Molecular Vision 2016; 22:294-310
Received 22 January 2016 | Accepted 31 March 2016 | Published 02 April 2016
Chulbul M. Ahmed, Manas R. Biswal, Hong Li, Pingyang Han, Cristhian J. Ildefonso, Alfred S. Lewin
Department of Molecular Genetics and Microbiology, University of Florida, Gainesville, FL
Correspondence to: Alfred S. Lewin, Department of Molecular Genetics and Microbiology, University of Florida, 1200 Newell Dr., Gainesville, FL, 32610-0266; Phone: (352) 392-0676; FAX: (352) 273-8905; email: email@example.com
Purpose: Chronic oxidative stress and subacute inflammation have been implicated as causes of age-related macular degeneration (AMD). In this study, we tested whether an orally available 5-OH-tryptamine (5HT) 1a receptor agonist, xaliproden, could protect against retinal pigment epithelium (RPE) cell damage in culture and in a mouse model of geographic atrophy.
Methods: Paraquat was used to create mitochondrial oxidative stress in ARPE-19 cells, and tumor necrosis factor-α (TNF-α) was used to stimulate the production of inflammatory cytokines in these cells. The production of antioxidant proteins, metallothionein, and inflammatory cytokines was assayed with quantitative real-time PCR. Cell survival was analyzed with microscopy and a cell titer assay. Integrity of the RPE monolayer was determined by measuring the transepithelial electrical resistance (TEER) and with immunocytochemistry with zona occludens protein 1 (ZO-1) antibody. RPE atrophy was studied in mice deleted for Sod2 (the gene for mitochondrial superoxide dismutase) specifically in the RPE. The mice were treated orally with daily doses of xaliproden at 0.5 and 3 mg/kg for 4 months. The retinal structure was analyzed with spectral domain optical coherence tomography (SD-OCT) and with light and electron microscopy. Retinal function was assessed with full-field electroretinography (ERG) and with optokinetic measurements.
Results: Xaliproden led to a dose-dependent increase in cell survival following treatment with paraquat. Synthesis of the antioxidant response genes NqO1, GSTM1, CAT, HO-1, and Nrf2 was increased in response to the drug, as was the zinc chaperone metallothionein. Treatment of cells with TNF-α led to increased production of IL-1β, IL-6, chemokine (C-C motif) ligand 20 (CCL20), and vascular endothelial growth factor (VEGF) by ARPE-19 cells, and this response was attenuated by treatment with xaliproden. TNF-α also led to a decrease in the TEER that was prevented by treatment with the 5HT1a agonist. Daily gavage with xaliproden at either dose induced the production of protective enzymes in the mouse retina, and treatment of the Sod2-deleted mice with the drug showed improved thickness of the outer nuclear layer and improved visual acuity relative to the control-treated mice. There was no significant difference in full-field scotopic ERG among the treatment groups, however. Vacuolization of the RPE and disorganization of the photoreceptor outer segments were reduced at both dose levels of xaliproden.
Conclusions: Xaliproden protected RPE cells from oxidative and inflammatory insults and protected the mouse RPE and retina from RPE atrophy in the face of excess mitochondrial oxidative stress. These results suggest that this drug, which had a reasonable safety profile in clinical trials, may be used to prevent the progression of geographic atrophy in humans.
The 5H1a receptor is a seven-membrane spanning domain G-protein-coupled receptor that is expressed in many cell types . Agonists of this receptor are currently in clinical use for treatment of anxiety, and such drugs have exhibited neuroprotective properties in spinal cord injury , Parkinson disease models [3,4], and excitotoxicity . In the eye, Collier and colleagues demonstrated that two different 5HT1a agonists provided nearly complete protection against phototoxic injury in albino rats when given systemically before and shortly after exposure to bright light . In a subsequent paper, the same group demonstrated that one of these compounds (AL-8309A) blocked the deposition of complement related proteins and the activation of microglial following severe photo-oxidative stress .
In earlier work, we showed that a prototype 5HT1a agonist, 8-hydroxy-2-dipropylaminotetralin (8-OH-DPAT), protected ARPE-19 cells from oxidative damage by hydrogen peroxide and by paraquat, and that daily administration of 8-OH-DPAT to mice prevented injury to the RPE and neural retina in two mouse models of geographic atrophy caused by depletion of manganese superoxide dismutase (MnSOD) in the RPE [8,9]. Those experiments required daily injections of the compound, which is not approved for human use. We wanted to determine whether a similar effect could be achieved using a drug that has been tested in patients and is available orally.
Xaliproden is a highly selective synthetic 5HT1a receptor agonist . It crosses the blood–brain barrier and can be taken by mouth. The drug increases the release of dopamine in the prefrontal cortex of rats and decreases the level of extracellular serotonin in the ventral hippocampus. Both effects are attributable to binding of the 5HT1a receptor and can be inhibited by pretreatment with a 5HT1a antagonist, WAY100635 . The activity of xaliproden appears to be mediated by the mitogen-activated kinases Erk1 and Erk2 . Xaliproden also has neurotrophic properties that are dependent on binding to the 5HT1a receptor. This drug has been tested in humans as a potential treatment for amyotrophic lateral sclerosis (ALS)  and Alzheimer disease (Clinicaltrials.gov NCT00103649). Although no efficacy was demonstrated in these studies, the drug appears to relieve allodynia associated with chemotherapeutic agents such as paclitaxel , and xaliproden has been tested for this indication in cancer patients (Clinicaltrials.gov NCT00272051).
Because of the protective effect of 8-OH-DPAT in our mouse model of retinal degeneration, we wanted to test xaliproden in the same model, particularly since the drug’s safety profile and side effects have been established in several clinical trials. We began by testing the compound on an RPE-derived cell line to determine the impact on the expression of antioxidant genes and on proinflammatory cytokines. We then determined that orally administered xaliproden preserved visual function and RPE structure in mice subjected to chronic mitochondrial oxidative stress.
We obtained validated ARPE-19 cells (CRL-2301, Appendix 1) from the American Type Culture Collection (Manassas, VA) and froze cells from the first culture immediately. We discarded the ARPE-19 cells after three subsequent passages. To extract RNA, cells were grown to 90% confluence and then placed in a 1:1 mixture of Dulbecco’s Modified Eagle Medium (Sigma-Aldrich, St. Louis, MO) and Ham’s F12 nutrient mixture (DMEM/F12) plus 1% fetal bovine serum (FBS). We treated the cells with 20 µM xaliproden (R&D Systems, Minneapolis, MN) plus or minus 20 ng/ml TNF-α (PeproTech, Rocky Hill, NJ) or 300 µM paraquat (Sigma-Aldrich, St. Louis, MO) for 24 h. We washed the cells with PBS (120 mM NaCl, 20 mM KCl, 10 mM NaPO4, 5 mM KPO4, pH 7.4), pH 7.4 and extracted RNA using the RNeasy Mini Kit (Qiagen, Valencia, CA). We then synthesized cDNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) and amplified PCR products (The PCR reaction mixture contained 3 µM gene specific primers. After denaturation at 95 °C for 2 min, 40 cycles of amplification (94 °C for 15 seconds; 60 °C for 30 seconds).) using the gene-specific primers listed in (Appendix 2) and the SsoFast EvaGreen Supermix (Bio-Rad). We amplified β-actin cDNA from each sample to serve as an internal standard. We used the system software from the CFX96 real-time thermal cycler (Bio-Rad) to compare transcript levels in the treated and untreated samples.
We grew ARPE-19 cells in DMEM/F12 plus 1% FBS for 4 weeks in eight-well chambered slides. By this time, the cells had assumed a cobblestone appearance characteristic of the RPE. We then subjected them to treatment with 300 µM paraquat, 20 µM xaliproden, or both for 48 h. We fixed the cells using freshly prepared 4% paraformaldehyde for 30 min and washed them once with PBS. Cells were made permeable by treatment with 1% Triton X-100 in PBS for 30 min. We exposed the cells to 10% normal goat serum in PBS plus 0.5% Triton X-100 for 30 min to reduce background antibody binding. We then washed the cells in 0.2% Triton X-100 in PBS and then incubated them in the same buffer with a 1:200 dilution of polyclonal antibody to ZO-1 (Invitrogen, Grand Island, NY) for 15 h at 4 °C. They were washed in PBS + 1% Triton X-100 three times and then incubated with secondary antibody (Cy-3 conjugated goat anti rabbit; Invitrogen) for 30 min at room temperature followed by three additional wash steps. After they were covered with a coverslip, we imaged the cells in a fluorescence microscope.
IL-1β secreted from ARPE-19 cells was assayed using an enzyme-linked immunosorbent assay (ELISA) kit (BioLegend, San Diego, CA) following the manufacturer’s protocol. Briefly, we coated microtiter plates with capture antibody by incubation overnight at 4 °C. The wells were washed and then incubated with a blocking buffer for 2 h at room temperature, followed by three washings. We added different dilutions of the supernatants from ARPE-19 cells to the wells in triplicate and incubated the cells for 2 h, followed by three washings with the wash buffer. We then added detection antibody and incubated the cells for 1 h, followed by three washings. Avidin-horseradish peroxidase (HRP) was added for 0.5 h, followed by three washes. Substrate was added for 0.5 h, and the reaction was stopped with the stop buffer. The absorbance was measured in a microplate reader at 450 nm.
We plated the ARPE-19 cells on Transwell inserts (33.6 mm2, pore size 0.4 µm; Greiner Bio-One, Monroe, NC) in DMEM/F12 including 1% FBS and allowed the cells to grow in 24-well plates for 4 weeks. We added paraquat plus or minus xaliproden at the concentrations used above, and the cells were incubated for an additional 48 h. We used an EVOM2 volt/ohm meter (World Precision Instruments, Sarasota, FL) to measure the transepithelial electrical resistance (TEER) at room temperature as specified by the manufacturer. We calculated the net resistance values by subtracting the measurement of a blank well from the measurements of the wells that contained cell monolayers.
We conducted all animal procedures in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. The University of Florida Institutional Animal Care and Use Committee reviewed and approved all animal procedures in advance. Details of the mouse model are provided in our earlier publication . Depletion of manganese superoxide dismutase either by ribozyme knockdown or by deletion of Sod2 in the RPE (retinal pigment epithelium) of mice leads to characteristic features of dry AMD including disruption of Bruch's membrane, increased accumulation of bisretinoid compounds, increase in RPE lipofuscin and CEP-modified proteins, localized death of RPE cells and overlaying photoreceptors [8,15-17]. For the current experiments we employed mice with a specific deletion of Sod2 in the RPE [9,15]. Briefly, we employed C57Bl/6J mice transgenic for PVMD2-rtTA and tetO-PhCMVcre  and homozygous for a “floxed” (flanked by loxP) of Sod2, the gene for MnSOD . We induced the expression of cre recombinase by feeding doxycycline containing chow (200 mg/kg in Harlan Rodent Diet 7012, Harlan Teklad, Indianapolis, IN) to nursing dams for 2 weeks after delivery. We refer to these animals as Sod2loxP/loxP RPE-cre mice. Of note, the first time point of ERG analysis was 2 weeks after the cessation of doxycycline, and by 1 month of age, cre recombinase is not detectable in the RPE.
Cohorts of 24 Sod2loxP/loxP RPE-cre mice received a daily dose of xaliproden by gavage at 0.5 mg/kg or 3 mg/kg of drug dissolved in 0.3% carboxymethylcellulose plus 0.25% Tween-20 (vehicle). An additional cohort of mice was treated with vehicle only. Sub-groups of 14 mice randomly selected were analyzed at monthly intervals with full-field scotopic electroretinography (ERG) and spectral domain optical coherence tomography (SD-OCT) with methods described in the paper by Mao et al. 
In eight randomly selected mice from each treatment group, we measured the visual acuities of treated and untreated eyes by observing the optokinetic responses of mice to rotating sinusoidal gratings (OptoMotry™) . This method measures the acuities of the left and right eyes independently based on their sensitivities to rotating patterns of bars: Right eyes are more sensitive to counterclockwise rotation, and left eyes are more sensitive to clockwise rotation. The methods we used to measure acuity are described by Pang et al. . Briefly, an unrestrained mouse was placed on a pedestal located in the center of four LCD computer monitor screens and was observed by an overhead video camera. A trial was initiated by presenting the mouse with the sinusoidal pattern rotating either clockwise or counterclockwise as determined randomly by the system software. Initially, the 100% contrast pattern had a spatial frequency of 0.200 cycles/degree for both directions of rotation. We determined the thresholds for each eye simultaneously using incremental functions for correct responses in both directions. We defined acuity as the highest spatial frequency yielding a threshold response at 100% contrast.
We performed full-field electroretinography as described in our earlier study on 8-OH-DPAT . After overnight dark adaptation and ketamine/xylazine anesthesia (95 mg/kg ketamine and 5 mg/kg xylazine in 100 µl given i.p.), we recorded responses from both eyes using an LKC UTAS Visual Electrodiagnostic System with a BigShot™ full-field dome (LKC, Gaithersburg, MD). Scotopic ERGs were elicited with 1 msec flashes of white light at 0 dB (2.68 cds/m2), −10 dB (0.18 cds/m2), and −20 dB (0.02 cds/m2). Ten scans were averaged at each light intensity. We then exposed mice to a 2 min white light bleach in the Ganzfeld dome and then to a white flash at 1.0 cd sec/m2 intensity to elicit a cone response. The a-wave amplitudes were measured from the baseline to the peak in the cornea-negative direction, and the b-wave amplitudes were measured from the cornea-negative peak to the major cornea-positive peak.
Mice were injected with 150 mg/kg of sodium pentobarbital (Euthasol) and then perfused with PBS containing 2% paraformaldehyde and 2.5% glutaraldehyde. Mice were enucleated, and the eyes were soaked overnight in 4% paraformaldehyde and 2% glutaraldehyde. We soaked the tissue in 0.1 M cacodylate buffer (pH7.4) for 10 min and incubated it in 1% osmium tetroxide for 4 h at 4 °C in the same buffer. We then soaked the tissue overnight in cacodylate buffer at 4 °C. Subsequently, we dehydrated the tissue in a series of ethanol baths and steeped the tissue in epoxy/propylene on a rotator for embedding. For light microscopy, we prepared 1 µm sections, while for electron microscopy, we prepared sections of 80–100 nm.
For immunohistochemistry, we euthanized mice with inhalation of carbon dioxide. For analysis of retinal sections, eyes were fixed in melting isopentane as described by Wolfrum . We placed cryosections on polylysine-coated slides and soaked them in PBS containing 0.01% Tween-20 for 30 min. We then soaked them in a blocking solution containing mouse immunoglobulin (IgG; Vector® M.O.M.™ Immuno detection kit, cat no. BMK-2202, Burlingame, CA) for 1 h followed by an overnight incubation at 4 °C with primary antibody against carboxyethylpyrrole (CEP)  at a dilution of 1:400 in PBS/Tween-20. We washed the sections in the same buffer three times and incubated them for 1 h at room temperature with fluorescently tagged secondary antibody (Alexa Fluor® 594 goat anti-mouse IgG from Life Technologies, Cat No: A11005) plus 4’,6-diamidino-2-phenylindole (DAPI) to stain the nuclei. We prepared digital micrographs using a Keyence (Itsaca, IL) fluorescence microscope.
We treated 2-month-old mice with 3 mg/kg of xaliproden by mouth for 4 days. The mice were euthanized with CO2 inhalation, and RNA was extracted from the dissected retinas using the RNeasy Mini Kit from Qiagen. We performed the cDNA synthesis and qPCR as described for cultured ARPE-19 cells.
To compare the mean values of several groups (e.g., ERG amplitudes over time), we employed ANOVA with the Newman-Keuls test for multiple comparisons. Except for ERG results, error bars represent the standard deviation. For ERG results, error bars indicate standard error of the mean. We took a significance level p<0.05 as significant. For comparison of RNA levels in treated and untreated cells or mice, we used the Student t test for unpaired samples.
Paraquat is an herbicide structurally similar to MPP+, which is known to induce Parkinsonism in animals and humans. Complex III of the mitochondrial electron transport chain (coenzyme Q-cytochrome c oxidoreductase) converts paraquat into a free radical that interacts with molecular oxygen to form superoxide . To model mitochondrial oxidative stress in the RPE, we exposed ARPE-19 cells, which are derived from human RPE cells, to paraquat in the presence or absence of 20 µM xaliproden. There is strong pharmacological evidence of 5HT1a receptors on human RPE cells . To determine whether the drug induced the expression of antioxidant genes, we extracted RNA after 15 h, and measured gene expression with quantitative RT–PCR (Table 1).The values in this table represent the RNA levels compared to untreated cells; a value of 1 indicates no change compared to the control. As previously observed using 8-OH-DPAT, we detected a significant induction of genes regulated by the antioxidant response element (NqO-1, GSTM1, HO-1) in response to paraquat or to paraquat plus xaliproden. The induction of NqO-1 and GSTM1 was much greater when cells were treated with both compounds, but xaliproden alone did not lead to increased expression of HO-1 and catalase (CAT). Interestingly, expression of the transcription factor Nrf2, which regulates these genes, was suppressed by either paraquat or xaliproden but was induced by the combination treatment relative to treatment with paraquat alone. Others have reported that paraquat treatment reduces the levels of Nrf2 in kidney and neuroblastoma cell lines [26,27]; nevertheless, we do not know the cause of the decreased expression of Nrf2 in ARPE-19 cells after paraquat or xaliproden treatment alone. The level of Nrf2 protein is regulated post-transcriptionally, but the promoter of the Nrf2 gene contains active antioxidant response elements (ARE) suggesting autoregulation of this gene . We also observed a substantial induction in the transcription of metallothionein 1 (a zinc chaperone protein) in the presence of paraquat (5.5-fold) or xaliproden (3.3-fold) and a 10-fold increase in expression by the combination treatment.
To determine whether upregulation of these protective genes correlated with increased cell survival, we measured cell viability in response to paraquat at different concentrations of xaliproden added simultaneously. At 300 µM, paraquat killed more than 80% of the cells after a 24 h incubation, but addition of xaliproden led to an increase in cell viability at all levels tested (Figure 1 and Appendix 3). In our earlier work, we observed a maximum of 60% protection from paraquat-induced oxidative stress with 8-OH-DPAT , while up to 80% protection was observed with xaliproden under the same conditions.
ARPE-19 cells form tight junctions similar to RPE monolayers when grown in culture for several weeks in low serum medium. We grew such cells on permeable Transwells and measured the cells’ TEER after treatment with paraquat in the presence or absence of xaliproden. Treatment with paraquat caused a greater than 50% drop in the resistance of confluent ARPE-19 cells that had been maintained 4 weeks in 1% FBS medium (Figure 2). Simultaneous treatment with paraquat and xaliproden resulted in no decrease in transepithelial resistance, indicating that the integrity of the monolayer had been maintained. Light micrographs of the ARPE-19 monolayers decorated with primary antibodies to ZO-1 confirmed that elevation of oxidative stress by paraquat disturbed the tight junctions, but they were preserved by simultaneous treatment with xaliproden (Figure 3).
TNF-α is increased in several retinal diseases involving chronic inflammation, including age-related macular degeneration (AMD) . This cytokine promotes migration and adhesion of RPE cells grown in culture , and TNF-α induces the expression of IL-1β, IL-6, and vascular endothelial growth factor (VEGF)-A by RPE cells . To determine whether xaliproden could modulate the expression of cytokines or chemokines induced by TNF-α, we treated ARPE-19 cells with TNF-α in the presence and absence of xaliproden. As expected, treatment of cells with TNF-α alone caused a massive increase in the production of mRNA for IL-1β (171-fold), IL-6 (10.3-fold), and the chemokine CCL-20 (119-fold) relative to untreated cells (Table 2). Treatment of the ARPE-19 cells with xaliproden reduced the impact of TNF-α on the production of inflammatory cytokines. In addition, TNF-α led to a 2.5-fold increase in the synthesis of VEGF-A mRNA, and this effect was abrogated by cotreatment with xaliproden. Members of the IL-6 family protect photoreceptors from light injury and retinal ganglion cells following injury to the optic nerve [32,33], but IL-6 is a proinflammatory cytokine that abrogates immune privilege of the eye . Therefore, reduction of IL-6 may diminish neuroprotection while also reducing inflammation.
To confirm that the increase in synthesis of IL-1β RNA corresponded to an increase in the production and release of IL-1β, we measured the levels of the cytokine in the culture medium using ELISA. TNF-α led to an 80-fold increase in IL-1β released into the medium, and this increase was reduced 58% by cotreatment with xaliproden (Figure 4). Treatment of confluent ARPE-19 cells with TNF-α led to a loss in cell count, and this loss was prevented by simultaneous treatment with xaliproden (Appendix 3).
Treatment of ARPE-19 cells that had been allowed to form tight junctions with TNF-α led to a 55% reduction in transepithelial resistance suggesting a disruption of tight junctions. Cotreatment with xaliproden mitigated this damage, and the TEER was preserved (Figure 5), corresponding to an increase in tight junctions revealed with ZO-1 immunohistochemistry (Figure 3). This result is consistent with an increase in the synthesis of the tight junction protein occludin in the presence of TNF-α and xaliproden (Table 2).
To determine whether xaliproden would lead to an elevation of protective enzymes in the retina, we fed C57Bl/6J mice xaliproden (3 mg/kg) for 4 days. Two hours after the final dose the mice were humanely sacrificed, and their retinas dissected for analysis of RNA levels with quantitative real-time PCR (Table 3). Relative to the levels of mRNA measured in the mice treated with vehicle alone, we detected a 3.9-fold increase in the level of HO-1 and Met-1 mRNA, a 3.5-fold increase in GSTM1 expression, and a 3.8-fold increase in NqO1. The extent of the increase in the expression of these genes was greater than we had observed earlier using 8-OH-DPAT . The level of catalase mRNA was elevated by 50% in the drug-treated mice, similar to the level of increase seen using 8-OH-DPAT.
CEP is an oxidation product of docosahexaenoic acid-containing lipids that covalently modifies proteins in Bruch’s membrane and drusen in AMD eyes [35,36]. We previously showed that CEP-adducted proteins are increased in the RPE of mice depleted for MnSOD in that cell layer . To determine whether the protection of RPE-like cells in culture would translate to reduced oxidative stress RPE cells in mice, we turned to our model of retinal degeneration caused by the RPE-specific deletion of the gene for MnSOD . These mice exhibit a gradual decline in ERG amplitudes that is statistically significant only after 6 months, but they show signs of RPE and photoreceptor atrophy by 4 months. We treated the mice at two dose levels (0.5 mg/kg and 3 mg/kg) by daily gavage for 4 months starting at 1 month of age (approximately 1 week post-weaning). These levels corresponds to doses 0.04 mg/kg and 0.24 mg/kg for a 60 kg human based on body surface area . The dose given in the ALS clinical trial of xaliproden was 2 mg per day, meaning that our low dose was in that range. We employed an antibody generated against CEP-conjugated human serum albumin  to visualize CEP-modified proteins by immunohistochemistry (Figure 6). We noted reduced staining for CEP in Sod2loxP/loxP RPE-cre mice treated at either level of xaliproden (0.5 or 3 mg/kg) relative to mice treated with the vehicle alone. This result suggests that the elevation of protective proteins in response to xaliproden treatment reduces the burden of oxidized lipids in the RPE. Renganathan and colleagues had previously reported that another 5HT1a agonist, AL-8309A, reduced the level of CEP adducts in albino rats subjected to injury by blue light exposure .
We analyzed retinal function in these mice using the optokinetic response by OptoMotry™. Measuring acuity with this approach does not average the response of the full retina, as does full-field ERG, so that acuity measurements appear to be more sensitive to the impact of injury  or, conversely, to rescue  in some cases. Treatment with 0.5 mg/kg xaliproden led to a 28% improvement in visual acuity and treatment with 3 mg/kg to a 31% improvement relative to mice treated with vehicle alone (Figure 7A). This improvement in behavior performance was paralleled by a slight, but statistically significant, increase in the thickness of the outer nuclear layer in mice treated with either dose level of xaliproden, suggesting improved survival of the photoreceptor cells (Figure 7B).
Despite the relative protection of the outer nuclear layer (ONL) thickness and visual acuity afforded by xaliproden, we observed no comparable impact of the drug on ERG amplitudes. After 4 months of treatment, we observed no significant difference in a-wave and b-wave amplitudes between Sod2loxP/loxP RPE-cre mice treated with vehicle or either dose of xaliproden (Figure 8). However, as expected from our earlier characterization of this model, the ERG amplitudes were not significantly changed at this time point. There was no significant difference between the amplitudes measured at 2 months and those measured at 4 months (Figure 8 and Appendix 4). Similarly, we observed no significant difference in photopic ERG b-wave amplitudes between mice treated with vehicle alone or with the low dose of xaliproden (0.5 mg/kg), but we noticed a depression of the photopic ERG response at 4 months in mice treated at the higher dose (3 mg/kg). This reduction was statistically significant relative to the low-dose-treated mice and may reflect a harmful effect of prolonged high-dose treatment that was not apparent with SD-OCT or optokinetic measurements (Figure 7).
At the conclusion of 4 months of treatment, the eyes of mice from each treatment group were prepared for electron microscopy. Even at low magnification (Figure 9), the difference between the control-treated Sod2loxP/loxP RPE-cre mice and those treated at either dose level of xaliproden was apparent. The RPE of vehicle-treated mice was thinner and exhibited abundant vacuoles that contained membrane fragments. The increase in RPE vacuolization extended along the entire periphery of the retina as revealed with light microscopy (Appendix 5). Nuclei of the RPE cells were irregular and condensed, typical of dying cells, and the in-foldings at the basal surface of the RPE were distended. Photoreceptor outer segments were broken and contained cystic spaces. The stacking of disc membranes in the rod outer segments appeared disrupted. In contrast, the RPE of xaliproden-treated mice did not contain massive vacuoles, and the basal in-foldings appeared condensed. Photoreceptor outer segments were continuous, and disc membranes were compact and regularly stacked. At higher magnification (Figure 10), the RPE of xaliproden-treated mice exhibited more mitochondria along the basal surface and vertically displaced melanosomes typical of normal RPE, features lacking in the control mice. Treatment with the drug also reduced the accumulation of lipofuscin-containing vesicles seen in the vehicle-treated mice.
The induction of antioxidant (Table 1) and anti-inflammatory (Table 2) genes by xaliproden may explain the protective properties of this drug in models of peripheral neuropathy  and the anti-inflammatory effects of the drug in the EAE model . These properties provided hope that xaliproden would be a useful therapeutic for ALS, a hope that was unfulfilled in clinical trials. Nevertheless, safe and tolerable doses of this drug were identified, and in phase III trials, the drug had a small positive effect on clinical outcome. In a clinical trial of xaliproden for peripheral neuropathy caused by chemotherapy for colon cancer, xaliproden decreased neurotoxicity by 39%. Side effects of the drug included diarrhea, dizziness, anxiety, tinnitus, and vertigo and are serotonin-related . These side effects are similar to the side effects of the 5HT1a agonists currently on the market. The fact that the compound can be taken by mouth and has an established clinical profile encouraged us to consider xaliproden for treatment of age-related macular degeneration.
We tested this compound on RPE-derived cells in culture subjected to oxidative and proinflammatory stimuli and in a mouse model of RPE and retinal atrophy based on mitochondrial oxidative stress in the RPE. RPE oxidative stress that increases with age and the resultant chronic subacute inflammation are considered important contributory factors for AMD. At 0.25 mM, paraquat actually extends the longevity of C. elegans grown in culture possibly with induction of Hif-1 and AMPK . In ARPE-19 cells, paraquat alone induced the production of metallothionein and the Nrf2-regulated genes NqO1 and GSTM1. Others have noted that AMPK activation stimulates Nrf2 signaling in mammalian cells [45,46]. Nevertheless, elevation of antioxidant pathways was not sufficient to prevent the toxic effects of paraquat on ARPE-19 cells, but increased stimulation of antioxidant genes and metallothionein by the combination of paraquat and xaliproden resulted in increased survival of the same cells (Figure 1).
We have reported similar results for another 5HT1a agonist, 8-OH-DPAT , but in that study we did not examine the response of cultured RPE cells to proinflammatory stress. Here we report treatment of ARPE-19 cells with TNF-α led to the increased synthesis of IL-1β, IL-6, CCL20, and VEGF-A. This impact was mitigated by xaliproden (Table 2). Furthermore, TNF-α led to a breakdown of the tight junctions between RPE cells, and the integrity of the monolayer was preserved by xaliproden (Figure 5). Terasaki et al. studied the impact of TNF-α in primary porcine RPE cells and reported a difference in outcome depending on whether the cells were allowed to become polarized or not . They reported that in both conditions, TNF-α stimulated the production of NF-κB p65 and that nuclear accumulation was enhanced in polarized cells. Chronic sub-acute inflammation in the RPE and retina is considered a contributory factor to the development of AMD , and our cell culture data suggest that xaliproden may be an effective treatment to curtail inflammatory and angiogenic signaling.
In our original description of the Sod2loxP/loxP RPE-cre mice as a model for geographic atrophy , we detected significant thinning of the ONL only at the 6-month time point, but we did not assess mice at 5 months, the age reported here (Figure 7B). At this age, treatment of mice with either dose level of xaliproden led to an increased thickness of the ONL relative to the control-treated mice, which we attribute to increased survival of photoreceptors. Increased thickness at this age correlated with improved visual acuity (Figure 7A) but not with improved scotopic ERG response (Figure 8). As noted above, optokinetic analysis appears to be more sensitive to small improvements in retinal function than does full-field ERG. We observed a depression in the photopic b-wave amplitude after 4 months of treatment at the high dose of xaliproden. This treatment did not correlate with damage evident by electron microscopy. Alterations in central serotonin neurotransmission can alter the implicit time of the photopic b-wave , and it is possible that the chronic exposure to high levels of xaliproden elicited an adaptive physiologic response that was not evident by changes in microscopic anatomy. Serotonin agonists change circadian responses to light, and chronic administration of the 5HT1a agonist buspirone caused significant changes in circadian patterns of locomotion . Since photoreceptor disc shedding is governed by circadian rhythm, it is conceivable that this process was also impacted by chronic administration of a 5HT1a agonist.
Histology and electron microscopy revealed a major impact of xaliproden on the Sod2-deleted mice (Figure 9). A masked observer correctly differentiated treated from untreated animals based on inspection of light micrographs (Appendix 5). What distinguished eyes from untreated mice from either treatment group was the presence of numerous vacuoles in the RPE layer. Some vacuoles or separations between RPE cells were present in all of the mice, but the extent of vacuolization was far greater in the mice treated with vehicle alone. We present these results qualitatively, because accurate measurement of vacuole numbers would require serial sectioning of eyes from several mice per group, and the workforce for this effort was not available. Study of higher magnification images of the RPE revealed the normal appearance of the basal surface of the RPE in xaliproden-treated mice, relative to the disorganized and osmophilic basal in-foldings of the vehicle-treated eyes. Similar damage to this structure has been reported in other mouse models of AMD based on inflammatory stimulation or oxidative stress [38,51-54].
Although we did not detect significant shortening of photoreceptor outer segments in the control mice (data not shown), the rod outer segments appeared broken and sometimes contained large vacuoles. Treating the mice with xaliproden prevented this photoreceptor damage. Accumulation of broken tips of photoreceptors may reflect a defect in the phagocytic function of RPE cells or in the function of RPE lysosomes in mice with unrelieved mitochondrial oxidative stress. We hope to test this hypothesis in primary RPE cultures derived from Sod2loxP/loxP RPE-cre and control mice, and to study the effects of 5HT1a agonists in this experimental system.
These data suggest that xaliproden may be an effective treatment for preventing RPE loss leading to geographic atrophy. Remaining questions concerning the use of this drug for this indication include (1) whom to treat and (2) whether the serotonergic side effects of the drug will be tolerable. We suggest that patients with mid-stage disease presenting with large irregular drusen would be candidates for treatment. Whether the side effects are acceptable depends on the dose and on the patient. One approach to avoiding side effects would be to deliver xaliproden locally instead of systemically. Recent advances in drug delivery to the back of the eye may make that route feasible [55-57].
Appendix 1. STR analysis.
Appendix 2. Primers for qPCR
Appendix 4. ERG wave forms
Appendix 6. Tukey’s multiple comparison test of data from Table 1.
Grant support came from the BrightFocus Foundation (M2012029) and from the NEI core grant to the University of Florida (P30 EY02172). Initial support for the mouse model came from the Macula Vision Research Foundation. Funding was also provided by the Shaler Richardson Professorship endowment. We thank Dr. Eduardo Candelario-Jalil for allowing us to use his voltohmmeter to measure transepithelial resistance and to Dr. John Crabb for the gift of CEP antibodies.