Pharmacological Inhibition of the VCP/Proteasome Axis Rescues Photoreceptor Degeneration in RHOP23H Rat Retinal Explants

Rhodopsin (RHO) misfolding mutations are a common cause of the blinding disease autosomal dominant retinitis pigmentosa (adRP). The most prevalent mutation, RHOP23H, results in its misfolding and retention in the endoplasmic reticulum (ER). Under homeostatic conditions, misfolded proteins are selectively identified, retained at the ER, and cleared via ER-associated degradation (ERAD). Overload of these degradation processes for a prolonged period leads to imbalanced proteostasis and may eventually result in cell death. ERAD of misfolded proteins, such as RHOP23H, includes the subsequent steps of protein recognition, targeting for ERAD, retrotranslocation, and proteasomal degradation. In the present study, we investigated and compared pharmacological modulation of ERAD at these four different major steps. We show that inhibition of the VCP/proteasome activity favors cell survival and suppresses P23H-mediated retinal degeneration in RHOP23H rat retinal explants. We suggest targeting this activity as a therapeutic approach for patients with currently untreatable adRP.


Introduction
Proteostasis requires a complex network of cellular factors, including protein synthesis, folding, and degradation [1]. Protein synthesis is not an error-free process, and approximately 5% of newly translated proteins contain a sequence error, which can induce protein misfolding and aggregation [2]. Cells are equipped with different physiological mechanisms that prevent or deal with stress caused by misfolded proteins. However, a prolonged imbalance between the folding demand and the cell folding capacity will disrupt cellular fitness, cause disease phenotypes, and eventually lead to cell death.
Retinitis pigmentosa (RP) is a group of inherited retinal degenerations (IRDs) characterized by a progressive loss of photoreceptor cells which ultimately leads to blindness [3,4]. In autosomal dominant RP (adRP), the most common genetic factors are linked to point mutations in the rhodopsin (RHO) gene, with known biochemical defects affecting synthesis, folding, or transport of RHO to the outer segment (OS) and initiation of disease despite the presence of wild-type (WT) RHO protein [5][6][7][8]. Since RHO is the most abundant protein in photoreceptors-making up 25% of total rod cell protein [6]-it is easy to Step 1: RHO recognition. RHO P23H (blue) is a misfolded glycoprotein that is first recognized by ER-resident chaperones, such as Hsp90, BiP, and PDI, to start the refolding cycle. The prolonged presence of these mutant forms in the ER activates the unfolded protein response (UPR).
Step 2: RHO targeting. (a) After the failure to refold, misfolded RHO is trapped in the quality control/folding cycle and becomes a target for ERM1 to remove mannose residues. (b) After mannose removal, misfolded RHO is targeted to the ERAD retrotranslocation site by EDEM1, a facilitator of glycoprotein disposal from the ER.
Step 3: RHO retrotranslocation. RHO P23H is ubiquitinated and presented to the VCP-Ufd1-Npl4 complex, recruited to the retrotranslocation site by the ER membrane proteins, and retrotranslocated into the cytosol.
Step 4: RHO proteasomal targeting and degradation. The VCP complex delivers mutant RHO proteins to the proteasome for their degradation. Red arrows show different inhibition targets to mod- Step 1: RHO recognition. RHO P23H (blue) is a misfolded glycoprotein that is first recognized by ER-resident chaperones, such as Hsp90, BiP, and PDI, to start the refolding cycle. The prolonged presence of these mutant forms in the ER activates the unfolded protein response (UPR).
Step 2: RHO targeting. (a) After the failure to refold, misfolded RHO is trapped in the quality control/folding cycle and becomes a target for ERM1 to remove mannose residues. (b) After mannose removal, misfolded RHO is targeted to the ERAD retrotranslocation site by EDEM1, a facilitator of glycoprotein disposal from the ER. Step 3: RHO retrotranslocation. RHO P23H is ubiquitinated and presented to the VCP-Ufd1-Npl4 complex, recruited to the retrotranslocation site by the ER membrane proteins, and retrotranslocated into the cytosol.
In the present study, we investigated and compared the pharmacological modulation effects of ERAD at four different major steps. These pathways were modulated using specific inhibitors in well-characterized RHO P23H transgenic rat organotypic retinal cultures. We utilized geldanamycin (GA) to inhibit the protein-folding mechanism in the ER by interrupting Hsp90 activity, the mannosidase inhibitor kifunensine (KIF) to block the processing of glycoproteins by ERM1, NMS-873 to interfere with the retrotranslocation of protein by inhibiting VCP ATPase activity, and the 26S proteasome inhibitor bortezomib (BO) to suppress proteasomal degradation. We show here that targeting the VCP/proteasome activity favors cell survival and suppresses degeneration in RHO P23H rat retinal explants. Therefore, we suggest targeting this activity as a therapeutic approach for patients with currently untreatable adRP.

Results
To analyze the impact of interference with different steps of ERAD on photoreceptor degeneration, we employed primary retinal culture from RHO P23H transgenic rats as a serum-free organotypic culture system preserving the entire mounted flat retina with its adherent RPE and the retinal ganglion cell layer uppermost [21]. RHO P23H transgenic rats, a genetically engineered RHO mutant model, closely mimic the expression pattern of the disease in humans. The RHO P23H rat presents an early and quick progression with a degeneration peak in heterozygous animals at postnatal day (PN) 15 [22]. We explanted RHO P23H rat retinae at PN9 and cultured them for six days until the peak of degeneration at PN15. Photoreceptors undergoing cell death were identified by the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay and calculation of the percentage of positive cells compared to the total outer nuclear layer (ONL) cell nuclei.

Hsp90 or ERM1 Inhibition Is Detrimental for RHO P23H Rat Retinae Ex Vivo
Recognition and targeting of misfolded proteins are the first two steps, determining their fate in the ERAD process. Therefore, we initially investigated whether the inhibition of key elements in these two steps will affect the degenerative course in the RHO P23H retinae. To inhibit the protein recognition, we selected geldanamycin (GA), which binds competitively with ATP to the N-terminal ATP-binding site of Hsp90 [23]. As an inhibitor for the second step of ERAD (protein targeting-mannose trimming), we used kifunensine (KIF), a class I α-mannosidase inhibitor, to inhibit ERM1 [24,25]. Organotypic retinal cultures allowed us to screen different GA and KIF doses, which were chosen according to several in vitro assays, as reported (Table S1), to inhibit Hsp90 and ERM1 proteins, respectively. After treating the organotypic retinal cultures with either GA or KIF, we evaluated the effect by calculating the percentage of TUNEL positive cells in the ONL of treated RHO P23H retinae.
GA treatment at 0.01 µM and 0.1 µM for 6 days did not protect the retinas. No significant effect in the percentage of TUNEL-positive cells in the ONL of RHO P23H rats was observed. Contrarily, retinae treated with 1 µM GA exhibited increased degeneration with a significant increase in the number of dying photoreceptor cells (Vehicle: 3.844% ± 1.2; 1 µM GA: 8.313% ± 2.7, p < 0.05; Figure 2A,B).

Hsp90 or ERM1 Inhibition Is Detrimental for RHO P23H Rat Retinae Ex Vivo
Recognition and targeting of misfolded proteins are the first two steps, determining their fate in the ERAD process. Therefore, we initially investigated whether the inhibition of key elements in these two steps will affect the degenerative course in the RHO P23H retinae. To inhibit the protein recognition, we selected geldanamycin (GA), which binds competitively with ATP to the N-terminal ATP-binding site of Hsp90 [23]. As an inhibitor for the second step of ERAD (protein targeting-mannose trimming), we used kifunensine (KIF), a class I α-mannosidase inhibitor, to inhibit ERM1 [24,25]. Organotypic retinal cultures allowed us to screen different GA and KIF doses, which were chosen according to several in vitro assays, as reported (Table S1), to inhibit Hsp90 and ERM1 proteins, respectively. After treating the organotypic retinal cultures with either GA or KIF, we evaluated the effect by calculating the percentage of TUNEL positive cells in the ONL of treated RHO P23H retinae.
GA treatment at 0.01 µM and 0.1 µM for 6 days did not protect the retinas. No significant effect in the percentage of TUNEL-positive cells in the ONL of RHO P23H rats was observed. Contrarily, retinae treated with 1 µM GA exhibited increased degeneration with a significant increase in the number of dying photoreceptor cells (Vehicle: 3.844% ± 1.2; 1 µM GA: 8.313% ± 2.7, p < 0.05; Figure 2A,B).  We next evaluated the role of mannose trimming in P23H photoreceptor degeneration by using the inhibitor KIF. Similarly to GA, retinae treated with 10 or 100 µM KIF showed induced cell death, detected by an increased percentage of TUNEL-positive cells in the ONL (Vehicle: 4.149% ± 0.8; 10 µM KIF: 5.674% ± 0.5, p < 0.05; and 100 µM KIF: 6.579% ± 0.2, p < 0.001; Figure 3A,B). Treatment with the lower dose of 1 µM KIF did not show any effect. at the evaluated concentrations ( Figure 2C). However, the observed increase in TUNELpositive cells with 100 µM KIF was reflected by a significantly reduced number of remaining cell rows in the ONL of KIF-treated RHO P23H retinae (Vehicle: 10.03 rows ± 0.5; 100 µM KIF: 7.628 rows ± 1.0, p < 0.01; Figure 3C).
These results indicate that inhibition of pathways involved in recognizing misfolded RHO or targeting RHO P23H to ERAD by the selected concentrations of GA or KIF do not provide neuroprotection in RHO P23H rat retinal explants.

VCP Inhibition Enhances Photoreceptor Cell Survival in RHO P23H Retinal Cultures
Following recognition and targeting, RHO P23H ERAD substrates are retrotranslocated across the ER membrane into the cytosol. In fact, for almost all ERAD substrates, VCP has an essential role in substrate retrotranslocation in yeast and mammals [26]. VCP extracts substrates from the ER to the cytosol in an ATP-dependent manner (retrotranslocation). As a second measurement to test for potential protective effects of Hsp90 or ERM1 inhibition on photoreceptor survival, we next counted photoreceptor cell rows in the ONL. We found that GA did not affect the number of remaining photoreceptor cell rows at the evaluated concentrations ( Figure 2C). However, the observed increase in TUNEL-positive cells with 100 µM KIF was reflected by a significantly reduced number of remaining cell rows in the ONL of KIF-treated RHO P23H retinae (Vehicle: 10.03 rows ± 0.5; 100 µM KIF: 7.628 rows ± 1.0, p < 0.01; Figure 3C).
These results indicate that inhibition of pathways involved in recognizing misfolded RHO or targeting RHO P23H to ERAD by the selected concentrations of GA or KIF do not provide neuroprotection in RHO P23H rat retinal explants.

VCP Inhibition Enhances Photoreceptor Cell Survival in RHO P23H Retinal Cultures
Following recognition and targeting, RHO P23H ERAD substrates are retrotranslocated across the ER membrane into the cytosol. In fact, for almost all ERAD substrates, VCP has an essential role in substrate retrotranslocation in yeast and mammals [26]. VCP extracts substrates from the ER to the cytosol in an ATP-dependent manner (retrotranslocation). Ubiquitin-labeled substrates are then directed to the proteasome [27]. Previous studies in cellular and Drosophila models indicate that excessive retrotranslocation activity causes cell loss in the presence of mutant RHO P23H and Rh1 P37H , respectively [10,14]. Also, VCP inhibition has been shown to be protective in RHO P23H animal models [17,19].
Here, we compare the inhibition of VCP-dependent retrotranslocation with the interference at different other ERAD steps under the same experimental conditions in RHO P23H rat retinal cultures. To test the impact of VCP inhibition, we employed the allosteric non-ATP-competitive VCP inhibitor NMS-873, one of the most potent and specific VCP inhibitors described to date, at three different doses. NMS-873 alters the binding affinity of the D1-D2 interdomain linker of VCP and inhibits ATPase activity [28]. When applied to RHO P23H rat retinal cultures under the same experimental conditions as described above, NMS-873 reduced photoreceptor cell death in a dose-dependent manner, showing minimum cell death and maximum photoreceptor cell protection at 5 µM. The percentage of TUNEL-positive cells decreased (Vehicle: 6.46% ± 2.4; 5 µM NMS-873: 2.043% ± 0.6, p < 0.001; Figure 4A,B) and the number of photoreceptor cell rows significantly increased (Vehicle: 7.51 rows ± 0.3; 1 µM NMS-873: 9.53 rows ± 0.2, p < 0.001; and 5 µM NMS-873: 11.56 rows ± 0.2, p < 0.001; Figure 4C). Ubiquitin-labeled substrates are then directed to the proteasome [27]. Previous studies in cellular and Drosophila models indicate that excessive retrotranslocation activity causes cell loss in the presence of mutant RHO P23H and Rh1 P37H , respectively [10,14]. Also, VCP inhibition has been shown to be protective in RHO P23H animal models [17,19].
Here, we compare the inhibition of VCP-dependent retrotranslocation with the interference at different other ERAD steps under the same experimental conditions in RHO P23H rat retinal cultures. To test the impact of VCP inhibition, we employed the allosteric non-ATP-competitive VCP inhibitor NMS-873, one of the most potent and specific VCP inhibitors described to date, at three different doses. NMS-873 alters the binding affinity of the D1-D2 interdomain linker of VCP and inhibits ATPase activity [28]. When applied to RHO P23H rat retinal cultures under the same experimental conditions as described above, NMS-873 reduced photoreceptor cell death in a dose-dependent manner, showing minimum cell death and maximum photoreceptor cell protection at 5 µM. The percentage of TUNEL-positive cells decreased (Vehicle: 6.46% ± 2.4; 5 µM NMS-873: 2.043% ± 0.6, p < 0.001; Figure 4A,B) and the number of photoreceptor cell rows significantly increased (Vehicle: 7.51 rows ± 0.3; 1 µM NMS-873: 9.53 rows ± 0.2, p < 0.001; and 5 µM NMS-873: 11.56 rows ± 0.2, p < 0.001; Figure 4C).

26S Proteasome Inhibition Reduces Photoreceptor Degeneration in RHO P23H Rat Explants
VCP activity and proteasomal activity are two interacting elements of substrate degradation during ERAD [10]. In addition, proteasome overload has been previously shown to contribute to photoreceptor cell death in the RHO P23H retinae [14,29].
We used the inhibitor bortezomib (BO) to see if proteasome inhibition could confer neuroprotection in RHO P23H rat retinal explants. BO blocks proteasome-targeted proteoly-sis by inhibiting the 26S proteasome, a large protease complex that degrades ubiquitinated proteins [30].
We used the inhibitor bortezomib (BO) to see if proteasome inhibition could confer neuroprotection in RHO P23H rat retinal explants. BO blocks proteasome-targeted proteolysis by inhibiting the 26S proteasome, a large protease complex that degrades ubiquitinated proteins [30].
Thus, pharmacological inhibition of the VCP/proteasome axis is protective for RHO P23H photoreceptors, further suggesting that excessive retrotranslocation and/or excessive degradation of RHO is a critical detrimental event in the RHO P23H retina.  Thus, pharmacological inhibition of the VCP/proteasome axis is protective for RHO P23H photoreceptors, further suggesting that excessive retrotranslocation and/or excessive degradation of RHO is a critical detrimental event in the RHO P23H retina.

Only VCP Inhibition Enhances RHO Trafficking to the Outer Segment in RHO P23H Rat Retinal Explants
RHO P23H mutation is characterized by RHO mislocalization even before photoreceptor degeneration starts. Impaired RHO transport is associated with disorganized and shortened photoreceptor OSs [31]. WT control retinae exhibit high RHO expression, and the protein localizes mainly to the photoreceptor's OS ( Figure S1). In contrast, RHO P23H retinae show  Figure S1). Therefore, we investigated whether different ER modulators (GA, KIF, NMS-873, and BO) could reconstitute proper physiological RHO trafficking to the OS.
In RHO P23H corresponding vehicle control retinae, immunofluorescence analysis showed that RHO was mislocalized, exhibiting an accumulation in the ONL (Figure 6). Using this as a reference, we found similar staining after treatment at all evaluated doses of GA, KIF, and BO. Only VCP inhibition by NMS-873 restored the distribution of RHO to the OS in a dose-dependent manner ( Figure 6). RHO P23H mutation is characterized by RHO mislocalization even before photoreceptor degeneration starts. Impaired RHO transport is associated with disorganized and shortened photoreceptor OSs [31]. WT control retinae exhibit high RHO expression, and the protein localizes mainly to the photoreceptor's OS ( Figure S1). In contrast, RHO P23H retinae show RHO staining distributed throughout the ONL [17] (Figure S1). Therefore, we investigated whether different ER modulators (GA, KIF, NMS-873, and BO) could reconstitute proper physiological RHO trafficking to the OS.
In RHO P23H corresponding vehicle control retinae, immunofluorescence analysis showed that RHO was mislocalized, exhibiting an accumulation in the ONL ( Figure 6). Using this as a reference, we found similar staining after treatment at all evaluated doses of GA, KIF, and BO. Only VCP inhibition by NMS-873 restored the distribution of RHO to the OS in a dose-dependent manner ( Figure 6).
In conclusion, modulation of ERAD either by inhibition of VCP or 26S proteasome activity decreases photoreceptor cell death and improves photoreceptor survival. However, only VCP inhibition enhances RHO localization to the OS.

Effects of Different ERAD Modulators on Retinal Stress and Inflammation in RHO P23H Retinal Organotypic Cultures
Next, we investigated possible retinal stress and inflammatory effects of different ERAD modulators in the RHO P23H retinae. Since Müller cells and microglia/macrophages activate due to retinal stress and inflammation [32,33], we evaluated these cell populations. We performed immunofluorescence staining for glial fibrillary acidic protein (GFAP) and ionized calcium-binding adapter molecule 1 (Iba1) to evaluate astrocytes, Müller cells, and microglia in RHO P23H retinae treated with GA, KIF, NMS-873, and BO by using their highest concentrations (1 µM GA, 100 µM KIF, 5 µM NMS-873, and 1 µM BO). In conclusion, modulation of ERAD either by inhibition of VCP or 26S proteasome activity decreases photoreceptor cell death and improves photoreceptor survival. However, only VCP inhibition enhances RHO localization to the OS.

Effects of Different ERAD Modulators on Retinal Stress and Inflammation in RHO P23H Retinal Organotypic Cultures
Next, we investigated possible retinal stress and inflammatory effects of different ERAD modulators in the RHO P23H retinae. Since Müller cells and microglia/macrophages activate due to retinal stress and inflammation [32,33], we evaluated these cell populations. We performed immunofluorescence staining for glial fibrillary acidic protein (GFAP) and ionized calcium-binding adapter molecule 1 (Iba1) to evaluate astrocytes, Müller cells, and microglia in RHO P23H retinae treated with GA, KIF, NMS-873, and BO by using their highest concentrations (1 µM GA, 100 µM KIF, 5 µM NMS-873, and 1 µM BO).
GFAP immunostaining is restricted under normal conditions to the astroglial cells and the endfeet of the Müller cells in the retina and is used to recognize gliosis [34]. In the RHO P23H transgenic rat, the number of astrocytes is higher than in the age-matched wild type (WT) control due to the retinal degeneration [33], and the GFAP immunoreactivity extends from the GCL through the ONL, indicating activation of Müller cells and retinal gliosis [32], as reproduced in our untreated RHO P23H explants (Figure 7, vehicle). The retinae treated with the inhibitors interfering with protein folding or integrity (GA or KIF, respectively) showed the same or even increased levels of retinal gliosis as the untreated controls ( Figure 7A). Interestingly, in the retinae treated with VCP or proteasome inhibitors, less GFAP immunoreactivity propagates from the GCL through the ONL, indicating less gliosis ( Figure 7B). GFAP immunostaining is restricted under normal conditions to the astroglial cells and the endfeet of the Müller cells in the retina and is used to recognize gliosis [34]. In the RHO P23H transgenic rat, the number of astrocytes is higher than in the age-matched wild type (WT) control due to the retinal degeneration [33], and the GFAP immunoreactivity extends from the GCL through the ONL, indicating activation of Müller cells and retinal gliosis [32], as reproduced in our untreated RHO P23H explants (Figure 7, vehicle). The retinae treated with the inhibitors interfering with protein folding or integrity (GA or KIF, respectively) showed the same or even increased levels of retinal gliosis as the untreated controls ( Figure 7A). Interestingly, in the retinae treated with VCP or proteasome inhibitors, less GFAP immunoreactivity propagates from the GCL through the ONL, indicating less gliosis ( Figure 7B). Iba-1 immunostaining allows the determination of microglia activation by their ameboid cell shape (rounded cell bodies) and migration from the GCL to the ONL, indicating a retinal inflammatory response [34,35] (Figure 8, vehicle). Retinae treated with VCP or proteasome inhibitors but not with GA or KIF showed abated Iba1 activation ( Figure 8A), as evidenced by significantly fewer round-shaped positive cells in the GCL and ONL (Figure 8B). These results allowed us to conclude that pharmacological modulation of the VCP/proteasome axis in RHO P23H reduces inflammation and retinal stress. Iba-1 immunostaining allows the determination of microglia activation by their ameboid cell shape (rounded cell bodies) and migration from the GCL to the ONL, indicating a retinal inflammatory response [34,35] (Figure 8, vehicle). Retinae treated with VCP or proteasome inhibitors but not with GA or KIF showed abated Iba1 activation ( Figure 8A), as evidenced by significantly fewer round-shaped positive cells in the GCL and ONL ( Figure 8B). These results allowed us to conclude that pharmacological modulation of the VCP/proteasome axis in RHO P23H reduces inflammation and retinal stress.

Discussion
Misfolded RHO P23H is retained in the ER and then identified by ER folding and chaperone systems, processed by ERAD, and retrotranslocated to the cytosol for degradation by the proteasome [6]. In this scenario, excessive ERAD combined with aggregation-associated proteotoxicity may exceed the capacity of cellular protein clearance, resulting in acute cellular stress and subsequent processes initiating cellular degeneration [36]. Several studies of pharmacological modulation of the cellular proteostasis network have been performed, looking to rebalance proteostasis and potentially unmask novel therapeutic targets. Here, we evaluated four substances able to interfere at the different ERAD steps and tested them under the same experimental conditions to compare their effects. Thus, we selected one of the most commonly used adRP models, the RHO P23H transgenic rat, using a serum-free retinal organotypic system as a unifying approach, and evaluated cell death, cell survival, and RHO immunostaining distribution, choosing VCP inhibition as a positive reference.
VCP is a key interactor of misfolded RHO P23H that allows its retrotranslocation and proteasomal clearance [10]. We have found that inhibition of VCP or proteasome individually attenuates cell death and preserves photoreceptor structure and function in an insect model of adRP [14] and shown that VCP inhibition [17,19] or silencing [18] mitigates disease progression in RHO P23H rodent models. Here, we were able to substantiate these previous findings when we tested VCP inhibition by NMS-873 in the RHO P23H transgenic model and confirmed that inhibition of retrotranslocation had a protective effect on degenerating photoreceptors. In addition, we found that the inhibition of 26S proteasome activity also enhanced photoreceptor survival in RHO P23H rat retinal explants, however, without the improved protein trafficking observed with VCP inhibition. Moreover, tar-

Discussion
Misfolded RHO P23H is retained in the ER and then identified by ER folding and chaperone systems, processed by ERAD, and retrotranslocated to the cytosol for degradation by the proteasome [6]. In this scenario, excessive ERAD combined with aggregation-associated proteotoxicity may exceed the capacity of cellular protein clearance, resulting in acute cellular stress and subsequent processes initiating cellular degeneration [36]. Several studies of pharmacological modulation of the cellular proteostasis network have been performed, looking to rebalance proteostasis and potentially unmask novel therapeutic targets. Here, we evaluated four substances able to interfere at the different ERAD steps and tested them under the same experimental conditions to compare their effects. Thus, we selected one of the most commonly used adRP models, the RHO P23H transgenic rat, using a serum-free retinal organotypic system as a unifying approach, and evaluated cell death, cell survival, and RHO immunostaining distribution, choosing VCP inhibition as a positive reference.
VCP is a key interactor of misfolded RHO P23H that allows its retrotranslocation and proteasomal clearance [10]. We have found that inhibition of VCP or proteasome individually attenuates cell death and preserves photoreceptor structure and function in an insect model of adRP [14] and shown that VCP inhibition [17,19] or silencing [18] mitigates disease progression in RHO P23H rodent models. Here, we were able to substantiate these previous findings when we tested VCP inhibition by NMS-873 in the RHO P23H transgenic model and confirmed that inhibition of retrotranslocation had a protective effect on degenerating photoreceptors. In addition, we found that the inhibition of 26S proteasome activity also enhanced photoreceptor survival in RHO P23H rat retinal explants, however, without the improved protein trafficking observed with VCP inhibition. Moreover, targeting the VCP/proteasome axis in RHO P23H reduces inflammation/stress, increases the organization of cone photoreceptors ( Figure S2), and improves general retinal structure ( Figure S3).
By contrast, inhibition of initial steps in ERAD (recognition and targeting of misfolded RHO) does not enhance photoreceptor cell survival. Indeed, our results indicate that this type of modulation may have the opposite effect and accelerate the degeneration process, suggesting that prolongation of protein folding attempts might be less harmful to the cell than excessive degradation. Regulation of the structure and folding of proteins is particularly important for maintaining cellular homeostasis [37]. In the past, several studies have attempted to directly modulate the folding capacity in the cell as a neuroprotective strategy. One of the targets has been the chaperone Hsp90. Hsp90 participates in protein folding and protein degradation, playing a role in the assembly and maintenance of the 26S proteasome [38].
Nevertheless, inhibition of Hsp90 has shown contradictory results. Single oral gavage of the Hsp90 inhibitor HSP990 has improved visual function and delayed photoreceptor degeneration in RHO P23H rats. However, its longer oral gavage adversely affected visual function [20]. Other reports have also shown that inhibition of Hsp90 induces a deleterious effect in mice [39] and human RPE cells in vitro [40] and exhibits retinal toxicity in dogs in vivo [41]. In clinical trials, adverse effects have been reported, including ocular toxicity associated with visual disturbances [42]. In line with that, upregulation of Hsp70 and Hsp90 by treatment with Arimoclomol resulted in photoreceptor protection and improved visual function [43], suggesting that potentiation of the heat shock response, not its inhibition, might relate to neuroprotection. This is supported by our results on Hsp90 inhibition with GA, in which photoreceptor degeneration was even increased in a dose-dependent manner.
ERM1 appears to have a major function in enhancing the degradation of misfolded RHO P23H by directing them to ERAD. Inhibition of ERM1 by KIF showed partial inhibition of enhanced degradation of RHO P23H in a neuroblastoma cell line [44]. However, the mannosidase inhibitor failed to increase cell survival in RHO P23H rat retinal explants in our study.
We found that targeting the last steps in ERAD (i.e., the VCP/proteasome axis) can have protective effects on P23H rat retinae. RHO WT is recruited into aggregates by RHO P23H , and these aggregates form a complex with VCP, leading to degradation of RHO WT [10]. Retention of these aggregates in the inner segment of photoreceptors restrains the amount of RHO reaching the OS. In addition, retrotranslocation of misfolded RHO P23H proteins via VCP and their subsequent degradation requires high energy levels (e.g., ATP hydrolysis) [16,45], and this could lead to photoreceptor cell death upon P23H-mediated adRP. This is why inhibition of the VCP/proteasome axis might prevent a severe energy imbalance and enhance photoreceptor cell survival [46]. However, the overall protection provided by proteasome inhibition is weaker compared to VCP inhibition. In addition, we observed proper localization of RHO only after VCP inhibition but not after proteasome inhibition, although both increased expression of mature RHO ( Figure S4). One possible explanation is that proteasome inhibition, even though it could compensate for the energy requirement, cannot decrease the RHO-VCP complex levels, thereby impeding the trafficking of RHO to the OS.
Inhibition of excessive ERAD and proteasome activities could, in addition, promote refolding. Nevertheless, approaches to promote RHO refolding alone (e.g., treatment with AMPK activator metformin [47], retinoids [48,49], overexpression of Calnexin [50], or BiP [51]) have proven challenging and indicate that promoting RHO P23H refolding alone does not result in a protective outcome. Metformin-rescued RHO P23H showed better protein trafficking but was still intrinsically unstable, increasing the rod OS's structural instability with consequently reduced photoreceptor function and increased photoreceptor cell death [47]. Enhanced folding of RHO P23H by overexpression of Calnexin enhanced the proper folding of RHO, but its loss of function showed a lack of effect in vitro [50]. Additionally, the neuroprotection obtained by BIP gene delivery in RHO P23H line-3 rats is more likely due to suppression of photoreceptor cell death rather than to enhanced RHO folding [51].
In conclusion, we report that that pharmacological modulation of the VCP/proteasome axis enhances photoreceptor cell survival and preserves retinal structure in the RHO P23H rat model of adRP in vitro, presenting a potential therapeutic strategy for adRP. In contrast, other proteostasis regulators in ERAD that control protein integrity or folding did not show any protective effects. Our study highlights the importance of further investigations to accurately characterize the mechanisms regulating protein folding, quality control, and degradation of RHO for the development of new therapeutic approaches. Furthermore, the importance of understanding these processes may not be limited to P23H-mediated retinal degeneration, but may be extendable to several other diseases caused by inefficient protein folding. To mimic the genetic background of adRP, we used heterozygous P23H rats obtained by crossing homozygous RHO P23H rats with RHO WT rats (CDH IGS Rat; Charles River, Germany). Animals were housed in the Institute for Ophthalmic Research animal facility under standard white cyclic lighting, with access to food and water.

Organotypic Retinal Explant Cultures of P23H Heterozygous Transgenic Rats
Retinae were isolated with the retinal pigment epithelium (RPE) attached as described previously [17,21]. Briefly, PN9 animals were sacrificed, the eyes were enucleated in an aseptic environment and pretreated with 12% proteinase K (MP Biomedicals, Illkirch-Graffenstaden, France, 0219350490) for 15 min at 37 • C in R16 serum-free culture medium (Invitrogen Life Technologies, Leicestershire, UK, 07490743A). The enzymatic digestion was stopped by 20% fetal bovine serum (Gibco, Waltham, MA, USA, F7524). Retina and RPE were dissected, and four radial cuts were made to flatten it. The tissue was transferred to a 0.4 µm polycarbonate membrane (Corning Life Sciences, Corning, NY, US, CLS3412), with the RPE facing the membrane. The retinal explants were cultured in supplemented R16 nutrient medium and maintained at 37 • C in a humidified 5% CO 2 atmosphere. The eyes were then randomly assigned to either vehicle control or inhibitors. For each inhibitor, vehicles (either water or DMSO) were applied according to the highest concentration of the inhibitors. Geldanamycin (Calbiochem, Darmstadt, Germany, 345805) was dissolved in DMSO, and different concentrations of geldanamycin were applied in 1 mL culture medium (0.01 µM, 0.1 µM, and 1 µM). 0.2% DMSO was used as vehicle control. Kifunensine (Calbiochem, Darmstadt, Germany, 422500) was dissolved in hot water (55 • C), and different concentrations of the drug were applied in 1 mL culture medium (1 µM, 10 µM, and 100 µM). ddH 2 O was applied as corresponding vehicle control. NMS-873 (Xcessbio, Chicago, IL, USA, M60165-b) was dissolved in DMSO, and different concentrations were applied in 1 mL culture medium (0.5 µM, 1 µM, and 5 µM). Bortezomib (Calbiochem, Darmstadt, Germany, 5.04314.0001) was dissolved in DMSO, and 0.01 µM, 0.1 µM, and 1 µM concentrations of bortezomib were applied in culture medium. For NMS-873 and bortezomib, 0.5% DMSO served as vehicle control. The medium was changed every second day. The PN9 cultures were fixed at DIV6, corresponding to PN15, the peak of degeneration in vivo in age-matched RHO P23H heterozygous rats.

TUNEL Assay
TUNEL assay was performed using an in situ cell death detection kit conjugated with fluorescein isothiocyanate (Roche, Mannheim, Germany, 11684795910). DAPI (Vectashield Antifade Mounting Medium with DAPI; Vector Laboratories, H-1200) was used as a nuclear counterstain.

Hematoxylin and Eosin Staining
Retinal sections were pre-rinsed in PBS then dipped in Harris Hematoxylin for 5 min followed by a wash in ddH 2 O. Slides were subsequently dipped in acid alcohol (70% ethanol 1% HCl) for 1 min, washed in ddH 2 O, then dipped in Eosin-Phloxyine for 1 min. Slides were then dehydrated in a series of 95% ethanol and 100% ethanol followed by 2 min in xylene. Once the slides were dried, mounting medium (Entellan, Merck Milipore, Darmstadt, Germany) was applied, and slides were cover-slipped.

Microscopy and Image Analysis
All samples were analyzed using a Zeiss Axio Imager Z1 ApoTome microscope, AxioCam MRm camera, and Zeiss Zen 2.3 software in Z-stack at 20× magnification. For quantitative analysis, positive cells in the ONL of at least three sections per group were manually counted. The TUNEL positive cells in the entire photoreceptor layer (ONL) of four cross-sections per culture were counted manually (4-10 images from at least three retinae per treatment were taken and shown as open circles). The percentage of positive cells was calculated by dividing the absolute number of TUNEL positive cells by the total number of cells in the ONL (total number of cells: ONL area/the size of a photoreceptor nucleus (17.3 µm 2 )). The number of remaining photoreceptor cell nuclei in a row in the ONL was quantified by counting DAPI-stained cells in a linear row in three different linear rows in four cross-sections per culture and averaging the counts. Graphs were prepared in GraphPad Prism 7.05 for Windows.

Statistics
The evaluation of TUNEL analysis, photoreceptor cell row counts, and rhodopsin band intensity were performed using GraphPad Prism 7.05 for Windows and one-way ANOVA testing, followed by Tukey multiple comparisons test. The n value for each experiment represents the number of retinae. Each closed circle represents one single retina in the graphs; thus, the number of closed circles corresponds to the n value. For GA treatment: the values were quantified by scoring several images from DMSO-treated retinae (n = 4) and GA-treated retinae at three different concentrations (0.01, 0.1, and 1 µM; n = 3 for each). For KIF treatment: the values were quantified by scoring several images from three retinae (n = 3) per treatment for vehicle control (H 2 O), 1 and 10 µM KIF, and five retinae for 100 µM KIF (n = 5). For NMS-873 treatment: the values were quantified by scoring several images from P23H retinae (n = 13 for TUNEL assay, n = 9 for cell rows) per treatment for DMSO 0.5%, three retinae (n = 3) for 0.5 and 1 µM NMS-873 and six retinae (n = 6) for 5 µM NMS-873. For BO treatment: the values were quantified by scoring several images from three retinae (n = 3) per treatment for DMSO and 0.01 µM BO, and four retinae (n = 4) for 0.1 and 1 µM BO.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/biom11101528/s1: Table S1: References list for the chosen concentrations in this study; Figure S1: Rhodopsin localization in RHO P23H and RHO WT retinal organotypic cultures; Figure S2: Pharmacological interference in the VCP/proteasome axis increases the organization of cone photoreceptors in RHO P23H retinal organotypic cultures; Figure S3: Modulation of the VCP/proteasome axis improves retinal structure in RHO P23H retinal organotypic cultures; Figure S4