New GABA modulators protect photoreceptor cells from light-induced degeneration in mouse models

Abstract

No clinically approved therapies are currently available that prevent the onset of photoreceptor death in retinal degeneration. Signaling between retinal neurons is regulated by the release and uptake of neurotransmitters, wherein GABA is the main inhibitory neurotransmitter. In this work, novel 3-chloropropiophenone derivatives and the clinical anticonvulsants tiagabine and vigabatrin were tested to modulate GABA signaling and protect against light-induced retinal degeneration. Abca4^−/−Rdh8^−/− mice, an accelerated model of retinal degeneration, were exposed to intense light after prophylactic injections of one of these compounds. Imaging and functional assessments of the retina indicated that these compounds successfully protected photoreceptor cells from degeneration to maintain a full-visual-field response. Furthermore, these compounds demonstrated a strong safety profile in wild-type mice and did not compromise visual function or damage the retina, despite repeated administration. These results indicate that modulating inhibitory GABA signaling can offer prophylactic protection against light-induced retinal degeneration.—Schur, R. M., Gao, S., Yu, G., Chen, Y., Maeda, A., Palczewski, K., Lu, Z.-R. New GABA modulators protect photoreceptor cells from light-induced degeneration in mouse models.

Age-related macular degeneration (AMD) is a leading cause of blindness in the elderly population (1). Currently, no treatments are available for individuals with dry AMD or other types of nonexudative retinal dystrophies, such as Stargardt disease. Early in the pathogenesis of retinal degeneration, drusen accumulate under the retina and alterations appear in the retinal pigment epithelium (RPE). Aberrant processing of retinoids in the visual cycle causes buildup of toxic byproducts of this pathway, such as all-trans retinal, which can also condense into bisretinoids and accumulate lipofuscin within the lysosomal compartments of the RPE (2). Toxic retinoid byproducts place continuous oxidative stress on these cells, fueling chronic inflammation. Ultimately, these processes trigger apoptosis of both photoreceptors and the RPE, leading to large areas of cell death in the form of geographic atrophy (3, 4). New treatment strategies for dry AMD focus on preventing the loss of RPE and photoreceptor cells (5). These approaches include antioxidant supplementation (6, 7); visual cycle modulators that slow the visual cycle to limit lipofuscin formation (8–11); and neuroprotection with growth factors (12, 13), antiamyloid-β therapies (14), and serotonin receptor agonists (15–17). Unfortunately, many of these compounds have failed in clinical trials or caused visual system toxicity (18–20). New approaches are needed to safely prevent retinal degeneration.

Communication between retinal neurons is dominated by the neurotransmitter-mediated chemical signaling (21, 22) that occurs at the synaptic terminals in the outer and inner plexiform layers. Excitatory signaling progresses vertically through the retina and is mediated primarily by glutamate at photoreceptor–bipolar cell and bipolar–ganglion cell junctions. Inhibitory signaling progresses laterally via horizontal cells and amacrine cells and is primarily mediated by GABA. Extracellular availability of neurotransmitters is further regulated by uptake and recycling in Müller glia (23). Additional neuroactive amino acids, including glutamate, glutamine, aspartate, taurine, and dopamine, are crucial for the metabolism of neurotransmitters and provide pools of monocarboxylates that feed into the tricarboxylic acid (TCA) cycle (22, 24, 25).

Cellular processes that regulate neurotransmitter availability and metabolism are disrupted during retinal degeneration. Glial cells, that sequester neurotransmitters to reduce their extracellular availability, become activated and undergo gliosis in response to retinal degeneration (23). Furthermore, cellular metabolism, fueled by the TCA cycle that obtains its monocarboxylate substrates from neurotransmitter pools via the GABA shunt, is affected during degeneration, resulting in oxidative and nonoxidative stress (26, 27). In conjunction with these processes, concentrations of neuroactive amino acids are also sensitive to exposure to light (28) and retinal degeneration (27, 29–33). Thus, we suspect that pharmacologic manipulation of retinal neurotransmitter function protects the retina under degenerative conditions.

GABA modulatory drugs are clinically available as anticonvulsants. In patients with epilepsy, overexcitation of neurons causes seizures, which are controlled pharmacologically by decreasing excitatory signaling and increasing inhibitory signaling (34). Glutamate excitotoxicity contributes to retinal degeneration during ischemic damage that has been implicated in the pathogenesis of diseases such as glaucoma and retinal ischemia (35), and reports have demonstrated that anticonvulsants can prevent excitotoxic damage to neural ganglion cells in the inner retina. For example, neuroprotection against NMDA- and monosodium glutamate–induced excitotoxicity in ganglion cells has been reported for clinically validated anticonvulsants, including tiagabine (TGB) (36), vigabatrin (VGB), topiramate (37), and valproic acid (38, 39), as well as endogenous neuroprotective agents, such as taurine (40) and pituitary adenylate cyclase–activating polypeptide (41, 42). Several of the agents that act on neurotransmitter pathways, including pregabalin (43), valproic acid (44), and tauroursodeoxycholic acid (45, 46), also have prevented photoreceptor loss in genetic models of retinal degeneration.

In this work, GABA modulatory compounds were evaluated for prevention of photoreceptor loss in mouse models of light-induced retinal degeneration. Clinical anticonvulsants and novel 3-chloropropiophenone derivatives provided structural and functional protection of the retina in a light-sensitive Abca4^−/−Rdh8^−/− mouse model of retinal degeneration. Moreover, the chloride derivatives demonstrated a safe profile and effective protection against light-induced retinal degeneration in wild-type BALB/c mice.

MATERIALS AND METHODS

Materials

The chemical names, abbreviations, and doses of the chloride derivatives are listed in Table 1, and their structures are shown in Fig. 1A. CL-1, -2, and -3 were purchased from MilliporeSigma (Billerica, MA, USA); CL-4 was purchased from Oakwood Chemical (Estill, SC, USA); and CL-5, TGB, and VGB were from Toronto Research Chemicals (Toronto, ON, Canada).

TABLE 1.

Compounds tested for protection from light-induced retinal degeneration in Abca4^−/−Rdh8^−/− double-knockout mice

No.	Compound name	Dose males (mg)	Dose females (mg)
CL-1	3-Chloropropiophenone	0.2	1
CL-2	(R)(+)3-Chloro-1-phenyl-1-propanol	0.2	1
CL-3	(S)(−)3-Chloro-1-phenyl-1-propanol	0.2	1
CL-4	3-Chloro-1-(3-methylphenyl)-1-oxopropanone	0.2	0.5
CL-5	3-Chloro-1-(4-hydroxyphenyl)-propan-1-one	0.5	1

Figure 1.

Chloride derivatives for inhibition of GABA-AT. A) Structures of the chloride derivatives. B) Michaelis-Menten kinetics of GABase in the presence of various concentrations of GABA substrate and 10 µM chloride derivatives. V₀ plotted as a function of GABA substrate concentration.

Enzyme studies

All materials for enzyme studies were purchased from MilliporeSigma. Enzyme conditions were modified from a published protocol (47, 48). Experiments were conducted at 25°C in Tris-HCl buffer (80 mM, pH 9.0), containing 750 mM sodium sulfate, 10 mM 2-ME, 2 mM α-ketoglutarate, 1.4 mM NAD⁺, 300 µg/ml GABase [1:1 mixture of GABA-aminotransferase (AT) and succinic semialdehyde dehydrogenase (SSADH)], and various concentrations of GABA. Chloride derivatives were dissolved in DMSO and added to the reaction at a final concentration of 10 μM (1% v/v DMSO) and were mixed in a 96-well plate (100 µl/well). Experiments were performed in triplicate. NAD⁺→NADH turnover was monitored spectrophotometrically every 30 s for 30 min (λ_ex = 340 nm, λ_em = 460 nm). To measure Michaelis-Menten parameters, the GABA concentration was varied from 0 to 20 mM, and V₀ was calculated from the slope of the first 10 min of enzymatic turnover. Enzyme kinetics were fit to the Michaelis-Menten equation (MatLab; MathWorks, Natick, MA, USA) (Eq. 1):

Spectrophotometric measurements were acquired in clear 96-well plates (Corning Costar; Thermo Fisher Scientific, Waltham, MA, USA) in a SpectraMax M5 plate reader with fluorescence output (Molecular Devices, CA, USA).

Animals

Male and female Abca4^−/−Rdh8^−/− (RRID:MGI:4410294) double-knockout mice were generated (43), and both male and female mice were genotyped, using published protocols, to ensure homozygosity for the Leu450 allele of Rpe65 (49) and the absence of Crb1/rd8 and Pde6/rd1 mutations (50). This mouse model is light sensitive, and retinal degeneration can be accelerated by exposure to intense light (2, 51, 52). Male wild-type BALB/c (RRID:IMSR_JAX:000651) and C57BL/6 (RRID:IMSR_JAX:000664) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). All animal interventions were initiated when mice were 4 wk old. The animals were housed and bred in the Animal Resource Center at Case Western Reserve University, and were cared for according to an approved protocol by the Case Western Reserve University Institutional Animal Care and Use Committee and in compliance with recommendations from the American Veterinary Medical Association Panel on Euthanasia and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Treatment experiments

For assessment of treatment efficacy, test compounds dissolved in 60 µl DMSO were administered by intraperitoneal injection into 4-wk-old mice 30 min before light exposure. Pilot dose-determination experiments were performed with injections ranging from 0.1 mg (6.67 mg/kg) to 2 mg (133.3 mg/kg) per mouse. Vehicle control animals were given 60 μl DMSO by the same procedures as were used for the drug-treated mice. After the application of 1% tropicamide for pupil dilation, the animals were placed in a white bucket. Abca4^−/−Rdh8^−/− mice were illuminated at 10,000 lux for either 30 min (males) or 45 min (females; Supplemental Fig. 1) (n = 5 mice/group; males and females evaluated separately). BALB/c mice were illuminated at 5000 lux for 2 h (males, n = 3/group). No more than 2 mice were placed in each bucket during the exposure to light.

Optical coherence tomography

Images of mouse retinal structures were obtained in vivo with ultra–high-resolution spectral domain–optical coherence tomography (OCT) (Envisu C2200; Bioptigen, Irvine, CA, USA) (53). Images were acquired 3 d after treatment and exposure to light, with follow-up images acquired after 8 d. Before imaging, mice were anesthetized by injection of a cocktail (10 µl/g body weight, i.p.) of ketamine (6 mg/ml) and xylazine (0.44 mg/ml) in PBS buffer [10 mM sodium phosphate (pH 7.2) and 100 mM NaCl], and treated with 1% tropicamide for pupil dilation. Twenty images acquired in the B-scan mode were registered and averaged in MatLab to construct a final spectral domain–OCT image.

Scanning laser ophthalmoscopy

In vivo whole-fundus and autofluorescence imaging of mouse retinas was performed by scanning laser ophthalmoscopic (SLO) imaging in the autofluorescence mode (54) (HRAII; Heidelberg Engineering, Heidelberg, Germany) 7 d after treatment and exposure to light. Mice were anesthetized and pupils were dilated, according to the protocol used for OCT imaging. The number of autofluorescent spots in each image was counted manually.

Electroretinograms

Electroretinograms (ERGs) were acquired (55) 10 d after the treatment and exposure to light. Mice were anesthetized and dark adapted for at least 2 d before ERG recording, which was also performed in a dark room. Three electrodes were placed on the animal: a contact lens electrode on the eye, a reference electrode underneath the skin between the ears, and a ground electrode underneath the skin of the tail. ERGs were recorded with the universal electrophysiologic system UTAS E-3000 (LKC Technologies, Gaithersburg, MD, USA). Light intensity, calibrated by the manufacturer, was computer controlled. The mice were placed in a Ganzfeld dome, and photopic and scotopic responses to flash stimuli were obtained from both eyes simultaneously. A- and b-wave amplitudes were calculated in MatLab.

Histology

Mice were euthanized after ERG experiments. Whole eyes were harvested, fixed in a solution of 4% paraformaldehyde and 1% glutaraldehyde overnight, and processed and embedded in paraffin in a Tissue-Tek VIP automatic processor (Sakura Finetek, Torrance, CA, USA). Sections, 5-µm thick, were cut and stained with hematoxylin and eosin (H&E). Stained slides were imaged on a BV100 bright-field microscope (Olympus, Center Valley, PA, USA).

Safety experiment

Male albino (BALB/c) 4-wk-old mice were injected daily for 28 d with 0.2 mg (10 mg/kg, i.p.) CL-4 or -5 dissolved in a mixture of 1:1 PBS:ethanol or vehicle control (n = 5/group). Body weight was measured daily after the injection. Mice were housed in a 12-h light–dark cyclic environment for the first 26 d. OCT images were acquired on d 26. Mice were then transferred to a dark room for 2 d and dark adapted before ERG recording on d 28, and all subsequent animal handling was performed in the dark. Mice were euthanized after ERG testing, and their eyes were collected and prepared for histology as previously described.

Data and statistical analyses

OCT images and ERG traces were visualized and analyzed in MatLab. From the OCT images, the outer nuclear layer (ONL) thickness was quantified at 150, 300, 450, and 600 µm from the optic nerve head (ONH) in the temporal–nasal and superior–inferior directions. A- and b-waves were calculated from ERG traces by using a custom MatLab script. Statistical analyses were performed using a Student’s t test, and statistical significance was set at P ≤ 0.05.

RESULTS

GABA-AT inhibition

Before in vivo testing for therapeutic efficacy, the inhibitory activity of the chloride derivatives on GABA-AT was first verified by an in vitro enzyme assay (Fig. 1B). Breakdown of GABA by GABA-AT was detected via the reporter enzyme SSADH, which couples the GABA-AT–catalyzed reaction to the fluorescent product of NAD⁺→NADH (λ_em = 460 nm). GABA-AT inhibition has been demonstrated for CL-5 (56). The chloride derivatives listed in Table 1 exhibited significant inhibition of GABA metabolism, as demonstrated with a commercially available GABase enzyme system (MilliporeSigma). Inhibition of the GABase system reduces the fluorescence output by lowering NADH accumulation during the reaction. The enzyme inhibition was monitored with various concentrations of the GABA substrate and 10 μM of test compound. Relative to the vehicle, all the chloride derivatives reduced the V_max in the GABase assay, as shown in Table 2. CL-5 evidenced the greatest inhibition, followed by CL-3, -2, -4, and -1.

TABLE 2.

Enzyme kinetic parameters calculated from Michaelis-Menten kinetics

Chloride derivative	Vmax	Km
No inhibitor	193.7	1.37
CL-1	101.7	0.13
CL-2	86.67	0.02
CL-3	67.97	0.06
CL-4	101.1	0.47
CL-5	52.21	0.26

Retinal protection by chloride derivatives

Before treatment efficacy studies, differences in susceptibility to light-induced retinal degeneration were observed in male and female mice. Abca4^−/−Rdh8^−/− male and female mice were injected with vehicle control (DMSO), and retinal degeneration was evaluated in response to exposure to light for different times at a constant intensity (10,000 lux) (Supplemental Fig. 1). In male mice, OCT images acquired 3 d after exposure to light demonstrated that 30 and 45 min of exposure caused ONL thinning and therefore photoreceptor degeneration. By contrast, OCT images of female mice demonstrated ONL thinning after 45 and 60 min of exposure, but not after 30 min of exposure. These results suggest that the male mice were more susceptible than female mice to retinal degeneration and must be evaluated separately for retinal protection therapies. Based on these results, the lowest amount of light exposure that would cause photoreceptor degeneration was used for subsequent therapeutic efficacy studies (30 min for males, 45 min for females).

To determine the dose range necessary for retinal protection, we performed pilot experiments independently in male and female mice for each compound. OCT imaging was performed 3 d after treatment and exposure to light to evaluate the extent of degeneration (Supplemental Fig. 2). For CL-1, -2, and -3, retinal protection was achieved by a dose of 0.2 mg (13.3 mg/kg) in male mice and 1 mg (66.7 mg/kg) in female mice. For CL-4, retinal protection was achieved with a dose of 0.2 mg (13.3 mg/kg) in males and 0.5 mg (33.3 mg/kg) in females. For CL-5, a dose of 0.5 mg (33.3 mg/kg) was necessary for retinal protection in males and 1 mg (66.7 mg/kg) for females. These lowest effective dosages were selected for further animal studies.

All the chloride derivatives showed effective retinal protection against light-induced degeneration, as revealed by OCT and histology evaluations (Fig. 2). The OCT images correlated well with the corresponding histologic images of the retina’s integrity and ONL thickness in all groups. Negative controls pretreated with DMSO before exposure to light demonstrated significant reductions in the thickness of the ONL containing the photoreceptor nuclei vs. the ONL of mice in the no-light-illumination (NLI) group (Fig. 2E). In contrast, all the layers of the retina remained intact, and the ONL thickness was similar to that of NLI mice in each of the treatment groups.

Figure 2.

Preservation of retinal structure in Abca4^−/−Rdh8^−/− mice pretreated with chloride derivatives before exposure to light. A) Experimental schedule for treatment, light exposure, and evaluations. B) OCT images of retinal cross-sections. C) H&E staining of retinal tissue slices. D) SLO images of autofluorescent granules in the subretinal space. E) ONL thickness measured from OCT images in the nasal–temporal and inferior–superior directions. NLI control data are shown as gray shading. P < 0.05, DMSO vs. NLI group. F) Counts of autofluorescent granules from SLO images. P < 1^E5 chloride-treated vs. DMSO-treated eyes. B, C) Asterisks denote the ONL. Error bars ± sd. Scale bars, 50 µm.

The ability of the chloride derivatives to protect the retina from oxidative damage was also tested. In DMSO-pretreated control mice, SLO images revealed bright autofluorescent granules that were highly concentrated and dispersed throughout the subretinal space (480 ± 187 spots/eye), indicating widespread retinal damage (Fig. 2D). By contrast, SLO images of NLI controls and mice pretreated with the chloride derivatives mainly lacked autofluorescent granules in the RPE, supporting the OCT assessments indicating protection from cellular damage during exposure to intense light (Fig. 2F). In the treated mice, a small number of autofluorescent granules (mean, 1–6 spots/eye;) were visible, mainly at the optic nerve site in the dark center of the image.

Visual function was also assessed, and in agreement with the imaging results, full protection of the retina was found in all treatment groups. Representative ERG recordings are shown in Fig. 3A. NLI control mice produced strong responses to light flashes during ERG testing, whereas the DMSO control mice produced comparatively weak responses. The a- and b-wave amplitudes, corresponding to the photoreceptor and inner retinal light responses, respectively, were quantified from these traces (Fig. 3B). Against a dark background, scotopic a- and b-wave ERG amplitudes in the chloride-treated mice were similar to those in the NLI controls, but significantly higher than the amplitudes in the DMSO-treated mice. Against a light background, photopic b-wave ERG amplitudes of treated groups were slightly lower than in the NLI controls, but were significantly higher than those in the DMSO-treated group. Photopic a-wave responses were similar across all groups.

Figure 3.

Preservation of visual function in Abca4^−/−Rdh8^−/− mice pretreated with chloride derivatives before exposure to light. A) Representative traces of ERG responses under scotopic and photopic conditions. B) Scotopic and photopic a- and b-waves quantified from ERG recordings. No illumination control data shaded between error bars. Error bars ± sd. P < 0.001, scotopic ERGs; P < 0.01, photopic ERGs, in chloride-treated vs. DMSO-treated mice.

Retinal protection by clinical anticonvulsants

Two clinically available anticonvulsant agents were also tested for their ability to protect the retina from light-induced damage. TGB and VGB, both of which promote GABAergic signaling, protected retinal structure and function during exposure to intense light (Fig. 4). OCT and histology assessments demonstrated that a thick ONL was preserved in animals pretreated with 0.05 mg TGB (3.33 mg/kg) or 1 mg VGB (66.7 mg/kg) before exposure to light. These OCT results were supported by SLO images which revealed few autofluorescent granules in the subretinal space. Finally, ERG recordings indicated that the visual response was preserved in mice pretreated with TGB or VGB (Fig. 5).

Figure 4.

Preservation of retinal structure in Abca4^−/−Rdh8^−/− mice pretreated with clinical anticonvulsants before light exposure. A) OCT images of retinal cross-sections. B) H&E staining of retinal tissue slices. Asterisks denote the ONL. C) SLO images of autofluorescent granules in the subretinal space. Images from control animals are repeated from Fig. 2 for clarity. Scale bars, 50 µm.

Figure 5.

Preservation of retinal structure in Abca4^−/−Rdh8^−/− mice pretreated with clinical anticonvulsants before exposure to light. A) Representative traces of ERG responses under scotopic and photopic conditions. B) Scotopic a-waves and photopic b-waves quantified from ERG recordings. Recordings from control animals are repeated from Fig. 2 for comparison. Error bars, ±sd.

Retinal protection in wild-type mice

To demonstrate the effectiveness of the chloride derivatives in a wild-type model of light-induced degeneration, the efficacy of pretreatment with CL-4 was tested in BALB/c mice. Similar to the Abca4^−/−Rdh8^−/− double-knockout model, BALB/c mice undergo retinal degeneration in response exposure to intense light (10). Retinal protection was evident in the albino BALB/c mice pretreated with 0.2 mg CL-4 (Fig. 6). OCT images also revealed an ONL with a thickness similar to that of NLI controls. By contrast, OCT images of control mice pretreated with DMSO demonstrated an inflamed, thinner, degraded ONL caused by exposure to light. ERG recordings demonstrated decreased scotopic a- and b-wave amplitudes in DMSO-treated controls, whereas mice treated with CL-4 produced a full response.

Figure 6.

Retinal protection in wild-type BALB/c mice pretreated with CL-4 before exposure to light. A) OCT images of NLI, DMSO-treated, and CL-4 treated mice. Asterisks denote the ONL. Representative ERG traces (B) and quantified a- and b-waves (C). Error bars ± sd.

Safety

Toxicity to the retina was assessed for the 2 leading chloride derivatives, CL-4 and -5, after single and repeated doses in wild-type mice. To determine whether the chloride derivatives compromised visual function, full-field ERG recordings were collected 30 min after the administration of a single high dose (1 mg; 66.7 mg/kg) of CL-4 or -5 in C57BL/6 mice. ERG responses in both drug treatment groups were similar to responses in vehicle-injected controls, and no significant differences in a- and b-wave amplitudes were measured under photopic or scotopic conditions (Fig. 7).

Figure 7.

Safety profiles of single injections of a high-dose (1 mg/mouse) of CL-4 or -5 in C57BL/6 mice. Representative full-field ERG traces (A) and quantified a- and b-wave amplitudes (B) under photopic and scotopic conditions recorded 30 min after injections. Error bars ± sd.

Repeated doses of CL-4 and -5 for 28 d had no deleterious effects on the retina or the overall well-being of the BALB/c mice. Body weight measurements recorded daily over the course of this study are shown in Fig. 8A. With the exception of 2 animals, DMSO control and drug-treated mice exhibited a modest weight gain during the study. One animal from the DMSO group rapidly lost body weight after d 3 and was euthanized at d 6. Likewise, 1 CL-5–treated mouse lost weight but stabilized by d 7; however, this animal was euthanized on d 15 because of its weak appearance. All other mice maintained a healthy weight during the study period.

Figure 8.

Safety profiles after daily administration of chloride derivatives by IP injection in male BALB/c mice for 28 d. A) Body weights of mice injected with vehicle, 0.2 mg CL-4, or 0.2 mg CL-5 for the course of the experiment. B) H&E staining of retinas harvested at d 28 of injections. C, D) In vivo OCT images (C) and quantified ONL thickness (D) of retinal cross-sections imaged at d 26 of injections. E, F) Representative ERG traces (E) and quantified a- and b-wave amplitudes (F) measured at d 28 of injections. ONH, optical nerve head. Error bars, ±sd.

OCT images taken on d 26 indicated that the structure of the retina remained intact during treatment (Fig. 8C, D). The ONL containing photoreceptor nuclei was of a normal thickness (∼60 μm), with no significant differences between treatment groups and the controls. Histologic images of retinal cross-sections collected at the conclusion of the study are shown in Fig. 8B. Histologic findings were in agreement with the OCT images and revealed fully preserved retinal structures in control and treatment groups. Full-field ERG recordings taken after 28 d of injections are shown in Fig. 8E, F. In both treatment groups and the control group, no significant differences were measured in a- or b-wave amplitudes. The ERG responses in the animals agreed with the maintenance of retinal structure observed in OCT and histology images.

DISCUSSION

There have been substantial efforts to develop effective therapeutics for preventing nonexudative retinal degenerative diseases. These include sequestration of toxic all-trans retinal by primary amines (43, 57); increasing clearance of retinoid metabolites with visual cycle inhibitors (10, 11); modulating GPCR signaling (54); and the use of neuroprotective agents (58), antioxidants (6), and gene therapy (59, 60). Many of these approaches have been successful in mice. However, in human clinical trials, inhibitors have slowed the visual cycle (61, 62) and antioxidants failed to prevent the onset of early symptoms of AMD. Neuroprotective agents, such as ciliary neurotrophic factor are in ongoing clinical trials, but their need for local administration by direct intraocular injection is invasive and not well suited for repeated administration. Complex approaches, such as gene or cell-based therapies, present many hurdles to production and clinical use, when compared to systemic administration of small-molecule drugs.

We present a library of small-molecule compounds that efficiently protect the retina from light-induced damage and can be administered repeatedly without causing any apparent damage to the retina. Most of the chloride compounds generated similar effects in in vitro enzyme assays and in in vivo treatment efficacy experiments. The 2 chloride derivatives with functional groups on the phenyl ring of the parent compound, CL-4 and -5, were outliers in their efficacy. Specifically, CL-4 showed highest in vivo efficacy, whereas CL-5 showed the highest in vitro efficacy, yet the lowest in vivo efficacy. We attribute this discrepancy to biodistribution of the compounds, where we anticipate that the less polar methylated CL-4 experiences better distribution than the more polar hydroxylated CL-5. In addition, we observed higher sensitivity to light-induced damage in male mice vs. female mice. In future work, pharmacokinetic studies may aid in understanding the relationship of doses and distribution with extent of degeneration.

In this work, the potential effectiveness of several novel and clinically validated agonists of GABA signaling were evaluated for their protection of photoreceptors from light-induced retinal degeneration. GABA is the main inhibitory neurotransmitter in the retina where it plays a dual role in modulating the transmission of visual signals through the retina while shunting substrates into the TCA cycle for cellular metabolism. GABA is released by horizontal cells, whereby it regulates synaptic communication between photoreceptor and bipolar cells in the outer plexiform layer. GABA is also released by amacrine cells to regulate bipolar–ganglion cell communication in the inner plexiform layer and is recycled by Müller cells that span the length of the retina. GABA metabolism feeds into the TCA cycle for cellular metabolism via GABA-AT–related breakdown of GABA into succinic semialdehyde and then into succinate (32).

Specifically, the use of new GABA-AT inhibitors and clinical anticonvulsants was explored as an alternative strategy for retinal protection against light-induced degeneration. Anticonvulsants act on a spectrum of targets to inhibit excitatory neurotransmitter action and protect neural cells from excitotoxic damage. These drugs act on glutamate pathways (e.g., felbamate and topiramate), ion channels (e.g., benzodiazepines, gabapentin, and phenytoin), and GABA pathways (e.g., VGB, TGB, and valproic acid) (34). Anticonvulsants have demonstrated neuroprotective effects on retinal ganglion cells that prevent excitotoxic damage during ischemic injury (36, 37, 39, 40). Models of excitotoxic retinal damage typically involve inner retinal disease, such as glaucoma, in which the ganglion cell layer is damaged by overexcitation of NMDA or kainic acid receptors, which can be counterbalanced by pharmacologic supplementation of inhibitory signaling.

When considering the translational potential of anticonvulsant used to prevent retinal degeneration, safety is a primary concern. Many anticonvulsants have the side effect of visual impairment, although this adverse effect differs in severity (63, 64). Whereas most of these drugs have side effects that are infrequent and reversible, VGB, a mechanism-based inactivator of GABA-AT, is known to cause irreversible visual field defects after repeated administration in animals (65, 66) and in patients (67, 68). TGB has less severe clinical side effects, such as altered color perception, and these effects are reversible once drug treatment is terminated. Because of these safety concerns, it was necessary to determine whether administration of our chloride derivatives would elicit similar adverse effects on the retina. ERG recordings after single or repeated doses of the lead chloride derivatives did not perturb retinal function. Furthermore, OCT and histologic images of retinal structure showed no signs of the degeneration or plasticity that were observed with VGB (65, 69). These promising results suggest that these chloride derivatives do not pose an acute risk to the retina. Future, comprehensive safety studies are necessary to obtain a full profile of their potential adverse effects.

In this model of light-induced retinal degeneration, a pathologic underpinning for excitotoxic damage remains to be elucidated. In one rodent model of retinitis pigmentosa, increased levels of GABA were found to accompany retinal degeneration (32). However, others have shown no difference in quantity or distribution of GABA as the retina degenerates (27, 28). Indeed, strict quantification of the amount of the various neurotransmitters in the retina does not necessarily correlate with their signaling activity, because of the complex metabolic interplay and rapid uptake by glial cells. However, there is precedent in the literature that implicates the GABAergic pathway in retinal damage. In patients with AMD, genome-wide association studies have identified mutations in SSADH, the enzyme following GABA-AT in the GABA shunt to the TCA cycle (70). Valproic acid, a clinically used anticonvulsant that increases GABAergic signaling, has entered clinical trials for treatment of retinitis pigmentosa with the promise of promoting correct folding of rhodopsin mutants in the dominant form of this disease (38, 39, 44). These trials have generated mixed results, with conflicting reports of visual acuity improvement in some patients, whereas others experienced visual toxicity related to the treatment (19, 71). Serotonin receptor (5-HT_1A) agonists, a class of drugs also known to possess neuroprotective functions against NMDA excitotoxicity, have also demonstrated a protective effect against light-induced retinal degeneration in animals (15, 72). Unfortunately, a recent phase III clinical trial, the Geographic Atrophy Treatment Evaluation (GATE) trial, showed no improvement in delaying lesion growth in patients with geographic atrophy secondary to AMD (20). However, this failure of drug efficacy may have been caused by factors other than the therapeutic strategy, such as the advanced stage of disease or transport problems with the topical formulation. Because of failures of neuromodulatory drugs to ameliorate AMD progression, future work should include a rigorous study of the mechanism of retinal protection by these drugs. Appropriate stage of disease, clinical endpoints for efficacy, and in vitro and animal models must be established before these drugs can achieve their translational potential.

In summary, this work demonstrates that modulation of inhibitory neurotransmitters in the retina could provide a new strategy for neuroprotection from light-induced damage. 3-Chloropropiophenone derivatives, a new class of GABA-AT inhibitors, effectively protected retinal structure and function in Abca4^−/−Rdh8^−/− mice that were treated before exposure to intense light. Clinical anticonvulsants that act on the GABA shunt were also shown to provide retinal protection in this model. Furthermore, the lead chloride derivative CL-4 replicated this protective effect in wild-type BALB/c mice. Preliminary evaluations revealed excellent safety profiles with little systemic or visual toxicity after single or repeated doses with the 2 lead compounds.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

AMD

age-related macular degeneration

ERG

electroretinography

GABA-AT

GABA aminotransferase

NLI

no light illumination

OCT

optical coherence tomography

ONH

optic nerve head

ONL

outer nuclear layer

RPE

retinal pigment epithelium

SLO

scanning laser ophthalmoscopy

SSADH

succinic semialdehyde dehydrogenase

TCA

tricarboxylic acid

TGB

tiagabine

VGB

vigabatrin

AUTHOR CONTRIBUTIONS

R. M. Schur was involved in all aspects of this work, including study design and execution and data analysis; S. Gao and Z.-R. Lu conceived the project; S. Gao, G. Yu, and Y. Chen performed pilot experiments; S. Gao performed ERG examinations; A. Maeda and K. Palczewski provided assistance with animal models and in vivo assays; R. M. Schur and Z.-R. Lu wrote the paper; and all authors contributed to the study design and manuscript preparation and have read and approved the final version.