Axin1 stabilizes S-opsin and maintains cone photoreceptor survival by inhibiting GSK3β activity

Abstract

To ensure high phototransduction efficiency in the retina, the precise subcellular localization of signaling molecules must be tightly orchestrated by scaffold proteins. Aberrant localization of these scaffold proteins not only disrupts the transition of photoelectrical signals but also triggers endoplasmic reticulum (ER) stress, which leads to photoreceptor apoptosis. However, it is unknown how these proteins are localized to specific subcellular compartments of photoreceptors or how protein mislocalization is coupled with apoptotic signaling. Herein, we observed a specific spatiotemporal expression pattern of the scaffold protein, Axin1, in the mouse retina. We found that Axin1 is essential for the retinal localization of S-opsin chromoprotein in the outer segment of photoreceptors. Moreover, retinal Axin1 deficiency disrupts light perception, accompanied by cone photoreceptor loss and ER stress. In addition, knockdown of Axin1 exacerbates ER stress-induced apoptosis of cone-derived 661W cells. Consistently, pharmacological elevation of Axin1 protein level alleviates tunicamycin-induced ER stress and apoptosis via inhibition of GSK3β activity. Thus, our findings demonstrate that Axin1 plays a pivotal role in organizing the phototransduction complex and ensuring photoreceptor survival in the retina.

Introduction

Retinal degeneration involves the irreversible apoptosis of retinal nerve cells or adjacent supporting tissues and leads to retinal degenerative diseases, such as age-related macular degeneration, retinitis pigmentosa, and Stargardt’s disease [1]. Of note, the progressive loss of photoreceptors is a primary contributor to retinal degeneration [2]. Photoreceptors are specialized neurons in the outer nuclear layer of the retina that sense and convert photons into electrochemical signals for visual perception [3]. This process involves two types of morphologically and functionally distinct photoreceptors: rods express rhodopsin for black-and-white vision and dim light, while cones express S-opsin and M-opsin for color vision and bright light, respectively [4]. Phototransduction within photoreceptors involves a series of intricate biological events, including activation of photosensitive pigments on the membrane and the downstream G protein, transducin; this leads to the hydrolysis of cGMP by phosphodiesterase (PDE) and changes in membrane potential to modulate neurotransmitter release at presynaptic terminals [5]. Because of their significant metabolic demands along with continuous long-term exposure to light, photoreceptors are particularly susceptible to endoplasmic reticulum (ER) stress and light toxicity [6].

Photoreceptor degeneration is an irreversible process believed to occur primarily via apoptosis [7]. ER stress-associated apoptosis is implicated in retinal degeneration. Moreover, ER stress-induced photoreceptor death is observed in numerous models of retinal degeneration. For instance, retinitis pigmentosa involving expression of P23H, T17M, or S334ter rhodopsin mutations leads to the misfolding and retention of rhodopsin [8]. In addition, excessive light exposure can generate reactive oxygen species, which damage cellular components, further exacerbating ER stress [9]. Thus, ER stress-induced cell death is a major mechanism of photoreceptor degeneration.

Scaffold proteins provide a platform for signaling molecules to dock, assemble, and coordinate signal transduction during phototransduction [10, 11]. The dysfunction of scaffold proteins results in the collapse of signaling cascades, which leads to retinal degradation [12, 13]. Axin1 (Axis inhibitor 1) is a scaffold protein that plays pivotal roles in intracellular events, such as the Wnt, TGFβ, SAPK/JNK, and p53 signaling pathways [14]. Disruption of these signaling pathways is often associated with photoreceptor dysfunction [15]. Activation of the Wnt/β-catenin signal pathway in the retina protects photoreceptors from light damage, hydrogen peroxide-induced oxidative stress, and N-methyl-N-nitrosourea (MNU)-mediated photoreceptor apoptosis [15, 16]. However, little is known about the role of Axin1 in the maintenance of photoreceptor survival.

Accordingly, in the present work, we demonstrate that Axin1 specifically accumulates in the outer segment of cone photoreceptors, where it colocalizes with S cones. Meanwhile, loss of Axin1 in the retina results in deficits of light perception in mice, accompanied by aberrant S-opsin distribution and ER stress-induced retinal degeneration. Furthermore, Axin1 deletion accelerates the ER stress and apoptosis of retina-derived 661W cells, whereas Axin1 stabilization alleviates photoreceptor degeneration via the inhibition of GSK3β activity. Thus, our study reveals a pivotal role of Axin1 in photoreceptor degeneration.

Results

Spatiotemporal expression of Axin1 in the mouse retina

While Axin proteins have been identified as key scaffold proteins that mediate signaling pathways in response to retinal development [17], their expression profiles and specific roles in the retina remain largely unknown. Immunofluorescence staining of flat mounts of retinas from 2–3-month-old mice showed that Axin1 protein expression increased gradually along the dorsoventral axis (Fig. 1A, B). In particular, Axin1 was sparsely expressed in the dorsal area but enriched in the ventral area (Fig. 1C). We confirmed the dorsal-to-ventral expression pattern of Axin1 by cross-sectional staining (Fig. 1D, E). In addition, Axin1 was mainly present in the retinal pigment layer and concentrated in the outer segment of the photoreceptors (Fig. 1F–H). We confirmed that Axin1 expression was restricted to cone photoreceptors by co-staining with cone arrestin, which is a photo-inhibitory component of cone cells but not rod cells (Supplementary Fig. 1A, B). Subsequent co-staining of Axin1 and markers representing different cone photoreceptors showed that Axin1 colocalized with S-opsin, which senses short wavelengths of light, and partially colocalized with M-opsin, which senses medium wavelengths of light (Fig. 1J–N, Supplementary Fig. 2A, B). We also observed a similar dorsal-to-ventral distribution of Axin1 with S-opsin but not M-opsin in flat-mounted mouse retinas labeled with the cone cell marker, PNA (peanut agglutinin) (Supplementary Fig. 1C–J). During retinal development, Axin1 and S-opsin exhibited similar spatiotemporal expression patterns: they increased gradually from a spot-shaped structure at the early stage (i.e., before postnatal day 10) to a rod-shaped structure at the outer segment of the retina (i.e., after postnatal day 15) (Supplementary Fig. 3). Accordingly, the well-controlled expression pattern of Axin1 in the outer segment of cone photoreceptors suggests a potential role in phototransduction.

Fig. 1. Axin1 is predominantly expressed in the outer segment of cone photoreceptors.

A Expression pattern of Axin1 in 2-month-old mouse retinas. Retinal wholemounts were immunostained with an antibody against Axin1 (red). Scale bar: 1000 μm. B Quantification of the fluorescence intensity of Axin1 signals in each square from the dorsal to ventral retina (n = 5 retinas). The x-axis represents the sampling location, while the y-axis denotes the fluorescence intensity. Dot size reflects the magnitude of fluorescence intensity. The R²-value and p-value of the regression model are displayed in the top-left corner. C Representative images of Axin1 staining in the dorsal and ventral regions of mouse retinal wholemounts at high magnification. Scale bar: 20 μm. D Immunostaining of Axin1 (green) and F-actin (red) in sagittal sections of mouse retinas. Scale bar: 500 μm. E Enlarged image showing Axin1 staining in the dorsal and ventral retinas of sagittal sections. Scale bar: 50 μm. F Co-staining of Axin1 and F-actin in sagittal sections of the mouse retina. Scale bar: 50 μm. ONL outer nuclear layer, OPL outer plexiform layer, INL inner nuclear layer, IPL inner plexiform layer, GCL ganglion cell layer, RPE retinal pigment epithelium, OS outer segment, IS inner segment. G Subcellular localization of Axin1 in photoreceptors. Scale bar: 5 μm. H, I Structural diagram of rod and cone photoreceptors. J Colocalization of Axin1 and S-opsin as well as S-opsin and M-opsin in both the dorsal and ventral regions of the mouse retina. Scale bar: 20 μm. K–N Quantification of the fluorescence intensity in (I).

Axin1 is required for light perception in the retina

The colocalization of Axin1 with S-opsin prompted us to examine the role of Axin1 during light response. Accordingly, we performed a pupillary light reflex (PLR) experiment to evaluate the light perception of cone photoreceptors in the retinas of Axin1-deficient mice. We delivered adeno-associated virus (AAV)-expressing Cre recombinase driven by mouse blue-opsin promoter into the retinas of Axin1^loxP/loxP mice at 7 weeks old [18]. We confirmed the knockdown efficiency of Axin1 in cone photoreceptors at 11 weeks (Supplementary Fig. 4). Next, we exposed the infected mouse eyes to light at 15 weeks (Fig. 2A) and determined the PLR by measuring the changes in the diameter of the pupils after 30 s light stimulation. Interestingly, the PLR of Axin1-deficient retinas in response to blue or green light was significantly slower than that in wild-type retinas, suggesting that loss of Axin1 in cone photoreceptors disrupts short-wave-light transduction (Fig. 2B–E, Supplementary Fig. 5).

Fig. 2. Axin1 is required for light perception in the retina.

A Schematic diagram of virus delivery and the pupillary light reflex (PLR) test. Axin1 was depleted by injection of adeno-associated virus (AAV) expressing Cre driven by blue pigment promoter (BP-Cre) into Axin1^loxP/loxP mouse retinas. B Representative image of mouse eyes after PLR by stimulation with blue and green light. C and D Normalized pupillary diameter after PLR. E Quantification of pupil constriction within 30 s upon light stimulation (Axin1^+/+: n = 14, Axin1^loxP/loxP: n = 14; **p < 0.01, ***p < 0.001, unpaired Student’s t-test). F Schematic diagram of light–dark transition test. G Normalized time spent in dark and light areas (Axin1^+/+: n = 13, Axin1^loxP/loxP: n = 15; *p < 0.05, n.s not significant, paired Student’s t-test). H Staining of c-Fos in retinas from wild-type and Axin1-deficient mice after light exposure. Scale bar: 20 μm. ONL outer nuclear layer, OPL outer plexiform layer, INL inner nuclear layer, IPL inner plexiform layer, GCL ganglion cell layer, RPE retinal pigment epithelium. I Quantification of the percentage of c-Fos⁺ cells in the ganglion cell layer (Axin1^+/+: n = 11, Axin1^loxP/loxP: n = 9; *p < 0.05, unpaired Student’s t-test).

To evaluate the visual function of Axin1-deficient mice, we performed a light–dark transition test. Compared to the wild-type mice, which preferred the dark chamber, Axin1-deficient mice exhibited no preference for the dark chamber during the light–dark transition (Fig. 2F, G), suggesting that Axin1-deficient mice exhibit disrupted light-mediated signal transduction during visual perception. During visual processing, the light-induced activation of photoreceptors converts light signals into electrical signals, which is followed by the activation of retinal ganglion cells [19]. The induced expression of the immediate early gene, FOS, which encodes c-Fos, in the ganglion layer is an indicator of the functional activation of photoreceptors. After the mice were exposed to light, the immunoreactivity of c-Fos in the retinas of wild-type mice increased significantly, whereas there were fewer c-Fos⁺ ganglion cells in Axin1-deficient mice (Fig. 2H, I). These findings collectively suggest that Axin1 is required for light signal transition and visual perception in the retina.

Loss of Axin1 leads to retinal degeneration

Given the spatial distribution of Axin1 and its role in light perception in the retina, we examined the distribution of opsin in cone photoreceptors in Axin1-deficient retinas. The retinal expression of S-opsin was significantly reduced in Axin1-deficient retinas; this was accompanied by altered S-opsin morphology and localization from a rod-shaped structure confined to the outer segment in wild-type retinas to small puncta in Axin1-deficient retinas (Fig. 3A–D). Quantitative PCR (qPCR) analysis showed that Axin1 deficiency resulted in decreased total S-opsin expression in these retinal tissues (Supplementary Fig. 6A). We also observed a moderate but significant reduction of M-opsin expression in Axin1-deficient retinas. These findings suggest that Axin1 is required for maintaining S-opsin at the outer segment of cone photoreceptors.

Fig. 3. Axin1 stabilizes S-opsin and M-opsin in the outer segment of cone photoreceptors.

A Reduction of Axin1, S-opsin, and M-opsin signals in Axin1-deficient photoreceptors. Scale bar: 50 μm. ONL outer nuclear layer, OPL outer plexiform layer, INL inner nuclear layer, IPL inner plexiform layer, GCL ganglion cell layer, RPE retinal pigment epithelium. B–D Quantification of the areas of the fluorescence signals for Axin1, S-opsin, and M-opsin in (A) (n = 12–13 retinas from 3 independent experiments; ***p < 0.001, unpaired Student’s t-test).

Photoreceptor abnormalities disrupt retinal function and contribute to degenerative diseases by affecting peripheral neurons, such as the retinal pigment epithelium (RPE), horizontal cells, and Müller cells. Among them, the RPE plays critical roles in nutrient delivery, waste management, and photoreceptor maintenance [20, 21]. Therefore, we investigated the integrity of the RPE in mice with Axin1 knockout in cones by performing immunostaining for zonula occludens-1 (ZO-1), an essential component of RPE tight junctions. In the wild-type retinas, ZO-1 formed a continuous line at the outer limiting membrane. In contrast, the retinas lacking Axin1 exhibited a loss of ZO-1 signal at the outer limiting membrane, indicating compromised integrity (Fig. 4A, B).

Fig. 4. Loss of Axin1 results in retinal degeneration.

A Immunostaining of retinal pigment epithelium (RPE) tight junctions for zonula occludens-1 (ZO-1) and synaptic marker synaptophysin (SYN) in horizontal cells, as well as astrocyte marker GFAP in Axin1-deficient mouse retinas. Scale bar: 50 μm. ONL outer nuclear layer, OPL outer plexiform layer, INL inner nuclear layer, IPL inner plexiform layer, GCL ganglion cell layer, RPE retinal pigment epithelium. B–D Quantification of the areas of the fluorescence signals for ZO-1, SYN, and GFAP (n = 5–10 retinas from three independent experiments; *p < 0.05, **p < 0.01, ***p < 0.001, unpaired Student’s t-test).

Synaptophysin (SYN) is a synaptic vesicle protein expressed in the synaptic vesicles of neurons. Increasing evidence indicates that the synaptic interactions between neurons in degraded retinas change substantially early on [22, 23]. Consistently, we found that Axin1 deletion in the retina led to reduced SYN expression in the outer plexiform layer (Fig. 4A, C), indicating that the absence of Axin1 in the cones disrupts the synaptic conduction between the cones and horizontal cells. Retinal degeneration is also associated with astroglial activation [24]. In wild-type mice, GFAP, a marker of astrocytes, was enriched in horizontal glial filaments and restricted to the ganglion layer (Fig. 4A). In contrast, in Axin1-deficient mouse retinas, the GFAP signal invaded other layers, such as the inner plexiform layer, inner nuclear layer, and outer plexiform layer, indicating that photoreceptor degeneration had led to extensive glial cell activation (Fig. 4A, D).

Axin1 deficiency leads to endoplasmic reticulum stress-associated apoptosis in retinal cells

Cone degeneration is associated with ER stress [25]. We also observed a significant increase in the accumulation of CHOP (C/EBP homologous) protein in the photoreceptor region of the retinas of Axin1-deficient mice, suggesting that ER stress occurs in Axin1-deficient photoreceptors (Fig. 5A, B). Given that loss of Axin1 induces ER stress and photoreceptor degeneration in the retina, we investigated the roles of Axin1 in ER stress using photoreceptor-derived 661W cells, a well-established cell model used in mechanistic studies of the retina [26]. Accordingly, we first assessed the transcriptional levels of genes associated with the ER stress by qPCR in 661W cells with siRNA-mediated Axin1 deletion exposed to 450-lx blue light (Fig. 5C). The expression of XBP1s, an early marker of ER stress, was significantly increased after 2 h of blue light insult in Axin1-knockdown cells in compared to the controls (Fig. 5D). Furthermore, we observed an increase in the expression CHOP in Axin1-deficient cells upon blue light exposure (Fig. 5E). Under prolonged blue light exposure, a slight reduction of S-opsin expression in Axin1-deficient cells we observed (Supplementary Fig. 6C). Meanwhile, there were no significant changes in the expression levels of ATF4, a transcriptional factor induced by PERK–eIF2α signaling, or ATF6, which translocate to the nucleus to activate chaperones and ER quality control genes, in Axin1-deficient cells upon blue light stimulation (Supplementary Fig. 6D–F). Furthermore, TUNEL staining of 661W cells with Axin1 deletion exposed to blue light revealed a significant increase in apoptotic cells (Fig. 5F, G). To further investigate the protective role of Axin1 in light-induced damage, we detected the levels of apoptosis-related proteins in 661W cells with Axin1 deletion. Cleaved caspase-3 expression was obviously increased in the 661W cells with Axin1 knockdown upon blue light exposure compared to the controls (Fig. 5H–K). These results demonstrate that Axin1 knockdown exacerbates light-induced ER stress and apoptosis.

Fig. 5. Axin1 alleviates endoplasmic reticulum stress and apoptosis caused by light damage.

A Immunostaining for the endoplasmic reticulum (ER) stress marker, CHOP, in Axin1-deficient mouse retinas. Scale bar: 50 μm. B Quantification of the areas of the fluorescence signals for CHOP (n = 6 retinas from 3 independent experiments; **p < 0.01, unpaired Student’s t-test). C Schematic diagram of blue light exposure to 661W cells. D, E The expression levels of XBP1s and CHOP in 661W cells were detected by qPCR analysis. GAPDH served as an internal control. Data are presented as mean ± SEM (from 4 independent experiments; **p < 0.01, ***p < 0.001 vs. control, unpaired Student’s t-test). F, G TUNEL staining showed that Axin1 protected 661W cells from damage caused by blue light exposure (n = 4 dishes from 3 independent experiments; *p < 0.05, unpaired Student’s t-test). H 661W cells were transfected with siRNA against Axin1, followed by blue light exposure. The treated cells were subjected to western blotting to detect Axin1, caspase-3, cleaved caspase-3, and GAPDH. I–K Quantification of immunoblots in (H) (from 3 independent experiments; *p < 0.05, unpaired Student’s t-test).

Stabilized Axin1 protects retinal cells from endoplasmic reticulum stress and apoptosis by inhibiting GSK3β activity

Tunicamycin (Tm) is widely used to trigger ER stress by inhibiting N-glycosylation [27]. Upon Tm treatment, Axin1 expression in 661W cells decreased in a dose-dependent manner, whereas the expression of CHOP, an ER stress marker, increased significantly (Fig. 6A, B). The inhibition of GSK3β activity can protect cells from ER stress-induced apoptosis [28]. Concordantly, GSK3β phosphorylation at serine 9 also decreased in 661W cells upon Tm treatment, suggesting that GSK3β activity is potentiated during Tm-induced ER stress (Fig. 6A, B). Meanwhile, siRNA knockdown of Axin1 exacerbated Tm-induced ER stress, which was followed by apoptosis, as indicated by increased expression of CHOP, pro-apoptotic Bax, and cleaved caspase-3 with concomitant decreases of GSK3β serine 9 phosphorylation and expression of anti-apoptotic Bcl-2 (Fig. 6C–K). Axin1-deficient cells also exhibited a much higher ratio of TUNEL⁺ cells upon Tm treatment (Fig. 6L, M), indicating that loss of Axin1 enhances cell susceptibility to ER stress-associated apoptosis. These findings collectively suggest that Axin1 has a protective effect against Tm-induced ER stress.

Fig. 6. Axin1 knockdown potentiates endoplasmic reticulum stress and apoptosis in retinal cells.

A, B 661W cells were treated with different doses of tunicamycin (Tm) for 12 h. The cells were collected to examine the expressions of Axin1, CHOP, and pSer9-GSK3β. B Quantification of immunoblots in (A) (from 3 independent experiments; ***p < 0.001, ns: not significant, unpaired Student’s t-test). The x-axis indicates the dose, and the y-axis indicates the corresponding relative expression levels. The R²-value and p-value of the regression models are shown in the bottom-left corner. *p < 0.05, **p < 0.01, ***p < 0.001 unpaired Student’s t-test). C 661W cells were transfected with siRNA against Axin1, followed by Tm treatment. The treated cells were collected at the indicated timepoints and subjected to western blotting to detect Axin1, CHOP, Bcl2, Bax, cleaved caspase-3, and pSer9-GSK3β. D–K Quantification of immunoblots in (C) (from 4 independent experiments; *p < 0.05, **p < 0.01, ***p < 0.001, unpaired Student’s t-test). L TUNEL staining revealed a significant increase in apoptotic cells among Axin1-knockdown 661W cells upon Tm treatment. Scale bar: 20 μm. M Quantification of the ratio of TUNEL⁺ cells from each condition (n = 9 dishes from 3 independent experiments; **p < 0.01, unpaired Student’s t-test).

XAV939 is a tankyrase inhibitor that pharmacologically stabilizes Axin1 protein in various cell lines and the developing mammalian brain [29]. Accordingly, we confirmed that XAV939 stabilizes Axin1 in 661W cells (Fig. 7A). In addition, XAV939 protected 661W cells from Tm-induced ER stress, as evidenced by decreased expression of CHOP, Bax, and cleaved caspase-3; increased expression of GSK3β serine 9 phosphorylation and Bcl-2 (Fig. 7A-C); and fewer TUNEL⁺ cells (Fig. 7D, E). Moreover, we treated 661W cells with LiCl, a potent inhibitor of GSK3β activity, before treating them with Tm. Accordingly, LiCl treatment significantly increased GSK3β phosphorylation and alleviated the induction of cleaved caspase-3 and CHOP caused by Axin1 knockdown (Fig. 7F–M). Notably, this protective effect was not limited to Tm-induced stress; LiCl also mitigated blue light-induced ER stress and apoptosis in Axin1-deficient cells (Supplementary Fig. 7A–F). These results collectively suggest that stabilizing Axin1 protects retinal cells from ER stress-induced apoptosis via inhibition of GSK3β activity.

Fig. 7. Stabilization of Axin1 by XAV939 treatment protects cells from endoplasmic reticulum stress.

A Increased Axin1 expression by XAV939 treatment alleviated endoplasmic reticulum (ER) stress and cell apoptosis. Cells were transfected with siRNA against Axin1 or treated with the Axin1 stabilizer, XAV939, for 16 h followed by stimulation with tunicamycin (Tm) for 8 h. ER stress markers, apoptotic proteins, and GSK3β activity were examined by western blotting. B, C Quantification of immunoblots in (A) (from 3 independent experiments; *p < 0.05, **p < 0.01, ***p < 0.001 unpaired Student’s t-test). D, E TUNEL staining showed that XAV939 treatment protected 661W cells from apoptosis (n = 8 dishes from 3 independent experiments; *p < 0.05, ***p < 0.001, unpaired Student’s t-test). F Inhibition of GSK3β activity by LiCl blocked ER stress and apoptosis in Tm-treated, Axin1-deficient cells. Axin1 knockdown or siRNA control cells were incubated with LiCl for 2 h before Tm treatment. ER stress marker, apoptotic markers, and GSK3β activity were examined by western blotting. G–M Quantification of immunoblots in (F) (from 3 independent experiments; *p < 0.05, **p < 0.01, ***p < 0.001, unpaired Student’s t-test).

Discussion

Retinal degeneration is an irreversible process for which there is no effective treatment [2]. Understanding how cell signaling molecules regulate retinal cell apoptosis may provide clues for preventing photoreceptor loss. In the present study, we demonstrate that Axin1 orchestrates the proper localization of the cone photoreceptor pigment, S-opsin, in the outer segment of mouse retinal cells, which is crucial for light perception and cone photoreceptor survival. Furthermore, Axin1 deletion accelerates ER stress and apoptosis after blue light exposure or Tm treatment. Meanwhile, stabilizing Axin1 mitigates ER stress and photoreceptor apoptosis by inhibiting GSK3β activity. These findings highlight the essential scaffolding role of Axin1 in maintaining S-opsin distribution and protecting cone photoreceptors from ER stress-induced apoptosis.

Spatial enrichment of Axin1 in cone photoreceptors

The fine-tuned distribution of distinct photoreceptors in the retina is essential because of the entry of light of varying wavelengths and brightness [5]. The major types of photoreceptors in the mouse retina are rods and cones, which have similar yet distinct properties of phototransduction. Unlike homogenously distributed rhodopsin in the retina, the cone opsins in photoreceptors exhibit dorsoventral counter-gradient distributions, with M-opsin and S-opsin being enriched in the dorsal and ventral retina, respectively [30]. The distributions of S-opsin and M-opsin determine the retinal receptive field. In particular, the high concentration of S-opsin in the ventral retina in mice generates a larger field of vision above the animal, which may help them detect predators from above [31].

However, the molecular mechanism underlying the distributions of M-opsin and S-opsin is incompletely understood. Previous studies suggest that thyroid hormone is the key molecular cue for generating the opsin gradient in the mouse retina during development by inhibiting S-opsin expression and promoting M-opsin expression [32]. Nevertheless, it remains unknown how the opsin gradient in the mouse retina is maintained and stabilized at the protein level. Here, we identified that Axin1 protein specifically accumulates in cone photoreceptors but not rod photoreceptors and that its enrichment during development closely overlaps with that of S-opsin in the outer segment of retinal cells. This indicates that Axin1 is a cone-specific regulator of the organization of S-opsin on cone photoreceptors. This is supported by the finding that Axin1 deficiency results in the loss of S-opsin in the retina. Furthermore, this observation is phenocopied in a retinal degeneration model involving deficiency of Rpe65, in which there is loss of S-opsin but no change in rhodopsin in the retina [33].

Axin1 protects photoreceptors from endoplasmic reticulum stress and apoptosis

ER stress-induced apoptosis is associated with various neurodegenerative diseases, including retinal degeneration [34]. The accumulation of misfolded proteins in the ER causes the unfolded protein response and hence ER stress, which often triggers cytotoxicity and cell death [35]. The mis-trafficking and accumulation of S-opsin in the outer segment of cone photoreceptors lead to degeneration by initiating the unfolded protein response, resulting in ER stress and subsequent apoptosis [25]. Thus, the orchestration of S-opsin in the outer segments of photoreceptors by Axin1 may be functionally crucial.

Scaffold proteins stabilize multiprotein complexes to maintain efficient signaling transduction [36]. Our data indicate that Axin1 stabilizes S-opsin at the outer segment. Loss of Axin1 causes S-opsin mislocalization and its accumulation as a punctate form (Fig. 3A) in the outer segment of cone photoreceptor. Under physiological conditions, S-opsin is constantly exposed to light stimulation; it is rapidly synthesized in the ER and efficiently transported to the outer segment to replace dysfunctional proteins. Mis-localized S-opsin in Axin1-deficient photoreceptor cells may fail to undergo the proper trafficking process, which may lead to S-opsin aggregation and trigger ER stress [25, 37]. Thus, as a scaffold protein, Axin1 is essential for maintaining the proper distribution of S-opsin. A similar molecular architecture is observed for the RGS9-1 anchor complex in rod cells, loss of which causes rhodopsin mislocalization leading to retinal degeneration in retinitis pigmentosa [38, 39].

On the other hand, Axin1 may modulate the hyperactivation of S-opsin signaling cascades. The enrichment of S-opsin in the ventral retina can facilitate the perception of short-wavelength light in rodents [40]. However, long-term exposure of mouse retinas to short-wavelength light causes cone photoreceptor degeneration due to overactivation of S-opsin and the downstream transducin, Gnat2. This triggers excessive hydrolysis of cGMP, which results in cyclic nucleotide-gated channel closure, reduced inward cation (i.e., Na⁺ and Ca²⁺) currents, and membrane hyperpolarization [41]. ER stress in photoreceptors can be evoked by cellular Ca²⁺ overload [42, 43]. As the crucial source of Ca²⁺ influx in the outer segment of photoreceptors, a lack of functional cyclic nucleotide-gated channels might affect calcium homeostasis, resulting in ER stress and cell death [7]. In contrast, S-opsin should be deactivated after light stimulation via phosphorylation by G protein-coupled receptor kinase 1 (GRK1), followed by arrestin binding. Persistent activation of transducin can also cause ER stress via the induction of the PERK signaling pathway [44]. Overall, the interplay among Axin1, S-opsin signaling, and calcium homeostasis is critical for maintaining cone photoreceptor health and preventing degeneration under prolonged light exposure.

Protective role of Axin1 in photoreceptor degeneration via inhibition of GSK3β activity

Axin1 is well known for its scaffolding role in the canonical Wnt/β-catenin signaling pathway, in which it assembles the destruction complex comprising APC, CK1, and GSK3β, thereby mediating the prolonged phosphorylation and degradation of β-catenin in the absence of Wnt activation [45]. GSK3β, a key serine/threonine kinase, is crucial for maintaining cell homeostasis and retinal photoreceptor function [45]. GSK3β dysfunction facilitates ER stress-induced apoptosis, contributing to neuronal death in retinal conditions such as retinitis pigmentosa and ischemic injury [28, 46]. Our results further support that Axin1 disruption exacerbates GSK3β-related apoptosis in photoreceptors. ER stress is a central driver of photoreceptor apoptosis in retinal degeneration [47]. In this study, we show that loss of Axin1 exacerbates ER stress upon tunicamycin treatment or blue light exposure, while LiCl or pharmacological stabilization of Axin1 attenuates apoptosis. Mechanistically, Axin1 regulates GSK3β, a kinase implicated in retinal degeneration and neuronal apoptosis [28, 46]. Disruption of the Axin1 scaffold likely leads to dysregulation of GSK3β, activation of which has been reported to exacerbate ER stress and enhance unfolded protein response signaling toward apoptosis. Thus, hyperactivated GSK3β in Axin1-deficient photoreceptors may be the key driver of ER stress and subsequent retinal degeneration.

In summary, our study highlights the critical role of Axin1 in the ER stress response in cone photoreceptors. Axin1 localization not only stabilizes S-opsin at the outer segment of cone photoreceptors but also regulates GSK3β activity to protect cone cells from light-induced ER stress and apoptosis. Hence, stabilization of Axin1 protein in retinas is a potential therapeutic avenue for alleviating retinal degeneration.

Materials and methods

Animals

Axin1^loxP/loxP conditional knockout mice were generously gifted by Professor Shengcai Lin of Xiamen University. The mice were housed at the Animal Facility of the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, at 22 °C under a 12-h light–dark cycle (lights on 7 a.m.–7 p.m.) with food and water ad libitum. Genomic DNA was extracted by ear tag or tail biopsy using a KAPA Mouse Genotyping Kit (KK7352, Kapa Biosystems) according to the manufacturer’s instructions. Genotyping was performed by PCR with the following primers: forward: 5′-ccagctgaaattgctgctgca-3′, primer: 5′-tccttcttctatgcacgaagaca-3′. Male Axin1^loxP/loxP and wildtype (WT) mice on a C57BL/6 background, aged 7–8 weeks, were utilized in the following experiments.

Immunofluorescence staining

Retinal sections were permeabilized with 0.2% Triton X-100 in PBS containing 10% donkey normal serum (017-000-121, Jackson ImmunoResearch) for 1 h at room temperature. The primary antibodies were diluted in PBS containing 5% donkey normal serum and incubated with retinal sections overnight at 4 °C. After washing with PBS, the retinal sections were incubated with corresponding secondary antibodies (diluted 1:500 in PBS) in 3% donkey normal serum for 2 h at room temperature. Tissue sections were incubated in DAPI staining buffer (15 μm, 1:5000, Sigma-Aldrich) for 5 min at room temperature and then mounted with Hydromount medium (0100-01, Southern Biotech). The primary antibodies and their dilution factors were as follows: cone arrestin (1:1,000, AB15282, Millipore), M-opsin (1:200, AB5405, Chemicon), GFAP (1:1,000, AB5804, Millipore), Axin1 (1:200, 06-1049, Millipore), S-opsin (1:200, AB5407, Millipore; 1:100, ab235274, Abcam), synaptophysin (1:200, SAB4200544, Sigma-Aldrich), c-Fos (1:500, 2250 s, Cell Signaling Technology), ZO-1 (1:200, 61-7300, Invitrogen), and rhodopsin (1:200, ab98887, Abcam). The secondary antibodies were as follows: donkey anti-rabbit Alexa Fluor 594 (A-21207, Invitrogen), donkey anti-rabbit Alexa Fluor 647 (A-31573, Invitrogen), donkey anti-goat Alexa Fluor 647 (A-21447, Invitrogen), donkey anti-rabbit Alexa Fluor 488 (A-21206, Invitrogen), and donkey anti-mouse Alexa Fluor 594 (A-21203, Invitrogen). All images were captured by confocal microscopy (Zeiss LSM880).

A Zenon Labeling Kit (Zenon Alexa Fluor 488 or 647 rabbit IgG1 Labeling Kit, Z25002/Z25308, Thermo Fisher Scientific) was used to pre-stain Axin1 or S-opsin in mouse retinas. Briefly, to form the labeling complex, 1 μg Axin1 antibody (1:200, 06-1049, Millipore) or 1 μg S-opsin (AB5407, Millipore) and 5 μL IgG conjugated with Alexa Fluor 488-labeling reagent (rabbit) in 10 μL PBS were incubated for 5 min at room temperature and then incubated with 5 μL blocking reagent for another 5 min. The labeling complex was diluted in 200 μL PBS containing 10% donkey normal serum, applied to the tissue slides, and allowed to incubate for 2 h at room temperature. After washing with PBS, the slides were subjected to cone arrestin, S-opsin, M-opsin, and ZO-1 staining, respectively.

AAV delivery into mouse retinas

AAV expressing Cre recombinase driven by S-opsin (blue pigment, BP) promoter (5.4 × 10¹² vg/mL for AAV-BP-Cre) and AAV-FLEX-tdTomato (6.26 × 10¹² vg/mL, 2 μL per eye) were injected into the vitreous body of anesthetized Axin1^loxP/loxP mice or control mice at 7–8 weeks of age. From 13 to 15 weeks after AAV injection, the virus-infected mice were subjected to pupillometry, light aversion testing, and protein staining.

Pupillometry

Mice were adapted to the dark for 12 h to maximize pupil dilation and were subsequently anesthetized. The mice were immobilized on a stereotaxic adapter and exposed to different wavelengths of light through a light guide (MN-V-200920-01, Chongqing Guote Medical Equipment) for 1 min with at least a 5-min interval between each measurement. The light device was usually aimed at one eye, while a microscope detected pupil constriction in the contralateral eye [48, 49]. An infrared LED was used for background lighting throughout the experiment, and images were captured by an infrared camera (B011, Supereyes). Pupil diameter was measured using ImageJ (version 2.0.0) before and after light exposure. Changes in pupil constriction are expressed as the difference in pupil diameter before and after light exposure.

Light–dark transition test and analysis of c-Fos activation

A light–dark transition test was conducted to evaluate the visual function of mouse retinas [50]. Briefly, the mice were adapted to a dim ambient light environment for at least 2 h and then placed in the center of a box with a dark chamber and a white chamber. The mice were allowed to explore both chambers for 10 min, and their tracks were recorded on camera. The time spent in each chamber was analyzed by ANY-maze software (version 7.4). To ensure objectivity, investigators were blinded to group assignments; those assigning the groups were separate from those assessing outcomes.

To examine c-Fos activation, the mice were adapted to dark conditions for 12 h, followed by exposure to 30-min light pulses with an LED (20 W). After light stimulation, the mice were transferred to a dark environment for 60 min. Their retinas were collected, fixed in 4% PFA for ~45 min, dehydrated in 30% sucrose for 3 h, and then embedded in optimal cutting temperature compound (Sakura) for sectioning and c-Fos staining.

Cell culture and transfection

Cone-derived 661W cells obtained from Guangzhou Genio Biotech Co., Ltd. were cultured in DMEM (Thermo Fisher Scientific) containing 10% heat-inactivated FBS as well as 1% penicillin and streptomycin in a humidified incubator with 5% CO₂ at 37 °C, and regular mycoplasma testing was performed to ensure cell line integrity. To knock down Axin1 in the 661W cell line, the cells were transfected with 20 μM siRNA against the mouse Axin1 gene using Lipofectamine RNAiMAX reagent according to the manufacturer’s instructions (Thermo Fisher Scientific). The following oligonucleotide for siRNAs was synthesized by Gene Pharm (Shanghai): 5′-GGCAUUGUGUCCAGACAATT-3′.

Western blot analysis

The 661W cells were washed once with cold PBS, collected in 150 μL RIPA lysis buffer (50 mM Tris–HCl [pH 7.4], 0.15 M NaCl, 2% NP-40, and 0.5% sodium deoxycholate containing protease inhibitors) with a cell scraper, and incubated on ice for 30 min. The cell lysate was centrifuged at 14,000 × g for 15 min at 4 °C. The resultant supernatant was subjected to SDS–PAGE and immunoblotted with the following antibodies: Axin1 (1:1000, 2087S, Cell Signaling Technology), Bax (1:1000, 2772S, Cell Signaling Technology), caspase-3 (1:5000, 9662s, Cell Signaling Technology), cleaved caspase-3 (1:1000, 9661T, Cell Signaling Technology), Bcl-2 (1:500, 3498S, Cell Signaling Technology), CHOP (1:1000, 2895s, Cell Signaling Technology), GAPDH (1:10,000, G8795, Sigma-Aldrich), β-catenin (1:1000, 610154, BD Transduction Laboratories), GSK3β (1:1000, 610202, BD Transduction Laboratories), and phospho-GSK3β (i.e., serine 9) (1:1000, 5558s, Cell Signaling Technology).

Drug treatment

For the dose-dependent experiment, 661W cells seeded at 1.5 × 10⁶ cells per 60-mm dish were incubated with 0–2 μg/mL Tm for 24 h. In the time-course experiment, the cells were treated with 5 μg/mL Tm for 0, 4, 8, or 12 h. To stabilize Axin1, the cells were incubated with 20 μM XAV939 for 8 h before Tm treatment. To inhibit GSK3β activity, the cells were incubated with 10 mM LiCl for 2 h before Tm treatment.

TUNEL staining

TUNEL staining was performed using a DeadEnd Fluorometric TUNEL System (Promega) according to the manufacturer’s instructions. Briefly, 661W cells were fixed in 4% PFA for 15 min at 4 °C and then permeabilized in 0.2% Triton X-100 solution in PBS for 5 min. The cells were incubated with equilibration buffer for 5 min at room temperature, followed by terminal deoxynucleotide transferase in a humidified chamber for 60 min at 37 °C. The cells were then immersed in 2× SSC buffer for 15 min at room temperature. After washing with PBS 3 times to remove unincorporated fluorescein-12-dUTP, the cells were mounted with Fluoromount-G (0100-01, Southern Biotech). Cells positive for TUNEL staining were considered apoptotic cells.

Blue light exposure

Six customized blue LED lamps (470 nm, 100 lx) were placed in a cell incubator. The wire lengths were adjusted to keep the distance between the lamps and the cells at ~5 cm. Each individual lamp was equipped with a conical lampshade that effectively covered a 60 mm diameter cell dish. The lampshades could be moved up and down to facilitate placement of the cell dishes and prevent cross-illumination. For light-induced photoreceptor injury, cells were seeded at 1.2 × 10⁶ per 60 mm dish. After transfection with siRNA, the cells were placed under the lampshades, exposed to blue light for 2–6 h, and then collected for subsequent TUNEL staining, qPCR, and western blot analysis.

Real-time PCR

Total RNA was extracted from 661W cells using TRIzol Reagent (Invitrogen Life Technologies) according to the manufacturer’s instructions. Eluted RNA (1 μg) was reverse-transcribed with a ReverAid First Strand CDNA Synthesis (K1622, Molecular Biology). The primers were assessed in 0.5 μL cDNA with Power SYBR Green PCR Master Mix (4367659, Applied Biosystems) on a Quant Studio Q5 Real-time PCR System according to the manufacturer’s recommendations. The AXIN1 primers were as follows: forward, 5’-ggaatccccccatacaggat-3’; reverse, 5’-tagaggtacccgcccattga-3’. The OPN1SW primers were as follows: forward, 5’-gtcgccatgtttgtgctctgga-3′; reverse, 5’-gcttggagttgaagcggatgct-3’. The real-time PCR primers for XBP1s, ATF4, CHOP, ATF6, and GAPDH have been reported previously [51].

Statistical analysis

All data are presented as mean ± SEM. Normality of the data distribution and homogeneity of variance were assessed, and statistical significance was determined using Student’s t-test or one-way ANOVA, where appropriate, using GraphPad Prism 8. The level of significance was set at p < 0.05 (2-tailed). For animal studies, power analysis was conducted to ensure sufficient statistical power. Samples were excluded only in cases of contamination, technical failure, or when identified as statistical outliers or measurement errors.

Author contributions

YC and Y-WC conceived of the project and designed the experiments. J-YX, J-HM, and Y-YF performed the experiments. YC, Y-WC, and J-YX drafted the manuscript.

Data availability

The datasets generated and/or analyzed during this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval

All animal experiments were approved by the Animal Care Committee of the Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences (SIAT-IACUC-230206-NS-LCY-A2248). This study did not involve human participants, human tissues, or clinical data. Therefore, informed consent and consent for publication were not applicable.

Supplementary information

The online version contains supplementary material available at 10.1038/s41420-026-02968-5.