescent proteins (magenta curve). Thus, cells coexpressing WT-YFP and G188R-mTq are predicted to generate FRET curves with a higher maximal FRET efficiency compared to cells coexpressing WT-mTq and G188R-YFP. This pattern mirrors the one observed experimentally (Fig. 7C), which suggests that only the small fraction of wild-type opsin that misfolds is capable of aggregating with misfolded G188R opsin.

Fig. 8.

Simulated curves assuming only misfolded wild-type opsin interact with G188R opsin. The binomial distribution was used to compute the fraction of mTq-tagged opsins that are in complex with YFP-tagged opsins, and therefore capable of FRET, at different ratios of YFP- and mTq-tagged receptors. Simulated curves were generated by plotting the relationship between the fraction of mTq-tagged receptors in a complex capable of FRET and the ratio of expressed mTq- and YFP-tagged receptors. A scenario where all expressed mTq-tagged opsins can form a complex with YFP-tagged opsins is represented by the black curve. Scenarios where only 10% of the expressed wild-type opsin misfolds and is capable of forming a complex with G188R opsin are represented by the green and magenta curves. The green curve results from the coexpression of WT-YFP and G188R-mTq. The magenta curve results from the coexpression of G188R-YFP and WT-mTq. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

If only a small fraction of wild-type opsin physically interacts with G188R opsin, then a majority of wild-type opsin is expected to traffic normally. We have shown previously that G188R opsin is retained in the endoplasmic reticulum whereas wild-type opsin can traffic to the plasma membrane [33]. The localization of coexpressed opsins was investigated by confocal microscopy. As expected, the differently tagged wild-type opsins alone or G188R opsins alone are predominantly colocalized when expressed at equal levels or when the YFP-labeled opsin is expressed at higher levels compared to the mTq-labeled opsin (Figs. 9A, B, 10A, B). In contrast, coexpression of wild-type and G188R opsins resulted in limited colocalization and a large fraction of wild-type opsin trafficked normally even under conditions where the G188R opsin was in excess (Figs. 9C, D, 10C, D). These confocal microscopy data are consistent with the FRET data, which indicate that the majority of wild-type opsin does not physically interact with G188R opsin. Only the small fraction of wild-type opsin that is misfolded interacts with G188R opsin in aggregates.

Fig. 9.

Confocal microscopy of HEK293 cells coexpressing YFP- and mTq-tagged opsins at equal expression levels. HEK293 cells were cotransfected with equal amounts of DNA coding for tagged opsins indicated in the figure. Each row of images shows YFP fluorescence (green), mTq fluorescence (magenta) and a merge of YFP and mTq fluorescence. DAPI fluorescence is shown in blue. Scale bar, 10 μM. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 10.

Confocal microscopy of HEK293 cells coexpressing YFP- and mTq-tagged opsins at unequal expression levels. HEK293 cells were cotransfected with DNA coding for tagged opsins indicated in the figure at a ratio of 7:1 in favor of the YFP-tagged opsin. Each row of images shows YFP fluorescence (green), mTq fluorescence (magenta) and a merge of YFP and mTq fluorescence. DAPI fluorescence is shown in blue. Scale bar, 10 μM. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

4. Discussion

The recognition that rhodopsin/opsin forms a quaternary structure in the form of oligomers under normal physiology has led us to consider here whether the G188R misfolding opsin mutant can physically interact with the wild-type receptor. Physical interactions between wild-type and misfolded opsin mutants have been proposed to underlie the dominant phenotype in adRP [25]. Previously, FRET has been used to demonstrate that the P23H misfolding opsin mutant can form a complex with wild-type opsin [25], however, the nature of the complex was unknown. The studies conducted in the current study provide a detailed and alternate picture of the processes predicted to occur when wild-type and mutant opsins are coexpressed. These insights were made possible by the ability to distinguish between normal oligomers of properly folded opsin and aggregates of misfolded opsin and the implementation of control experiments.

In the current study, we demonstrated that wild-type opsin can physically interact with G188R opsin, which is consistent with a previous study examining interactions between wild-type and P23H opsins [25]. The complex formed between wild-type and G188R opsins represents aggregates rather than normal oligomers (Fig. 7). The interaction detected in our study, however, only represents the behavior of a minor fraction of the expressed wild-type receptor. When wild-type opsin is expressed alone, a minor fraction appears to misfold and aggregate (Fig. 3C), perhaps due to overexpression in the HEK293 cell. This small fraction of receptor is the species that aggregates with G188R opsin. Thus, the mutant receptor does not appear to promote misfolding and aggregation of wild-type opsin, but rather, aggregates with wild-type receptor that is already misfolded.

The majority of wild-type opsin appears to properly fold and traffic (Figs. 9 and 10). Thus, properly folded wild-type opsin does not aggregate with the misfolded opsin mutant, pointing to protein specificity in aggregation. Specificity in the aggregation of P23H opsin has been demonstrated as unrelated misfolded opsin and aggregation-prone proteins, including a transmembrane protein, were shown to be unable to aggregate with the misfolded opsin [50]. Misfolding of both P23H and G188R opsins only partially disrupts the structure of the receptor, which results in an increase in β-sheet structure at the expense of some α-helical structure [33]. Unfolding of wild-type rhodopsin also results in increased β-sheet structure and reduced α-helical structure [51]. The attainment of β-sheet structure may be a prerequisite for aggregation. Thus, wild-type opsin that does not misfold and attain β-sheet structure will be unaffected by G188R opsin.

The inability of properly folded wild-type opsin to interact with misfolded mutants is consistent with observations in animal models of adRP expressing the P23H opsin mutant. Increasing the expression level of wild-type opsin does not improve the transport of P23H opsin to the rod outer segment [29]. This observation is consistent with studies here on G188R opsin, where the mutant was unable to form oligomers with wild-type opsin and was unable to traffic to the plasma membrane when coexpressed with the wild-type receptor (Figs. 9 and 10). The biosynthesis and transport of wild-type opsin appears to be unaffected by misfolded opsins, as rod photoreceptor cells in heterozygous P23H opsin knockin mice express about half the normal amount of the light receptor [19]. The lack of effect on wild-type opsin would be predicted if the properly formed wild-type opsin did not aggregate with misfolded opsin, which is observed here by the lack of robust FRET between wild-type and G188R opsins (Fig. 7).

In the absence of interactions between wild-type opsin and misfolded mutants, the pathogenesis of adRP is predicted to occur due to the accumulation of toxic aggregates in the endoplasmic reticulum or the presence of the mutant receptor in the rod outer segment disk membranes (Fig. 1). G188R opsin is a severe misfolded mutant, and therefore all mutants cannot bind chromophore and are unable to traffic to the plasma membrane in cultured cells [2,33,34]. Thus, this mutant is predicted to be unable to traffic to the rod outer segment. These observations contrast with P23H opsin, which is a partially misfolded mutant where a small fraction of the mutant can traffic to the rod outer segment in photoreceptor cells [18,20,52]. Thus, the pathogenesis of the P23H mutant may be more complex compared to that of the G188R mutant. Future studies are required to investigate a number of different misfolding opsin mutants that cause adRP to determine the extent to which different mutants are similar to each other in terms of aggregation and interactions with wild-type opsin.

The nature of the aggregates that are toxic to cells is unclear. Two distinct types of aggregates have been reported: large aggregates in inclusion bodies called aggresomes and smaller aggregates that are diffusely localized in the endoplasmic reticulum [4]. Aggresomes can be distinguished by light microscopy whereas smaller aggregates cannot. In the current study, the aggregates detected by FRET are predominantly small aggregates since aggresomes are rarely observed by confocal microscopy (Figs. 9 and 10). Aggresomes are also not observed in rod photoreceptor cells in adRP animal models expressing P23H opsin [20,52,53]. Thus, small aggregates are the likely toxic species rather than aggresomes. Whether or not aggregates detected in heterologous expression systems like HEK283 cells are the same as those formed in photoreceptor cells of the retina is unclear. Approaches that reduce the aggregation of misfolded opsin mutants in heterologous expression systems also appear to reduce opsin aggregation in photoreceptor cells and improve the health of the retina in adRP animal models expressing P23H opsin [10–12,54]. Thus, aggregates of opsin formed in heterologous expression systems like HEK293 cells are likely similar to those formed in photoreceptor cells of the retina. Based on the current study, therapeutic strategies that are predicted to be effective for patients expressing the G188R opsin mutant include those that suppress expression of the mutant receptor (e.g., [55–59]), prevent aggregation of the mutant receptor (e.g., [10–12,14,41,54]), or general approaches that target endoplasmic reticulum stress [60–63].

Abbreviations

A:D

acceptor to donor

CFP

cyan fluorescent protein

DM

n-dodecyl-β-D-maltoside

DMSO

dimethyl sulfoxide

FRET

Förster resonance energy transfer

FTIR

Fourier transform infrared

G188R-mTq

G188R opsin tagged with mTq

G188R-YFP

G188R opsin tagged with YFP

GPCR

G protein-coupled receptor

mTq

mTurquoise

PCR

polymerase chain reaction

adRP

autosomal dominant retinitis pigmentosa

WT

wild-type

WT-mTq

wild-type opsin tagged with mTq

WT-YFP

wild-type opsin tagged with YFP

YFP

yellow fluorescent protein