Reduced Inspired Oxygen Decreases Retinal Superoxide Radicals and Promotes Cone Function and Survival in a Model of Retinitis Pigmentosa

Abstract

Retinitis pigmentosa (RP) is caused by many different mutations that promote the degeneration of rod photoreceptors and have no direct effect on cones. After the majority of rods have died cone photoreceptors begin to slowly degenerate. Oxidative damage contributes to cone cell death and it has been hypothesized that tissue hyperoxia due to reduced oxygen consumption from the loss of rods is what initiates oxidative stress. Herein, we demonstrate in animal models of RP that reduction of retinal hyperoxia by reducing inspired oxygen to continuous breathing of 11% O₂ reduced the generation of superoxide radicals in the retina and preserved cone structure and function. These data indicate that retinal hyperoxia is the initiating event that promotes oxidative damage, loss of cone function, and cone degeneration in the RP retina.

Graphical Abstract

Introduction

Retinitis Pigmentosa is an inherited eye disorder that affects 1 in 4000 worldwide, affecting two million people, and resulting in the loss of photoreceptors [1]. This disease shows high genetic heterogeneity with 90 genes identified and about 3100 mutations have been reported in these genes [2, 3]

Multiple animal models have been developed to understand the mechanisms that cause photoreceptor loss and explore therapies to prolong their survival [4].

In 3 models of retinitis pigmentosa (RP), Royal College of Surgeons (RCS) rats, P23H rats, and the Abyssinian cat after substantial death of rods oxygen levels are increased in the outer retina [5-7]. After rods degenerate in a pig model of RP there is progressive oxidative damage in cone photoreceptors and gradual degeneration of cones [8]. Oxidative damage contributes to cone cell death because its reduction preserves cone function and promotes cone survival [9-12]. It has been hypothesized that retinal hyperoxia is the inciting event that begins the production of reactive oxygen species initiating oxidative stress in the RP retina but it has not been proven [5, 6]. In this study, we sought to test that hypothesis.

Materials and Methods

Study Approval

Both rd10 [13] and Q344ter mice [14] were treated in accordance with the Association for Research in Vision and Ophthalmology and approved by the Johns Hopkins Animal Care and Use Committee (approval number MO20M118). The mice were housed in core barrier housing in barrier plastic cages with corn cob bedding and a wire bar lid that serves as a food hopper and handled in biosafety cabinets under aseptic conditions using protective clothing. Food and water are provided ad libidum. Not more than 5 mice were maintained per cage. The lighting inside the hypoxia or normoxia box was about 1.9 Watt/m² and in the room was 2.1 Watt/m².

Study Design

The primary objective of the study was to determine if hypoxia promotes cone functional and structural survival after rod death in the rd10 mouse model of RP. The animals were subjected to hypoxia (11.5% FiO₂) or normoxia (21% FiO₂) at P35, a time point when most rod photoreceptors are gone but cones persist. The prespecified primary endpoint was at P50 when the functional protection in cones was evaluated by measuring the mean photopic b-wave amplitude by electroretinograms (ERG). If functional protection was observed, a second photopic ERG was done to determine the mean photopic b-wave amplitude at P65. Important secondary outcome measurements were the structural protection assessment by measuring cone density in immunohistochemistry, western blots, and retinal flat mounts by S opsins and cone arrestin staining. Exploratory outcomes were oxidative stress assessment by Dihydroethidium (DHE) staining on retinal sections and NADPH oxidase assays using lucigenin chemiluminescence.

Electroretinogram recordings (ERG)

Electroretinograms were done in accordance with previously published protocols [15, 16].

Immunohistochemistry

Retinal sections and retinal flat mounts were subject to immunohistochemistry in accordance with previously published protocols [15, 16]. The sections were stained with the primary antibodies, anti-Rhodopsin (1:500, Cat# MA1-722, ThermoFischer Scientific, anti-S opsin (1:250, Cat# ABN1660, EMD Millipore), and anti-cone arrestin (1:250, Cat# 15282, EMD Millipore). The secondary antibody used was donkey anti-rabbit Alexa Fluor^™ 555 (1:300, Cat# A32794, Invitrogen). For S opsin expressing cone counting in retinal flat mounts, S cone numbers were counted in the entire 40X visual field in 3 independent areas on the inferior (ventral) retina and averaged to get the S cone number/retina. For cone arrestin counting in retinal flat mounts, cone arrestin numbers were counted in the entire 40X visual field in the superior, inferior, nasal, and temporal regions of the retina using the fluorescent microscope Zeiss Axioskop 2.

Western blot

Western blots were done in accordance with previously published protocol [15]. The primary antibody used was anti-S opsin (1:1000, Cat# ABN1660, EMD Millipore). The secondary antibody used was anti-rabbit FIRP (1:1000, Cat# 7074S, Cell Signaling). The membranes were visualized with SuperSignal^™ West Duration Extended Duration Substrate (Cat# 3407S, Thermo Fisher Scientific) and imaged using ChemiDoc^™ XRS Molecular Imager (BIO-RAD). The protein bands were quantified using Image Lab Software (BIO-RAD).

Dihydoethidium staining

Animals were injected intraperitoneally with 50 mg/kg dihydroethidium (Cat# D1168, Invitrogen) and the hypoxia animals were put back in the hypoxia chamber for 1 hour. After the hour, the normoxia and hypoxia-treated animals were euthanized and eyes were removed and frozen unfixed in Tissue-Tek^® O.C.T. Compound (Cat#4583, Sakura). Sections were cut to 10 μm thickness using Microm HM 500 OM Microtome (ThermoFischer Scientific) and counterstained with DAPI Solution (1:1000, Cat# 62248, ThermoFischer Scientific). Sections were imaged using fluorescence microscope Zeiss Axioskop 2 and the Corrected total fluorescence (CTF) was obtained by ImageJ (NIH) [17] on the entire retinal section according to the formula: CTF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings).

NADPH Oxidase Assays by Lucigenin chemiluminescence.

Retinas from hypoxia and normoxia were suspended in 250 μl of Krebs-HEPES buffer. The samples were sonicated for 10 seconds followed by centrifugation at 800 g/10 min. Then, 5 μM of lucigenin was added (Cat# 14872, Cayman Chemicals) and luminescence was measured until luminescence was below 100. Then, 250 μM NADPH (Cat#481973, EMD Millipore) was added and luminescence was recorded every 30 seconds for 2 minutes. The values were then averaged for each animal. Protein estimation was done on the retinas using Bradford reagent (Cat# 5000006, BIO-RAD). The NADPH oxidase assay values were plotted as luminescence counts/mg protein.

Results

At postnatal day 35 (P35), rods in the rd10 retinas have no detectable rhodopsin staining in the outer segments (OS) and fewer rows of nuclei in the ONL layer, indicating advanced degeneration (Supplementary Figure 1B right panel) compared to wild-type C57Bl/6 mice (Supplementary Figure 1A right panel). However, cone arrestin staining in rd10 retinas at P35 is robust indicating that most cones have survived (Supplementary Figure 1B left panel) compared to wild-type C57Bl/6 mice (Supplementary Figure 1A left panel). At P50 however, cone arrestin staining in the rd10 retinas has dramatically reduced indicating cone degeneration (Supplementary Figure 1C left panel). To prevent cone degeneration between P35 and P50, rd10 mice were placed in a sealed chamber with a low oxygen environment in which the fraction of inspired oxygen (FiO2) was 11.5%, or they were placed in a normal oxygen environment (room air, 21% FiO2) starting at P35 when most of the rods have degenerated.

At P50, after light adaptation, electroretinograms (ERGs) showed significantly higher mean photopic b-wave amplitudes at two stimulus intensities in mice exposed to the hypoxic environment compared with those exposed to a normoxic environment (Figure 1A). This suggests that reducing retinal hyperoxia by reducing inspired oxygen concentration preserves cone function. To investigate how long the protection is maintained, electroretinograms were done at P65, when there is advanced cone degeneration in normoxia-exposed rd10 mice. At P65, mean photopic b-wave amplitudes remained significantly higher in hypoxia-exposed mice at 2 stimulus intensities (Figure 1B). In a second animal model of RP, mice with the autosomal dominant Q344ter mutation in rhodopsin [18] in which photoreceptor degeneration is more rapid than that seen in rd10 mice, reduced inspired oxygen also promoted maintenance of cone function (Supplementary Figure 2).

Figure 1. Hypoxia preserves cone function in the rd10 mouse model of Retinitis Pigmentosa.

At P35, rd10 mice were placed in 11.5% oxygen (hypoxia) or 21% oxygen (normoxia). At P50, hypoxia-exposed mice had significantly higher mean photopic b-wave amplitudes at stimulus intensities of 1 cd-s/m² (p=0.0143) and 1.39 cd-s/m² (p<0.0001) (A, left panel). Examples of photopic ERG waveform responses at P50 for stimulus intensities of 0.6, 1.0, and 1.39 cd-s/m² (A, right panel). At P65, hypoxia-exposed mice continued to show higher mean photopic b-wave amplitudes at stimulus intensities of 1 cd-s/m² (p=0.0008) and 1.39 cd-s/m² (p=0.0416) (B, left panel). Examples of photopic ERG waveform responses at P65 for stimulus intensities 0.6, 1.0, and 1.39 cd-s/m² (B, right panel).

Mann-Whitney non-parametric analyses were done for all data and p values ≤0.05 were considered significant. * < 0.05, ** < 0.01, *** < 0.001 and **** < 0.0001, n=number of mice used in the experiment.

There was significantly more S-opsin in the retinas of P50 rd10 mice exposed to a hypoxic environment starting at P35 compared with those exposed to a normoxic environment (Figure 2A, top, and bottom panels), indicating more surviving S cones. Ocular sections immunostained for S-opsin confirmed that at P50 there were more surviving S cones in the inferior retinas of hypoxia-exposed rd10 mice compared to those exposed to normoxia (Figure 2B, left top and bottom panels). In RP, S cones are preferentially lost in the inferior retina [19], and low-magnification views of retinal flat mounts to show entire retinas immunostained for S opsin, showed more homogeneous staining throughout the inferior retina of hypoxia-exposed versus normoxia-exposed rd10 mice at P50 (Figure 2B, right top and bottom panels). High magnification images of identical regions of inferior retina demonstrated significantly more surviving S cones in the retinas of rd10 mice that had been exposed to hypoxia versus those that had been exposed to normoxia at P50 (Figure 2C top and bottom panel and Figure 2D) and P65 (Figure 2E), a stage when most S cones had degenerated in normoxia raised rd10 mice (Figure 2E). Immunohistochemistry for another cone-specific protein, cone arrestin, showed that even as late as P71, there were significantly more surviving cones in the inferior and nasal quadrants of the retinas of rd10 mice that had been exposed to hypoxia compared to normoxia-raised rd10 mice (Figure 2F).

Figure 2. Hypoxia slows down cone degeneration in mouse models of retinitis pigmentosa.

At P50, immunoblots of retinal lysates probed with anti-S opsin and anti-actin antibodies (A, top panel) were assessed by densitometry (A, bottom panel) which showed significantly higher S opsin/actin ratio for hypoxia-exposed versus normoxia-exposed rd10 mice. (B) Immunostaining with anti-S opsin antibody showed more staining in ocular sections of hypoxia-exposed (top left panel) versus those from normoxia-exposed (bottom left panel) rd10 mice at P50. Whole retinas were immunostained for S opsin and flat mounted. A low magnification view to show the entire retina shows more homogeneous staining in the inferior retinas from hypoxia-exposed P50 rd10 mice (top right panel) versus those from normoxia-exposed P50 rd10 mice (bottom right panel). (C) High 40X magnification view of identical locations of inferior retina of flat-mounted S-opsin-stained retinas from hypoxia-exposed (top panel) and normoxia-exposed (bottom panel) rd10 mice at P50. Measurements of S opsin expressing cone cell density in identical locations of inferior retina on flat mounts showed significantly higher S-cone cell density in hypoxia-exposed retinas at P50 (D) and P65 (E). (F) At P71, identical locations of superior, inferior, nasal, and temporal retina on cone arrestin-stained retinal flat mounts from hypoxia-exposed (left top panel) and normoxia-exposed (left bottom panel) rd10 mice showed significantly greater cone density in the nasal and inferior retina of hypoxia-exposed rd10 mice (right panel). Mann-Whitney non-parametric analyses were done for all data and p values ≤0.05 were considered significant. * < 0.05, ** < 0.01, *** < 0.001 and **** < 0.0001, n=number of mice used in the experiment. Scale bar shown in all images is 50 μm.

Dihydroethidine (DHE) reacts with superoxide radicals to form 2-hydroxyethidium, which binds to DNA and fluoresces. P35 rd10 mice that had been placed in a hypoxic or normoxic environment were given a 50 mg/kg subcutaneous injection of DHE an hour prior to tissue collection at P43. Fluorescence microscopy of ocular sections showed strong fluorescence in sections from normoxia-exposed mice and minimal fluorescence in sections from hypoxia-exposed mice (Figures 3A and B). When rd10 mice were put in 11.5% oxygen at P21 rather than P35, there was even less DHE-induced fluorescence in ocular sections at P43 (Figure 3C). Compared with normoxia-exposed rd10 mice, those exposed to hypoxia between P35 and P50, showed a reduction in NADPH oxidase activity (Figure 3D), indicating NADPH oxidase as one source of superoxide radicals causing oxidative stress in normoxia-raised rd10 mice.

Figure 3. Reduced levels of ROS detected in hypoxic retinas.

P35 rd10 mice (A, B) or P21 rd10 mice (C), were placed in 11.5% oxygen (hypoxia) or 21% oxygen (normoxia). At P43, hypoxia-exposed and normoxia-exposed mice were given a subcutaneous injection of 50 mg/kg dihydroethidium (DHE), and after 1-hour mice were euthanized and ocular sections were examined by fluorescence microscopy. There was less fluorescence in retinas from hypoxia-exposed versus normoxia-exposed mice (A) which was confirmed by measurements of fluorescence intensity (B, p=0.0227). Similar results were obtained when rd10 mice were placed in hypoxia or normoxia starting at P21 and DHE was injected at P43 (C, p=0.0027). P35 rd10 mice were placed in hypoxia or normoxia and euthanized at P50 and NADPH oxidase was assayed by Lucigenin chemiluminescence and it was significantly reduced in retinal homogenates from hypoxia exposed versus normoxia-exposed mice (D, p=0.0206). Red fluorescence is DHE and blue florescence is DAPI.

Mann-Whitney non-parametric analysis was done for all data and p values ≤0.05 were considered significant and exact p values are shown for significant data points.

* < 0.05 and ** < 0.01, n=number of mice used in the experiment. Scale bar shown in the images is 100 μm.

Discussion

RP is a genetically heterogeneous disease that preferentially affects rod photoreceptors resulting in primary death of rods and the secondary death of cones.

What causes secondary death of cones is not completely understood. One theory is the decreased abundance of rod-derived cone viability factor (RdCVF) after rod death which provides survival of cones [20]. However, the knockout of RdCVF causes only a 17% reduction in cone density [21] suggesting that RdCVF may not be the main cause of cone death in RP.

A major observation in RP retinas after rod death is that oxygen levels in the retina are markedly increased [5-7] as cones begin to die. Several lines of evidence have indicated that oxidative damage contributes to cone cell death [9-12], and in this study, we have demonstrated that retinal hyperoxia after substantial rod death initiates oxidative damage through the generation of superoxide radicals. Reducing retinal hyperoxia by decreasing inspired oxygen decreased superoxide radical production, reduced cone cell death, and preserved cone function. This demonstrated a link between hyperoxia, oxidative stress, and cone cell death provides a clearer picture of the pathogenesis of cone degeneration in RP.

One way that hyperoxia increases superoxide radicals is through increased activity of NADPH oxidase which is consistent with the previous observation that suppression of NADPH oxidase promotes cone function and survival in a mouse model of RP [10].

A limitation of our study is that we are unable to measure the decrease in retinal PO2 with an oxygen microelectrode while RP mice are exposed to 11.5 % oxygen without exposing them to room air. Therefore, the magnitude of the reduction in retinal PO₂ due to exposure to 11.5% oxygen that causes structural and functional rescue of cones is unknown. Another limitation is that reducing inspired oxygen is not a good strategy for the treatment of RP, because it would have undesirable extraocular effects.

However, this study helps to elucidate the mechanism by which oxidative stress is initiated in RP retinas and adds to the body of evidence implicating oxidative damage in cone cell death. These findings support many prior studies demonstrating that antioxidants promote cone survival and function in animal models of RP [9, 22-24] and strengthen the rationale for testing antioxidant treatments in patients with RP.

In a small study of patients with RP, oral antioxidant NAC treatment for 6 months provided small improvements in measures of visual function [25]. This suggests that long-term treatment with NAC could be better by potentially slowing cone cell death and maintaining cone function in patients with RP, a hypothesis that will be tested in a multicenter, placebo-controlled trial that will begin early in 2023.

Conclusion

Cone photoreceptors in RP experience hyperoxia due to excess unused oxygen after rod photoreceptor death. As a result, cones experience oxidative stress which causes functional and structural damage causing cones to ultimately die. Our studies show that lowering the oxidative stress by exposing cones to hypoxia, decreases the oxidative stress on cones and rescues cone structure and function.

Highlights.

• In RP, after rods die from a mutation, cones are exposed to hyperoxia.

• Hyperoxia causes activation of NADPH oxidase, promoting oxidative damage in cones.

• Oxidative damage causes loss of function and eventual death of cones.

• Reduced inspired oxygen in mice with RP reduces tissue hyperoxia.

• Reduced Hyperoxia reduces oxidative damage and promotes cone survival and function.

Abbreviations

DHE

Dihydroethidium

NAC

N-Acetyl Cysteine

NADPH

Nicotinamide adenine dinucleotide phosphate

RCS

Royal College of Surgeons

RP

Retinitis Pigmentosa