Mutations in the dim light photoreceptor protein rod opsin cause autosomal dominant retinitis pigmentosa. The majority of these mutations (class II) lead to protein misfolding. For example, the common class II rod opsin mutation P23H misfolds and is retained in the ER, prior to retrotranslocation and degradation by the proteasome. If degradation fails then the protein can aggregate to form intracellular inclusions. Furthermore, mutant opsin exerts a dominant negative effect on the wild-type (WT) protein. Here we show that the toxic gain of function and dominant negative properties of misfolded rod opsin in cells can be alleviated by drug treatments targeted against a range of cellular pathways. P23H rod opsin aggregation, inclusion formation with associated caspase activation and cell death were reduced by kosmotropes, molecular chaperone inducers and mToR inhibition. But these treatments did not enhance mutant opsin folding or reduce the dominant negative effect of P23H rod opsin. In contrast, retinoids acted as pharmacological chaperones to enhance P23H folding and reduce the dominant negative effect on WT rod opsin processing, as well as reducing toxic gains of function. Therefore, the suppression of the dominant negative effects of protein misfolding required enhanced folding of the mutant protein, whereas suppression of toxic gain of function effects did not require improved folding per se . These studies suggest that some forms of rhodopsin RP may be treated by targeting protein folding and reducing protein aggregation.
Retinitis pigmentosa (RP) is a heterogeneous group of inherited retinal degenerative diseases which are characterized by the progressive loss of rod and cone photoreceptors and for which no therapies are currently available. Mutations in the rod visual pigment rhodopsin were first described in 1990 ( 1 ) and are the most common cause of autosomal dominant retinitis pigmentosa (ADRP; OMIM 180380). Rhodopsin is formed from the rod opsin protein and the chromophore 11 -cis- retinal. Soon after the identification of mutations in rod opsin, several research groups heterologously expressed rod opsin in mammalian cell culture in order to characterize the properties of these mutations ( 2–4 ). Further detailed characterization of mutant rod opsin in animal models revealed two major classes of mutations. Class I mutants fold normally in cell culture but are not correctly transported to the outer segment in vivo . In contrast, class II mutants caused rod opsin misfolding resulting in retention in the ER ( 5 ).
Despite a wealth of data describing the biochemical properties of Class II mutants, the pathogenic mechanism by which these mutations cause retinal degeneration remains elusive. Dominance in ADRP could be due to a loss-of-function, gain-of-function, dominant-negative mechanism or a combination of the three ( 6 ). There are several lines of evidence suggesting that the loss of function (haploinsufficiency) is not likely to be the mechanism leading to photoreceptor cell death, as heterozygous rhodopsin-knockout mice showed little photoreceptor degeneration ( 7 , 8 ) suggesting that the mechanism of cell death induced by Class II mutations is either due to a toxic gain-of-function or dominant-negative effect of the mutant protein.
Gain-of-function cell death mechanisms related to protein misfolding, similar to several neurodegenerative diseases, are supported by studies of the P23H class II mutation. This mutant, unlike wild-type (WT) protein, caused retention in the ER, induction of the unfolded protein response (UPR), inhibition of the proteasome and aggregation into oligomeric high molecular weight species that form intracellular inclusions ( 9–12 ).
Several studies have also suggested a potential dominant-negative effect related to class II mutants. Co-expression of WT and P23H rod opsin resulted in the formation of inclusions which contained the WT protein ( 9 ). Class II mutants also prevented the correct processing and enhanced proteasome-mediated degradation of the WT protein suggesting a potential dominant-negative mechanism ( 13 ).
The pathology of a variety of protein-folding disorders involves the intracellular aggregation of proteins and formation of inclusion bodies. Indeed, it has been suggested that saturation of the proteolytic machinery and disruption of protein homeostasis (proteostasis) by misfolded proteins leads to a common cellular response of accumulation of misfolded protein in inclusions ( 14 ). Potential pharmacological therapies for these diseases are based either on promoting correct folding, inhibiting aggregation, increasing degradation or protection from cell death ( 15 ).
Problems of protein folding can be addressed either by directly targeting the protein structure or by manipulating the cellular milieu. Pharmacological chaperones and kosmotropes target protein structure, whereas chaperone inducers and stimulation of degradation manipulate the cellular quality control machinery. The term ‘pharmacological chaperones’ refers to small molecules that bind specific sites within a protein’s native or quasi-native structure, thereby shifting the folding equilibrium towards the native structure and away from off pathway intermediates. Pharmacological chaperones are often ligands, agonists or antagonists, which bind directly to substrate polypeptides in their ligand-binding pocket and contact several sites simultaneously. For example, previous studies have suggested that the rod opsin chromophore (11 -cis- retinal) and retinal analogues (e.g. 9 -cis- retinal) can act as pharmacological chaperones to stabilize class II rod opsin to improve translocation through the secretory pathway and produce more correctly folded rhodopsin ( 9 , 16–18 ). However, the effect of retinoids on the gain of function and dominant negative effects of class II rod opsin mutations have not been studied. Nevertheless, the data suggest that class II rod opsin may be amenable to other forms of pharmacological manipulation.
‘Kosmotrope’ (order-maker), as opposed to ‘chaotrope’ (disorder-maker), denotes solutes that stabilize proteins and membranes; thus, kosmotropes stabilize proteins and hydrophobic aggregates. Kosmotropes are small, low molecular weight compounds, also known as ‘chemical chaperones’, which enhance protein folding and reduce aggregation of many proteins, such as CFTR ( 19 ). This class of compounds include: polyols such as glycerol; solvents such as dimethyl sulfoxide (DMSO); methylamines such as trimethylamine- N -oxide (TMAO); fatty acids such as 4-phenylbutyric acid (4-PBA); and sugars such as trehalose among others. A key difference between pharmacological chaperones and kosmotropes is their specificity. Pharmacological chaperones are specific to a particular protein conformation, whereas kosmotropes are non-specific.
The cell’s natural response to protein misfolding exploits the molecular chaperone machinery. This ‘machine’ assists protein folding and regulates protein quality control and degradation ( 20 ). There is now substantial evidence from many model systems that the chaperone machinery can be manipulated to counteract protein misfolding, either to stimulate correct folding, reduce aggregation, enhance degradation and/or promote cell viability. For example, celastrol, a quinone methide triterpene which is an active component from Chinese herbal medicine has potent anti-inflammatory and anti-oxidative effects in addition to activating heat shock factor 1 (HSF1) ( 21 ). Celastrol extended the life span of transgenic mice in a model of ALS ( 22 ). An alternative approach to stimulate molecular chaperone expression is to use inhibitors of Hsp90. Geldanamycin, radicicol and 17-allylamino-17-demethoxygeldanamycin (17-AAG) bind Hsp90, thereby disrupting the chaperone complex with HSF1, which in turn leads to the activation of HSF1 and consequent expression of molecular chaperones. Treatment of COS-1 cells with geldanamycin induced the expression of Hsp40, Hsp70 and Hsp90 and inhibited aggregation of polyglutamine expanded Huntingtin exon 1 protein ( 23 ). Similarly, 17-AAG ameliorated motor impairments in the spinal and bulbar muscular atrophy (SBMA) transgenic mouse model without detectable toxicity, by reducing amounts of monomeric and aggregated mutant androgen receptor ( 24 ).
Autophagy is an important part of the intracellular quality control systems and has been directly linked to proteostasis disorders. The induction of autophagy can be achieved by inhibiting the mammalian target of rapamycin which negatively regulates autophagy ( 25 ). Treatment with rapamycin protected against neurodegeneration and the rapamycin analogue CCI-779 improved the performance in behavioural tasks and decreased the size and frequency of inclusions in a mouse model of Huntington disease ( 26 ).
In this study, we investigated the effect of these classes of drugs in a cellular model that mimics the gain of function and dominant negative mechanisms in ADRP patients. Pharmacological chaperones alleviated both the gain-of-function and the dominant-negative mechanisms of cell death through promoting the folding of P23H opsin. In contrast, other compounds alleviated the gain-of-function mechanism of cell death but did not reduce either the dominant-negative effect of the P23H opsin on the WT protein or promote mutant protein folding.
Development of cell model of rod opsin retinitis pigmentosa
A cell model was developed to investigate potential mechanisms of cell death in rod opsin RP and test the effect of manipulating rod opsin traffic and protein aggregation. This model was based on expression of the class II mutant P23H rod opsin or P23H-GFP rod opsin in human SK-N-SH neuroblastoma cells. SK-N-SH cells have been used extensively as a model for other protein aggregation diseases, e.g. polyglutamine expansions ( 27 ). As previously reported ( 9 , 10 ), P23H-GFP rod opsin was retained in the ER and formed intracellular inclusions, whereas WT-GFP trafficked to the plasma membrane ( Supplementary Material, Fig. S1 ). The incidence of inclusions in P23H-GFP transfected cells was 27% after 24 h compared to 3% for the WT-GFP rod opsin (Fig. 1 A and B). Inclusion incidence increased over time (data not shown) but cell death (see below) became a confounding factor after 24 h so this time-point was used for the inclusion and protein aggregation assays. The formation of intracellular inclusions can be used as a surrogate marker of protein aggregation ( 28 ), and protein aggregation was also measured directly through a sedimentation assay for rod opsin GFP. The increase in inclusion incidence correlated with an increase in the amount of insoluble rod opsin that was sedimentable, 61% of total P23H-GFP compared to 18% for WT-GFP after 24 h (Fig. 1 C). The expression of P23H rod opsin was also associated with a significant ( P ≤ 0.001) increase in cell death and apoptosis after 48 h. Cell death/cytolysis was measured by lactate dehydrogenase (LDH) release into the tissue culture supernatant, and for P23H, it was 0.52 absorbance units compared to 0.25 for the WT (Fig. 1 D). Apoptosis was assessed by caspase-3 activation, which was measured by the cleavage of pNA from DEVD-pNA. P23H expression led to increased caspase activation, 0.97 compared to 0.32 absorbance units for the WT control (Fig. 1 E). This model was used to test the effect of several classes of drug treatment on mutant rod opsin trafficking, aggregation and cell death.
Treatment with pharmacological chaperones
Previous studies have suggested that the rod opsin chromophore (11 -cis- retinal) and retinal analogues (e.g. 9 -cis- retinal) can act as pharmacological chaperones for rod opsin ( 9 , 16–18 ). Therefore, we confirmed this effect on rod opsin localization and investigated how this correlated with rod opsin aggregation and cell death associated with P23H rod opsin expression.
The most effective concentration, determined in terms of reduction of inclusion incidence, was 10 µM for both 9 -cis- retinal and 11 -cis- retinal. Treatment with 9 -cis- retinal reduced inclusion incidence from 27% ± 5 in the controls to 10% ± 7 ( P ≤ 0.005; Fig. 1 B). Treatment with 11 -cis- retinal reduced the inclusion incidence to 10% ± 2 ( P ≤ 0.001) of transfected cells (Fig. 1 B). In addition to a reduction in inclusion incidence, there was a significant change in rod opsin localization (Fig. 1 A) as 58% ± 10 ( P ≤ 0.001) of P23H-GFP cells treated with 9 -cis- retinal and 42% ± 11 ( P ≤ 0.001) of cells treated with 11 -cis- retinal showed predominant plasma membrane staining. In contrast, vehicle-treated P23H-GFP resulted in no cells with predominant plasma membrane staining ( Supplementary Material, Fig. S1 ). At these doses, retinoids had no effect on WT rod opsin ( Supplementary Material, Fig. S1 ). This effect was also not seen in cells expressing the rod opsin Schiff base mutant K296E, where no morphological changes were observed as a result of treatment with retinoids ( Supplementary Material, Fig. S2 ). These data suggest that direct binding to P23H rod opsin via the Schiff base is required for retinoids to mediate their effect on mutant opsin traffic and aggregation.
Following treatment with 9 -cis- retinal and 11 -cis- retinal, increased amounts of soluble EndoH resistant P23H rod opsin were detected by western blotting with the rod opsin C-terminal monoclonal antibody 1D4 ( Supplementary Material, Fig. S3 ). These data confirmed that the retinoids stimulated increased opsin translocation through the secretory pathway.
Retinoids reduced the percentage of insoluble P23H-GFP rod opsin from 61% ± 4 in the vehicle-treated controls to 39% ± 2 ( P ≤ 0.005) following treatment with 9 -cis- retinal and 32% ± 2 ( P ≤ 0.005) following treatment with 11 -cis- retinal (Fig. 1 C). These reduced levels of insoluble P23H rod opsin were in agreement with the reduction in inclusion incidence observed following retinoid treatment.
Importantly, both retinoids reduced P23H-GFP associated cell death. 9 -cis- retinal reduced LDH release from 0.52 ± 0.04 absorbance units to 0.17 ± 0.01 ( P ≤ 0.001), whereas 11 -cis- retinal was slightly less efficient, reducing LDH release to 0.32 ± 0.03 ( P ≤ 0.005) (Fig. 1 D). Treatment with either 9 -cis- retinal or 11 -cis- retinal reduced caspase activity to levels similar to those of the WT protein. Caspase activation levels were 0.97 ± 0.23 absorbance units in the vehicle-treated controls and this was reduced to 0.37 ± 0.14 ( P ≤ 0.001) by 9 -cis- retinal and 0.27 ± 0.10 ( P ≤ 0.005) by 11 -cis- retinal (Fig. 1 E).
The effect of the kosmotropes DMSO, TMAO, 4-PBA and trehalose on P23H rod opsin expression, localization, aggregation and cell death was assessed. DMSO had a positive effect on P23H-GFP rod opsin inclusion incidence (Fig. 2 A). The largest decrease in inclusion incidence was achieved using 0.25% DMSO which reduced inclusion incidence from 27% ± 6 in the untreated controls to 7% ± 1. This reduction corresponded to 24% ± 2 of the vehicle treated P23H-GFP ( P ≤ 0.001) (Fig. 2 B). Using concentrations higher than 0.25% DMSO failed to reduce the incidence of inclusions further and compromised cell morphology. Treatment with TMAO resulted in a dose-dependent decrease in inclusion incidence in cells expressing P23H-GFP rod opsin (Fig. 2 A). The most effective concentration of TMAO was 5 m m which resulted in a reduction of inclusion incidence to 24% ± 2 of vehicle treated levels ( P ≤ 0.001) (Fig. 2 B). The effect of 4-PBA on inclusion incidence in cells expressing P23H rod opsin was assessed (Fig. 2 B) and the concentration which resulted in the largest decrease in inclusion incidence was 10 m m . This led to a reduction in inclusion incidence to 24% ± 3 ( P ≤ 0.001) of vehicle treated controls (Fig. 2 A and B). Treatment with trehalose resulted in a milder decrease in inclusion incidence in cells expressing P23H rod opsin (Fig. 2 A and B), when compared to the other kosmotropes DMSO, TMAO and 4-PBA. The most effective concentration was 50 m m which resulted in a decrease of P23H-GFP inclusion incidence to 46% ± 6 ( P ≤ 0.001) of the untreated controls (Fig. 2 A and B). None of the compounds altered WT-opsin localization ( Supplementary Material, Fig. S1 ).
The effect of the kosmotropes on P23H-GFP solubility was confirmed by a sedimentation assay. Treatment with DMSO, TMAO, 4-PBA and trehalose did not affect the total levels of either WT or P23H rod opsin expressed in SK-N-SH cells (data not shown). In contrast, all of the compounds reduced the percentage of insoluble rod opsin (Fig. 2 C). Treatment with 0.25% DMSO and 5 m m TMAO reduced the percentage of insoluble P23H opsin from 60% ± 5 in the untreated controls to 41% ± 3 ( P ≤ 0.01) and 38% ± 4 ( P ≤ 0.01), respectively (Fig. 2 C). Ten millimolar 4-PBA and 50 m m trehalose appeared slightly more effective reducing the percentage of insoluble mutant opsin to 26% ± 1 ( P ≤ 0.005) and 26% ± 2, respectively ( P ≤ 0.005) (Fig. 2 C).
We investigated whether DMSO and TMAO could affect inclusion formation in cells expressing K296E mutant rod opsin. In contrast to retinoids, treatment with DMSO resulted in a decrease in inclusion incidence in K296E cells to 40% ± 1 of vehicle treated controls ( P ≤ 0.001) ( Supplementary Material, Fig. S2 ). A similar reduction in inclusion incidence was observed with TMAO as it resulted in a decrease to 40% ± 8 of vehicle treated controls ( P ≤ 0.001).
Despite the effects on inclusion incidence and aggregation, none of the kosmotropes tested appeared to promote the translocation of P23H-GFP rod opsin to the plasma membrane (Fig. 2 A). The effect of these kosmotropes on the levels of soluble and EndoH resistant P23H rod opsin species was analysed by SDS–PAGE followed by immunoblotting with 1D4 ( Supplementary Material, Fig. S3 ). Treatment with DMSO, TMAO, 4-PBA and trehalose did not increase the level of soluble P23H rod opsin detected by 1D4. Furthermore, none of the kosmotropes promoted increased EndoH resistance of mature opsin species indicating that these compounds were unable to assist the folding of the mutant protein so it could exit from ER to the Golgi.
The effects of kosmotropes on cell death and apoptosis were investigated. Significantly, all the kosmotropes reduced LDH release in P23H expressing cells with DMSO ( P ≤ 0.01) and TMAO ( P ≤ 0.01), reducing LDH release by ∼60% when compared with the untreated P23H controls, whereas 4-PBA ( P ≤ 0.05) and trehalose ( P ≤ 0.05) were slightly less effective, reducing LDH release only by ∼40% (Fig. 2 D). Kosmotropes also had a positive effect in protecting P23H-GFP expressing cells against caspase activation as DMSO ( P ≤ 0.05), TMAO ( P ≤ 0.05), 4-PBA ( P ≤ 0.001) and trehalose ( P ≤ 0.05) all reduced caspase-3 activity by ∼30% when compared with the vehicle-treated P23H controls (Fig. 2 D). Therefore, kosmotropes appear to counteract some of the dominant gain of function properties of mutant opsin expression without enhancing the folding of the mutant opsin. These data are summarized in Table 1 .
|Compound||Inclusions||EndoH resistance||Insoluble protein||LDH release||Caspase-3 activity||Dominant-negative|
|9- cis -retinal||++||Yes||++||+++||+++||+++|
|11- cis -retinal||++||Yes||++||++||+++||+++|
|Compound||Inclusions||EndoH resistance||Insoluble protein||LDH release||Caspase-3 activity||Dominant-negative|
|9- cis -retinal||++||Yes||++||+++||+++||+++|
|11- cis -retinal||++||Yes||++||++||+++||+++|
Molecular chaperone inducers
Celastrol caused a moderate decrease in inclusion incidence in P23H-transfected cells (Fig. 3 A and B). The most efficient concentration at reducing inclusion incidence was 2 µM which led to a reduction of inclusion incidence to 88% ± 13 of the untreated control P23H-GFP cells ( P ≤ 0.05) (Fig. 3 B). However, despite this moderate reduction in inclusion incidence treatment with 2 µM celastrol appeared to have a detrimental rather than a beneficial effect on the P23H expressing cells as observations using fluorescent and confocal microscopy indicated a high level of dead or abnormally shaped cells when compared with the untreated controls (Fig. 3 A). The levels of abnormal cells were reduced when lower concentrations of celastrol were used; however, at these doses celastrol did not appear to reduce inclusion incidence.
The use of the prototypical Hsp90 inhibitor geldanamycin resulted in a reduction of inclusion incidence in cells expressing P23H rod opsin (Fig. 3 A). The concentration of geldanamycin with the greatest effect on inclusion incidence was 2 µM which resulted in a decrease in inclusion incidence to 63% ± 18 of untreated control levels ( P ≤ 0.05) (Fig. 3 B). Treatment with another Hsp90 inhibitor, radicicol, resulted in a reduction of inclusion incidence in cells expressing P23H rod opsin (Fig. 3 B) to levels similar to those achieved with geldanamycin. The largest decrease in inclusion incidence was observed when using 5 µM radicicol which caused the inclusion incidence to decrease to 37% ± 13 of the untreated controls ( P ≤ 0.001). 17-AAG led to a dose-dependent reduction of inclusion incidence in cells expressing P23H rod opsin ( Supplementary Material, Fig. S3 ). 17-AAG had maximal effect at a concentration of 1 µM and reduced inclusion incidence to 29% ± 5 of the untreated control level ( P ≤ 0.001). Therefore, 17-AAG appeared to be more effective at reducing inclusion incidence than other Hsp90 inhibitors tested.
Geldanamycin, radicicol and 17-AAG, but not celastrol, had a positive effect in reducing the amount of insoluble rod opsin. 17-AAG and radicicol were the most efficient and resulted in a reduction of insoluble P23H rod opsin from 59% ± 2 in the untreated controls to 26% ± 1 ( P ≤ 0.01), and 30% ± 3 ( P ≤ 0.01), respectively. Geldanamycin was slightly less efficient at reducing the levels of insoluble opsin to 44% ± 1 ( P ≤ 0.001).
Despite the decrease in inclusion incidence geldanamycin, radicicol and 17-AAG did not appear to promote the translocation of the mutant protein to the plasma membrane (Fig. 3 A). Treatment with geldanamycin and radicicol did not increase the levels of soluble P23H protein and did not appear to promote EndoH resistance of soluble mutant protein species as detected by western blotting. Treatment with 17-AAG appeared to mediate a small increase in the levels of soluble P23H protein, although it did not promote detectable levels of EndoH resistant P23H rod opsin species ( Supplementary Material, Fig. S3 ).
The most effective molecular chaperone inducers at reducing cell death were radicicol ( P ≤ 0.01) and 17-AAG ( P ≤ 0.01) which reduced LDH release back to WT levels (Fig. 3 D). Geldanamycin significantly reduced LDH release albeit to a lesser degree and celastrol did not reduce LDH release confirming initial morphological observations that it was not beneficial to P23H cells. Geldanamycin, radicicol and 17-AAG were equally effective at reducing caspase activation in P23H-GFP expressing cells (Fig. 3 E). Celastrol, on the other hand, increased caspase activity levels by ∼20% in agreement with previous morphological observations. These data are summarized in Table 1 .
Stimulation of macroautophagy
Misfolded class II rod opsin is degraded by the proteasome ( 9 , 10 ), another major pathway for protein and organelle degradation in eukaryotic cells is autophagy, which is stimulated by rapamycin. Treatment with different concentrations of rapamycin resulted in a dose-dependent reduction of inclusion incidence in cells expressing P23H-GFP rod opsin ( Supplementary Material, Fig. S4 ). The most effective concentration of rapamycin at reducing inclusion incidence was 1 µM which resulted in a decrease in inclusion incidence to 42% ± 6 of the untreated control levels ( P ≤ 0.005) (Fig. 4 A and B). Despite the reduction of inclusion incidence, rapamycin did not appear to promote the translocation of P23H-GFP rod opsin to the plasma membrane (Fig. 4 A).
The autophagic clearance induced by rapamycin of other autophagy substrates, such as mutant huntingtin and mutant α-synuclein transfected into COS-7 cells, was further enhanced when trehalose was used in conjunction with rapamycin ( 29 ). Therefore, trehalose was added to the rapamycin treatments to test if there was any additional effect. The addition of trehalose with rapamycin resulted in a slight enhancement of the rapamycin effect (Fig. 4 A and B). The most effective concentrations tested were 5 µM rapamycin and 100 m m trehalose and resulted in a decrease in inclusion incidence to 25% ± 2 of the untreated controls ( P ≤ 0.005). Interestingly, the optimal concentrations tolerated were higher than when the compounds were used together than when they were used separately and this may reflect a mutual protective effect of the compounds.
The percentage of insoluble P23H-GFP rod opsin was reduced by treatment with rapamycin from 61% ± 3 in the untreated controls to 39 ± 5, whereas rapamycin plus trehalose was more efficient, reducing the percentage of insoluble mutant protein to 32% ± 3 (Fig. 4 C).
LDH release and caspase activity were measured in cells expressing P23H-GFP rod opsin treated with rapamycin or rapamycin plus trehalose. Treatment with rapamycin or rapamycin plus trehalose reduced LDH release in P23H-GFP rod opsin expressing cells to levels similar to those observed with the WT protein (Fig. 4 D). These treatments were also effective at reducing caspase activity (Fig. 4 E). Rapamycin reduced caspase activity from 0.80 ± 0.18 in the untreated controls to 0.56 ± 0.05, whereas rapamycin plus trehalose reduced caspase activity to 0.51 ± 0.02. The degree of protection against caspase activation by these compounds was similar to that observed with kosmotropes and molecular chaperones inducers, with the exception of celastrol, but was less than that seen with retinoids (Table 1 ).
Modulation of dominant negative effects
In order to test the effect of these drugs on the potential dominant-negative effects of the mutant rod opsin on the WT protein, equal amounts of untagged WT rod opsin and WT-GFP or untagged WT opsin and P23H-GFP were expressed in SK-N-SH cells. Cells were fixed after 24 h and immunostained with the anti-rhodopsin antibody 1D4. This antibody only detected the untagged rod opsin species as the GFP-tag blocks the antibody epitope at the C-terminus of rod opsin. When cells expressed untagged and GFP-tagged WT rod opsin, the vast majority of the cells showed predominantly plasma membrane staining (Fig. 5 A). In contrast, when WT opsin was expressed with P23H-GFP opsin, there was a change in the localization of the WT protein. Co-transfected cells showed a co-localization of the two rod opsin species in the ER (arrowhead) or in large intracellular inclusions (asterisk), confirming earlier reports that the mutant rod opsin can prevent the normal translocation of the WT protein ( 9 , 13 ).
When the rod opsin localization was quantified in co-transfected cells, ∼65% showed ER staining and ∼15% had both opsin species accumulated in intracellular inclusions (Fig. 5 A). However, some co-transfected cells (∼15%) had a predominantly plasma membrane staining suggesting that the effect of P23H rod opsin on WT rod opsin, while significant, was not absolute. The ER retention coupled to enhanced degradation and recruitment to the large intracellular inclusions correlated with a reduction in the levels of soluble WT rod opsin that were observed in an immunoblot with 1D4 when the WT protein was expressed with P23H-GFP compared to WT-GFP ( Supplementary Material, Fig. S3 ).
Importantly, treatment with 10 µM 9 -cis- retinal (Fig. 5 A and B) or 10 µM 11 -cis- retinal (Fig. 5 A and B) appeared to suppress this effect. An increased number of co-transfected cells had the WT protein present on the plasma membrane. 9 -cis- retinal stimulated an increase in the percentage of co-transfected cells with predominant plasma membrane localization of the WT protein from 15% ± 4 in the vehicle-treated controls to 57% ± 5 ( P ≤ 0.001). 11 -cis- retinal also alleviated the dominant-negative effect of the P23H rod opsin on the WT protein, albeit not to the extent of 9 -cis- retinal, as 42% ± 9 ( P ≤ 0.005) of the co-transfected cells had WT rod opsin on the plasma membrane. These results correlated with the effect of retinoids on P23H rod opsin alone, where 9 -cis- retinal and 11 -cis- retinal promoted the translocation of the P23H protein to the plasma membrane of 58 and 42% of the cells, respectively. In addition, both retinoids reduced the percentage of co-transfected cells with intracellular inclusions containing the WT protein from 15% ± 3 in the vehicle-treated controls to 8% ± 2 ( P ≤ 0.01) after treatment with 9 -cis- retinal and 10% ± 4 after treatment with 11 -cis- retinal (not significant).
In contrast, kosmotropes (Fig. 5 C), chaperone inducers (Fig. 5 D) or rapamycin and trehalose (Fig. 5 E) did not appear to promote a major increase in the translocation of WT rod opsin to the plasma membrane in the dominant-negative context. However, a small, but significant, increase in the percentage of co-transfected cells showing predominant WT plasma membrane staining was observed with geldanamycin, radicicol, 17-AAG and rapamycin. Importantly, all of the drugs, with the exception of celastrol, significantly reduced the incidence of WT protein inclusion formation in the presence of P23H-GFP as observed for the expression of P23H-GFP alone. Therefore, there was a correlation between the ability of the drugs to stimulate mutant protein folding and the suppression of the dominant negative effect.
This study provides ‘proof of principle’ that P23H mutant rod opsin can be pharmacologically manipulated to alleviate toxic gain of function and dominant negative effects. Rod opsin misfolding presents an excellent paradigm to study the role of protein misfolding in neurodegeneration. The function of the protein is well defined, the structure is known and a range of animal models exist. This offers several advantages over other forms of neurodegeneration where frequently the normal function or structure of the disease protein is unknown. The knowledge of rod opsin structure and function can, therefore, be exploited to illuminate the mechanisms of cell death associated with protein misfolding and therapeutic approaches.
This study confirms earlier reports that 9 -cis- retinal and 11 -cis- retinal can act as pharmacological chaperones to increase the amount of P23H rod opsin that can exit the ER and transit to the plasma membrane ( 9 , 16 , 18 ) to form functional rhodopsin ( 18 ). Importantly, here we show that this improvement in folding had a positive effect on both the gain-of-function and dominant-negative mechanisms induced by the P23H rod opsin. For the gain of function mechanism of P23H rod opsin, this was demonstrated by a reduction in inclusion incidence, reduced protein aggregation, the presence of mature glycosylated rod opsin species and protection against caspase activation and cell death. The ER retention of the WT rod opsin by P23H was suppressed resulting in the increased presence of WT protein in the plasma membrane of co-transfected cells and a reduction in inclusions which contained the WT protein showing a reduction of the dominant-negative effects.
In contrast, kosmotropes, chaperone inducers and autophagy inducers were effective at reducing rod opsin aggregation and inclusion formation, but did not appear to enhance correct rod opsin folding and transit through the secretory pathway, or alleviate the dominant negative effect on the WT protein. The most effective drugs in each class (4-PBA, 17-AAG and rapamycin plus trehalose) were more effective than retinoids at reducing rod opsin aggregation and inclusion formation and this correlated with a reduction of cell death and caspase activation close to levels observed with the WT protein alone. Therefore, the toxic gain of function effects can be alleviated without correcting the protein folding defect but by reducing protein aggregation and promoting degradation of the aggregation prone species.
The correction of folding for a mutated polytopic transmembrane protein, such as rod opsin, may be challenging and require a specific small molecule to stabilize a quasi-native conformation and shift the folding equilibrium towards the native state. Retinoids that interact with multiple transmembrane helices in the core of the opsin molecule appear ideal for this purpose. The potential use of these compounds in ADRP patients is questionable, however, as they lose efficacy when bleached, and could also affect cone opsin function. Furthermore, it is likely that in patients with a healthy diet, there would be an adequate supply of 11 -cis- retinal to the photoreceptors produced by the enzymes and isomerases of the visual cycle. Dietary supplements may not enhance this further. Trials have already been conducted based on giving high doses of vitamin A to RP patients ( 30 ). A randomized, controlled clinical trial in RP patients receiving 15000 IU/day of vitamin A resulted in a slower rate of decline of retinal function measured by the ERG than the two groups not receiving this dosage. However, no significant difference was observed in the loss of visual field with time between the groups of patients receiving vitamin A or placebo. Although prolonged daily consumption of <25000 IU of vitamin A was considered safe in a clinical trial of adults aged 18–54 with RP ( 31 ), other studies reported side-effects such as increased intracranial pressure, hepatomegaly and elevated blood lipids with these doses of vitamin A ( 32 ). There was no available genotype information on the RP patients recruited to this clinical trial and differences in the genetic basis of RP may account for the high patient variability observed in the study.
The availability of 11 -cis- retinal within photoreceptors highlights that in vivo the folding efficiency of class II mutant rhodopsins may be much higher than cell-based assays of the rod opsin apoprotein. In addition, rod photoreceptors produce millions of rod opsin molecules each day and may have developed specialized chaperones to facilitate rod opsin biogenesis. We would also predict that dominant negative effects of class II rod opsins would be reduced by insertion of 11 -cis- retinal early in biosynthesis and the dominant negative effect may not be revealed without light exposure that would bleach the 11- cis -retinal to all- trans -retinal. Similarly, light exposure could have a deleterious effect on rod opsin folding by reducing the amount of 11 -cis- retinal that is available during the biosynthesis and traffic of the mutant opsin and may, in part, explain the protective effects that have been associated with dark rearing of Xenopus P23H rod opsin models ( 33 ). Another consideration is that high 11 -cis- retinal may even be detrimental. For example, polymorphisms in RPE65 that have lower levels of isomerase activity ( RPE65450Met ) and, hence reduced 11- cis -retinal, have slower disease progression for class II rod opsin mutations ( 34 ).
Non-isomeraseable retinoid analogues may also be an attractive therapeutic goal. For example, Noorwez et al. ( 17 ) successfully rescued P23H rod opsin with a ‘locked’ form of retinal, 11 -cis- 7-ring retinal. Addition of 11 -cis- 7-ring retinal to cell culture increased correctly folded P23H protein. This type of compound, however, must effectively discriminate between rod and cone opsins, or risk adversely affecting residual cone function in RP patients. Therefore, in order to promote the correct folding of class II rod opsin mutations, the identification of other small molecules that can act as pharmacological chaperones to stabilize folding is needed. A recent study has shown that NSC45012 (1-(3,5-dimethyl-1H-pyrazol-4-yl)ethanone), a weak inhibitor of opsin regeneration, could stimulate a 40% increase in folded P23H opsin ( 35 ). Consequently, it is likely that other small molecules that bind close to the ligand-binding pocket or in the intradiscal domain can be identified that could be applied to treat patients.
Nevertheless, the data suggest it may not be necessary to improve rod opsin folding to improve photoreceptor viability. Importantly, drugs that reduced aggregation also improved cell viability and this is likely to be further enhanced in the presence of 11 -cis- retinal in vivo . They may also have a broader applicability. For example, kosmotropes, unlike retinoids, also reduced inclusion incidence in cells expressing K296E rod opsin which did not respond to retinoids. There is also evidence that these compounds may be clinically applicable.
4-PBA (Buphenyl) is a low molecular weight fatty acid with FDA approval for clinical use in patients with urea cycle disorders ( 36 ). It was also the most effective kosmotrope in our study, which might suggest that it is the most applicable kosmotrope to rhodopsin RP patients. 4-PBA induced CFTR channel function on the plasma membrane of ΔF508-CFTR airway epithelial cells in vitro ( 37 ) and was tested in a randomized, double-blind and placebo-controlled trial of 18 patients with mutations in CFTR ( 19 ). Subjects in the 4-PBA group demonstrated small, but statistically significant improvements of the nasal potential difference (NPD) response patterns, but did not demonstrate significantly reduced sweat chloride concentrations or alterations in the amiloride-sensitive NPD. Side effects due to drug therapy were minimal and comparable in the two groups. Therefore, doses of 4-PBA needed to affect protein aggregation may be well tolerated in patients. Furthermore, as the major effect of 4-PBA in our study was to suppress rod opsin aggregation but not induce rod opsin folding, this kosmotrope may be more successfully targeted against a toxic gain of function disease as opposed to enhancing protein folding in recessive disease, such as cystic fibrosis.
Pharmacological intervention with compounds that cause the induction of heat shock proteins and molecular chaperones has been suggested to have an enormous therapeutic potential for protein misfolding diseases. Late onset neurodegenerative misfolding diseases may be triggered by an imbalance between the cellular chaperone machinery and the amount of misfolded protein species, and this approach may address this imbalance. In this study, three inducers of molecular chaperones (geldanamycin, radicicol and 17-AAG) were identified as potential pharmacological candidates for patients with class II rhodopsin RP, as they were able to alleviate the gain of function mechanisms induced by P23H rod opsin. Despite their very different Kd for Hsp90, they were all effective at similar concentrations, highlighting that bioavailability was important for efficacy in vitro and will also be critical in vivo . Our data would suggest that 17-AAG is the most appropriate candidate. 17-AAG is currently in clinical trials as an anti-tumour agent and it may be that its use can be adapted at lower doses for diseases of proteostasis. Alternatively, other Hsp90 inhibitors may be better tolerated or have improved bioavailability, and other mechanisms of inducing or co-inducing chaperones may be more appropriate.
The stimulation of autophagy by rapamycin was previously found to have positive effects on cell and animal models of Huntington disease ( 29 , 38 ). In our study, rapamycin was effective against the toxic gain of function but not the dominant-negative effects of mutant rod opsin. Kaushal ( 39 ) reported that rapamycin appeared to enhance degradation of P23H rod opsin and stimulate co-localization of P23H rod opsin with the autophagy marker proteins, Atg7, Atg8 (LC3) and LAMP-1 ( 39 ). Therefore, stimulation of autophagic clearance of rod opsin aggregates may also be a viable therapeutic option for rhodopsin RP patients. We observed that the effect of rapamycin was enhanced by the addition of trehalose. Similarly, combination therapy with drugs that target different pathways in the cellular response to misfolded protein and stimulation of improved folding by pharmacological chaperones may be the most effective approach to treat rhodopsin RP.
This study has identified a range of compounds that target different cellular pathways to combat the gain of function and dominant-negative mechanisms of cell death induced by a class II rod opsin mutation. It is now important to establish the efficacy of these pharmacological agents in animal models prior to a potential translation into the clinic.
MATERIALS AND METHODS
9 -cis- retinal, dimethyl sulfoxide (DMSO), trimethylamine N-oxide (TMAO), trehalose, 17-Allylamino-17-demethoxygeldanamycin (17-AAG), radicicol, rapamycin and 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) were purchased from Sigma. 11 -cis- retinal was a gift from Professor Paula Booth (University of Bristol). Celastrol was from Cayman Chemical Company. Lipofectamine and plus reagent were from Invitrogen. EndoH and PNGaseF were from New England Biolabs. The rhodopsin monoclonal antibody 1D4 was a gift of Professor Robert Molday (University of British Columbia). Cy3 goat anti-mouse Cy3 conjugated secondary antibody was from Jackson Immuno Research Laboratories and the goat anti-mouse antibody conjugated to horseradish peroxidase was from Pierce. Rod opsin expression plasmids were as previously described ( 9 ).
SK-N-SH human neuroblastoma cells were acquired from the European Collection of Cell Cultures (ECACC) and used at low passage numbers (<P20). Cells were grown in DMEM/F12 with Glutamax-I+10% heat inactivated foetal bovine serum (FBS) and 100 units/ml penicillin and 100 µg/ml streptomycin with an atmosphere of 6% CO 2 at 37°C. Twenty-four hours after seeding eight-well permanox chamber slides with 2.5 × 10 4 cells per well, the cells were transfected with 0.2 µg DNA per well+0.5 µl lipofectamine+1 µl plus reagent in a serum-free media for 3 h according to the manufacturer’s instructions. After 3 h, 10% FBS cell culture media was re-established. All drug treatments were added at this stage. Inclusion incidence was measured by scoring 400 transfected cells from three separate experiments for the presence of inclusions by an observer blind to the transfection conditions and drug treatments. Rod opsin localization analysis in co-transfected cells was obtained by scoring the predominant localization of WT-opsin in 400 co-transfected cells from three separate experiments. Means and standard errors were calculated to two decimal points and have been rounded up or down to whole numbers for presentation in the text. Example images showing representative control, vehicle and drug-treated WT-opsin and P23H opsin localization are presented in Supplementary Material, Figure S1 . For 96- and 6-well dishes, a similar protocol was followed and the transfection conditions and the amount of DNA were scaled down/up. The cells were transfected 24 h after seeding with 5 × 10 5 cells per well. Paired and unpaired Student’s t -tests were carried out where appropriate. The sets of values were considered statistically significant if P ≤ 0.05.
Fluorescence and immunofluorescence microscopy
SK-N-SH cells transfected with GFP-tagged opsin were washed three times with 4°C PBS (PBS: 137 m m NaCl, 2.7 m m KCl, 8.1 m m Na 2 HPO 4 , 1.5 m m KH 2 PO 4, pH 7.3 at 25°C) 24 h after transfection, followed by fixation with 3.7% paraformaldehyde for 15 min at 22°C. Cells were then washed four times in PBS with DAPI (2 µg/ml) for 5 min in the third wash.
Twenty-four hours after transfection, cells were washed three times with 4°C PBS and fixed using 3.7% paraformaldehyde for 15 min followed by two washes with PBS. Permeabilization was carried out using 0.1% Triton X-100 for 5 min, followed by two washes in PBS. Prior to the incubation with antibodies, the wells were blocked for 30 min at 22°C with a blocking solution consisting of 3% bovine serum albumin and 10% normal donkey serum in PBS. Cells were incubated with 1D4 rhodopsin antibody (0.5 µg/ml) in blocking solution for 1 h at 22°C followed by three washes in PBS and incubation for 1 h at 22°C with donkey anti-mouse Cy3 conjugated secondary antibody (1:100) in blocking solution. Cells were consequently washed four times in PBS and DAPI (2 µg/ml) in PBS for 5 min in the third wash. Fluorescence and immunofluorescence were visualized using a Zeiss LSM 510 laser scanning confocal microscope. The following excitation/emission conditions were used in separate channels with the x63 oil immersion objective: DAPI 364/475–525 nm, GFP 488/505–530 nm and Cy3 543/560 nm.
Preparation of cell extracts for western blotting
Twenty-four hours after transfection, cells were washed twice with 4°C PBS and lysed using 190 µl 1% Triton X-100 in PBS+10 µl protease inhibitor cocktail. Cells lysates were scraped and collected in 1.5 ml microcentrifuge tubes on ice prior to centrifugation at 17500 g for 30 min at 4°C. Soluble cell lysates were normalized for total protein and 10 µg of protein mixed with an equal volume of 2x Laemmli sample buffer (1x sample buffer: 50 m m Tris–HCl pH 6.8, 2% w/v SDS, 10% v/v Glycerol, 2.5% v/v 2-mercaptoethanol and 0.1% w/v bromophenol blue) prior to being separated by 10% SDS–PAGE and electroblotted onto nitrocellulose. Immunodetection of opsin was carried out by using 1D4 rhodopsin antibody (1.33 µg/ml) followed by goat anti-mouse HRP (1:30 000) in PBS+1% Marvel+0.1% Tween-20. The ECL plus chemiluminescent detection reagent was used according to the manufacturer’s instructions. The mobility of different opsin glycoforms were determined by digesting 10 µg of soluble cell lysates using EndoH (500 units) and PNGaseF (500 units) for 2 h at 37°C.
Rod opsin sedimentation solubility assay
SK-N-SH cells were seeded on six-well dishes and transfected with GFP-tagged opsin as described above. Twenty-four hours after transfection, cells were lysed with 190 µl of RIPA buffer+10 µl of protease inhibitor cocktail (RIPA buffer: 50 m m Tris–HCl pH 8.0, 150 m m w/v NaCl, 1 m m EDTA, 1% v/v NP-40, 0.1% w/v SDS and 0.05% w/v sodium deoxycholate). Soluble and insoluble fractions were obtained by centrifugation at 17 500 g for 30 min at 4°C followed by removal of the supernatant and resuspension of the pellet in 190 µl of RIPA buffer+10 µl protease inhibitor cocktail. Fifty microgram of total protein was loaded onto a 96-well dish and the volume of each condition was normalized to 200 µl with RIPA buffer. GFP fluorescence was measured by using a Tecan Safire microplate reader with 488/509 nm emission/excitation wavelengths and a bandwidth of ± 2.5 nm. Paired and unpaired Student’s t -tests were carried out. The sets of values were considered statistically significant if P ≤ 0.05.
Lactate dehydrogenase and caspase-3 activation assays
Quantification of cytotoxicity/cytolysis associated with mutant opsin expression was based on the measurement of cytoplasmic LDH release into the tissue culture supernatant from damaged cells. SK-N-SH cells were seeded on a 96-well dish at a concentration of 5 × 10 3 cells/well. Twenty-four hours later, cells were transfected with GFP-tagged opsin and the LDH assay was carried out 48 h after transfection according to the manufacturer’s instructions (Roche). LDH activity was determined in a coupled enzymatic reaction that reduced a tetrazolium salt to formazan dye. LDH enzyme activity in the supernatant increased as the number of dead cells (or cells with damaged plasma membranes) increased. The increase in supernatant LDH activity directly correlated to the amount of formazan formed over time which was measured by absorbance at 492 nm using a Tecan Safire microplate reader. LDH activity was expressed in absorbance units.
Caspase-3 activation assay
Activation of apoptosis was measured by assessing the activation of caspase-3 using a colorimetric assay based on the protocol of Novoselova et al . ( 40 ). SK-N-SH cells were seeded in a six-well dish at a concentration of 5 × 10 5 cells/well followed by transfection with GFP-tagged opsin 24 h later. Twenty-four or 48 h later, transfection cells were washed twice with 4°C PBS followed by lysis with 190 µl buffer A (25 m m HEPES pH 7.5, 1 m m EDTA, 5 m m EGTA, 50 m m NaCl)+10 µl protease inhibitor cocktail. Hundred microlitre of buffer B (50 m m HEPES pH 7.5, 20% v/v glycerol, 1 m m EDTA, 5 m m DTT containing 200 µM DEVD-pNA caspase-3 substrate (Calbiochem)) was added to 50 µg of total protein followed by incubation for 4 h at 37°C. Cleavage of pNA, and hence caspase activity, was measured by the increase in absorbance at 405 nm using a Tecan Safire microplate reader. For both the LDH and caspase-3 assays, experiments were carried out at least three times in triplicates ( n = 3) and the data shown are representative of one experiment. Paired and unpaired Student’s t -tests were carried out where appropriate. The sets of values were considered statistically significant if P ≤ 0.05.
This work was supported by the British Retinitis Pigmentosa Society.
We would like to thank Professor R. Molday (University of British Columbia) for the gift of the 1D4 antibody and Professor P. Booth (University of Bristol) for the gift of 11 -cis- retinal. We also thank Alison Hardcastle, Maria Kosmaoglou and John Bett for critical reading of the manuscript.
Conflict of Interest statement. None declared.