Acid-Sensing Ion Channel 3 in Retinal Function and Survival

Abstract

purpose. Changes in extracellular pH occur in the retina and directly affect retinal activity and phototransduction. The authors analyzed the expression in rodent retina of ASIC3, a sensor of extracellular acidosis, and used ASIC3 knockout mice to explore its role in retinal function and survival.

methods. The expression and the role of ASIC3 were examined by immunolocalization and by comparing retinas from wild-type and knockout mice at different ages through electroretinography, retinal histology (light and electron microscopy), expression of glial fibrillary acidic protein (GFAP), analysis of cell apoptosis (TUNEL assay), and patch-clamp recordings in primary cultures of retinal ganglion cells (RGCs).

results. ASIC3 is present in the rod inner segment of photoreceptors and in horizontal and some amacrine cells. ASIC3 is also detected in RGCs but does not significantly contribute to ASIC currents recorded in cultured RGCs. At 2 to 3 months, knockout mice experience a 19% enhancement of scotopic electroretinogram a-wave amplitude and a concomitant increase of b-wave amplitude without alteration of retinal structure. Older (8-month-old) knockout mice have 69% and 64% reductions in scotopic a- and b-waves, respectively, and reductions in oscillatory potential amplitudes associated with complete disorganization of the retina and degenerating rod inner segments. GFAP and TUNEL staining performed at 8 and 12 months of age revealed an upregulation of GFAP expression in Müller cells and the presence of apoptotic cells in the inner and outer retina.

conclusions. Inactivation of ASIC3 enhances visual transduction at 2 to 3 months but induces late-onset rod photoreceptor death, suggesting an important role for ASIC3 in maintaining retinal integrity.

Neuronal signaling within the vertebrate retina is highly dependent on the pH of the extracellular fluid. For example, alteration of the external pH from 7.4 to 7.0 has been shown to suppress almost completely the light-induced responses of second-order neurons in the retina. ^{1 2 3} On the other hand, the light-sensitive current of rods is generally suppressed or slowed down at extracellular pH 6.5 and lower. ^{4 5 6} Stimulation of the retina by light can induce significant changes in extracellular pH, raising the possibility that normal light-induced alteration in the extracellular H⁺ concentration might play an important role in modulating the flow of visual information within the retina. In darkness, protons that are co-released with glutamate into photoreceptor ribbon synapses act as feedback inhibitors of L-type Ca²⁺ channels in cones but not in rods. ^{3 7 8} Furthermore, the pH of the retina follows a circadian rhythm, ⁹ and pH changes may play a role in the adaptation of the retinal response to different light intensities. 

The retina expresses acid-sensing ion channels (ASICs), ^{10 11 12 13 14 15} neuronal voltage-insensitive cation channels gated by extracellular protons. ¹⁶ The ASIC family comprises four different genes encoding at least six isoforms ^{17 18 19} : ASIC1a and ASIC1b, ASIC2a and ASIC2b, ASIC3 and ASIC4. ^{16 20 21 22 23 24} ASIC subunits assemble into homotrimers or heterotrimers ²⁵ with different activation, inactivation, and reactivation kinetics, pH sensitivity, and ion selectivity. ^{20 26 27}  

The participation of ASICs in visual transduction has been recently investigated with mice deficient in the ASIC2 gene ¹² and with in vivo blockade of ASIC1a activity by antisense oligonucleotides and the specific peptide toxin PcTx1. ¹³ The effect of amiloride, a poorly selective blocker of ASICs, has also been investigated on the mammalian electroretinogram (ERG). ¹⁴ ASIC2 appears as a negative modulator of rod phototransduction and protects from light-induced retinal degeneration. ¹² ASIC1a, which is present primarily in inner segments of cone, participates in gain adaptation to ambient light. ¹³ The presence in the retina of ASIC3, a subunit largely expressed in sensory neurons, ²⁸ in which it has been associated with nociception ^{29 30 31} (including acidic and primary inflammatory pain ²⁹ ) and mechanoperception, ³¹ has also been described, ^{11 15} but its distribution and functional role in vision have not been studied. We have analyzed here, with the use of ASIC3 knockout mice, ³² the expression of ASIC3 in rodent retina to explore its role in retinal function and survival. 

Methods

Experimental Animals

The generation of the ASIC3 knockout mice and their littermates (C57BL/6) has been described. ³² Experiments were carried out with males, and animals were housed under controlled temperature conditions (21°C) and a 12-hour light/dark cycle. All animal experiments fully conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Light and Electron Microscopy

For light microscopy, eyes from wild-type and ASIC3 knockout mice, previously killed by asphyxiation with CO₂, were enucleated. The anterior part of each eye was removed, and the posterior segments were fixed for 24 hours in 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), then rinsed with buffer and postfixed in osmium tetroxide (1% in 0.1 M phosphate buffer) for 1 hour. After washing in water, they were dehydrated by incubation in acetone and embedded in resin (Epon; Hexion Specialty Chemicals, Aubervilliers, France). We prepared, respectively, semithin (1-μm) and ultrathin (80-nm) sections with the use of an ultracut S ultramicrotome (Leica, Wetzlar, Germany). Semithin sections were counterstained with methylene blue-azur II for light microscopy, and ultrathin sections were contrasted with uranyl acetate and lead citrate for electron microscopy. Ultrathin sections were viewed and photographed with an electron microscope (CM12; Philips, Eindhoven, Netherlands) operating at 80 kV and a high resolution digital camera (Morada; Olympus, Tokyo, Japan). 

Immunochemistry

Mice were dark adapted overnight, anesthetized, and killed by cervical dislocation. Eyes were removed, punctured at the limbus, and perfused with ice-cold 2% paraformaldehyde in phosphate-buffered saline (PBS). These suboptimal conditions to fix the retina were imposed by the nature of the anti–ASIC3 antibody used (guinea pig anti–ASIC3 antibody [Acris GmbH, Herford, Germany]; the only reliable antibody available) because, as described by Molliver et al., ³³ the quality of the staining was reduced with stronger fixation. After 15 minutes of fixation, the cornea, the lens, and the vitreous were removed, and the eyecups were cryoprotected in 20% sucrose in PBS for 1 hour and then embedded (Tissue-Tek; Sakura Finetek, Bayer Diagnostic, France) and rapidly frozen in isopentane cooled by liquid nitrogen. Frozen 10-μm sections were then prepared. 

For immunochemistry and double labeling, the frozen retinal sections were immediately permeabilized with 0.1% Triton X-100 in PBS for 30 minutes, subsequently washed in 1× PBS three times, and blocked with 10% horse serum in PBS to avoid nonspecific staining. ASIC3 was detected with the guinea pig anti–ASIC3 antibody (1:200; Acris) incubated on sections overnight. Detection was achieved with a biotinylated secondary antibody (anti–guinea pig, 1:100; Vector Laboratories, Burlingame, CA) and FITC-coupled streptavidin (1:80; DakoCytomation, Trappes, France). In double-immunostaining experiments, cell-type marker antibodies were incubated overnight at 4°C and were detected with an indocarbocyanine (Cy-3)–labeled anti–mouse IgG (1:300; Invitrogen, Cergy Pontoise, France) or anti–rat IgG (1:400; Invitrogen). Control experiments were conducted without primary antibody. Rod and cone inner segments were localized, respectively, with anti–recoverin antibody (1:200; Abcam, Cambridge, UK) and peanut agglutinin antibody (PNA; 1:200; Vector Laboratories), horizontal cells with anti–calbindin monoclonal D-28K antibody (1:600; clone CB-955; Sigma, St. Quentin Fallavier, France), bipolar cells with PKCα (1:600; Sigma), amacrine cells with anti–syntaxin monoclonal antibody (1:600; clone HPC-1; Sigma), and ganglion cells with rat anti–Thy1.2 antibody (1:200; rat monoclonal [30-H12] to Thy1.2; Abcam). Immunostained retinas on slides were mounted in antifade solution containing DAPI for nuclear staining (Vectashield; Vector Laboratories). Negative controls were used without primary or secondary antibodies (data not shown). Images of retinal sections were acquired and photographed with an imaging microscope (Axioplan2; Zeiss, Thornwood, NY) and with ×63, ×40, or ×25 oil immersion objective. 

Electroretinography

Mice were allowed to adapt to the dark overnight and were subsequently manipulated under dim red light. They were anesthetized with ketamine (66.7 mg/kg, intraperitoneally) and xylazine (11 mg/kg, intraperitoneally) and placed on a heating pad so that body temperature could be maintained near 38°C. The pupil of the test eye was dilated with 0.5% tropicamide (Mydriaticum; Thea, Clermont-Ferrand, France). A single drop of methylcellulose gel (Civigel; Novartis-Pharma, Huningue, France) was applied to the eye, and the mice were kept in the dark for 10 minutes. Dark-adapted responses (scotopic ERG) were subsequently recorded with an electroretinogram (ERG) test system (UTAS 2000; LKC Technologies, Gaithersburg, MD), as described. ¹² Light stimuli were produced by a xenon flash (PS 22; Grass) positioned 10 cm from the eye. Flash intensities were controlled with neutral density filters and were calibrated with a radiometer-photometer (IL 1700; International Light, Peabody, MA). Stimuli were presented in order of increasing intensity and ranged from −3.52 to 3.52 log scotopic trolands per second (log scot td/s). Background illumination during recording of the scotopic ERGs was 1.2 lux. Responses to six successive flashes were averaged. The interflash interval was 15 seconds for low (≤ 0.42 log scot td/s) and 30 seconds for higher (≥1.26 log scot td/s) stimulus intensities. 

The amplitude of the a-wave was measured from the prestimulus baseline to the apex negative peak of the a-wave. The b-wave amplitude was measured from the a-wave negative peak to the b-wave positive peak and not to the peak oscillations, which can be higher than the apex. Oscillatory potentials (OPs), which are superimposed over the ascending phase of the ERG b-wave, were extracted at the highest intensities 

Statistical analysis of ERG data was performed with statistical software (Prism 4; GraphPad San Diego, CA). ERG results were analyzed by two-way ANOVA. Bonferroni post hoc testing was used to evaluate amplitude differences among intensities. Data were considered significant if P < 0.05. 

RGC Primary Cultures and Patch Clamp Experiments

RGC primary cultures were prepared as previously described. ¹³ Cells were plated onto coated petri dishes (Biocoat; Becton Dickinson, Stuttgart, Germany) and incubated at 37°C in culture medium plus supplement (Neurobasal + N2; Invitrogen), 5% fetal bovine serum (ICN Biomedicals, Illkirch, France), 10 ng/mL bFGF (Boehringer Mannheim, Meylan, France), and 1% penicillin-streptomycin (Invitrogen). We used the whole-cell configuration of the patch clamp technique to measure RGC membrane currents (voltage clamp) 1 to 2 days after plating. Recordings were made with an amplifier (RK-400; Bio-Logic Science Instruments, Claix, France) with a 3-kHz low-pass filter (Krohn-Hite, Brockton, MA). Data were sampled at 10 kHz, digitized by an A-D/D-A converter (Digidata 1322A; Axon Instruments), and recorded on a hard disk using electrophysiology software (pClamp; version 9.2.0.11; Axon Instruments, Foster City, CA). Patch pipettes (5–7 MΩ) contained (135 mM KCl, 2.5 mM Na₂-ATP, 2 mM MgCl₂, 2.1 mM CaCl₂, 5 mM EGTA, 10 mM HEPES, pH 7.25, with KOH). Various pH-buffered external solutions were applied to individual patch-clamped cells with a homemade microperfusion system driven by microsolenoid valves (Sirai, Bussero, Italy) allowing rapid solution changes. The control bath solution contained 145 mM NaCl, 5 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, 10 mM HEPES, pH 7.4, with NaOH and 10 mM glucose). 2-N-morpholinoethanesulfonic acid (MES) or acetate was used instead of HEPES to buffer solution pH, ranging from 6 to 5 and from 4.5 to 4, respectively. ASIC currents were induced by shifting 1 of 8 outlets of the microperfusion system from the pH 7.4 control solution to an acidic test solution. Experiments were performed at room temperature, and ganglion cells were recognized by their ability to bind anti–Thy1.2 antibody (rat mAb to Thy1.2; Abcam, Cambridge, MA). Briefly, living cells were first incubated with anti–Thy1.2 antibody, carefully washed, and then incubated with a secondary anti-rat antibody (Invitrogen) coupled to Alexa Fluor 488. 

Results

Localization of ASIC3 in the Mouse Retina

Immunohistologic analysis revealed ASIC3 in inner segments of photoreceptors and in the inner nuclear and ganglion cell layers of the ASIC3^+/+ retina (Fig. 1A) . No immunoreactivity was detected in ASIC3^−/− retina (Fig. 1B) . More detailed localization was performed using specific markers of the different retinal cell types (Figs. 1C 1D 1E 1F 1G 1H 1I) . Colabeling with anti–recoverin antibody, a specific marker for inner and outer segments, showed that ASIC3 was present in inner segments of photoreceptors (Fig. 1C) . Given that no expression was detected in the inner segments of cone photoreceptors, as shown by the lack of colocalization with PNA, a specific marker for inner and outer cone segments (Fig. 1D) , we concluded that ASIC3 was expressed in rod inner segments. Double labeling with anti–calbindin D-28K antibody, a specific marker of horizontal cells, showed the expression of ASIC3 in the soma and dense processes of horizontal cells (Fig. 1E) , but no expression in the soma of bipolar cells was detected with PKCα as a marker (Fig. 1F) . Immunoreactivity observed at the proximal portion of the inner nuclear layer corresponded to the presence of ASIC3 in the soma of amacrine and some AII amacrine cells and their dendrites in inner plexiform layer, as shown by colocalization with an anti–syntaxin antibody (Fig. 1G) . Double labeling with an anti–Thy1.2 antibody, a specific marker of mouse ganglion cells, revealed a very low expression of ASIC3 in these cells (Fig. 1H) . In the mouse retina, the ganglion cell layer contains approximately 50% of large ganglion and small ganglion cells and 50% of displaced amacrine cells. ^{34 35} ASIC3 immunoreactivity observed in ganglion cells seemed to include displaced amacrine cells, as shown with the anti–syntaxin antibody (Fig. 1I) . ASIC3 is, therefore, predominantly present in rod inner segments and in horizontal and amacrine cells. 

ASIC3 Contribution to Ganglion Cell ASIC Currents

Because a small expression of ASIC3 was found in ganglion cells (Fig. 1H) , we decided to test whether ASIC3 participates in the ASIC current in RGC cells. We compared the properties of acid-evoked transient currents recorded from wild-type and ASIC3^−/− cultured RGCs (Fig. 2A) . Ganglion cells were recognized by their ability to bind the Thy1.2 antibody. Wild-type and ASIC3^−/− RGCs have the same membrane capacitance (8.40 ± 0.54 pF [n = 15] vs. 9.17 ± 0.63 pF [n = 12], respectively; P = 0.36; unpaired Student’s t-test) and resting potential (−50.87 ± 2.28 mV [n = 15] vs. −52.25 ± 2.69 mV [n = 12], respectively; P = 0.70; unpaired Student’s t-test). The absence of ASIC3 did not alter the overall properties of the ASIC current recorded from ASIC3^−/− RGCs compared with wild-type. Both currents have the same pH activation threshold (approximately pH 6.0) and were not maximally activated at pH 4.0 (Figs. 2A 2B) , indicating a low pH dependence. The current densities appeared to be smaller in ASIC3^−/− RGCs, but the differences were not statistically significant (Fig. 2B ; P > 0.05; Kruskal-Wallis test followed by Dunn post hoc test). Analysis of current inactivation rates did not reveal a difference between wild-type and ASIC3^−/− RGCs (Fig. 2C ; P > 0.05; Kruskal-Wallis test followed by Dunn post hoc test). Electrophysiological data together with the low level of expression of ASIC3 in ganglion cells indicated that ASIC3 had no major role in mouse RGC ASIC currents. 

Physiological Responses of the Retina in ASIC3 Knockout Mice

To explore the functional status of the retina in ASIC3^−/− mice, we recorded the full-field ERG response to increasing intensities of flash light. The scotopic ERG a-wave is an indicator for the decrease of the dark current between inner and outer segments and arises at dim light intensities from the suppression of rod circulating current after photoreceptor activation. The amplitude of scotopic a-wave significantly increased in the retinas of 3-month ASIC3^−/− mice (P < 0.01; Figs. 3B 3E ). The enhancement became significant at 1.68 log scot td/s (Fig. 3E) . The maximum scotopic a-wave amplitude (438.2 ± 38 μV) corresponded to a 19.3% increase compared with wild-type littermates (367 ± 20 μV; Figs. 3A 3E ). Scotopic a-wave amplitudes then decreased progressively and significantly with age. ASIC3^−/− mice of 8, 12, and 16 months showed decreases in maximum scotopic a-wave amplitude compared with wild-type littermates of almost 18% (304 ± 21 μV), 40% (222 ± 22 μV), and 69% (113 ± 21 μV), respectively (Figs. 3C 3D 3E) . No significant alteration with age of a-wave amplitudes was noted in wild-type mice (data not shown). In the rodent retina, only 3% of photoreceptors are cones, ³⁴ and the cone a-wave amplitude primarily affects the amplitude of the scotopic b-wave. ³⁶ Maximum scotopic b-wave amplitude was increased by 17.6% (721 ± 53 μV) in retinas of 3-month-old ASIC3^−/− mice compared with those of control wild-type littermate mice (594 ± 34 μV) but was significantly reduced by 21% (474 ± 68 μV), 32% (406 ± 28 μV), and 64% (219 ± 20 μV) in 8-, 12- and 16-month-old ASIC3^−/− mice, respectively (Fig. 3F) . These data indicate an important dysfunction of the rod system with age in ASIC3^−/− mice. 

Recording of photopic ERG at 2.2 Hz in the light-adapted retina (10 minutes of rhodopsin bleaching) did not reveal a change in cone a-wave amplitudes at all ages tested (Fig. 3G , right upper panel). The b-wave amplitude became moderately but not significantly affected at 16 months for the highest intensities (Fig. 3G , right lower panel). These data show the lack of effect of ASIC3 on cone phototransduction and further support, together with the absence of expression in cones, a role for ASIC3 in rods but not in cones. 

To determine whether phototransduction sensitivity was affected in ASIC3^−/− mice, the kinetics of the scotopic a-wave were compared with those of wild-type mice. ERG profiles were obtained at intense flash intensities from ASIC3^−/− mice at different ages (Figs. 4A 4B 4C 4D) , and ERG waveforms were normalized to those of control to compare the time course of the a-wave. Leading edges of the a-wave overlapped for the four groups (3-month-old wild-type mice; 3-, 8- and 16-month-old ASIC3^−/− mice; Fig. 4E ). These results suggest that phototransduction sensitivity is not altered in the ASIC3^−/− retina despite a significantly reduced maximum response. 

The ratio of the scotopic b- to a-wave amplitude was not affected in the retinas of 3-month-old ASIC3^−/− mice (1.65 ± 0.2; n = 12) compared with 3-month-old ASIC3^+/+ mice (1.56 ± 0.10; n = 12). These results suggest that rod bipolar cell function is not affected. The ratio was also not changed in 8-month-old null mice (1.55 ± 0.19; n = 10), whereas it became slightly but not significantly larger at 12 months (1.83 ± 0.2; n = 10) and 16 months (1.94 ± 0 0.45; n = 8). 

Enhancement of the b-wave amplitude did not affect OPs in 3-month-old ASIC3^−/− mice (685 ± 29 μV; n = 12) compared with wild-type mice (645 ± 65 μV; n = 12; Fig. 4F ). However, significant impairment of the OP amplitudes was observed in the retinas of 8-month-old (387 ± 45 μV; n = 10), 12-month-old (365 ± 32 μV; n = 10), and 16-month-old (160 ± 41 μV; n = 8) ASIC3^−/− mice (Fig. 4F) . These data suggest that some amacrine or AII amacrine cells are affected after 3 months in the ASIC3^−/− retina. 

Importance of ASIC3 for the Long-Term Structural Integrity of the Retina

The altered physiological responses of the retina in ASIC3 knockout mice led us to investigate the retinal structural integrity in these animals. Development of the retina appeared normal in ASIC3^−/− mice at 3 months compared with wild-type animals (Figs. 5A 5B) , and no apparent change was observed in the thickness of the inner and outer retinal layers. Electron micrographs (Figs. 5C 5D 5E 5F)showed no apparent morphologic changes of inner segments (IS; Figs. 5C 5D ) or outer segments (OS; Figs. 5E 5F ) of rod photoreceptors. However, the situation was dramatically changed at 12 months (Figs. 6A 6B) . ASIC3 ^−/− retina displayed an apparent disorganization of the outer nuclear layer (ONL) and of the OS and IS. The thickness of rod OS was increased in ASIC3^−/− mice, and some vacuolated areas appeared in OS and IS as well as holes in the outer plexiform layer (OPL). The number of rows in the ONL was reduced to eight (Fig. 6B) . At 16 months, the OS was more vacuolated and the IS exhibited many vacuolated areas, suggesting massive cell death (Fig. 6C) . The ONL, inner nuclear layer (INL), and OPL were thinner. To determine the cause of retinal disorganization, we examined the ultrastructure of the retina in ASIC3^+/+ and ASIC3^−/− mice. No significant difference between 12-month-old and 3-month-old retinas was observed in wild-type animals because each OS was flat, well organized, and closely aligned with the other (compare Figs. 6D and 5E ). At 12 months, the retinas of ASIC3^−/− mice showed gross disorganization of OS and IS (Figs. 6E 6H)in rod photoreceptors, with many gap area degradation products around the OS and around the mitochondria in the IS. At 16 months, the OS and IS from ASIC3^−/− mice appeared consistently affected by an enhancement of the gap areas, suggesting the degeneration of IS and the concomitant degradation of OS (Figs. 6F 6I) . The INL was also affected and showed some pyknotic nuclei with condensed chromatin (Fig. 6C) . The ganglion cell layer exhibited some displaced amacrine cells, ganglion cells, or both, in a degenerating state (compared Figs. 6K and 6Lwith 6J). These observations are in good agreement with the functional data showing altered retinal physiological responses in knockout mice at 12 and 16 months, and they demonstrate an important role for ASIC3 in the long-term structural integrity of the retina. 

Apoptosis in the Retina of ASIC3 Knockout Mice

To better analyze the mechanisms associated with the loss of photoreceptors and some inner retinal cells in ASIC3^−/− mice, we performed glial fibrillary acidic protein (GFAP) staining and TUNEL staining on retinal sections of 12-month-old ASIC3^+/+ mice and 8- and 12-month-old ASIC3^−/− mice. GFAP activation in Müller cells has been associated with retinal injury caused by a wide variety of insults and is a sensitive indicator of photoreceptor stress. ^{37 38 39} We observed the upregulation of GFAP expression in Müller cells from the retinas of 8-month-old ASIC3^−/− mice and strong expression at 12 months (Figs. 7A 7B 7C) . This result suggested that photoreceptors were affected in ASIC3 knockout mice. In situ labeling of the fragmented nuclear DNA using the TUNEL method on retinal sections of 12-month-old ASIC3^+/+ mice and 8- and 12-month-old ASIC3^−/− mice showed the presence of TUNEL-positive nuclei in the ONL and INL, corresponding to photoreceptor and inner nuclear cell death through apoptosis in ASIC3^−/− retinas of mice older than 8 months (Figs. 7E 7F) . The ganglion cell layer also revealed the presence of many apoptotic cells in temporal and nasal regions in the retinas of 12-month-old ASIC3^−/− mice (Fig. 7F)that were absent in the retinas of 8-month-old ASIC3^−/− and wild-type retinas (Figs. 7D 7E) . These apoptotic cells corresponded to ganglion and displaced amacrine cells, as shown on the retinal sections of 12-month-old ASIC3^−/− mice with TUNEL assay and labeling with, respectively, anti–Thy 1.2 and anti–syntaxin antibody (Figs. 7G 7H) . These data indicated that photoreceptors and cells in the inner nuclear layer and ganglion cell layer underwent apoptosis in the retinas of 8-month-old and older ASIC3 knockout mice. 

Discussion

We have shown that ASIC3 is widely expressed in the mouse retina, where it negatively modulates rod phototransduction in young animals and plays a crucial role for the long-term maintenance of rod photoreceptors, some inner retinal cells, and ganglion cells in older mice. This is the most severe phenotype associated with ASIC channels described thus far in the retina. ^{12 13}  

ASIC3 is strongly expressed in the rod inner segment of the photoreceptor, the region containing all the mitochondria, providing energy and nucleotides, and performing protein synthesis and rhodopsin renewal required for phototransduction. The high metabolic activity associated with the inner segment produces a significant increase in the extracellular proton concentration during subjective night. ⁹ The extracellular acidosis that occurs in the dark-adapted retina has been proposed to modulate many cellular phenomena now known to be regulated by a circadian clock. ^{40 41} Activation of ASIC3-containing channels could participate in these processes. ASIC3 has a high sensitivity to protons, with pH_0.5 close to 6.5, and it responds to an extracellular drop in pH by a biphasic current with a fast desensitizing phase followed by a late sustained current. ^{16 42} The persistent, noninactivating current is activated even in response to modest acidification (pH 6.6–6.9). ^{29 43} In addition, ASIC3 sensitivity to H⁺ is enhanced by lactate, ⁴⁴ which is massively released in the extracellular space of photoreceptors in the dark retina to maintain the high level of energy. ⁴⁵ ASIC3 is also upmodulated by nitric oxide, ⁴⁶ which is abundantly synthesized in the retina. ⁴⁷ Interestingly, a reduction or suppression of the dark current by protons has been described in isolated frog photoreceptors. ⁵ ASIC3 sustained or transient currents could, therefore, participate in the negative modulation of the dark current circulating between rod outer and inner segments in the absence of any light. ⁴⁸ Inactivation of the channel in knockout mice, by removing the inhibitory action of ASIC3, could lead to an increase of the dark current and an increase of the amplitude of the a-wave after light-stimulated cGMP breakdown in young animals. ⁴⁹ Their ability to exert inhibitory roles would therefore emerge as an interesting feature of ASIC channels, as recently demonstrated with the protective function of ASIC1a in seizure progression. ⁵⁰  

As with ASIC3, the inactivation of the ASIC2 gene in mice also leads to an increase in rod a- and b-waves in young animals, ¹² and it has been proposed that ASIC2 is a negative modulator of rod phototransduction. In such cases, however, the b-wave was more strongly affected than the a-wave, and ASIC2 knockout mice were more sensitive to light-induced retinal degeneration than wild-type animals. ¹² Inactivation of ASIC3 does not significantly affect the b-wave amplitude and the b/a ratio, an indicator for bipolar neuron activity, suggesting that ASIC3, contrary to ASIC2, does not play a substantial role in the modulation of neurotransmission between photoreceptors and bipolar cells or in the activity of bipolar neurons. This is in good agreement with the lack of significant expression of ASIC3 in these cells. In addition, ASIC3 knockout mice have no modification in their sensitivity to light-induced retinal degeneration (Supplementary Fig. S1), suggesting no or only modest contribution of ASIC3 to the adaptation of the retina to light. This suggests differences in the role of ASIC2a and ASIC3 in rods, the role of ASIC3 being probably restricted to the inner segment. 

We have shown the expression of ASIC3 in displaced amacrine cells and, to a lesser extent, in ganglion cells. The retinal ganglion cell layers exhibit a high level of metabolic activity, which has been shown to cause transient interstitial acidification around the optic nerve during normal neurotransmission ⁵¹ and acidification in the synaptic cleft. ⁵² ASIC-like currents have been described in primary cultures of rat RGC. ^{13 15} We have suggested in a previous work that these currents could be supported by heteromeric channels containing the ASIC1a, ASIC2a, and ASIC3 subunits, ¹³ probably assembled in trimers, as indicated by the recent determination of the tridimensional structure of a chicken ASIC1 deletion mutant. ²⁵ The data shown here do not seem to support a significant participation of ASIC3 in the main ASIC current of RGC, raising questions about the role of this subunit in retinal ganglion cells. The very acidic pH dependency and the inactivation rate of RGC ASIC current suggest participation of ASIC2a-containing heteromeric channels ²⁷ or, alternatively, of a not yet described ASIC channel isoform. ¹³  

Functional and morphologic analysis of retinas in ASIC3^−/− mice at 8 months and later demonstrated an essential role for ASIC3 in the maintenance of retinal integrity. The progressive reduction of the scotopic a- and b-wave amplitudes is well correlated with morphologic and ultrastructural analysis showing a progressive disorganization and degeneration of rod photoreceptor inner and outer segments. How ASIC3 deletion affects the integrity of rod photoreceptors remains to be established. This could be a consequence of the enhanced visual transduction observed in the first months in knockout mice. ASIC2 knockout mice have similarly increased activity but do not display retinal degeneration up to 16 months (data not shown). Therefore, the ASIC3 modulation of rod phototransduction probably involves a different mechanism restricted to the inner segment. In addition, the progressive loss of rod photoreceptor cells was concomitant to or followed the loss of cells in the INL. ASIC3 is expressed in some amacrine and AII amacrine cells. These cells are critical elements in the primary rod pathway because they act as an obligatory conduit of rod signals to on- and off-center ganglion cells. The reduction of oscillatory potentials strongly suggests the death, impairment, or both of specific amacrine cells in ASIC3^−/− mice after 3 months. However, in addition to reflecting the activity of amacrine cells, ⁵³ OPs are indicators of neuronal dysfunction, affecting inhibitory feedback pathways or revealing pathologic microcirculation in the inner retina, or both. Secondary vascular changes in ASIC3^−/− retina cannot, therefore, be completely ruled out. ASIC3 is also expressed in horizontal cells, and the reduction of the outer plexiform layer thickness and the appearance of holes suggest some effect on these cells. Exclusive loss of horizontal cells causes selective b-wave reductions. ⁵⁴ Only scotopic, but not photopic, b-wave was markedly reduced in older ASIC3^−/− mice, suggesting a minor impact of ASIC3 in horizontal cells. 

The inactivation of ASIC3 mimics the deleterious effects observed in glaucoma and in chronic ischemia associated with inner segment defects. The main characteristic feature of glaucoma and ischemia is the progressive and relatively slow loss of retinal-displaced amacrine and ganglion cells. ⁵⁵ The abnormalities observed in ASIC3 knockout mice are reminiscent of some human retinal diseases such as Usher syndrome, a frequent cause, combined with retinitis pigmentosa, of hereditary deafness and blindness in humans. ^{56 57} Interestingly, ASIC3 null mice also experience mild to moderate hearing loss at 4 months of age. ⁵⁸ The cell death observed in ASIC3^−/− mice appeared to occur by apoptosis, as previously seen in a variety of mice models of retinal degeneration. ⁵⁹ No link between ASIC3 and retinal dystrophies has been described thus far in humans. However, this new channel protein essential for retinal integrity will now have to be considered in human visual disorders associated with retinal degeneration. Diclofenac, a nonsteroidal anti-inflammatory drug widely used in eye drops to treat inflammation after cataract extraction, ⁶⁰ is also a direct inhibitor of ASIC3. ⁶¹ Results presented in this work suggest that prolonged treatment with diclofenac may mimic, to a certain extent, partial deletion of the ASIC3 gene and may have deleterious effects. ASIC3 knockout mice also provide a new valuable model of visual loss associated with night blindness. 

 Supported by the Centre National de la Recherche Scientifique and the Agence Nationale de la Recherche (ANR-05-NEUR-031).

 Submitted for publication October 17, 2008; revised December 5, 2008; accepted March 11, 2009.

 Disclosure: M. Ettaiche, None; E. Deval, None; S. Pagnotta, None; M. Lazdunski, None; E. Lingueglia, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Immunolocalization of the ASIC3 protein in mouse retina. (A) Immunolabeling of wild-type retina with an anti–ASIC3 antibody revealed the presence of the ASIC3 protein (green) in IS, in cells at the distal and proximal borders of the INL, in inner plexiform layer (IPL), and in the ganglion cell layer (GCL). Sections were counterstained with DAPI. (B) Immunolabeling of ASIC3^−/− retina with the anti–ASIC3 antibody gave no detectable signal except faint, nonspecific labeling in the ganglion cell layer. (C–I) Double labeling of ASIC3 (green) with markers specific for the different cell types in the retina (red); arrowheads indicate signals in the soma, and arrows indicate signals in the dendritic buttons. (C) Colocalization with recoverin antibody, a marker of rod and cone inner segments. (D) Lack of expression in cone inner segments detected with peanut agglutinin antibody (PNA). (E) ASIC3 is expressed in the soma of horizontal cells, as detected with anti–calbindin monoclonal D-28K antibody. (F) No expression of ASIC3 in bipolar cells, as demonstrated with monoclonal phospho-kinase Cα antibody (PKCα). (G) Colocalization with syntaxin in amacrine cells. (H) Little colocalization was detected in ganglion cells with rat anti–Thy1.2 antibody. (I) Expression in displaced amacrine cells in the GCL detected with monoclonal syntaxin antibody; Scale bars: (A, B) 40 μm. (C–I) 20 μm.

Figure 2.

 

ASIC currents in wild-type and ASIC3^−/− cultured RGCs. (A) Typical ASIC-type currents recorded at −80 mV from wild-type and ASIC3^−/− RGCs. Dotted lines: zero current level. The pH at which currents were elicited are indicated (pH drop from pH 7.4). (B) Mean ASIC current densities plotted as a function of extracellular pH, indicating that pH dependency and current amplitudes were the same in wild-type and ASIC3^−/− RGCs (n = 9–13 and n = 11–12 for wild-type and ASIC3^−/−, respectively; P > 0.05). (C) Statistical analysis of the inactivation rates (fitted with a monoexponential) measured from wild-type and ASIC3^−/− RGC ASIC currents as a function of extracellular pH. Wild-type and ASIC3^−/− RGC ASIC currents have the same inactivation rates at pH 6.0, 5.0, and 4.0 (n is indicated above each bar; P > 0.05).

Figure 3.

 

Effect of the ASIC3 inactivation on scotopic and photopic ERG. (A–D) ERG profiles obtained at flash intensities ranging from −3.52 to 3.54 log scot td/s. Each trace is the average of at least eight mice. (E, F) Intensity response curve of the dark-adapted a-waves and b-waves in the retinas of 3-month-old ASIC^+/+ and in 3-, 8-, 12-, and 16-month-old ASIC3^−/− mice. Flash intensities are expressed in log scot td/s. Symbols: mean ± SD. Asterisks: statistically significant difference from wild-type (P < 0.05). The amplitudes of scotopic a- and b-waves significantly increased in retinas of 3-month-old ASIC3^−/− mice and decreased progressively and significantly with age. (G) Effect of the ASIC3 inactivation on light-adapted retina ERG waveform extracted at 2.2 Hz. Left: photopic ERG traces elicited at high intensity (3.54 log scot td/s) in retinas of 3-month-old ASIC^+/+ and in 3-, 8-, 12-, and 16-month-old ASIC3^−/− mice. Each photopic ERG trace represents the average of at least eight mice. Right: intensity-response curve of the light-adapted ERG a-waves (upper) and b-waves (lower) in retinas of 3-month-old ASIC^+/+ (black squares) and 3- (gray circles), 8- (gray triangles), 12- (gray inverted triangles), and 16- (gray diamond) month-old ASIC3^−/− mice, respectively. Photopic a- and b-waves remained unaffected.

Figure 4.

 

Time course of scotopic a-wave and oscillatory potentials extracted from scotopic ERG waveform. (A–D) Initial negative deflection as the saturated a-wave to intense flash intensities ranging from 1.68 to 3.54 log scot td/s in 3-month-old ASIC^+/+ mice and in ASIC3^−/− mice at 3, 8, and 16 months. (E) Time course of saturated a-wave. Normalization of the ERG waveforms elicited by 3.5 (left) and 3.28 (right) log scot td/s as shown in (A–D) (n = 12 for 3-month-old wild-type mice; n = 12, 10, and 8 for 3-, 8-, and 16-month-old ASIC3^−/− mice, respectively). Arrow: flash onset. Time course of the a-wave was not affected in the ASIC3^−/− retinas. (F, left) Oscillatory potential (OP) traces of 3-month-old ASIC3^+/+ and 3-, 8-, 12-, and 16-month-old ASIC3^−/− mice extracted at the highest intensity (3.54 log scot td/s). Each trace represents the average of at least eight mice. Right: intensity-response curve obtained in 3-month-old ASIC^+/+ and in 3-, 8-, 12-, and 16-month-old ASIC3^−/− mice were plotted as a function of intensity ranging from 1.68 to 3.54 log scot td/s. Average OP amplitudes were not affected at 3 months in knockout mice, but significant impairment was observed in older animals. Symbols: mean ± SD. Asterisks: statistically significant difference from wild-type animals (P < 0.05).

Figure 5.

 

Morphology of the retina in 3-month-old ASIC3^+/+ and ASIC3^−/− mice. (A, B) Semithin retinal sections (1 μm) counterstained with methylene blue-azur II. ASIC3^−/− mice have normal retinas in terms of outer nuclear layer (ONL) and in INL thickness. (C–F) Electron microscopy of 80 nm–thick retinal sections counterstained with uranyl acetate showed a similar morphology of inner and outer segments in ASIC^+/+ and ASIC^−/− mice. Scale bars: (A, B) 20 μm; (C–F) 2 μm. OPL, outer plexiform layer; OS, outer segments; RPE, retinal pigment epithelium, MI, mitochondria, N, nucleus.

Figure 6.

 

Retinal morphology in 12- and 16-month-old mice. (A–C) Semithin retinal sections (1 μm) from ASIC3^+/+ and ASIC3^−/− mice of various ages, counterstained with cresyl violet acetate. (A) At 12 months, the retina remained unchanged in ASIC3^+/+ mice compared with younger animals (Fig. 5A) . (B) Retinas of ASIC3^−/− mice at 12 months showed more elongated OS and IS with vacuolated areas (black arrows) and holes in the OPL (white arrows), and the ONL was reduced in terms of number of photoreceptor rows. (C) At 16 months, retinas in ASIC3^−/− mice were in a degenerating state. Note the increased number and size of vacuolated areas (black arrows), primarily in IS and OS, and the reduced thickness of the entire retina as well as the presence of pyknotic nuclei in the distal and proximal portions of the INL (white arrows). (D–L) Representative electron micrographs of 80-nm ultrathin retinal sections showing the degenerating state of OS, IS, and GCL at 12 months (E, H, K) and 16 months (F, I, L) in ASIC3^−/− mice compared with OS, IS, and GCL of ASIC3^+/+ mice (D, G, J). Vacuolated areas were observed (black arrows) between OS (E, F) and in IS between mitochondria (MI; indicated by white circle) (H, I). The 16-month-old ASIC3^−/− retina showed severely shortened OS (F) and IS (I) and the absence of nucleus in displaced amacrine cells, ganglion cells, or both (L). IS (H, I) contains vacuolated degenerated mitochondria (dm) and many vacuolated areas (black arrows). Vb, vitreous body. Scale bars: (A–C) 20 μm; (D–F, J–L) 2 μm; (G–I) 1 μm.

Figure 7.

 

GFAP expression and apoptosis in ASIC3^+/+ and ASIC3^−/− retina. (A) In the 12-month-old ASIC3^+/+ retina, GFAP (red) was predominantly expressed in astrocytes in the GCL and around blood vessel in the OPL, as expected. (B, C) In the retinas of 8- and 12-month-old ASIC3^−/− mice, GFAP expression was significantly increased in Müller cells (arrows). Immunoreactivity was observed along the Müller cell from the end foot membranes facing the vitreous body (Vb) to the outer surface of the retina in the microvillus process. At 12 months, GFAP expression was observed along the Müller cell except in the outer surface of the ASIC3^−/− mouse retina. (D–F) TUNEL labeling of wild-type and ASIC3^−/− mouse retina. (D) Absence of apoptotic nuclei (green) in 12-month-old wild-type mouse retina. (E) In the 8-month-old ASIC3^−/− mouse retina, apoptotic nuclei (green) were localized in INL and ONL. (F) In 12-month-old ASIC3^−/− mouse retina, apoptotic nuclei were localized in ONL and appeared in GCL. (G, H) Characterization of apoptotic cells in GCL. (G) Some apoptotic nuclei (green) corresponded to ganglion cells (white arrows) detected with anti–Thy1.2 antibody. (H) Apoptotic nuclei corresponded with some displaced amacrine cells (white arrow) detected with syntaxin antibody. Scale bars: (A–F) 20 μm; (G, H) 30 μm.

Supplementary Materials

Supplementary Figure S1 - (PDF)  

Effect of light exposure on scotopic ERG. Left, ERG traces recorded before illumination (day 0) and 1, 2 and 7 days after light exposure (15 Klux, 2 hours, after 16 hours dark-adaptation) of 2 old-months ASIC3^+/+ and ASIC3^-/- mice and extracted at high intensity. Each trace represents the mean of 14 traces from 7 mice. Right, values of a-wave (top) and b-wave (bottom) amplitudes as a function of age. Values represent means ± SD of the a- and b-wave amplitude recorded before (day 0) and after light exposure (1, 2 and 7 days). * p <_0.05 nonparametric="nonparametric" mann-whitney="mann-whitney" test.="test." the="the" effect="effect" of="of" illumination="illumination" was="was" same="same" in="in" asic3="asic3" knock-out="knock-out" mice="mice" and="and" wild-type="wild-type" with="with" no="no" modification="modification" a-="a-" b-wave="b-wave" suggesting="suggesting" very="very" modest="modest" to="to" contribution="contribution" adaptation="adaptation" retina="retina" light="light" on="on" light-induced="light-induced" retinal="retinal" degeneration.--="degeneration.--" end="end" desc="desc" dc1="dc1" _1="_1" _--="_--"></_0.05> 

The authors thank Rainer Waldmann and Nicolas Guy for providing the ASIC3^−/− mice, Xavier Gasull for comments on the manuscript, and Valérie Friend for technical assistance.