VIRMA modulates function of photoreceptor cells through m⁶A modification and alternative splicing

Two morphologically distinct types of photoreceptors in the retina, rod cells and cone cells, are central to visual perception, converting light stimuli into electrical signals that are transmitted to the brain for further processing (1). Inherited retinal dystrophies (IRDs) are a group of retina disorders characterized by photoreceptor death, leading to progressive vision loss (2–4). Retinitis pigmentosa (RP) is the most common IRD, characterized by primary degeneration of rod photoreceptors followed by the loss of cone photoreceptors, with a prevalence of approximately 1 in 4,000 individuals (5, 6). The currently identified RP-causing genes account for only about 70% of cases (7), suggesting that there are yet unknown genetic factors that can cause RP. This highlights the need for in-depth exploration of other regulator layers, including gene regulation and posttranscriptional modifications, in the pathogenesis of RP.

Epigenetic regulation has been identified to play a crucial role in the regulation of numerous physiological and pathophysiological processes (8, 9). N⁶-methyladenosine (m⁶A) is the most prevalent and conserved posttranscriptional epigenetic modification in eukaryotic RNAs, especially within higher eukaryotic RNAs (8, 10, 11). The dynamic m⁶A modification is installed by a methyltransferase complex (known as writers), including METTL3, METTL14, WTAP, HAKAI, VIRMA, METTL16, RBM15, and ZC3H13, and removed by demethylases (known as erasers), such as FTO and ALKBH5 (12–20). Specific RNA-binding proteins (known as readers), including YTHDF1/2/3, YTHDC1/2, IGF2BP1/2/3, and HNRNPC/A2B1, can selectively recognize the m⁶A modified mRNAs and, consequently, affect RNA export, stability, splicing, and translation (21–26). Genetic knockout and gene knockdown studies on m⁶A writers, erasers, and readers have shown that m⁶A plays important roles in regulating various aspects of organism development, physiology, and disease progression (27–29). In recent years, the role of m⁶A methylation in the development and homeostasis of retina has been well-established. For instance, Mettl3 deficiency in retinal progenitor cells distorts late-stage retinogenesis (29). Conditional knockout of Mettl14 in retinal photoreceptors causes impaired visual function and photoreceptor death (30). In addition, YTHDC1 directly binds to the lncRNA IPW (imprinted in Prader–Willi syndrome) to regulate retinal pigment epithelium (RPE) apical–basal polarization in patients with AMD (31). VIRMA is the largest known protein in the methyltransferase complex, and it is responsible for recruiting the catalytic core components METTL3/METTL14/WTAP to guide region-selective m⁶A methylation (32, 33). This underscores the pivotal role of VIRMA in the assembly of the m⁶A writer complex. In addition, VIRMA depletion led to the most obvious reduction in m⁶A levels compared with METTL3/14/WTAP knockdown cells, suggesting the central role of VIRMA in m⁶A modification (12, 34–36). Recently, VIRMA has been proven to play a key role in central nervous system diseases (37, 38). However, whether VIRMA-mediated m⁶A methylation, like other m⁶A modification components, also plays a crucial role in maintaining neuronal retina function remains unclear.

In addition, VIRMA has been reported to be involved in hepatocellular carcinoma, breast cancer, and mouse follicular development by regulating alternative splicing (AS) events (39–41). AS is a common mechanism for tissue-specific and development-specific regulation of gene expression in eukaryotes (42–44). Particularly high levels of AS are found in neuronal tissues, as these tissues maintain delicate cell subcellular diversity (45). Retina development relies on finely regulated cell type–specific gene expression, and AS plays a crucial role in contributing to this intricate development process (45, 46). It is no surprise that aberrant AS can impact retinal cell survival and function, resulting numerous IRDs, including RP, cone-rod dystrophy, and Usher syndrome (47). Whether VIRMA also plays a role in modulating AS events in photoreceptor remain unclear.

Here, we investigated the function of VIRMA in photoreceptors by generating retina-specific Virma-knockout mice using the RHO-Cre and HRGP-Cre lines. Our study found that loss of Virma in rods led to impaired visual function and degeneration of rods. Analogously, deficiency of Virma in cones caused death of cone cells. Further investigation revealed that Virma deficiency resulted in reduced global retina m⁶A methylation levels. Mechanistically, a comprehensive multi-omics analysis illustrated that decreased m⁶A methylation level via inactivation of VIRMA resulted in reduced expression level of multiple visual perception–associated genes. VIRMA also mediates AS events in photoreceptors, contributing to photoreceptor degeneration in RKO mice. Reintroduction of functional Virma expression in rods successfully rescued the photoreceptor degeneration. Therefore, our study revealed important mechanisms by which VIRMA regulates photoreceptor function and survival, highlighting key roles of VIRMA-mediated m⁶A modification and splicing in the retina.

VIRMA deficiency in rod cells causes impaired visual function. m⁶A modification has been reported to be involved in maintaining retinal homeostasis in several mouse models, as described in the Introduction. Although previous study illustrated that conditional knockout of the Mettl3/14 or Ythdc2 impaired photoreceptor function and survival in mice (30, 48, 49), whether other components involved in m⁶A modification could modulate photoreceptor function and survival remains unclear. To address this question, we focused on VIRMA, a structural component of the m⁶A methyltransferase complex, as its depletion leads to the biggest downregulation of m⁶A modification compared with other key members such as METTL3, METTL14, and WTAP (12). Immunofluorescence staining demonstrated that VIRMA was broadly expressed in nucleus of outer nuclear layer (ONL), inner nuclear layer, and ganglion cell layer in retina labeled with anti-VIRMA antibody, suggesting a potential functional role of VIRMA in retina (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.197880DS1). To determine the function of VIRMA in photoreceptor cells, we crossed Virma-floxed mice with RHO-Cre (50) and HRGP-Cre transgenic mice (51), which express Cre specifically in the rod and cone cells, respectively (Supplemental Figure 1A). Genotyping confirmed the deletion of exon 2 in homozygotes, and the littermates of the progeny were distinguished as follows: Virma^loxp/loxp; RHO-Cre (hereafter called RKO), Virma^loxp/loxp; HRGP-Cre (hereafter called HKO), and Virma^loxp/loxp (used as control [Ctrl]) mice (Supplemental Figure 1B). To confirm the cell type–specific expression of our Cre lines, we introduced ROSA26-tdTomato reporter mice to visualize the specific expression of RHO-Cre and HRGP-Cre in retina rod and cone photoreceptor cells, respectively. As expected, tdTomato fluorescence was detected in the ONL of RHO-Cre mice (Supplemental Figure 1, C and D). Given that rod cells constitute 97% in murine photoreceptors (52), the widespread tdTomato signal in the ONL of RHO-Cre mice supports its specificity for rod cells (Supplemental Figure 1D). As revealed by Western blot, RT-qPCR, and immunofluorescence staining, the expression level of Virma was drastically reduced in the retina, confirming the successful generation of Virma photoreceptor-specific knockout mice (Supplemental Figure 1, E and F). In HRGP-Cre mice, tdTomato fluorescence was also observed in the ONL (Supplemental Figure 1G). High-magnification costaining with the cone-specific marker cone arrestin (cArr) revealed precise colocalization of the tdTomato signal within cArr-positive cells, suggesting the precise excision.

To assess the visual function in Virma RKO mice, scotopic electroretinograms (ERGs) were performed on Ctrl and RKO mice at 4 weeks of age. Scotopic ERG responses in the RKO retinas exhibited diminished amplitudes for both a-waves and b-waves when compared with those of age-matched controls at each flash intensity (Figure 1, A and B). The average amplitudes of the a- and b-waves of RKO mice were reduced by approximately 69% and 62%, respectively. In contrast, photopic ERG traces, reflecting cone photoreceptor function, appeared normal in RKO mice at this time point (Figure 1, C and D). These results indicate that the rod photoreceptor function was significantly compromised by the loss of Virma. We then carried out behavioral assessments for visual function via the light-dark transition test and optomotor response test. The preferred movements in dark and light were intuitively presented in traveling trajectories (Figure 1E), and heatmaps documented the time spent at in different regions of the light-dark box (Figure 1F). RKO mice showed an apparent preference for staying in the dark chamber for over 55% of time at 2,000 lux luminance and traveled around 50% of distance in the dark, both of which were lower than either of those in Ctrl mice (Figure 1G). Meanwhile, the minimal duration time in dark was significant decreased, while the transition time increased (Figure 1G). In the C57BL/6J mouse strain, the ratio of the number of times of head movement responded in the correct and incorrect directions, namely optomotor reflex (OMR), is most significantly observed at a spatial frequency of 0.2 cycles/° (53). Therefore, a spatial frequency of 0.2 cycles/° was chosen to evaluate the effect of Virma deficiency on visual function in mice as previously described (53). The overall OMR values in Ctrl context exhibited equal or higher than 2.0, while in RKO mice, the OMR value was apparently reduced to 1.5 (Figure 1, H–J). These results indicate that the m⁶A writer VIRMA positively regulates rod photoreceptor function in vivo and Virma deletion compromise photoreceptor function.

Figure 1

Impaired visual function in rod-specific Virma-depleted mice. (A) Scotopic ERGs were recorded with increasing light intensities from dark-adapted Ctrl and RKO mice at 4 weeks of age. (B) Statistical analysis for the amplitudes of the a-wave and b-wave under scotopic conditions (Student’s t test, n = 3). (C) Photopic ERGs were recorded from Ctrl and RKO mice at 4 weeks of age. (D) Statistical analysis for the amplitudes of the a-wave and b-wave under photopic conditions (Student’s t test, n = 3). (E and F) The representative traveling trajectories (E) and heatmap recordings for time spent in distinct regions (F) of the light-dark box at 2,000 lux luminance. (G) Statistical analysis based on behavioral parameters, including time, distance traveled, and duration as well as transition between the chambers (Student’s t test, n = 6). (H) Representative heatmaps of the visual stimuli–driven optomotor responses at 0.2 cycles/° spatial frequency. (I) Representative images of OMR values. The response time in either stimulus direction (positive value, light green window) or opposite direction (negative value, light magenta window) at a specific velocity threshold was normalized to the maximal response time, set as 1, and presented as a blue bar. (J) Statistical analysis of optomotor response (Student’s t test, n = 6). Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. A lack of significant difference is indicated by #.

Rod cell degeneration in RKO mice. To confirm morphological changes in mice, histological analysis was performed on the retinas of Ctrl and RKO mice at different ages. At 3 weeks of age, the cellular layers and retinal thickness in RKO mice were well maintained (Figure 2, A and B). In contrast, RKO mice exhibited gradual reduction in the outer segments (OS), inner segments (IS), and ONL of photoreceptors at 4 weeks, with these changes becoming more evident with age as a consequence of the Virma deletion (Figure 2, A and B). Intriguingly, the retinas of Virma RKO mice exhibited much more rapid degeneration compared with Mettl3/14 RKO mice. The pathological phenotypes in the retinas of Mettl3/14 RKO mice first appeared at 3 months and 3.5 months of age, respectively (Supplemental Figure 2), highlighting the key role of VIRMA in retina function.

Figure 2

Virma deficiency leads to progressive rod cell loss in the retina. (A) H&E staining of paraffin sections of Ctrl and RKO retinas at the ages of 3 weeks, 4 weeks, and 8 weeks. Scale bars: 50 μm. (B) Quantitative analysis of the ONL thickness of the Ctrl and RKO retinas at defined ages (n = 3). Analyzed by multiple 2-tailed t tests with the Holm-Šídák method to correct for multiple comparisons. (C–F) Retinal cryosections from 4-week-old mice were stained with OS markers rhodopsin, PRPH2, GRK1, PDE6B, and CNGA1 and IS marker NaK ATPase. Scale bars: 25 μm. (G) Representative transmission electron microscopy images of photoreceptor outer segments in 4-week-old Ctrl and RKO mice. Red dashed line indicates the boundary between the RPE layer and the OS. Red arrowheads indicate the disorder of OS disk. RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. A lack of significant difference is indicated by #.

The OS house a large amount of phototransduction proteins, all of which are synthesized in the IS and transported to the OS via the photoreceptor cilium. To elucidate the effect of Virma deficiency on the function of photoreceptor cells, we conducted immunofluorescence staining to assess the subcellular localization and expression of multiple key OS proteins in the RKO retina, including rhodopsin, PRPH2, GRK1, PDE6B, and CNGA1. In RKO retinas, these proteins retained their normal OS localization, but the OS appeared markedly shortened (Figure 2, C–F). This morphological phenotype may result from decreased protein synthesis or defects in cilium transport. To investigate this, we evaluated the morphology and structural changes of cilia using immunostaining and transmission electron microscopy (TEM). Cryosections of Ctrl and RKO animals were stained with centriole marker CEP164 and the ciliary marker acetylated α-tubulin. The results illustrated that, similar to those in Ctrl mice, the photoreceptor cilia in RKO mice appeared slender in their arrangement without noticeable length differences (Supplemental Figure 3, A–C). TEM results further corroborated these findings, showing that cilia in RKO mice also exhibited a typical 9+0 structure (Supplemental Figure 3D). As evidenced by TEM, a palisade pattern formed by the OS was observed on the RPE, displaying highly ordered OS disc in the 4-week-old Ctrl mice. However, age-matched RKO mice showed disorganized OS discs in photoreceptors cells, indicating the damage of the OS (Figure 2G).

In addition, the activation of resident retinal glial and immune cells is a key event in the progression of photoreceptor degeneration (54–56). We thus stained retinal sections from Ctrl and RKO mice for activated Müller and microglia markers. The abnormal migratory behavior stained with glial fibrillary acidic protein (GFAP) provides obvious evidence for Müller cell proliferation in RKO mice in response to retinal injury or stress (Supplemental Figure 4A). The ionized calcium-binding adapter molecule 1–immunoreactive (IBA1-immunoreactive) microglial cells had features typical of the resting state in Ctrl mice. In contrast, the IBA1-immunoreactive cells in RKO mice transformed into an amoeboid shape and infiltrated the ONL and both the inner and outer segments of photoreceptors (Supplemental Figure 4B). Western blot analysis further confirmed these results, as revealed by increased GFAP and IBA1 protein level in the RKO retinas (Supplemental Figure 4C). TUNEL staining suggested that gradual loss of photoreceptors in RKO mice primarily occurred via apoptosis (Supplemental Figure 4D).

Owing to the loss of trophic factors and structural support provided by neighboring rod photoreceptor cells, the degeneration of rods frequently leads to secondary morphological and physiological changes of cones, ultimately resulting in cell death (5, 57). While rods die first at 4 weeks, with ONL shrinking over time, cone structure begins to deteriorate at 8 weeks of age in RKO mice (Supplemental Figure 5, A–G). From 8 to 15 weeks of age in RKO mice, cones exhibited pronounced morphological alterations, mislocalization of cone OS proteins, and gradual cell death as a consequence of the progressive rod cell loss (Supplemental Figure 5, E–I). These results indicated that specific deletion of VIRMA in retina rod photoreceptor cells contributed to progressive photoreceptor cell loss.

Cone cell defects in HKO mice. Rod cells detect light intensity, contributing to night and peripheral vision, while cone cells perceive color, supporting daylight and central vision (58). To further examine the role of VIRMA in photoreceptor function, we specifically knocked out Virma in cone cells using HRGP-Cre. Photopic ERG showed decreased a- and b-wave amplitudes in 8-week-old HKO mice compared with controls (Figure 3A). At this time point, scotopic a-wave amplitude was unaffected (Figure 3B). At 8 weeks of age, HKO mice showed marked and clear changes in cones morphology, along with a slight decrease in cone density (Figure 3, C–F). In the both dorsal and ventral retinas of HKO mice, the M-opsin–positive cones exhibited a significant decrease in number, accompanied by morphological disorganization and misshapen appearance (Figure 3, C–E). In contrast, the number of S-opsin–positive cones showed no significant change, with only minor morphological distortions (Supplemental Figure 6). Subsequently, starting from week 16, a sharp decline in the number of cones was observed in HKO mice, which became more evident with increasing age (Supplemental Figure 7). Degeneration of cones also triggered glial cell activation in the retina (Figure 3G). Misshaped and reduced cone cells observed in HKO mice suggested that Virma depletion resulted in the degeneration of cone cells.

Figure 3

Virma depletion in cone cells leads to cone defects in the retina. (A) Representative photopic ERG traces and statistical analysis at identified flash intensities in mice at 8 weeks of age (Student’s t test, n = 3). (B) Representative scotopic ERG traces and statistical analysis at identified flash intensities in mice at 8 weeks of age (Student’s t test, n = 3). (C) Immunostaining of retina flat mount from 8-week-old Ctrl and HKO mice with M-opsin and PNA (red). Scale bars: 50 μm. Representative images from the dorsal and ventral retinal quadrant are shown. Scale bars: 20 μm. Inset images showed a cropped and zoomed image. Scale bars: 5 μm. White arrowheads indicate the lost and misshaped M-opsin–positive cones. (D) Retinal cryosections from 8-week-old Ctrl and HKO mice were labeled with M-opsin. Scale bars: 20 μm. (E) Quantification of the number of M-opsin–positive cones at both dorsal and ventral side of the retina per field (Student’s t test, n = 5). (F) Representative immunofluorescence images of cArr (green) of retina sections from 8-week-old Ctrl and HKO mice, and the number of cArr-positive cones per 500 μm field was quantified. Scale bars: 25 μm. Analyzed by multiple 2-tailed t tests with the Holm-Šídák method to correct for multiple comparisons. (G) Retinal sections from 8-week-old mice were stained with GFAP (green). Scale bars: 20 μm. Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. A lack of significant difference is indicated by #.

VIRMA depletion decreases the abundance of m⁶A methylation in the RKO retina. VIRMA serves as an interaction platform for the assembly of MACOM complex (including VIRMA, WTAP, HAKAI, RBM15, and ZC3H13), which is required for mRNA m⁶A methylation (18). We thus assessed the global profiles of m⁶A methylation levels in retina via dot blot assay, and the result revealed a 50% reduction of overall m⁶A levels in the RKO mice compared with the Ctrl mice (Figure 4A). Immunofluorescence staining further confirmed the aberrant downregulation of m⁶A levels in ONL of RKO mice (Figure 4B). We then explored whether the loss of VIRMA affects the expression and subcellular distribution of other MACOM members. To this end, we detect the protein levels of these members, and the data showed that only WTAP expression was downregulated upon VIRMA deletion, while the expression of other members remained unaffected, which consistent with findings from a previous report (32) (Figure 4C).

Figure 4

Loss of Virma led to reduction of global m⁶A modification abundance in the retina. (A) m⁶A dot blot assay of global m⁶A abundance in retinas from 4-week-old Ctrl and RKO mice using 250 ng or 500 ng mRNAs. Methylene blue staining was used as a loading control (Student’s t test, n = 4). (B) Immunofluorescence staining of m⁶A (red) of retina cryosections from 4-week-old mice. Scale bars: 20 μm; 5 μm (higher-magnification images). (C) Western blot and quantification of expression of m⁶A writers in 4-week-old mouse retinas (Student’s t test, n = 3). (D) Volcano plots of the significantly differentially expressed genes between Ctrl and RKO (fold change ≥1.2 and P < 0.05). (E) Pathway analysis in downregulated genes after Virma deletion. (F) Top consensus m⁶A motif identified by HOMER in Ctrl and RKO retinas. (G) Number of m⁶A peaks identified from MeRIP-seq in Ctrl and RKO retinas. (H) Number of m⁶A-modified genes identified from MeRIP-seq in Ctrl and RKO retinas. (I) Graphs illustrating the proportion distribution of the m⁶A peaks in mRNA transcripts from Ctrl and RKO retinas. (J) Distribution of m⁶A peaks across 5′ UTR, CDS, and 3′ UTR of mRNA in Ctrl and RKO retinas. CDS, code sequence. (K) A Venn diagram showing the shared genes with reduced m⁶A modification levels and mRNA expression levels, along with enriched GO terms for these 71 selected genes.

Previous studies suggest that m⁶A methyltransferases are predominately localized in the nuclear speckles, where m⁶A methylation occurs (15–17, 34). MACOM complex showed no obvious change in cellular localization upon VIRMA depletion (Supplemental Figure 8). These results suggest that VIRMA is not required for the nuclear localization of the MACOM complex, except for its effect on WTAP stabilization within the nucleus.

To investigate whether the functional defects in photoreceptor cells of RKO mice were attributable to decreased m⁶A modification and to elucidate the underlying molecular mechanisms, we comprehensively analyzed the retinal mRNA m⁶A modification landscape via methylated RNA immunoprecipitation–qPCR (MeRIP-qPCR) and the corresponding gene expression profiles via RNA-seq. The data revealed 600 downregulated genes and 1,417 upregulated genes with significant differences (fold change >1.2, P < 0.05) in transcript levels between Ctrl and RKO mice, and these genes were primarily enriched in the phototransduction signaling pathway (Figure 4, D and E). The m⁶A “GGACU” consensus motif was highly enriched in both Ctrl and RKO groups using the HOMER program (59, 60) (Figure 4F). In total, MeRIP-seq identified 26,324 m⁶A peaks from 3,601 m⁶A-modified transcripts in Ctrl retinas and 25,193 m⁶A peaks from 3,546 m⁶A-modified transcripts in RKO retinas. Following Virma deletion in rods, 539 genes gained m⁶A modification , 582 genes lost m⁶A modification, and 3,007 genes were common to both groups (Figure 4, G and H). Peak distribution analysis showed a similar pattern, with m⁶A sites enriched in both CDS and 3′ UTRs. The highest enrichment of m⁶A residues was located near the stop codon and 3′ UTRs (Figure 4, I and J), which is consistent with results from previous studies (32, 41). As is typical, less enrichment of m⁶A peaks appeared at 3′ UTR and near the stop codon in RKO retinas (Figure 4J). By intersecting the 600 downregulated genes identified from RNA-seq and 418 m⁶A-downregulated genes identified from MeRIP-seq, 71 candidate genes were selected (Figure 4K). Gene ontology (GO) analysis of these genes yielded the enrichment signaling pathways of biological process, and notable gene sets include visual perception and photoreceptor cell maintenance (Figure 4K). Therefore, we focused primarily on the alterations of these two signaling cascades in the subsequent investigation.

Virma regulates the expression of visual perception–related genes through m⁶A modification. Heatmaps of the visual perception–related genes showed significant downregulation at the mRNA level (Figure 5A), which was further validated by RT-qPCR showing reduced expression of Gnat1, Guca1b, Prph2, Pde6b, Pde6g, Rgs9bp, Rs1, Prph2, Slc24a1, Rho, Mdm1, Cc2d2a, and Rgs9 in the retinas of RKO mice (Figure 5B). These genes also exhibited significant loss of m⁶A peak signals in response to Virma knockout, as confirmed by MeRIP-qPCR (Figure 5, C and D). Compared with the previously reported Mett13/14 RKO mice, a greater number of visual perception–related genes were downregulated in the retinas of Virma RKO mice (30, 49). To further explore the mechanism underlying VIRMA’s regulation of photoreceptor function, we used a proteomic approach on retinas from Ctrl and RKO mice. GO enrichment analysis identified substantially altered categories of proteins following Virma deletion. The differently expressed proteins were mainly involved in retina homeostasis, visual perception, and protein deglutamylation (Figure 5E), which was closely aligned with the RNA-seq data. Western blot analysis also revealed a reduction in RHO, GNAT1, PDE6B, PRPH2, and RDH12 (Figure 5F). To determine whether m⁶A hypomethylation directly caused the downregulation of visual perception–related genes, we analyzed global m⁶A levels at postnatal week 3, prior to the onset of photoreceptor degeneration. Dot blot analysis revealed a significant reduction in overall m⁶A levels in the RKO mice compared with the Ctrl mice (Supplemental Figure 9A). At this early stage, H&E staining and immunostaining for key photoreceptor function or structural proteins showed no obvious difference between RKO and Ctrl retina sections (Figure 2A and Supplemental Figure 9, B–D). GFAP and TUNEL staining also showed no cellular stress or apoptosis in RKO photoreceptors (Supplemental Figure 9, E and F). Furthermore, RT-qPCR analysis revealed no significant difference in the expression of visual perception–related genes between RKO and Ctrl mice (Supplemental Figure 9G). Thus, m⁶A methylation reduction precedes both transcriptional decline and phenotypic changes, confirming its primary role in disease initiation. Collectively, these findings shed light on the epigenetic underpinnings that drive the progress of photoreceptor degeneration, with a marked decrease in m⁶A methylation level of visual perception–associated key molecules.

Figure 5

Virma regulated visual perception–associated gene expression via m⁶A modification in rods. (A) Heatmap of several visual perception–related genes identified in Figure 4. (B) RT-qPCR verified the mRNA expression for the indicated genes (Student’s t test, n = 3). (C) Integrative Genomics Viewer tracks of MeRIP-seq reads along indicated mRNAs. Normalized reads density levels are shown as blue (input), red (Ctrl), and green (RKO) shades, respectively. Three replicates are shown. (D) MeRIP-qPCR assay indicating the m⁶A modification levels for the indicated genes (Student’s t test, n = 3). (E) Chord plot of differentially expressed proteins in the RKO group compared with Ctrl groups. (F) Western blot and quantification of visual perception–related protein expression in retinas from 4-week-old mice. β-Actin served as control. Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Virma modulates the AS of Imphd1 and Polg2. VIRMA, like other known nuclear enzymes related to m⁶A modification, is localized to the nuclear speckles that enrich pre-mRNA processing factors (12, 16, 17, 34, 61) (Supplemental Figure 8). SRSF3, a member of serine/arginine rich protein family, is involved in regulating AS (62). It has been reported that the m⁶A-binding protein YTHDC1 can recruit SRSF3 to modulate the AS of target genes (63). Immunofluorescence showed that VIRMA colocalized with SRSF3 in the nuclear speckles, while coimmunoprecipitation confirmed the interaction between VIRMA and SRSF3 (Figure 6A), suggesting that VIRMA may also play a role in regulating AS. To confirm this, high-throughput transcriptomic was applied to measure the AS events in mRNAs potentially affected by VIRMA depletion in rods. A total of 452 significant AS events were identified (P < 0.05) in 205 genes using SUPPA2 (64), including 109 in alternative 5’ splice sites (A5SS), 108 in alternative 3’ splice sites (A3SS), 89 in skipped exons (SE), 69 in alternative first exon (AF), 62 in retained introns (RI), 9 in alternative last exon (AL), and 6 in mutually exclusive exons (MXE) under VIRMA deficiency (Figure 6B). Among them, 29 events were directly related to retina functions, and 19 gene were most attractive, as they have been shown to regulate nucleotides metabolism, visual phototransduction, ciliogenesis, mitochondrial dysfunction, and ubiquitination modification in retina (Supplemental Table 1). The results revealed that significant AS events in these genes, in response to VIRMA deletion, tended to promote exon skipping and intron retention (Supplemental Table 1). We next validated the reliability of high-throughput transcriptomic data by RT-PCR analysis of key retina function–related genes in Ctrl and RKO retinas: Impdh1 (associated with purine metabolism), Polg2 (mitochondrial DNA [mtDNA] replication related), and Actg1 (engaged in synaptic function). Compare with controls, RKO mice exhibited increased skipping of exon 2 in Impdh1 (Figure 6, C–E) and increased retention of intron 2 in Polg2 (Figure 6, F–H). Retinal inosine monophosphate dehydrogenase 1(IMPDH1) encodes enzyme IMP dehydrogenase type 1, an enzyme that catalyzes the rate-limiting step of de novo guanine synthesis by converting inosine monophosphate (IMP) to xanthosine monophosphate (XMP) with the reduction of NAD (65). Mutations in IMPDH1 were previously reported in patients with autosomal dominant RP 10 (66, 67), highlighting the critical role of this enzyme in maintaining photoreceptor function.

Figure 6

Virma is involved in alternative splicing of transcripts in retina. (A) Colocalization and coimmunoprecipitation analysis of VIRMA and SRSF3. Scale bar: 10 μm. (B) Seven types of alternative splicing events were identified in RKO retains. A3, alternative 3′ splice site; A5, alternative 5′ splice site; MX, mutually exclusive exon; SE, skipped exon; RI, retained intron; AF, alternative first exon; AL, alternative last exon. (C) Sashimi plot of Impdh1 gene in Ctrl and RKO retinas. (D and E) RT-PCR validation of exon 2 skipping of Impdh1 gene (Student’s t test, n = 3). (F) Wiggle plots showing intron retention of Polg2. (G and H) RT-PCR validation of intron retention of Polg2 gene in RKO retinas (Student’s t test, n = 3). (I) Western blot and RT-qPCR analysis of POLG2 expression in Ctrl and RKO mice (Student’s t test, n = 3). (J) HSP60 (a mitochondrial marker) immunofluorescence staining of shCtrl and shVIRMA 293STF cells. Scale bars: 20 μm; 5 μm (higher-magnification images). (K) ATP levels in fresh retains from 4-week-old Ctrl and RKO mice were measured using luciferase assay (Student’s t test, n = 5). (L) RT-qPCR analysis of the expression of mitochondrial function–related genes (Student’s t test, n = 4). Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Intron retention level is negatively correlated with the steady-state expression of genes (68, 69), and RT-qPCR and Western blot further confirmed reduced expression of Polg2 at both the mRNA and protein levels, respectively (Figure 6I). Polg2, encoding the α subunit of mtDNA polγ, has been identified to be associated with progressive external ophthalmoplegia accompanied by vision and hearing loss (70). Mitochondrial dysfunction is implicated in multiple common pathologies, including neurodegeneration, metabolic syndrome, and cancer (71–73). mtDNA plays a crucial role in mitochondria function, and any alterations in mtDNA replication can effect the normal function of mitochondria. Therefore, we first assessed the status of mitochondrial in the absence of VIRMA by immunostaining with heat shock protein (HSP60, a mitochondrial chaperonin) and found abnormal mitochondrial morphology following VIRMA deletion (Figure 6J). We also found that Virma loss decreased ATP production in RKO retinas (Figure 6K). Bcl2a, an antiapoptotic factor in the mitochondrial death pathway, showed decreased expression in RKO retina. Additionally, the mitochondrial detoxification enzymes Sod1 and Sod2 were downregulated in RKO retina as well (Figure 6L). These data suggested the presence of mitochondrial damage in RKO mice. In summary, the m⁶A methyltransferase VIRMA participates in posttranscriptional regulation by regulating AS in rod cells.

Srsf3 depletion in retina rods causes impaired visual function. Retinal tissue exhibits one of the highest levels of AS, and AS makes a major contribution to gene expression during retinal development and homeostasis. Mutations in general splicing factors have been reported to be associated with RP in humans (74–76). Previous studies on AS focused on the spatial and temporal patterns of protein isoforms during retina development (77). Here, we generated rod cell–specific Srsf3 knockout mice (Srsf3-RKO) to study the AS dysregulation in photoreceptor function. We first confirmed a greater reduction in SRSF3 expression in the retinas of Srsf3-RKO mice compared with Ctrl mice using immunofluorescence staining (Figure 7A). Scotopic ERG recordings revealed reduced amplitudes of both a- and b-waves in Srsf3-RKO mice compared with controls at each flash intensity at 3 weeks of age (Figure 7B). H&E staining showed a shorter photoreceptor segment in the Srsf3-RKO mice, with obvious photoreceptor loss at 3 weeks of age. However, Srsf3-RKO mice underwent rapid pathological changes: by 5 weeks of age, the ONL was reduced to approximately 50%, and by 8 weeks of age, the ONL was nearly completely lost (Figure 7C). Immunofluorescence staining (Figure 7, D–G) and Western blot (Figure 7H) further revealed reduced expression of OS structural and functional proteins in Srsf3-RKO retinas at 3 weeks of age, including rhodopsin, PDE6B, CNGA1, PRPH2, and GRK1. These data highlight the importance of a stable splicing process in maintaining retinal photoreceptor function.

Figure 7

SRSF3-mediated alternative splicing is crucial for retina function. (A) Immunofluorescence staining of SRSF3 in Ctrl and Srsf3-RKO mice. Scale bars: 20 μm; 5 μm (higher-magnification images). (B) Representative scotopic ERG traces and statistical analysis at identified flash intensities in mice at 3 weeks of age (Student’s t test, n = 3). (C) H&E staining was performed on paraffin sections from Srsf3-RKO and Ctrl mice at 4, 5, and 8 weeks. The thickness of the photoreceptor segments (PRS) and ONL was quantified at defined ages (n = 3). Scale bar: 25 μm. Analyzed by multiple 2-tailed t tests with the Holm-Šídák method to correct for multiple comparisons. (D–G) Retinal cryosections from 3-week-old mice were stained with OS markers rhodopsin, PRPH2, GRK1, PDE6B, and CNGA1 and IS marker NaK ATPase. Scale bars: 25 μm. (H) Western blot and quantification analysis of OS key functional proteins (Student’s t test, n = 3). OS, outer segment; IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer, RPS, photoreceptor segment. Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. A lack of significant difference is indicated by #.

Rescue effect of Virma in rod cells on photoreceptor degeneration in RKO mice. The m⁶A modification is a highly dynamic and reversible process, with both upregulation and downregulation of m⁶A modification reported to be involved in various cancers, suggesting that the m⁶A modification is crucial for the proper tissue function. We attempted to rescue the photoreceptor phenotypes in RKO mice by restoring m⁶A modification with a Virma overexpression allele (Virma^loxp/loxp; CAG-Virma-3xFlag; RHO-Cre, hereafter named Rescue). Remarkably, reexpression of Virma in rods of RKO mice significantly improved visual perception and ameliorated degeneration of photoreceptors cells at 8 weeks of age (Figure 8). In Virma-rescued retinas, the global m⁶A methylation levels were significantly restored compared with those in RKO mice, as quantified by dot blot assay (Figure 8A). Scotopic ERG recordings revealed elevated amplitudes of a-waves and b-waves in Rescue mice compared with age-matched RKO mice, indicating the improved rod-driven circuit responses (Figure 8B). H&E staining confirmed the rescue effect on retinal morphology: Rescue mice showed significantly thickened photoreceptor segments and ONL compared with RKO mice, with values nearly restored to levels similar to those of Ctrl mice (Figure 8C). Immunofluorescence staining (Figure 8D) and Western blot analysis (Figure 8E) revealed increased expression of OS structural and functional proteins. In conclusion, these data provided additional evidence that VIRMA loss compromised photoreceptor function, highlighting the critical role of m⁶A methylation photoreceptor cells.

Figure 8

Replenishing Virma can attenuate photoreceptor degeneration in RKO mice. (A) m⁶A dot blot assay of global m⁶A abundance in retinas from 8-week-old Ctrl, RKO, and Rescue mice. Methylene blue staining was used as a loading control (1-way ANOVA, n = 4). (B) Representative scotopic ERG traces and statistical analysis at identified flash intensities in mice at 8 weeks of age (n = 3). Analyzed by 2-way ANOVA with post hoc Bonferroni’s multiple-comparisons test. (C) H&E staining of paraffin sections of RKO mice, Rescue mice, and Ctrl mice at 8 weeks, and quantification analysis of the photoreceptor segments (PRS) and ONL thickness of the groups (n = 3). Scale bar: 25 μm. Analyzed by multiple 2-tailed t tests with the Holm-Šídák method to correct for multiple comparisons. (D) Retina cryosections from 8-week-old mice were stained with OS marker rhodopsin and IS marker NaK ATPase. Scale bars: 25 μm. (E) Western blot and quantification analysis of visual perception–associated proteins (n = 3). OS, outer segment; IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer, RPS, photoreceptor segment. Data are presented as the mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

m⁶A methylation modification is the most abundant internal mRNA modification in mammals and plays a vital role in retina development and homeostasis maintenance. As a critical scaffold component, VIRMA guides region-selective m⁶A methylation by recruiting catalytic core components (32). Notably, the deletion of VIRMA led to the biggest reduction in m⁶A level but not in METTL3/14 or WTAP knockdown cells (12, 36), highlighting its critical role in m⁶A modification. VIRMA has been reported to be involved in numerous malignant tumors (19, 40, 78–80) and nonneoplastic diseases, including cardiovascular system disease (81), respiratory system disease (82, 83), orthopedic diseases (84), reproductive system disease (41), and central nervous system diseases (37, 38). However, whether VIRMA, like other m⁶A writers (29, 30, 85–87), is involved in occurrence of neuron retina disease is yet to be elucidated. Given the link between m⁶A modification and retinal function, we generated and characterized rod- and cone-specific Virma-knockout mice (RKO and HKO) and showed that Virma-catalyzed m⁶A modification is essential for the function and survival of both rods and cones (Figures 1–3). When Virma-mediated m⁶A was depleted in rods, the whole profile m⁶A modification level of retina was dramatically decreased (Figure 4). Using an unbiased combined analysis of MeRIP-seq and RNA-seq data, we found that genes exhibiting simultaneous downregulation in both m⁶A modification and transcriptional levels in RKO mice were predominantly enriched in the visual perception cascade (Figures 4 and 5). Notably, RKO mice displayed visual impairment and photoreceptor loss following Virma depletion. Furthermore, reintroduction of Virma expression in rods successfully alleviated retinal lesions (Figure 8). Therefore, our results elucidate the key pathogenesis of Virma-associated retinal degeneration.

An interesting finding of our study is that the pathological phenotypes of retinal damage caused by Virma deficiency occur earlier and progress more rapidly than those resulting from deficiencies in other m⁶A writers, including core components Mettl3 and Mettl14 (Supplemental Figure 2). This finding highlights the crucial role of VIRMA in m⁶A methylation modification and is consistent with prior research (30). On one hand, this may be attributed to the downregulation of m⁶A modification levels on a larger number of target genes in the retina following Virma depletion. On the other hand, the nuclear speckle localization of VIRMA and its interaction with serine/arginine-rich splicing factor 3 (SRSF3) led us to investigate whether VIRMA regulates AS events in retina. SUPPA2 analysis demonstrated that Virma was involved in AS regulation in the retina, with the most pronounced changes in AS events after Virma deletion occurring primarily in A5SS, A3SS, SE, and RI. This analysis uncovered disparities in contrast to a previous study, which observed a higher percentage of SE events (78% in oocytes and 48% in HeLa cells) following VIRMA depletion (41). These discrepancies may be attributed to the unique role of VIRMA in different cell types and enhanced accuracy of data from high-throughput transcriptomic. After validation, we identified increased exon skipping in Impdh1 and increased intron retention in Pogl2 in the RKO mouse retinas (Figure 6). Previous studies have shown that IMPDH1 catalyzes the rate-limiting step of de novo guanine synthesis and involved in RP development. Furthermore, intron retention directly contributes to the regulation of gene expression and plays a particularly important role in the synaptic plasticity of neuronal cells (88). Pogl2, as the processivity subunit of the mtDNA polymerase γ, is solely responsible for mtDNA replication and has been associated with mtDNA depletion syndrome (89). Aberrant transcript appears and, consequently, mitochondrial dysfunction occurs in RKO mice, as illustrated in Figure 6. Furthermore, visual impairment and photoreceptor degeneration in rod-specific Srsf3-knockout mice provide additional evidence for the key role of stable splicing in retina homeostasis (Figure 7).

In summary, we demonstrated that Virma is required for photoreceptor function and survival, likely via the posttranslational regulation of visual perception signaling and AS in photoreceptors. With the decrease in m⁶A levels in the retina, Virma deficiency alters the expression pattern of visual perception cascade that orchestrates photoreceptor function, resulting in impaired visual function and photoreceptor degeneration. Reintroduction of Virma in rods can ameliorate Virma-associated retina degeneration. Notably, we found that VIRMA is localized in the nuclear speckles and interacts with SRSF3, implying an essential role of VIRMA in AS. Of importance, compared with those previously reported in oocytes and HeLa cells following VIRMA deletion, the differences in the proportion of significant AS event types in retina suggest the cell-specific role of VIRMA in mRNA metabolism. These data provide a comprehensive understanding of the role of Virma in retina and highlight the contribution of RNA epigenetic modification to nervous system function. Given the high conservation of VIRMA between mice and humans (>90% sequence homology, https://blast.ncbi.nlm.nih.gov/Blast.cgi), the findings from our model may be insightful for the development of future therapeutic strategies for related human retinal diseases. However, whether the homeostasis of m⁶A modification affects retinal function remains to be further explored.

Sex as a biological variable. Sex was not considered as a biological variable. Both male and female mice were used in this study.

Experimental animals. Mice with Virma and Srsf3 deletion specifically in retinal rod or cone cells were generated using the Cre-loxP system. The Virma-floxed mice and Srsf3-floxed mice in a C57BL/6J genomic background were constructed by Cyagen Biosciences. RHO-Cre mice were purchased from The Jackson Laboratory (stock no. 015850). The HRGP-Cre mice were a gift from Yunzheng Le at the University of Oklahoma Health Sciences Center, Norman, Oklahoma. Virma-floxed mice were mated with RHO-Cre or HRGP-Cre transgenic mice to obtain rod- or cone-specific Virma-knockout mice (named Virma^flox/flox; RHO-Cre [RKO] or Virma^flox/flox; HRGP-Cre [HKO] mice). To determine the specificity of Cre expression, we employed a tdTomato reporter allele (strain Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J; stock 007914). In this system, Cre-mediated recombination excises a loxp-flanked stop codon before the tdTomato expression cassette, allowing for the expression of tdTomato (red fluorescence) in Cre-positive cells.

Additionally, we acquired a knockin mouse model that carries the SApolyA-CAG-LSL-Virma-3×Flag-IRES-EGFP-WPRE-pA expression cassette inserted into the Rosa26 locus on chromosome 6 (named CAG-Virma) from the Shanghai Model Organisms Center. Virma RKO mice were then mated with CAG-Virma mice to generate Virma^flox/flox; RHO-Cre; CAG-Virma (named Rescue) mice. In the presence of Cre enzyme, the STOP cassette was removed, and Flag-tagged Virma was expressed for genetic rescue. Mouse genomic DNA samples were extracted from mouse tails and genotyped using the corresponding primers (Supplemental Table 2).

All animals were kept in specific pathogen–free grade rooms with controlled conditions: temperature at 25°C, relative humidity at approximately 40%–60%, and a 12-hour light/dark cycle.

ERGs in mice. ERG recordings were performed as described in a previous study (90). Briefly, mice were dark-adapted for at least 8 hours, and all subsequent procedures were performed under dim red light. Mice were anesthetized using ketamine (16 mg/kg body weight) and chlorpromazine (80 mg/kg body weight) mixed with saline. A drop of tropicamide was administered to the eyes of the mice to dilate the pupils. Dark-adapted ERGs in response to flashes of light with intensities ranging from 0.01 to 10.0 cd-s/m² and light-adapted with intensities of 3.0 and 10.0 cd-s/m² were recorded in mice using the Espion Visual Electrophysiology System (Diagnosys LLC).

Histological analysis. For H&E staining, eyes were extracted from mice of different ages and fixed overnight in FAS solution (0.08 M phosphate buffer containing 1.22% glutaraldehyde and 0.8% paraformaldehyde). Eyes were embedded in paraffin (in the same direction as the labeled embedding) and then cut to 5 μm sections. Sections containing the optic nerve were selected for H&E staining according to standard protocols.

Immunohistochemistry. For retinal section immunofluorescence, eyes were removed and fixed in 4 % PFA for 2 hours. After fixation, the cornea and lens were removed, and the eyes were dehydrated in 30% sucrose for 2 hours. The eyes were then embedded in OCT, and 10 μm thick sections were cut. Following permeabilization in PBS containing 5% FBS and 0.1% Triton X-100 for 1 hour, the sections were incubated overnight at 4°C with a primary antibody. After 3 washes in PBS, the sections were incubated for 2 hours at room temperature with Alexa Fluor 594/488–conjugated goat anti-mouse/rabbit secondary antibodies and DAPI. Images were acquired using a Zeiss LSM 900 confocal laser scanning microscope.

For retinal whole-mount staining, eyes were dissected and flattened as previously described (91). After fixing, permeabilization, and staining, the whole mounts were visualized with a Zeiss LSM 900 confocal laser scanning microscope. The primary antibodies used for immunohistochemistry are shown in Supplemental Table 3.

Western blotting. Retinas or cells were lysed in RIPA buffer (89900, Thermo) with a mixture of protease and phosphatase inhibitors. After sonication and centrifugation, the supernatant was diluted with 4x SDS loading buffer. Equal amounts of protein were separated on SDS-PAGE gels and transferred to NC membranes. The membranes were incubated using 8% skim milk in Tris buffer containing 0.1% Tween 20 (TBST) at room temperature for 1 hour and then primary antibodies were added overnight at 4°C. The membranes were washed 3 times in TBST and then incubated with horseradish peroxidase–coupled antibody at room temperature for 2 hours. Signals were developed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fischer). ImageJ (NIH) was employed to determine the relative protein density. Relative quantification was performed using β-actin as an internal reference. The primary antibodies used for Western blotting are shown in Supplemental Table 2.

Light-dark box test. The device used for the light-dark box test (XR-XB110, XINRUN) consisted of a box divided into a dark compartment and an illuminated compartment with a luminance of 2,000 lux, connected to the acquisition device through a small opening. Mice were first placed in the light compartment with its back to the opening and moved freely between the 2 compartments for 5 minutes. Based on the behavior recorded by the system’s video camera, the time spent in the light and dark compartments, the distance traveled, the total number of transitions, and the average speed of the mice were automatically analyzed using VisuTrack software. After each experiment, the compartments were cleaned with 75% ethanol to remove olfactory cues.

Optomotor response test. The optomotor response test was performed as described in previous study (53). Briefly, mice were placed on an elevated platform in the center to move freely and were presented with vertical sinusoidal grating stimuli, rotated horizontally in either a clockwise or counterclockwise direction, which was randomly generated by virtual cylinders projected onto 4 surrounding LCD displays. The visual stimuli were driven at a spatial frequency of 0.2 cycles/°. Head movements triggered by the visual stimuli were recorded by a video camera placed above the animal and analyzed by an algorithm that automatically tracked the mouse head position (OptoTrack XR-OT101 system, OptoTrack Version 4, XINRUN). Each mouse was tested at least 10 times, and the average of its performance was taken for analysis. After each mouse was tested, the compartment was cleaned with 75% ethanol to remove olfactory cues.

m⁶A dot blot assay. Total RNA was extracted from mouse retinas, and the mRNA was purified using PolyATtract mRNA Isolation System III (Z5300, Promega) following the standard procedure. Following adjustment of mRNA concentrations to uniform levels, both experimental and control groups were denatured at 95°C for 2 minutes. Subsequently, 2 μL mRNA was transferred to a nitrocellulose (NC) membrane, which was air-dried at room temperature and crosslinked under 1,200 W UV light for 50 seconds. The membrane was then washed with TBST buffer for 5 minutes, followed by 0.02% methylene blue staining to ensure consistency across the groups, and the resulting images were recorded. To block nonspecific binding, the membrane was incubated with 5% BSA for 1 hour at room temperature. It was then probed with an m⁶A antibody overnight at 4°C. The following day, the membrane was washed 3 times with TBST and incubated with a horseradish peroxidase–conjugated secondary antibody for 2 hours. The images were captured using an e-BLOT imager.

RNA-seq and m⁶A MeRIP-seq. For RNA-seq, retinas from 1-month-old Ctrl and RKO mice were harvested, and total RNA was extracted. mRNA was then purified and fragmented into 60–200 nt segments, followed by library construction. Sequencing was performed on the Illumina NovaSeq 6000 platform, and differential expression analysis was conducted using the DESeq2 R package. Gene expression values were transformed to log₂ scale for further analysis, with P ≤ 0.05 considered statistically significant.

For MeRIP-seq, retinas from 1-month-old Ctrl and RKO mice were also collected, and total RNA was extracted. The RNA quality and integrity were assessed, and the RNA was fragmented into 200 nt segments. The fragmented RNA was then incubated with an m⁶A antibody for 2 hours at 4°C to perform immunoprecipitation. Library construction was carried out, and sequencing was performed on the Illumina NovaSeq 6000 platform using 3 independent biological replicates. The reads per kilobase per million mapped reads values for the 5′ UTR, coding sequence (CDS), and 3′ UTR of all genes were calculated using deepTools software. Differential peaks were identified using the ExomePeak R package, with P ≤ 0.05 considered statistically significant.

MeRIP-qPCR and RT-qPCR. Total RNA was extracted from 1-month-old Ctrl and RKO mice using TRIzol (ET111-01, TransGen). The MeRIP assay was carried out with a GenSeq m⁶A MeRIP kit (GS-ET-001, CloudSeq) in accordance with the manufacturer’s instruction. Genes were quantified for m⁶A modification level by RT-qPCR using specific primers, and gene expression level also quantified by RT-qPCR using specific primers (Supplemental Table 4).

Gene knockdown and overexpression strategies. HEK-293T cells (purchased from ATCC) were seeded onto 6 cm dishes and cultured overnight to allow cell attachment. The following day, the cells were transfected with lentivirus containing human VIRMA shRNA (5′-AAGACCTTCGTGAAGTATA-3′, Genechem) or a negative control shRNA (5′-TTCTCCGAACGTGTCACGT-3′). After 72 hours of culture, puromycin selection was applied to isolate successfully transfected cells, which were subsequently used for further experiments.

For exogenous protein overexpression, HEK-293T cells were seeded onto 6 cm dishes and cultured overnight for attachment. The next day, cells were transfected with corresponding plasmids using Lipofectamine 3000 (L3000015, Invitrogen) following the manufacturer’s protocol, including pTriEx 1.1-Neo-VIRMA-3×Flag and pcDNA3.1-SRSF3-HA plasmids. The pTriEx 1.1-Neo-VIRMA-3×Flag plasmid was a gift from Jianzhao Liu (Zhejiang University, Zhejiang, China) and Chuan He (Howard Hughes Medical Institute, The University of Chicago, Chicago, Illinois, USA). After 36–48 hours of incubation, the cells were harvested for subsequent experiments.

Statistics. GraphPad Prism 8.0 software was used for statistical analysis. All data are presented as mean ± SD. The data sets were tested for normal distribution using Shapiro-Wilk test. Statistical significance was determined by unpaired 2-tailed Student’s t test or by 1- or 2-way ANOVA followed by Tukey’s, Dunnett’s, or Šídák’s multiple-comparisons test as appropriate. At least 3 independent experiments were performed. Results were considered significant when P < 0.05.

Study approval. All experimental protocols were approved by the Ethics Committee of Sichuan Provincial People’s Hospital (LS-2023-046) and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Data availability. All data produced or analyzed in the present study are included in the manuscript and supplemental files. MeRIP-seq and high-throughput transcriptomic data have been uploaded to the Genome Sequence Archive (https://ngdc.cncb.ac.cn/gsa/) under accession numbers CRA036966 and CRA036993, respectively. The proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) with the dataset identifier PXD073210. Values for all data points in graphs are reported in the Supporting Data Values file. All data needed to evaluate the conclusions in the manuscript are present in the manuscript and/or the supplemental materials.