Systemic administration of the di-apocarotenoid norbixin (BIO201) is neuroprotective, preserves photoreceptor function and inhibits A2E and lipofuscin accumulation in animal models of age-related macular degeneration and Stargardt disease

Introduction

Age-related macular degeneration (AMD) is the commonest cause of severe visual loss and blindness in developed countries among individuals aged 60 and older [1]. AMD is a major unmet medical need as it is estimated that more than 20 million patients will be affected by 2050 in the US alone. STGD is the most common hereditary macular dystrophy, mostly affecting young patients aged between 6 and 15 years old with a prevalence of 1/8,000-1/10,000 [2, 3]. It has an autosomal recessive mode of inheritance and may lead to registered blindness within the second or third decade of life. STGD is caused by mutations in the ABCR gene encoding the ATP-binding cassette gene, subfamily A, member 4 (ABCA4) transporter which is expressed by photoreceptors and retinal pigmented epithelium (RPE) cells and plays an important role in the visual cycle [4, 5]. Polymorphisms of the gene coding for ABCA4 have also been associated with increased risk of developing AMD [6]. Other genetic polymorphisms, especially in the complement factor H (CFH) gene [7] have also been associated with AMD, but risk factors are mostly linked to age [8, 9] and environmental such as smoking [10]. AMD can either evolve towards neovascular AMD, also called wet AMD, characterized by the growth of new choroidal blood vessels in the subretinal space, or towards geographic atrophy also named dry AMD, characterized by RPE and photoreceptor degeneration. In STGD patients, neovascularization is extremely rare and RPE and photoreceptor atrophy occurs in the vast majority of cases [11]. Thus, despite some differences, dry AMD and STGD share similar pathophysiological mechanisms [11]. In both pathologies, early signs of evolution are characterized by subretinal accumulation of lipids and proteins forming drusen in AMD and flecks in STGD [2, 12]. AMD and STGD evolution are associated with Bruch’s membrane thickening, RPE alterations and ultimately to RPE and photoreceptor degeneration. Rod photoreceptors, responsible for scotopic/mesopic vision (i.e. under dim light conditions) are the first visual cells dying in the retina. The cones mediating colored photopic vision under normal light conditions are essentially preserved until late stages of AMD and STGD [13, 14]. Therefore, both AMD and STGD induce the progressive loss of night vision followed by loss of color vision and central vision [3]. To date, no treatment is available for either STGD [11] or dry AMD [15].

Both drusen in AMD and flecks in STGD contain N-retinylidene-N-retinylethanolamine (A2E), which is a toxic by-product of the visual cycle [16]. It is formed by the reaction of 2 all-trans retinal molecules with phosphatidylethanolamine generating N-retinylidene-PE (A2E precursor: A2-PE), as a detoxication mechanism of retinal isomers including all-trans and 11-cis-retinal [17]. Under normal conditions the ABCA4 protein participates in the elimination of A2-PE from the photoreceptors and inhibition of this clearance increases the accumulation of A2E and all-trans-retinal dimer in the RPE [18, 19]. Recently, it has been shown that ABCA4 is also expressed in RPE cells where it would participate in the recycling of retinaldehyde released during proteolysis of rhodopsin in endolysosomes following phagocytosis of photoreceptor outer segments [5]. The authors also demonstrated that A2E accumulates at similar rates in RPE from Abca4^-/- mice reared under cyclic light or total darkness suggesting that de novo bis-retinoids formation within RPE endolysosomes contributes more to lipofuscin build-up than do bis-retinoids formed in outer segment discs during light exposure. Therefore, they propose that the clearance of retinaldehydes from RPE phagolysosomes may be more critical for photoreceptors viability than the clearance of retinaldehydes from outer segment discs [5]. In vitro, it has been shown that the combination of blue-light illumination and A2E is toxic for RPE cells; either the ARPE19 cell line or primary porcine RPE cells [20–22]. Indeed, in the presence of blue light and oxygen, A2E undergoes photo-oxidation as evidenced by the appearance of toxic oxygen adducts [23]. It generates small amounts of singlet oxygen and is finally cleaved to small reactive aldehydes, which contribute to its deleterious effects on RPE cells [23]. A2E photo-oxidation products also damage DNA [20, 21] and activate the complement system [24]. Moreover, photosensitization of A2E triggers telomere dysfunction and accelerates RPE senescence [25]. In the absence of illumination, A2E alone affects normal RPE functions by inducing membrane permeabilization and thereby impairing lysosomal function [26]. It also impairs mitochondrial homeostasis and function resulting in the reduction of ATP production [27–30]. Furthermore, high A2E concentrations increase oxidative stress [31] and secretion of inflammatory cytokines by RPE cells in vitro [25, 32, 33]. In addition, A2E induces the expression of vascular endothelial growth factor in vitro [34] and in vivo [32, 35].

A rapid accumulation of A2E and of lipofuscin is observed in 3-month-old Abca4^-/- Rdh8^-/- mice [36] and increases in 6-month-old mice [36]. In these mice, progressive A2E accumulation is associated with retinal degeneration and loss of both scotopic and photopic full-field electroretinogram (ERG) responses indicate that both rods and cones are dysfunctional (rod-cone dystrophy) [36, 37]. Indeed, it has been shown that as early as in 3-month-old Abca4^-/- Rdh8^-/- mice, the amplitudes of scotopic A and B waves and flicker ERG are reduced [36]. Abca4^-/- Rdh8^-/- mice recapitulate most phenotypic retinal alterations observed during STGD and AMD [36, 37], and represent a chronic model of these diseases.

Acute white-light exposure is known to induce apoptosis of photoreceptors in albino mice retinas [38]. The model of light-induced retinopathy is used as an amenable model of AMD and has been used to test several neuroprotective molecules [39].

Norbixin (9’-cis-norbixin) is a 6,6’-di-apo-carotenoid extracted from annatto (Bixa orellana) seeds [40]. Tolerability of norbixin is well known, based on animal and human studies, and supports its use as food additive/dye [41]. Norbixin protects primary porcine RPE cells against phototoxicity induced by A2E and blue-light illumination in vitro [22]. Norbixin also reduces the accumulation of A2E by primary porcine RPE cells in vitro [22]. We have previously reported that a 3-month supplementation with norbixin in water reduced A2E ocular accumulation in vivo in Abca4^-/- Rdh8^-/- double-knockout mice [22]. In the same mouse model, the local treatment with norbixin via intraocular injections inhibited retinal degeneration and the loss of full-field ERG, induced following blue-light illumination [22]. In addition, norbixin was neuroprotective against blue-light damage (BLD) in rats [22].

Here we used an acute BLD model in BALB/c mice modified from our previous article [22] and that is used to rapidly study some features of AMD and the chronic model of AMD and STGD in aging Abca4^-/- Rdh8^-/- mice to evaluate the protective efficacy of norbixin. In the present article we demonstrate that norbixin administered systemically is neuroprotective against blue-light-induced retinal degeneration in BALB/c mice. Moreover, we report for the first time that 6-month oral supplementation of Abca4^-/- Rdh8^-/- double-knockout mice with chow containing norbixin is neuroprotective and partially preserves the function of both rods and cones in vivo. Oral supplementation with norbixin also reduces A2E and lipofuscin accumulation in RPE cells. We define the therapeutic window during which oral supplementation with norbixin is the most effective in relation to progressive photoreceptor loss of function and A2E accumulation.

Results

Discussion

We have previously shown that 9’-cis-norbixin, a 6,6’-di-apo-carotenoid extracted from annatto (Bixa orellana) seeds, protects primary porcine RPE cells against phototoxicity induced by A2E and blue-light illumination in vitro [22]. Norbixin was also shown to reduce the uptake of A2E by primary porcine RPE cells in vitro [22]. In addition, 3 months of oral supplementation with norbixin in drinking water reduced the A2E ocular accumulation in vivo in Abca4^-/-Rdh8^-/- mice [22]. In the same mouse model, a single norbixin intraocular injection inhibited retinal degeneration and the loss of full-field ERG, induced following blue-light illumination [22]. In addition, when 50 to 100 mg/mL of norbixin was injected intraperitoneally, almost complete neuroprotection was achieved in albino rats subjected to BLD [22]. In the present study, we tested on one hand, the effect of four intraperitoneal injections of 10 mg/kg of norbixin administered before and during the course of a modified BLD exposure in BALB/c mice, an acute model used to rapidly study the neurodegeneration symptomatic of AMD, and, on the other hand, the effect of norbixin supplementation for 5 to 6 months in Abca4^-/- Rdh8^-/- mice of different ages, a chronic model of AMD and STGD.

Here, we demonstrate that norbixin IP injections are partially neuroprotective and protect photoreceptor function following retinal degeneration induced by blue-light exposure. The protective effect of norbixin in the blue-light model is probably associated with its antioxidative properties as previously demonstrated in vitro [22, 42] and in vivo [43]. Despite the fact that we haven’t measured the specific effect of norbixin on expression of inflammatory markers in retina of Balb/c mice subjected to BLD we quantified sub-retinal infiltration in this model. We observed a small trend toward reduction of macrophages infiltration by norbixin, however this was not statistically significant (data not shown). The absence of a significant effect of norbixin on macrophages infiltration in the BLD model could be explained by the limited dose of norbixin used in this experiment. Indeed, the dose of norbixin only induced a 30 % neuroprotective effect in the retina. It could be hypothesized that a higher dose of norbixin would induce a significant reduction of macrophages infiltration in the sub-retinal space. Further experiments to confirm this hypothesis and to decipher the neuroprotective mechanism of norbixin in this model are therefore still required.

Abca4^-/- Rdh8^-/- mice are a model of STGD but also a broader model of retinal degeneration such as AMD [36, 37]. Indeed, both STGD and AMD are characterized by progressive accumulation of A2E and lipofuscin, RPE cell and photoreceptor degeneration eventually leading to the gradual “night” and “day” vision loss that are mediated by rods and cones, respectively [13, 14]. Structural and morphological alterations over time of the retina of Abca4^-/- Rdh8^-/- mice has been extensively described [37]. It is noteworthy that despite the fact that the mice we used also carried the rd8 mutation in the Crb1 gene, we haven’t observed the formation of photoreceptor “rosettes” in the retina nor focal inferior retinal degeneration characterized by the formation of local retinal folds representative of rd8^-/- mouse retinal phenotype [44] (data not shown). It has been shown that Crb1^rd8/rd8 mutation is necessary but not sufficient for the development of these degenerative features and that the influence of small genetic background difference is involved in the resulting phenotype [45]. Therefore, we can expect that our strain does not express gene mutations necessary to induce a degenerative phenotype as it was already the case in our previous publication [22].

In the present study, we describe the kinetics of expression of cathepsin D, RPE65 and GFAP expression and the progressive loss of visual function in relation with photoreceptor degeneration, and of the accumulation of A2E in eyes of Abca4^-/- Rdh8^-/- mice as they aged between 2 months and 18 months. Ocular A2E accumulation in the eyes of Abca4^-/- Rdh8^-/- mice appeared to be biphasic with a maximum observed at 9 months followed by a plateau afterwards. The rapid increase of ocular A2E concentration has been previously reported in “young” Abca4^-/- Rdh8^-/- mice up to the age of 6 months [36]. Interestingly, while retinal degeneration and loss of ERGs appeared constant during aging, we noted that photoreceptor loss and decrease in retinal function was more pronounced between 2 and 6 months and 12 and 18 months than between 9 and 15 months. Evolution of photoreceptor function over time might be related to the kinetics of A2E accumulation during the early phase in young mice and following its stabilization during the late phase in old animals. A2E and other toxic all-trans-retinal derivatives accumulate with age in the retina and RPE as by-products of the visual cycle even in non-pathological conditions [46]. In Abca4^-/- Rdh8^-/- mice, lacking two proteins essential for avoiding accumulation of all-trans-retinal derivatives in photoreceptors and RPE, high amounts of A2E are found in the RPE. Similarly, in the retina of AMD and STGD patients, A2E also accumulates faster and leads to retinal degeneration [46]. A2E promotes the release of Ca²⁺ from the endoplasmic reticulum leading to the production of reactive oxygen species (ROS) by NADPH oxidase [46, 47]. Excessive ROS production and oxidative stress in general contributes to photoreceptor cell death and retinal degeneration [23] and is recognized as a major risk factor for dry AMD development [48–50]. It could be hypothesized that in young Abca4^-/- Rdh8^-/- mice, rapid accumulation of toxic A2E is responsible for the early decline in photoreceptor counts and visual function that we observed. Then, approximately around 9 months of age, due to the reduction of the number of photoreceptors and reduction in the visual cycle activation, demonstrated by reduced ERG amplitudes, the production of A2E may be progressively slowed, reaching an equilibrium illustrated by the plateau observed thereafter. During this intermediate period, when A2E is produced at a slower rate, A2E may also be more sensitive to photodegradation. Indeed, due to the reduced number of photoreceptor nuclei, the retina is thinner, allowing increased light intensity to reach A2E internalized in RPE cells [51]. In the meantime, apoptosis of photoreceptors continues at a slower rate, probably secondary to the early phase of photoreceptor degeneration progressively displacing the equilibrium between A2E concentrations and numbers of photoreceptors established at 9 months. It is to note that A2E ocular concentrations measured in our experiments are slightly higher and retinal degeneration slightly milder when compared to results found in the article by Maeda and colleagues [36], but are consistent with reported values by others [52]. Moreover, during the long-term studies presented here we used the same eyes to dose A2E and norbixin. Since norbixin could have been present in ocular fluids and to avoid losing vitreous/aqueous humor, we did not perform dissections of eyes from Abca4^-/- Rdh8^-/- mice to determine the exact location of A2E in these mice. However, we assume that A2E is mostly present in the RPE layer. Indeed, in previous dosing experiments performed on mice and dog eyes we noted that A2E is almost exclusively found in the RPE (data not shown). This observation is also consistent with the publication from Lenis et al. [5]. Altogether, the observations that the kinetics of evolution of ERG amplitudes, A2E accumulation and photoreceptor loss were not constant over time, suggest that the efficacy of a systemic treatment might be dependent on pathological course.

In the present study, we attempted to determine the optimal therapeutic window during which oral supplementation with norbixin is the most effective in relation to kinetics of A2E accumulation and of loss of visual function in Abca4^-/- Rdh8^-/- mice. Here we report that 6-month oral supplementation of 9-month-old Abca4^-/- Rdh8^-/- mice with chow containing norbixin totally preserves cone function and limits loss of function in rods, as well as photoreceptor degeneration and A2E accumulation. By contrast, during norbixin supplementation of younger mice (starting at 1.5 months), we only observed a slight but significant protective effect of norbixin on scotopic A wave ERG and on retinal accumulation of A2E. This might be related to the fact that sharp decrease in scotopic A wave amplitude and an important A2E accumulation occur during the early retinal degeneration phase. The neuroprotective effect of norbixin in Abca4^-/- Rdh8^-/- mice during the early curative supplementation confirms our previous observations in acute blue-light induced retinal degeneration model in vivo in the same mice [22]. In the three supplementation protocols (preventive, early and late curative), norbixin prevented the loss of scotopic A wave amplitude measured by ERGs. This suggests a protective effect on rod photoreceptors by norbixin. Rods are the first neurons affected during the early stages of AMD and STGD [13, 14]. They are also much more abundant in the mouse retina than cones [13, 14]. In the preventive supplementation the protective effect of norbixin on rod function, was associated with a slight reduction of A2E concentration (Figure 2). This slight reduction however appears insufficient to prevent the loss of cone function as demonstrated by the absence of effect on photopic B wave. The loss of rod’s function is a progressive phenomenon observed even in 15-month-old Abca4^{-/ -}Rdh8^-/- mice and was also reduced by norbixin supplementation in the late curative supplementation protocol. It is important to note that the late curative experiment presented here, was performed in female mice rather than male mice in the other protocols. However, we compared ERG results obtained in 17 months male mice treated with norbixin for 5 months (between 12 and 17 months) and ERG amplitudes data obtained in male mice aged 18 months from the experimental group used in Figure 2. We observed a protection of scotopic A wave ERG (but not scotopic B wave) linked with rod function in norbixin complemented male mice (data not shown). Due to the difference of age between the two groups of male mice we cannot conclude that these effects are solely related with norbixin treatment with the exclusion of age, but this observation is consistent with the preferential preservation of rod photoreceptors reported for female mice in the late curative experiment (Figure 5B). Therefore, we are confident that the results reported here on the late curative complementation in female are representative of a phenomenon also occurring in male mice. The protective action of norbixin extended to cone photoreceptor function in mice supplemented between 9 and 15 months of age. Norbixin treatment during the early curative supplementation reduced by 40 % the A2E retinal concentration. This suggests that a sharp reduction of A2E concentration is necessary to protect cone photoreceptor function, whereas a more limited reduction of A2E concentration appears to be sufficient to protect rod photoreceptor function. Other compounds including retinylamine [36, 53], C20-d3-vitamin A, [54], Emixustat [55], primary amines [56], omega-3 fatty acids [57] and the selective oestrogen receptor modulator raloxifene [47], among others, have been shown to reduce A2E accumulation and concomitantly to “rescue” retinal degeneration in Abca4^-/- Rdh8^-/- mice or Abca4^-/-in vivo. Altogether, this supports the hypothesis that norbixin effectiveness is at least partly linked to the level of reduction of ocular A2E concentration.

The exact mechanism(s) of reduction of A2E concentration in vitro and in vivo by norbixin remain(s) to be determined. But, based on our ERG results, it seems unlikely that norbixin supplementation limits A2E production in vivo by slowing down the visual cycle. We also showed previously in vitro that norbixin reduces the amounts of A2E internalized by porcine RPE cells as determined by HPLC-MS/MS [22]. Therefore, norbixin-induced reduction of intraocular A2E concentration in Abca4^-/- Rdh8^-/- mice could be due to direct effects on RPE cells in vivo. We could also hypothesize that in the BLD experiments reported here, the neuroprotective effect of norbixin observed in Abca4^{-/ -}Rdh8^-/- mice in vivo may be due to a reduction of oxidative stress and ROS production. Indeed, norbixin has been shown to reduce oxidative stress in rats [42] and humans [43] following a high-fat meal diet. Further experiments to confirm the antioxidant properties of norbixin in the context of retinal degeneration in vivo and in RPE cells in vitro are still required. In the present article which focuses on the effect of norbixin in vivo in mice models of AMD and STGD we haven’t described any mechanistic processes of norbixin. However, experiments in vitro on primary porcine RPE cells are actually performed to decipher the potential mechanism of action of norbixin (V. Fontaine et al. in preparation). Indeed, in addition with reduction of A2E accumulation, anti-apoptotic, antioxidant and anti-inflammatory properties of norbixin could play a role in the protection of visual function and in the neuroprotective effect of norbixin that we report in the present study in vivo. In order to determine if norbixin displayed an anti-inflammatory effect in the late curative experiment we quantified macrophages infiltration in the subretinal space of Abca4^-/- Rdh8^-/- mice (data not shown) in vivo but we did not observe any significative differences. It has been proposed that A2E may potentiate subretinal macrophage accumulation [58]. Therefore, the slight reduction of A2E accumulation obtained in our model by norbixin may not be sufficient to significantly reduce macrophage recruitment. However, it could be hypothesized that norbixin may alter macrophage activation or differentiation in vivo. This will be the aim of future experiments. In order to better understand the effects of norbixin in the retina in vivo and to determine whether norbixin effects target specifically the retina or RPE or both, we performed GFAP, RPE65 and cathepsin D stainings on retinal sections of eyes obtained from Abca4^-/- Rdh8^-/- mice treated with norbixin versus untreated in the early curative supplementation experiment. However, we did not observe any differences if RPE65 nor cathepsin D expression in both groups. Surprisingly, we observed enhanced GFAP staining of Müller cells in the retina of norbixin supplemented mice. Further experiments are required to decipher the mechanisms behind the protective effects of norbixin.

To our knowledge, this is the first report demonstrating the efficiency of treatment given orally through food complementation on the photoreceptor function in the aging Abca4^-/-Rdh8^-/- mouse model. Most of the previous studies performed to measure the efficacy of molecules on the retinal function, using this Abca4^-/- Rdh8^-/- mouse model, were done by intraperitoneal or intravitreal injections and after light damage in young mice [36, 47, 53, 56, 59–61]. Maeda and colleagues also showed the beneficial effect of 9-cis-retinal administrated by monthly oral gavage over 6 or 10 months in 4-month-old C57BL/6 mice on scotopic A and B waves, as well as flicker ERG [62]. Similar observations were reported following weekly gavage for one month in 12-month-old Rdh5^-/-Rdh11^-/- mice [63]. In another study, the same group showed that 3 months treatment with primary amines in 1-month-old Abca4^-/- Rdh8^-/- mice force-fed daily reduced A2E accumulation, photoreceptor degeneration, and retinal functions [56].

Based on our previous results showing that norbixin increases RPE survival and reduces A2E internalization in vitro [22], we assume that norbixin’s neuroprotective effect is local. However, free norbixin concentrations in the eyes were below detection limits in all supplementation experiments. It could be hypothesized that limited amounts of free norbixin present in the plasma access the eye and are consumed in a continuous fashion. Oral supplementation exposure of norbixin in mice depends on the quantity of food absorbed by each mouse and also depends on the time of day of the analysis of plasma concentrations relative to food consumption which occurs mainly at night time. Therefore, we cannot assume that all mice received the same amount of norbixin during supplementations. Nevertheless, at 3 and 6 months, norbixin was detected in the plasma of all supplemented mice across all experimental designs, confirming the exposure of mice to norbixin administered via food as early as 3 months. Interestingly, in the early curative supplementation experiment, plasma norbixin and norbixin glucuronide conjugate concentration was the lowest. Nevertheless, the observed partial neuroprotection, and full preservation of scotopic and photopic ERG amplitudes, suggest that this level of plasma exposure is sufficient to support norbixin biological effects. As a mean, we calculated that norbixin consumption represents approximately a daily intake of 47.5 mg +/- 5 mg per kg/mouse. This value is in the same range of administration in the BLD study, where mice received four intraperitoneal injections of 10 mg/kg each. Nevertheless, we have reported previously in acute BLD model in albino rats a more profound effect of our tested compound. The partial efficacy reported here in acute and long-term models of retinal degeneration may be due to the small doses used and might be improved by increasing the doses of norbixin administered. Increasing bio-availability of norbixin into the eye following systemic administration may also improve efficacy and we are actually working on these two options.

In conclusion, our present study demonstrates that systemic administration of norbixin in the acute BLD model of dry AMD is neuroprotective and partially preserves photoreceptor function. In addition, 6 months of oral supplementation with norbixin is effective in Abca4^-/-Rdh8^-/- mice. We show that chronic norbixin supplementation reduces the concentration of A2E in the eye, that norbixin is neuroprotective, and preserves visual function of Abca4^-/-Rdh8^-/- mice, modelling retinal degenerative conditions such as STGD and dry AMD. We believe that treatment using norbixin could potentially preserve “night” and “day” visual acuity in humans affected by dry AMD and STGD. It is essential for patient care to develop drugs that are effective on visual function following oral administration rather than by repeated local intraocular injections. These results demonstrated the effectiveness of the norbixin in a chronic and acute model of retinal degeneration and could offer a new therapeutic strategy, alone or in combination with gene therapies, for AMD and/or STGD patients. Thus, norbixin is a good drug candidate to treat patients and may provide a cure for these very debilitating diseases.

Materials and Methods