Ubap1l Knockout Mice Model Recapitulates Retinal Degeneration Phenotype Observed in Patients and Exhibits Irregular Photoreceptor Morphology

Abstract

Purpose

UBAP1L is a newly discovered gene related to recessive inherited retinal degeneration (IRD) with an unknown pathogenic mechanism. This study aims to investigate whether biallelic pathogenic variants in UBAP1L contribute to the unsolved cases in Chinese IRD families and to explore the impact of the Ubap1l genetic defect in a mouse model.

Methods

Next-generation sequencing was used to detect variants in families with unknown pathogenic variants. Biallelic variants in UBAP1L, evaluated through multistep bioinformatic analysis, comparison across multiple databases, and related phenotype analysis based on thoroughly reviewed clinical data, were confirmed by Sanger sequencing and cosegregation analysis. The expression profiles of UBAP1L in human and Ubap1l in mouse were demonstrated using quantitative RT-PCR and RNAscope. Phenotype and photoreceptor morphology were analyzed in the Ubap1l knockout mouse model.

Results

Biallelic novel pathogenic variants in UBAP1L were exclusively detected in two unrelated families with retinitis pigmentosa (RP), including a homozygous c.[278C>A;280dup]/p.[(T93K;I94fs)] variant in one family and compound heterozygous variants (i.e., c.[278C>A;280dup]/p.[(T93K;I94fs)] and c.1051A>G/p.M351V) in the other family. Quantitative RT-PCR showed that UBAP1L was highly expressed in human and mouse retina. RNAscope analysis revealed that UBAP1L was specifically expressed in the outer nuclear and inner segments in the human and mouse retina. Notably, in the human retina, UBAP1L was more highly expressed in cones, with low but detectable expression in the RPE, whereas this expression pattern was not observed in the mouse retina. Ubap1l knockout mice exhibited mottled retinal degeneration, impaired photoreceptor function, and deformation of the rod outer segments.

Conclusions

For the first time, biallelic pathogenic variants in UBAP1L were reported to explain the genetic defect among East Asian families with RP. The expression profile and precise localization of UBAP1L were demonstrated. A Ubap1l knockout mouse model replicating key hallmarks of IRD was established. These findings provide valuable insights into the essential role of UBAP1L in maintaining photoreceptor function and offer potential avenues for therapeutic development targeting IRD.

Inherited retinal degeneration (IRD) is a highly heterogeneous group of disorders characterized by the progressive degeneration of photoreceptors, with an estimated prevalence of 1 in 2000 individuals worldwide.¹ Retinitis pigmentosa (RP) and cone-rod dystrophy (CRD) are two major forms of IRD, with RP being the most prevalent subtype, accounting for more than 50% of cases. RP primarily manifests with symptoms related to rod dysfunction, followed by secondary cone dystrophy in advanced stages.^2–4 Conversely, CRD typically begin with progressive loss of central vision since cones are the first to be affected.⁵ With the rapid development of sequencing technology, genetic testing has become the most important approach for the precise diagnosis of IRD.⁶ To date, more than 300 IRD causative genes have been identified (https://retnet.org/). Despite these advances, approximately 40% of patients with IRD remain without a confirmed genetic cause.^4,7,8 Exploration and identification of previously undiscovered IRD-related causative gene advances our understanding of disease pathogenesis and provides critical insights into the molecular pathways governing retinal degeneration.⁹

The ubiquitin-associated protein 1–like (UBAP1L) gene encodes a 381–amino acid protein containing two distinct domains: UBAP1-MVB12-associated (UMA) domain in the N-terminal region and solenoid of overlapping ubiquitin associated (SOUBA) domain in the C-terminal region.^10,11 The biological function of the UBAP1L protein remains poorly characterized, and its underlying pathogenic mechanisms are completely unexplored. Given that the protein encoded by its paralog gene, UBAP1, has been shown to be a subunit of the endosomal sorting complexes required for transport (ESCRT)–I complex involved in endosomal–lysosomal trafficking and clearance of misfolded proteins,^12,13 UBAP1L has been hypothesized to function in regulating the transport of ubiquitinated cargo through the endosomal pathway.¹⁴ Recent studies established biallelic variants in UBAP1L as a novel genetic cause of CRD or RP in Caucasian individuals.^15–17 Notably, IRD families contributed to UBAP1L variants that have been absent from East Asian populations to date. Transcriptomic and single-cell RNA sequencing public databases analyses revealed retina-specific expression of UBAP1L.¹⁷ However, the precise localization of UBAP1L within the retina and functional evidence supporting the relationship between UBAP1L and retinal function still require experimental validation.

Our study employed next-generation sequencing and a comprehensive process involving variant pathogenicity analysis to evaluate UBAP1L variants in a cohort of Chinese IRD families with unresolved genetic causes. Two novel variants of UBAP1L were identified in two unrelated RP families, including one in homozygous status and the other in compound heterozygous form. These represent the first UBAP1L-associated RP cases reported in East Asian populations. Spatiotemporal expression profiling of UBAP1L was performed through quantitative RT-PCR and RNAscope in situ hybridization, which confirmed its high expression in the retina and revealed its specific localization in the outer nuclear layer and inner segments. In particular, our experiments demonstrated the specific expression pattern of UBAP1L in the human retina: higher expression in cones than in rods, with relatively low expression in RPE, consistent with previous findings.^15–17 Interspecies differences in UBAP1L expression patterns were also confirmed: specifically, Ubap1l exhibited diffuse localization in mouse retina, lacking preferential enrichment in specific photoreceptor subtypes. A Ubap1l knockout mouse model (Ubap1l^−/−) was successfully established, which recapitulates features of human IRD pathology. Longitudinal phenotypic observations revealed depigmented white spot retinal degeneration and significant dysfunction of photoreceptors. Structural analysis highlighted disorganized rod outer segments in the Ubap1l^−/− mouse. These findings conclusively demonstrate the essential role of UBAP1L in photoreceptor function as well as structural integrity, providing valuable insights for future mechanistic studies.

Materials and Methods

Ethics Statement

After institutional review board approval at the Zhongshan Ophthalmic Center, individuals with various eye conditions were recruited from the Pediatric and Genetic Clinic. Written informed consent was obtained from the probands or their guardians prior to the collection of clinical records and peripheral blood samples, consistent with the principles of the Declaration of Helsinki.

Postmortem human ocular tissues were sourced from the Eye Bank of Guangdong Province. All procedures conformed to the Declaration of Helsinki established by the World Medical Association and the ethics principles of the International Ethical Guidelines for biomedical research involving human subjects developed by the Council for International Organizations of Medical Sciences. All animal experiments were approved by the Animal Care and Use Committee of the Zhongshan Ophthalmic Center, Sun Yat-Sen University, and were conducted in accordance with its guidelines and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Patient Cohorts, Clinical Examination, and Variant Detection

The large cohort enrolled families with various inherited ocular diseases, including IRD families and other inherited ocular conditions (e.g., early-onset high myopia, hereditary optic neuropathies), as well as normal controls. DNA samples and detailed clinical data of probands and available family members were collected. Diagnosis of IRD was based on detailed clinical history, symptoms, family history, and comprehensive ophthalmic examination, including best-corrected visual acuity (BCVA), slit-lamp examination, direct ophthalmoscopy, refraction examination, axial length assessment via IOL Master biometry (Carl Zeiss, Germany), posterior pole fundus photography, scanning laser ophthalmoscopy (SLO), fundus autofluorescence (FAF), optical coherence tomography (OCT), and electroretinogram (ERG).

Genomic DNA was extracted from peripheral blood leukocytes using the method described earlier to perform whole-exome sequencing (WES) or whole-genome sequencing.¹⁸ Based on sequencing data, probands detected with biallelic variants in UBAP1L were selected for further analysis. First, variants with low sequencing quality of a coverage depth below 5 or a high minor allele frequency greater than 0.01 in the gnomAD database were excluded. Then, variants in UBAP1L were assessed by comparing multiple databases, including the in-house IRD cohort, the in-house control cohort, and the gnomAD database (https://gnomad.broadinstitute.org/). The missense variants were evaluated through in silico prediction tools, including SIFT (http://sift.jcvi.org/), Polyphen-2 (http://genetics.bwh.harvard.edu/pph2/), PROVEAN (http://provean.jcvi.org/index.php), MutationTaster (https://www.mutationtaster.org), CADD (http://cadd.gs.washington.edu), REVEL (https://sites.google.com/site/revelgenomics/), and AlphaMissense (https://palhamissense.hegelab.org). Sanger sequencing and cosegregation analysis were subsequently conducted.¹⁹

Quantitative Expression and Localization Analysis of UBAP1L/Ubap1l

Human RNA expression profiling analysis encompassed a total of 29 anatomically distinct tissues, including 6 ocular tissues (retina, sclera, lens, choroid, iris, and ciliary body) dissected from donor eyes under RNase-free conditions and 23 extraocular samples obtained from Clontech Laboratories (Mountain View, CA, USA), which was collected from 12 Caucasian individuals who died suddenly at ages ranging from 18 to 54 years. For wild-type C57BL/6J mice, 9 ocular tissues (retina, choroid, cornea, lens, iris, ciliary body, optic nerve, sclera, and ocular muscle) and 21 extraocular tissues were collected to analyze the RNA expression profile of Ubap1l in mice. Total RNA from the human ocular tissues and all mouse tissues was extracted using RNAiso Plus reagent (No. 9109; TaKaRa, Shiga, Japan). RNA concentration of all human and mouse tissues was measured using a NanoDrop One spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Total RNA of these tissues was reverse-transcribed into cDNA using the PrimeScript RT Reagent Kit (RR047A; TaKaRa). Quantitative RT-PCR amplification was performed in triplicate utilizing PowerUp SYBR Green Master Mix (A25742; Applied Biosystems, Waltham, MA, USA) on a QuantStudio Dx real-time fluorescence quantification PCR instrument. Relative quantification analysis in RNA levels employed the ΔΔCt method with GAPDH as the reference gene, as previously described.²⁰ All primers for quantitative RT-PCR, designed using Primer 3 (http://primer3.ut.ee/), are listed in Supplementary Table S1.

Spatiotemporal localization analysis of UBAP1L transcripts in human and mouse retinas was performed using RNAscope in situ hybridization technology (Advanced Cell Diagnostics, Hayward, CA, USA) following standardized protocols. The preparation of paraffin-embedded sections involved deparaffinization, pretreatment with hydrogen peroxide, and antigen retrieval in antigen retrieval buffer at 95°C for 15 minutes. Hydrophobic borders were drawn using an ImmEdge Hydrophobic Barrier Pen (cat. 310018; ACD, Brecksville, OH, USA), and tissue sections were then treated with proteinase K for 30 minutes at 40°C in a humidified chamber. After rinsing, mRNA probes were hybridized for 2 hours at 40°C, followed by a sequential amplification process: AMP1 and AMP2 were each incubated for 30 minutes, while AMP3 and HRP-C1 were incubated for 15 minutes. Signal amplification was achieved using Opal 690 reagent (ASOP690; Asbio, Guangzhou, China). The RNAscope staining slides were further costained with rhodopsin (Ab5417; Abcam, Cambridge, UK), GNAT2 (Code No. PM075; MBL, Beijing, China), opsin (AB5405; Merck Millipore, Burlington, MA, USA), PDE6B (sc-377486, Santa Cruz Biotechnology, Texas, USA), and RPE65 (Ab231782-10; Abcam) antibodies to confirm the specific location. High-resolution confocal microscopy was performed using a Zeiss LSM980 system (Carl Zeiss, Baden-Württemberg, Germany) with 20× or 63× objective magnification at 1.0× zoom, controlled by Zen Blue acquisition software.

Functional Validation of the Missense Variant in UBAP1L

Plasmids containing either the wild-type coding sequence of UBAP1L or the mutant UBAP1L sequence carrying the novel c.1051A>G/p.M351V missense variant were constructed by Beijing Tsingke Biotech (Beijing, China). Each plasmid was fused with enhanced green fluorescent protein (eGFP) at the N-terminus, and both constructs were verified by Sanger sequencing. Human embryonic kidney HEK293T cells were cultured in high-glucose Dulbecco's modified Eagle's medium (Gibco; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (Gibco) in a humidified incubator with 5% CO₂ at 37°C. For transient transfection, HEK293T cells were replated and grown to 70% to 80% confluence approximately 24 hours later. Transfection was performed using Lipofectamine 3000 transfection reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Following 24-hour transfection, total RNA was extracted from cells using RNAiso Plus reagent, and cDNA was synthesized with the PrimeScript RT Reagent Kit. Quantitative RT-PCR was performed using PowerUp SYBR Green Master Mix on a QuantStudio Dx Real-Time PCR System. For protein analysis, cells were harvested and lysed on ice in RIPA lysis buffer supplemented with 1% phenylmethylsulfonyl fluoride and 1% protease inhibitor cocktail (Beyotime; Shanghai, China). Protein concentration was quantified using the BCA Protein Assay Kit (Thermo Fisher Scientific). For simple western analysis (ProteinSimple, San Jose, CA, USA), 1 µg total protein per sample was mixed with Sample Buffer and Master Mix (ProteinSimple), heated at 95°C for 5 minutes, and loaded onto a 12- to 230-kDa simple western analysis Separation Module. Detection was performed using a GFP-tag polyclonal antibody (Proteintech; California, USA) and horseradish peroxidase–conjugated secondary antibody (ProteinSimple; California, USA). GAPDH (1:50 dilution) served as the loading control. Compass Software 6.0 (ProteinSimple) was applied to analyze the Western blot digital images and quantify the protein abundance of the bands. Both quantitative RT-PCR and Western blot results are from three independent experiments, each performed in triplicate.

Generation and Phenotype Examination of Mouse Models

The Ubap1l knockout mouse model in this study was established on a C57BL/6JGpt congenic background through 10 generations of backcrossing to ensure genetic homogeneity. Mice were maintained under specific pathogen-free conditions with controlled environmental parameters (22°C ± 1°C, 55% ± 5% humidity, 12-hour light/dark cycle) and provided with sufficient food and water. Germline transmission of the Ubap1l knockout (Ubap1l^−/−) mice was successfully achieved through CRISPR/Cas9-mediated deletion (GemPharmatech, Jiangsu, China). This targeted deletion spanned g. 65277569-65285931 (GRCh39, NM_001111145.2), precisely excising an 8.3-kb region encompassing exon 3 to exon 6 of the ENSMUST00000147185.2 transcript. Gene-edited zygotes generated through pronuclear microinjection of single guide RNA (sgRNA1: 5′-TGGGTGTCACAGGCAACCTC-3′ and sgRNA2: 5′-AACGTTCTTGGGTTCACTCT-3′) and Cas9 mRNA, which were immediately transferred into the uteri of pseudo-pregnant females. Positive founder mice were validated by Sanger sequencing of tail-derived genomic DNA using flanking primers listed in Supplementary Table S1. The knockout efficiency was rigorously quantified using quantitative RT-PCR assays with technical triplicates, demonstrating nearly 100% reduction in Ubap1l mRNA expression in Ubap1l^−/− mice compared to wild-type, normalized to GAPDH. RNAscope in situ hybridization experiments utilizing the Ubap1l probe were conducted on retinas from both Ubap1l^+/+ and Ubap1l^–/– mice to confirm the absence of Ubap1l localization in knockout mice. These founders were then bred with C57BL/6JGp mice to establish stable first-generation mouse models (Ubap1l^+/−). The first-generation Ubap1l heterozygous mice, at 1 month of age, were housed in breeding pairs (one male/one to two females) under controlled conditions. The genotypes of all neonatal mice were validated between postnatal days 14 and 21 through genomic DNA extraction from tail samples. PCR amplification was performed using knockout and wild-type allele primers, as listed in Supplementary Table S1. The amplification products from both wild-type and knockout-specific primer pairs were resolved by 2% agarose gel electrophoresis, and genotypic determination of the mouse progeny was achieved through comparative band pattern analysis.

Both Ubap1l^−/− mice and their background-matched wild-type (Ubap1l^+/+) littermates, aged 1, 2, 3, 6, and 9 months, underwent ophthalmic evaluations, including fundus photography and ERG. Each genotype group consisted of at least three animals. For full-field ERG testing, mice were dark-adapted overnight. Anesthesia was induced with an intraperitoneal injection of 10 µL/g of 1% amobarbital, and bilateral pupil dilation was achieved using compound tropicamide eye drops (SINQI, Shenyang, China). ERG responses were recorded using the CELERIS system (D430; Diagnosys LLC, Lowell, USA), with the stimulating electrode maintained in contact with the cornea, aided by hypromellose eye drops (ZOC, Guangzhou, China) to stabilize the electrode position. Flash stimuli of varying intensities were employed to assess both scotopic and photopic ERG responses, with the intensity ranges set at 0.003 to 10 cd·s/m² for scotopic conditions and 0.3 to 30 cd·s/m² for photopic conditions. To ensure accurate response measurement, light adaptation at 30 cd/m² was performed for 5 minutes between the scotopic and photopic recordings, allowing for proper transition of the retina to light conditions. For fundus photography, anesthesia and pupil dilation were performed following the same procedures used in the ERG testing. Retinal images were acquired utilizing the Retinal Imaging Microscope (Micron IV; Phoenix, New York, USA), with the mice positioned on the imaging stage. To maintain a clear view of the cornea during the imaging process, hypromellose eye drops were applied to prevent desiccation of the cornea. Fundus images were captured with the optic nerve head as the central reference point, and horizontal or radial scans were performed for each eye to obtain comprehensive retinal views. Retinal images were obtained for nine anatomically defined quadrants (central, superior, inferior, nasal, temporal, superior nasal, superior temporal, inferior nasal, and inferior temporal) in each eye.

Retinal Vibratome Sectioning and Immunofluorescence

Following perfusion with ice-cold PBS (pH 7.4), eyes from 3-, 6-, and 9-month-old mice (n = 3/group) were enucleated and micro-dissected to isolate neural retina from the RPE/choroid layer under stereomicroscopic guidance. The isolated retinas were immersion-fixed in 4% paraformaldehyde solution and mixed with an electron microscope fixative at a 1:1 ratio (C23328; Acros, Beijing, China). The retina was then embedded in 4% agarose gel to provide support for sectioning. Once solidified, the tissue was carefully sectioned into 100-µm-thick slices using a vibratome. Immunofluorescence was then performed following a structured protocol involving multiple steps.²¹ Briefly, the sections were permeabilized and blocked with 0.5% Triton X-100 and 5% bovine serum albumin to prevent nonspecific binding. Then, the sections were incubated overnight at 4°C with rhodopsin antibody (Ab5417; Abcam), followed by staining with the appropriate fluorescent secondary antibody and DAPI (C1006; Beyotime). PBS wash steps were performed between each incubation to remove any excess reagents. Finally, the sections were mounted with a mounting medium and imaged using a Zeiss LSM 980 confocal microscope. To rule out experimental confounding factors, ImageJ software (National Institutes of Health, Bethesda, MD, USA) was applied to quantify photoreceptors exhibiting normal or abnormal morphology in vibratome sections derived from Ubap1l^+/+ and Ubap1l^−/− mice across all three age groups, under maximum magnification.

Statistical Analysis

All statistical analyses were performed using GraphPad Prism v10.0.2 (GraphPad Software, La Jolla, CA, USA). ERG parameters and the incidence of photoreceptors with aberrant morphology in vibratome sections were compared between the Ubap1l^+/+ and Ubap1l^−/− cohorts using two-tailed unpaired Student's t-tests, with post hoc analysis conducted when necessary. The unpaired t-test was also employed to compare mRNA expression levels and protein abundance between 293T cells transfected with the wild-type plasmid and those transfected with the missense mutant plasmid. Data are presented as the mean ± SEM. A P value of <0.05 was considered statistically significant, and error bars represent ± SE.

Results

Biallelic Pathogenic UBAP1L Variants Identified in Chinese RP Families

In our large-scale Chinese cohort of inherited ocular disorders, biallelic pathogenic variants in the UBAP1L gene were detected exclusively in two unrelated families diagnosed with RP and were absent in cases with other eye conditions or in normal controls. The proband from family 1 harbored c.[278C>A;280dup]/p.[(T93K;I94fs)] and c.1051A>G/p.M351V in a compound heterozygous configuration, while the proband from family 2 was identified with the c.[278C>A;280dup]/p.[(T93K;I94fs)] variant in homozygous status. Both variants were first reported and classified as pathogenic or likely pathogenic according to American College of Medical Genetics and Genomics guidelines. The c.[278C>A;280dup] variant was absent in the gnomAD database. The c.1051A>G variant exhibited an extremely low heterozygote allele frequency (3.87E-06; 6/1549828) in the general population, with no homozygous cases recorded. The c.1051A>G/p.M351V missense variant was predicted to have a damaging effect by all seven bioinformatics tools, particularly AlphaMissense—a model that assesses the pathogenicity of variants, building upon highly accurate protein structure modeling (Supplementary Table S2). Furthermore, in vitro experiments confirmed that mRNA expression levels were significantly increased in cells transfected with the plasmid harboring the c.1051A>G/p.M351V missense variant, compared to those transfected with the wild-type UBAP1L plasmid (P = 0.0085). For the protein level, an approximately twofold elevation in protein abundance in mutant-transfected cells relative to wild-type–transfected cells was revealed through Western blotting experiments (P = 0.0015) (Supplementary Fig. S1). Both variants have been confirmed via Sanger sequencing. The transconfiguration of the two heterozygous variants has been established through cosegregation analysis performed on accessible members of family 1 (Fig. 1A).

Figure 1.

Biallelic variants in UBAP1L identified in two unrelated families with autosomal recessive retinitis pigmentosa. (A) Pedigrees and Sanger sequencing results for the two families with retinitis pigmentosa caused by biallelic pathogenic/likely pathogenic variants in UBAP1L. The transconfiguration of the two heterozygous variants in UBAP1L from the proband of family 1 has been confirmed in accessible family members. The blackened symbols represent affected individuals. Circles represent females, while squares represent males. Mx, mutant alleles; +, wild-type allele. (B, D, F) Examination results of the proband (II-2) from family 1. Widefield fundus photography shows extensive tapetoretinal atrophy in the mid-peripheral retina (hypo-autofluorescence change in fundus autofluorescence image) and macular atrophy, characterized by a hypo-autofluorescence central area surrounded by a hyper-autofluorescence ring. OCT results revealed retinal thinning and hyperreflective materials deposited beyond the sub-RPE layer. (C, E, G, H) More severe ocular manifestations were observed in the 53-year-old patient from family 2. In multimodal imaging, widefield tapetoretinal degeneration accompanied by bone-spicule pigment deposits was observed on scanning laser ophthalmoscope examination, with hypoautofluorescence changes on FAF. Obvious thinning of the retina, structural disorders of outer segments, and hyperreflective material depositions were observed in OCT. Undetectable rod and cone responses were recorded in the ERG examination. DA, dark adapted; Hz, Hertz; LA, light adapted; NC, normal control; OD, right eye (oculus dexter); OS, left eye (oculus sinister).

The proband of family 1 experienced bilateral progressive visual deterioration with constricted visual fields and blurred vision since age 28, leading to an RP diagnosis. Vision examination at age 31 demonstrated a BCVA of 32/100. Cycloplegic refraction quantified myopic errors of −2.25 diopters (D) of the right eye and −2.50 D of the left eye. Fundoscopy combined with color fundus photography revealed characteristic tapetoretinal degeneration accompanied by macular degeneration and bone-spicule pigmentary deposits in the peripheral retina. SLO high-resolution images documented pronounced pigmentary disturbances in the far-peripheral retina. Widefield FAF examination revealed a hypo-autofluorescent macular area accompanied by a hyper-autofluorescent perivascular ring encircling the arcades, and peripheral autofluorescence showed decreased alterations with pronounced involvement of the inferior retinal quadrants. OCT presented with extensive outer retinal layer disintegration and hyperreflective material accumulation beyond the RPE layer. Goldmann perimetry showed a diffuse decrease in light sensitivity and an absolute dark spot in the central ring (Figs. 1B, 1D, 1F).

The proband (II-1) of family 2 reported blurred vision at the age of 43 and presented for ophthalmic evaluation at 46. BCVA was measured at 20/100 bilaterally. Refractive examination demonstrated myopic errors of –1.75 D (right eye) and −1.50 D (left eye), respectively. Slit-lamp biomicroscopy revealed cortical lens opacities with characteristic dust-like dispersion patterns. Direct ophthalmoscopy observed pale optic discs, arteriolar attenuation, mid-peripheral retinal pigmentary deposits, and macular degeneration. Full-field ERG displayed extinguished rod and cone responses, establishing an RP diagnosis. At the age of 53, the patient reported photophobia with stable BCVA (20/100). Multimodal imaging examination was performed: SLO imaging demonstrated widefield tapetoretinal degeneration with characteristic bone-spicule pigmentary deposits. FAF showed diffuse hypo-autofluorescence throughout the retina with preserved autofluorescence islands predominantly in the superotemporal quadrant. OCT examination revealed retinal thinning, structural disorganization from the outer nuclear layer to the ellipsoid zone, and scattered deposits of a few hyperreflective materials at the RPE layer. Goldmann perimetry detected temporal-inferior visual field islands bilaterally (Figs. 1C, 1E, 1G, 1H).

Expression Profile of UBAP1L in Human and Mouse Tissue

To systematically investigate tissue-specific expression profiles of UBAP1L, quantitative RT-PCR analysis across 29 human tissues was performed. The encoding transcript NM_001163692.2 was selected for primer design in this experiment. Our analysis revealed that the UBAP1L transcript exhibited retina-specific expression, with mRNA levels at least 24-fold higher in the retina compared to other tissues. Expression was detectable but significantly reduced in several central nervous system tissues (cerebellum, 4.1% of retinal levels; fetal brain, 2.8%; and cerebrum, 1.5%). No detectable expression in other ocular tissues was recorded, including the choroid, sclera, iris, lens, and ciliary body, with Ct values over 35 cycles (Fig. 2A). Expression profiling of the murine homolog Ubap1l in wild-type mice also revealed a retina-specific expression pattern. Quantitative RT-PCR analysis demonstrated predominant expression in retinal tissue, with expression levels approximately 10-fold higher than in neural tissues (midbrain, 9.6% of retinal levels; spinal bulb, 7.9%; bone marrow, 7.8%; cerebellum, 7.0%; cerebrum, 1.6%; olfactory bulb, 1.1%; and forebrain, <1%). Notably, other whole-body tissues (i.e., liver, kidney, muscle) and other ocular tissues (i.e., cornea, lens, iris) showed no detectable expression, as indicated by Ct values >35 cycles (Fig. 2A; Supplementary Fig. S2; Supplementary Table S3). An RNAscope experiment was performed to precisely define UBAP1L mRNA localization in the human retina. UBAP1L transcript probe fluorescence was observed to localize in the outer nuclear layer (ONL) and inner segments of the photoreceptor. Notably, aggregation of UBAP1L probe fluorescence was detected in the ONL of GNAT2-labeled cones, suggesting that UBAP1L is predominantly expressed in cones and less so in rods. In addition, a relatively lower but detectable UBAP1L expression, labeled by the Cy5 fluorescent dye, was also observed in the RPE layer of the human retina (Figs. 2B, 2D; Supplementary Fig. S2). RNAscope analysis in the mouse retina revealed diffuse and scattered localization of Ubap1l mRNA within the photoreceptor ONL and inner segments in both cones and rods. Compared to humans, aggregation of the Ubap1l probe was not observed in the mouse ONL, with no detectable expression of Ubap1l in the mouse RPE layer (Fig. 2C; Supplementary Fig. S2).

Figure 2.

The expression profile and localization of UBAP1L/Ubap1l in human and mouse tissues. (A) Retina-specific expression of UBAP1L/Ubap1l was confirmed through quantitative real-time PCR in RNA from various human and mouse tissues. Central nervous tissues from both humans and mice exhibited relatively lower expression of this gene. (B) RNAscope in situ hybridization experiments utilizing a custom-designed UBAP1L probe were performed on human retinal sections, revealing specific UBAP1L mRNA expression in the ONL and inner segment (IS) layer of photoreceptors, with relatively low but detectable expression in the RPE layer. To minimize nonspecific autofluorescence interference from RPE cells, the UBAP1L probe was labeled with Cy5 fluorescent dye (emission wavelength >650 nm). Notably, aggregates of fluorescence from the UBAP1L probe were observed in the ONL of cones specifically labeled by the cone outer segment marker GNAT2. (C) RNAscope experiments in mouse retinal sections revealed a diffuse and scattered Ubap1l probe fluorescence signal within the ONL and IS layer of photoreceptors, with no detectable Ubap1l mRNA fluorescence in the RPE layer. (D) A magnified view of the human retinal RNAscope assay enabled clearer visualization that the aggregated regions of the UBAP1L probe in the ONL were primarily localized to cone photoreceptors, whose outer segments were labeled by the cone outer segment–specific antibody GNAT2. GCL, ganglion cell layer; INL, inner nuclear layer; OS, outer segments of photoreceptors. Scale bar: 50 µm in 20× magnification images and 10 µm in 63× magnification images.

Phenotype Characterization of Ubap1l Knockout Mice

CRISPR/Cas9-mediated deletion junctions spanning introns 2 to 6 of Ubap1l in knockout mice were confirmed through Sanger sequencing, with loss of Ubap1l expression in the retina of knockout mice demonstrated by both quantitative RT-PCR and the RNAscope experiment (Supplementary Fig. S3). Longitudinal ophthalmic evaluations, including nine-directional fundus photography and ERG, were conducted on both knockout and wild-type littermates at the 1-, 2-, 3-, 6-, and 9-month time points (n ≥ 3 per group). No degenerative changes were observed in the retinas of 1-month-old Ubap1l^−/− mice, with fundus performance indistinguishable from the age-matched Ubap1l^+/+ group (Supplementary Fig. S4A). Pigmentary disturbances, along with diffuse white spot degeneration and vascular attenuation change, were first found in 2-month-old Ubap1l^−/− mice, which exhibited progressive worsening over subsequent time points (3, 6, and 9 months), compared to Ubap1l^+/+ mice (Fig. 3, Supplementary Fig. S4B; Supplementary Fig. S5).

Figure 3.

Fundus photography and electroretinogram recordings of Ubap1l knockout (Ubap1l^−/−) and wild-type (Ubap1l^+/+) mice at different time points. At 3-month (A), 6-month (B), and 9-month (C) time points, the Ubap1l^−/− mice exhibited fundus alterations characterized by pigmentary disturbances, diffuse white spot degeneration, and vascular attenuation. These pathologic changes became more pronounced with advancing age compared to the normal fundus appearance observed in Ubap1l^+/+ mice. Electroretinogram analysis revealed significantly reduced scotopic and photopic ERG b-wave amplitudes in Ubap1l^−/− mice relative to wild-type controls. As the mice aged, the scotopic responses under maximal dark-adapted stimulation remained below 100 µV, while photopic responses under maximal light-adapted stimulation did not exceed 30 µV, demonstrating sustained functional impairment in both rod and cone pathways. In the lower left corner of the ERG waveform, the scale bar indicates an amplitude of 200 µV and a recording time of 100 ms. P values are indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001. DA, dark adapted; OD, right eye (oculus dexter); OS, left eye (oculus sinister).

The function of the photoreceptor was quantified through international society for clinical electrophysiology and vision (ISCEV)-standardized ERG protocols to assess both scotopic (rod-dominated) and photopic (cone-mediated) responses under different stimulus intensities. At 1 month of age, no significant differences in b-wave amplitudes were detected between Ubap1l^−/− and Ubap1l^+/+ mice under either dark-adapted or light-adapted conditions. Compared to Ubap1l^+/+ controls, ERG abnormalities first emerged at 2 months of age in Ubap1l^−/− mice and exhibited progressive dysfunction thereafter. Notably, the temporal sequence of rod versus cone degeneration could not be determined, even at the initial stage of abnormality onset at 2 months, as both rod- and cone-mediated responses showed concurrent impairment (Supplementary Fig. S4). In 3-month-old Ubap1l^−/− mice, a significant reduction in amplitude across all stimulus intensities (0.003–10 cd·s/m², P < 0.05) under dark adaptation was detected, with the maximal b-wave peak limited to 121.8 ± 21.8 µV. In contrast, Ubap1l^+/+ mice displayed a maximal b-wave peak of 349.7 ± 133.0 µV. The progressive reduction of the ERG scotopic b-wave amplitude exhibited a stimulus intensity–dependent pattern, with a 2.96-fold reduction at 0.003 cd·s/m² stimulation compared to a 3.11-fold decrease at 10 cd·s/m² (P < 0.05, linear mixed model). At 6 and 9 months of age, the scotopic amplitude of the Ubap1l^−/− mice remained severely impaired, with the b-wave peak showing statistically significant differences across all stimulus intensities. The maximal b-wave peak of Ubap1l^−/− did not exceed 100 µV, even with higher stimulus intensities. For the photopic responses, the light-adapted b-wave of Ubap1l^−/− presented with a decrease from the beginning intensity of 0.3 cd.s/m², with a reduction observed across all stimulus intensities, except for the 1.2 cd.s/m² stimulus intensities. At ages of 6 and 9 months, the photopic b-wave amplitude remained significantly lower compared to Ubap1l^+/+ mice, not exceeding 28.3 ± 4.2 µV, even at the highest stimulation intensity (Fig. 3).

Alternated Morphology of Rods in Ubap1l Knockout Mice

To systematically evaluate photoreceptor structural alterations resulting from the deletion of Ubap1l in the mouse model, retinal vibratome sections from 3-, 6-, and 9-month-old Ubap1l^−/− and Ubap1l^+/+ littermates were performed with rod photoreceptor-specific immunofluorescence staining following high-resolution confocal imaging. In contrast to the regularly arranged and uniformly long rod photoreceptors in Ubap1l^+/+ mice, the rod photoreceptors in Ubap1l^−/− mice exhibited varying degrees of structural abnormalities that worsened with age. For three age groups, the incidence of photoreceptors with abnormal morphology was significantly higher in Ubap1l^−/− mice compared with Ubap1l^+/+ mice, with statistically significant differences (P < 0.0001). Specifically, Ubap1l^+/+ mice exhibited an approximately 20% incidence of photoreceptors with abnormal morphology (likely attributable to experimental artifacts), whereas almost all photoreceptors in Ubap1l^−/− mice showed aberrant morphology. At 3 months of age, slight irregularities in the overall uniformity and length of the rod photoreceptors were observed. By 6 months, significant distortion was observed at the tips of the outer segments, accompanied by more pronounced variability in rod lengths. By 9 months, structural changes were more evident, with noticeable twisting and deformation of the outer segments (Fig. 4; Supplementary Fig. S6).

Figure 4.

Morphologic abnormalities in rod photoreceptors of Ubap1l knockout mice. Immunofluorescence staining of rod photoreceptor outer segments (rhodopsin antibody) in 3-, 6-, and 9-month-old Ubap1l knockout (Ubap1l^−/−) and wild-type (Ubap1l^+/+) mice revealed significant structural alterations. Compared to the well-organized, tightly stacked outer segments in Ubap1l^+/+ mice, Ubap1l^−/− mice exhibited progressive twisting, disorganization, and deformation of rod outer segments. These morphologic defects became increasingly severe with age, suggesting a time-dependent degenerative process (n = 3 per group). Quantitative analysis revealed that the proportion of abnormally shaped rod photoreceptors was significantly higher in Ubap1l^−/− mice compared to Ubap1l^+/+ controls, with statistically significant differences (P < 0.0001). The ∼20% rod photoreceptors with abnormal morphology in controls were likely attributed to technical artifacts introduced during vibratome sectioning. The scale bars in the images represent 20 µm (left) and 10 µm (right), respectively. ****P < 0.0001.

Discussion

UBAP1L, a newly discovered gene associated with inherited retinal diseases, has been predicted to be involved in the ubiquitin-dependent protein catabolic process and the ESCRT complex-mediated endolysosomal trafficking.^16,17 This study, for the first time, reports the correlation between UBAP1L and IRD in an East Asian ethnic group through the identification of two unrelated autosomal recessive RP families caused by biallelic pathogenic variants of UBAP1L, thereby expanding the mutation spectrum of this gene with two novel variants supplied. The retina-specific expression pattern of the UBAP1L gene was demonstrated in both human and mouse tissues. Furthermore, UBAP1L gene transcripts were specifically localized in the outer nuclear and inner segment layers of photoreceptors in both species. Notably, UBAP1L was revealed to exhibit higher expression in cones than in rods in the human retina, with relatively low expression uniquely detected in the human RPE layer—findings consistent with previous studies.^16,17 In contrast, no specific photoreceptor-enriched expression pattern or detectable RPE expression of Ubap1l was observed in the mouse retina. Functional experiments confirmed the essential role of UBAP1L in maintaining the normal function and structural integrity of photoreceptor cells.

All reported UBAP1L-associated IRD families prior to this investigation exclusively resulted from biallelic loss-of-function variants.^15–17 In this study, a compound heterozygous variant, including a novel missense variant c.1051A>G/p.M351V and a nonsense variant c.[278C>A;280dup], was identified in family 1. Functional validation experiments demonstrated that the c.1051A>G/p.M351V variant induced significantly increased transcriptional expression and significantly elevated protein levels. The aberrant mutant protein aggregation arising from the missense variant may involve mechanisms such as disruption of protein degradation homeostasis, abnormal cross-linking with interacting proteins, or cellular stress response.^22–24 Since the protein function of UBAP1L has not yet been fully elucidated, it is challenging to clarify how this missense variant induces functional alteration of UBAP1L, which requires further investigation in future studies. In addition, ethnic-specific UBAP1L variants have been documented, with the recurrent splice-site c.910-7G>A variant demonstrating particular prevalence in Middle Eastern populations (8/22 reported cases).^15–17 Our cohort identified the complex variant c.[278C>A;280dup] in two distinct East Asian pedigrees, further supporting the ethnic-specific characteristics of UBAP1L, which contributes to the development of ethnicity-tailored diagnostic panels and the optimization of variant interpretation guidelines.

UBAP1L-associated hereditary retinal disorders demonstrate highly phenotypic heterogeneity. Among the 22 families identified with biallelic pathogenic variants in UBAP1L, including families reported in this research, distinct phenotypic forms emerged: nine families presented with rod-predominant symptoms in early stage, meeting diagnostic criteria for RP, and eight families exhibited predominantly cone involvement or macular abnormalities leading to diagnoses of cone dystrophy (two families), cone-rod dystrophy (four families), or macular degeneration (two families). The remaining five families were described only as having retinal degeneration (two families) or were undiagnosed (three families).^15–17 Remarkable variability in age of onset and fundus manifestation was displayed among these patients, with age of onset ranging from early childhood to the sixth decade of life.^15–17 Notably, it was confirmed in our investigation that, despite the variable degree of fundus pigment accumulation and regional retinal atrophy, early macular involvement was exhibited in almost all patients diagnosed with RP resulting from variants in UBAP1L, characterized by a central hypo-autofluorescence area and surrounding hyper-autofluorescent rings in the macular region of the fundus.^15–17 OCT imaging revealed extensive retinal thinning and disruption of the outer nuclear layer. Impressively, hyperreflective deposits in a sub-RPE layer were consistently observed in both our patients and the published probands, which suggested a potential pathogenic mechanism linked to the degradation of the outer segments. ERG serves as a crucial diagnostic criterion for phenotypic stratification in UBAP1L-related retinopathies, in which diminished rod and cone responses were observed in patients with RP, while residual responses under scotopic conditions, along with severely reduced photopic responses, emerged as a characteristic of CRD. The sequential involvement of rod and cone cells in UBAP1L-related IRD requires elucidation through longitudinal pedigree studies and mechanistic investigations, and potential influencing factors warrant further exploration.

Consistent with previous findings based on analyses of public transcriptomic and single-cell RNA sequencing databases, our study confirmed the retina-specific expression profile of UBAP1L across all accessible human and mouse tissues.¹⁷ Moreover, the establishment of a custom-designed UBAP1L (for human)/Ubap1l (for mouse) probe and subsequent RNAscope in situ hybridization experiments enabled precise mapping of UBAP1L mRNA localization. In both human and mouse retinas, mRNA of UBAP1L was specifically detected in the ONL and inner segment regions of photoreceptors. The localization of UBAP1L within photoreceptor inner segments suggested a potential functional site of UBAP1L, as the inner segments are rich in ubiquitin and process complex protein metabolic processes.²⁵ Notably, aggregation of UBAP1L probe fluorescence was observed in the ONL layer of the human retina, especially in cones, as demonstrated by a cone outer segment marker of the GNAT2 antibody. This finding corroborates the higher expression levels of UBAP1L in cones compared to rod cells, as reported in previous studies.^15–17 Furthermore, the UBAP1L mRNA probe was stained with Cy5 fluorescent dye (emission wavelength >650 nm) to minimize autofluorescence interference.²⁶ As a result, a relatively low but detectable UBAP1L expression was observed in the human RPE layer. Conversely, the Ubap1l probe showed diffuse expression in the mouse ONL and inner segments in both cones and rods, and it was completely absent in the mouse RPE layer. Such species-specific expression patterns may underlie the phenotypic differences observed between humans and mice.

The knockout mouse model is an important approach for investigating the pathogenesis of the IRD-related causative gene.²⁷ In this study, a C57BL/6 mouse model targeting the Ubap1l genomic sequences was successfully established, specifically spanning the deletion of introns 2 through 6, utilizing CRISPR/Cas9-mediated technology. This model provides valuable functional evidence supporting the pathogenicity of UBAP1L loss-of-function mutations. Although the knockout models deleted the exon 5 through exon 7 region of Ubap1l (analogous to human c.910-7G>A variant effects), described in previous research that failed to observe the retinal degeneration phenotype until the age of 15 months,¹⁵ our model demonstrated early-onset pathologic manifestations that replicate the human IRD phenotype to a certain extent, including fundoscopic evidence of mottled retinal degeneration, a significant reduction in scotopic and photopic ERG amplitudes, and structural abnormalities in rod photoreceptor outer segments. The distinct retinal manifestations between the two knockout models may be accounted for by the differential disruption of structural domains of UBAP1L. Whereas the exon 5 to 7 deletion in the previous model exclusively impairs the SOUBA domain of Ubap1l, the deletion region in our model affects the integrity of both the SOUBA domain and its adjacent UMA-connecting region, which suggests the integrative requirement of both the SOUBA domain and its connecting region between UMA and the SOUBA region for maintaining protein function. Importantly, our model successfully recapitulates pathologic hallmarks of human IRD, providing essential experimental evidence for the role of UBAP1L in maintaining retinal function and serving as a valuable tool for the investigation of its underlying mechanisms. Clinical observations in patients carrying compound heterozygous pathogenic variants in UBAP1L revealed significant involvement of cone cells, with early-onset macular degeneration.^15–17 In this study, the ERG deficits identified in Ubap1l knockout mice confirmed that the deficiency of Ubap1l impairs the phototransduction and subsequent physiologic processes. The temporal sequence of rod versus cone abnormalities in mice could not be determined, which may be attributed to a range of differences between mice and humans, including divergent UBAP1L expression patterns (e.g., cone-enriched localization in the human retina versus diffuse distribution in the mouse retina, as confirmed by RNAscope experiments) and interspecies variations in cone to rod ratios, in which many more cones are presented in human eyes compared to those in mice.^28,29 Furthermore, morphologic disorganization of rod photoreceptor outer segments was observed in the Ubap1l knockout mouse via retinal vibratome sectioning. Due to the lack of a macular structure in mice and a significantly lower proportion of cone cells compared to that in humans, it was very difficult to observe obvious structural changes in cone cells in mice.^30,31 How the loss of function of Ubap1l disrupts visual signal transduction in photoreceptors and impairs their structure and biogenesis requires further validation in future studies.

Current functional hypotheses of UBAP1L remain speculative, inferred primarily from its conserved domains shared with the homolog gene UBAP1, lacking definitive experimental evidence. Since UBAP1 mediates endosomal sorting of ubiquitinated cargo and ubiquitination metabolism as a component of the ESCRT-I system, it is proposed that UBAP1L variants may cause RP through disruption of ESCRT-I system function.^{10–12,32–36} Emerging evidence highlights the ESCRT system as a critical regulator of retinal morphology and function.^37,38 Notably, IRD-associated PRPH2 localizes to late endosome (LE) lumens and uniquely regulates LE/ciliary targeting by binding ESCRT component Hrs.³⁹ However, the precise molecular role of UBAP1L, its direct involvement in ubiquitination pathways, and the mechanistic link between UBAP1L variants, ESCRT-I system impairment, and retinal photoreceptor dysfunction remain unvalidated and require experimental investigation to confirm.

In conclusion, the identification of two novel UBAP1L pathogenic variants in two unrelated Chinese RP families further supports the correlation between UBAP1L variants and RP in Asian populations, thereby broadening the mutational spectrum. For the first time, the precise localization of UBAP1L in the human retina, as well as its ortholog Ubap1l in the mouse species, was clearly shown to be enriched in the outer nuclear and inner segment layers. Rod-predominant phenotypes recapitulated in our Ubap1l knockout mouse model strongly support the gene’s significant role in retinal function. This model may be an effective approach for therapeutic discovery targeting ubiquitination dysregulation in IRDs. Elucidating the molecular cascades related to UBAP1L-mediated pathogenesis represents a promising strategy for developing novel therapeutic targets in IRD therapy aimed at promoting photoreceptor survival.