The IMPC ^,,,,,, (

Home

) consists of 13 phenotyping centers worldwide, aiming to generate and phenotype single-gene KO mouse lines for all coding genes in the mouse genome. Phenotyping data for each line are gathered using a set of tests from the International Mouse Phenotyping Resource of Standardised Screens (IMPReSS; https://www.mousephenotype.org/impress/index ). All lines were generated and maintained on a C57BL/6N genetic background. The technology adopted for generating each single-gene KO line was initially publicly available targeted mouse embryonic stem (ES) cells and then replaced by Cas9-mediated deletion strategies in 2014. For each gene, the Consortium produced and phenotyped 7 males and 7 females for most tests as defined in IMPReSS. For gene KOs affecting viability, with subviability defined as <12.5% live homozygotes (HOM) from heterozygote (HET) crosses, HET mice were phenotyped. KO and corresponding age- and sex-matched wild-type (WT) control mice were phenotyped as young adults between 9 and 16 weeks of age. At 14 weeks, hearing ability was evaluated in at least 4 mutant mice, both males and females in most cases, by analyzing auditory brainstem response (ABR) at frequencies of 6, 12, 18, 24, and 30 kHz. A click-evoked ABR was optional. A complete ophthalmic examination was performed at week 15 or 16. For the primary screen of retinal traits, almost all phenotyping centers used slit lamp examination (except for Baylor College of Medicine and HMGU Helmholtz Munich) and indirect fundus examination (except for Baylor College of Medicine, HMGU Helmholtz Munich, and Institut Clinique de la Souris—PHENOMIN-ICS, which instead conducted routine optical coherence tomography [OCT] cross-sectional and funduscopic imaging) as primary screen tools. Hematoxylin and eosin staining on formalin-fixed, paraffin-embedded tissues was also used by The Center for Phenogenomics and University of California Davis. Depending on the test and/or instrumentation used, the secondary screen studies varied across the phenotyping centers and included electroretinogram (ERG), OCT, fundus imaging, and transmission electron microscopy (TEM). Phenotyping data obtained from HET or HOM mice were compared with those from WT control mice from the same center with genotypes masked to the evaluators. If the phenotypic difference was statistically significant, an ontology term from the Mammalian Phenotype ontology was used to describe the phenotype.

The IMPC guidelines related to housing and husbandry met the requirements of the ARRIVE guidelines, the Gold Standard publication Checklist reporting Guidelines, and the Genetically Altered (GA) Passport to maintain animal welfare. All procedures complied with local, state, and national regulatory guidelines and were reviewed and approved by associated institutional animal care and use committees (IACUC), animal care committees (ACC), or equivalent.

The IMPC KO mouse line ^,,,,,, database (version 23, released on April 23, 2025) was queried for 11 retinal and 2 auditory traits. Retinal traits included abnormal retina morphology, retina pigmentation, eye electrophysiology, outer nuclear layer morphology, blood vessel morphology, vasculature morphology, blood vessel pattern, inner nuclear layer morphology, optic vesicle formation, decreased total retina thickness, and increased total retina thickness, but auditory traits were of abnormal ABR and abnormal otic vesicle morphology.

To identify potential candidate genes for USH that are not part of the 16 genes mentioned previously, the genes correlated with hearing and retina abnormality were screened for those that had at least 1 retinal and 1 auditory trait concurrently. To increase the stringency of assessing disease causation, only candidate genes causing bilateral retinopathy in most mice (a minimum of half of the KO mice affected in males and/or females) and/or were validated by histopathology were included. Mouse lines with retinal findings that did not meet these additional criteria were excluded, despite achieving statistical significance by IMPC standards.

We used the GeneCards database ( https://www.genecards.org ) version 5.25 (updated: July 23, 2025) ^, to search for human orthologues of the candidate mouse genes for multiple analyses, including protein interactions, molecular functions, and signaling pathways. Furthermore, mouse orthologues of inherited DSI genes were used to assess whether mouse lines with those genes disrupted recapitulate any DSI phenotypes. Using this database, we also screened the locus of the human orthologue of each candidate gene to see whether any were positioned in proximity to any of the previously mentioned 3 known mapped USH loci.

A protein interaction analysis was performed between the proteins encoded by USH genes, other established inherited DSI genes, ciliopathy genes, and our candidate genes using STRING-db v.12.0 released on July 26, 2023, built into the Cytoscape 3.10.3 App, StringApp. A confidence threshold of 0.9 was used to visualize any interaction between the proteins encoded by members of each group. To delve deeper into possible interactions among the 4 gene sets, an analysis with a confidence level of 0.7 was also performed.

An analysis based on v.19.0 of the PANTHER classification system, released on June 20, 2024 ( https://www.pantherdb.org/ ), was conducted to investigate the molecular functions contributed by each gene group. ^, Associated signaling pathways were analyzed by KEGG, Reactome, and WIKI pathways through the Database for Annotation, Visualization, and Integrated Discovery (DAVID) knowledge base v.2025_1, released on April 17, 2025 ( https://davidbioinformatics.nih.gov/ ). ^,

Gene expression in the retina was analyzed using the expression plot of Platform for Analysis of scEiad (single cell eye in a disk) or Plae v.0.95 ( https://plae.nei.nih.gov/ ). We used the “CellType_predict” cell grouping and set the facet on “Gene,” in addition to “organism” and “CellType_predict” as filter categories in a minimum of 50 cells per group to narrow our analysis to include the data from Homo sapiens and Mus musculus separately. As RP in USH predominantly affects rod photoreceptors and/or retinal pigment epithelium (RPE) cells, we checked the expression of the candidate genes and USH genes only in these cells, based on “Mean Log2(Counts+1)” values which were colored based on “study_accession.” Consequently, if the presented value for RPE cells and/or rod photoreceptors was >0, then the gene was considered to be expressed in the retina. For values equal to 0, the expression was considered as not detected. Similarly, we used the Gene Expression Analysis Resource (gEAR) portal v.2 ( https://umgear.org/ ) to analyze the expression of each candidate and USH gene in the cochlea and/or vestibular HCs of humans and mice. Expression of the gene in either the cochlea or vestibular HCs was considered as the expression of the gene in the inner ear. The Human Inner Ear and Inner Ear Organoid (van der Valk and associates ) profile was used for detection of gene expression in humans. In this profile, single-nucleus RNA sequencing (snRNA-seq) of inner ear HCs at fetal age weeks 7.5 and 9.2 exhibits the transcription of each gene in vestibular HCs, as the mature cochlea is still in development. The single-cell RNA sequencing (scRNA-seq) data of vestibular and inner HCs/outer HCs (IHC/OHC) demo profiles were used to identify the expression of candidates in the utricle and cochlear HCs of mice, respectively.

We used the retinal information network (RetNet, https://retnet.org/ ), accessed on September 5, 2025, to find any known association of human orthologs of candidate genes with retinopathies or mapped loci associated with retinopathies. Furthermore, we carried out a similar study for hearing impairment using the Shared Harvard Inner-Ear Laboratory database (SHIELD, https://shield.hms.harvard.edu/ ), accessed on September 5, 2025, and the Deafness Variation Database (DVD, https://deafnessvariationdatabase.org/ ) v.9.2, accessed on September 30, 2025. In DVD, we searched through the variant table of each represented gene using the value of hearing impairment in the phenotype field. The 263 genes related to ciliopathies, a combination of Syscilia and Ciliacarta, and 39 genes of other inherited DSI illnesses were derived from Higgins and associates and Guimaraes and associates, respectively. In addition, previously established associations of the candidate genes with retinal, hearing, and other clinical phenotypes in humans were retrieved from Online Mendelian Inheritance in Man, OMIM. McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University (Baltimore, MD), https://omim.org/ , accessed in September 2025.

To investigate the retinal and auditory phenotypes of variant mouse lines in greater detail, we performed a literature search on KO and/or HET mouse lines for each candidate and mouse orthologue of inherited DSI genes, which were generated and studied for inner ear and retinal abnormalities. The auditory and retinal phenotypes of these lines were individually investigated ( https://pubmed.ncbi.nlm.nih.gov/ and https://scholar.google.com/ ) by searching “gene symbol,” “knockout,” or “deletion” or “null,” “mouse,” “ear,” “retina,” “hearing,” and “eye” terms to make associations between the genes and the aforementioned abnormal traits.

The study was approved by the institutional review board of Partners HealthCare System and adhered to the Declaration of Helsinki. Informed consent was obtained from all individuals on whom genetic testing and further molecular evaluations were performed.

Patients and available family members from 230 families with syndromic or nonsyndromic IRD were ascertained at Massachusetts Eye and Ear and enrolled in a study assessing elusive causality. Clinical evaluation was performed by experienced ophthalmologists according to previously published protocols and included functional and structural assessment. ^,,,

Blood samples were obtained from probands and, when possible, their family members. DNA was isolated from peripheral blood lymphocytes by standard procedures. Genome sequencing (GS) was performed at Massachusetts Eye and Ear; data were aligned to hg38, and variant calling was performed using DRAGEN-GATK4 best practices. GATK-structural variant (SV) cohort mode using 5 different algorithms was used for structural variant calling.

Statistical analysis was done using GraphPad Prism v.10.5.0. ERG data were presented as mean ± SD. Owing to the different variances among the groups (53 WT vs 4 Mpdz ^−/− and 3 Rnf10 ^−/− mice), Welch’s t test was used to compare the ERG wave amplitudes. A P <.05 was considered statistically significant. For OCT data, presented as mean ± SD, retinal thickness was compared between Idua ^−/− and WT mice using the Wilcoxon rank sum test, with significance defined as P <.05. The ABR test results in KO lines, conducted by IMPC, with P values equal to or lower than 10 ⁻⁴ ( P ≤.0001) were marked as significant.

We queried KO mouse lines phenotyped by the IMPC for retinopathy and hearing abnormalities. We found 813 genes with at least 1 of 11 distinct abnormal retinal phenotypes and 338 genes with at least 1 of 2 distinct ear abnormalities. A total of 50 genes were identified that caused both auditory and retinal abnormalities in the same mouse line ( Figure 1 , A). Among these, 18 genes were considered high-likelihood candidates for USH because the resultant retinopathy was either confirmed by histopathology and/or was documented with bilaterally symmetric retinal abnormalities in most males and/or females of each mutant mouse line: AA986860 (MMRRC:066881-UCD), Adipor1 (EM:08491), Ankrd11 (retina phenotype: EM:00380, ear phenotype: EM:07651), Atp8b1 (MMRRC:067151-UCD), Chsy1 (MMRRC:067396-UCD), Cstb (EM:09566), Dyrk1b (EM:10419), Eef1d (MMRRC:066978-UCD), Fer (MMRRC:066693-UCD), Idua (EM:15473), Mpdz (MMRRC:048630-UCD), Pigq (MMRRC:043931-UCD), Rnf10 (retina phenotype: MMRRC:049477-UCD and MMRRC:049478-UCD, ear phenotype: MMRRC:049477-UCD), Slc20a2 (retina phenotype: EM:08067, ear phenotype: EM:05549), Spred1 (EM:12627), Sun1 (EM:09532), Tmem145 (MMRRC:065589-UCD), and Xrcc5 (EM:12867). ^,,,,,, Interestingly, genes such as Ankrd11, Dyrk1b, Eef1d , and Slc20a2 induced relevant USH phenotypes in heterozygous individuals, whereas for the other genes, these phenotypes were only observed in homozygous mutants. The situation for homozygous mutant genes is comparable to the AR inheritance of USH mutations, whereas the others may correspond in humans to a combination of partial loss of function. These 18 new genes, which are not part of the USH gene list, were considered candidate genes for human USH. Human orthologues for 11 of these genes ( SPRED1, PIGQ, IDUA, ANKRD11, ATP8B1, CHSY1, MPDZ, EEF1D, CSTB, DYRK1B, SLC20A2 ) were associated with other clinical phenotypes in humans (Supplementary Table 1).

In addition to hearing abnormality (Supplementary Figure 1) and retinopathy, many of these 18 mutant lines had abnormalities in other physiological systems as well, resulting in an expansion of the focus of our study to other inherited forms of DSI. If a particular mutant mouse line affected all the physiological systems similar to a known form of a human DSI syndrome, that gene was considered to be a candidate for the associated DSI disease. Some mouse lines reported here had significant but not complete similarity with DSI syndromes. As a result, the mutant lines that manifested phenotypes in at least 2 physiological systems in addition to visual and auditory systems were also considered positive hits for these DSI disorders, as summarized in Table 1 (all phenotyping data are available in Supplementary Data Sheet 1).

We wished to determine whether any of the 18 candidate genes had published mouse models with retinal or hearing phenotypes. Through literature search, we found that published papers from independent laboratories revealed all 6 candidate genes studied for retinopathy in KO mice ( Adipor1 , Chsy1 , Idua , Rnf10 , Sun1 , and Xrcc5 ) to have retinal abnormalities and all 4 candidate genes studied for inner ear abnormalities in KO mice ( Atp8b1 , Idua , Mpdz , and Sun1 ) to have the respective abnormalities. We did not find reports of negative retinal or hearing phenotypes in any KO mouse models of the 18 candidate genes.

Furthermore, we wished to assess the degree to which mouse models of USH genes and other established inherited DSI syndromes recapitulate the spectrum of findings found in humans. To this end, we queried the literature for reports of KO mouse models of 15 human USH genes (10 confirmed genes and 5 known human candidates). ^, We also queried the IMPC database for any retinal and/or auditory abnormality using the same set of 15 USH genes ( Table 2 , A). The IMPC data included studies on the retinal phenotype in only 9 KO and/or HET mouse lines among the 15 USH genes: Adgrv1, Arsg, Cep250, Cep78, Clrn1, Espn, Myo7a, Ush1c , and Whrn . The other 6 genes were either not tested or not fully analyzed. The IMPC also studied the hearing phenotype in 4 KO mouse lines of the 15 USH genes: Adgrv1, Cep250, Cep78 , and Clrn1 . Only these 4 lines had complete phenotyping of the eye and ear. The other 11 genes were not tested for hearing, or the results of the tests were not analyzed. Of the 4 lines, 3 had abnormal ABR results. None of the 9 lines with ocular phenotyping data had retinopathy.

Analysis of peer-reviewed articles for mouse models of the same 15 human USH genes revealed that 14 of the 15 had hearing abnormalities. Only 13 of the 15 mouse models had published evaluations of the retina, and 9 of the 13 had a retinal phenotype, but 4 of the 13 had a normal retina.

To compare IMPC screening to independent laboratories in a head-to-head comparison, we found just 4 genes that had retinal and hearing evaluations reported by both screening paradigms. Four genes ( Adgrv1, Cep250, Cep78 , and Clrn1 ) had mouse models studied for both retinal and inner ear phenotypes by independent laboratories, revealing concurrent retina and inner ear abnormalities in 3 of 4 lines ( Cep250, Cep78 , and Clrn1 ). The IMPC screening of these 4 lines revealed no retinopathy in any of them but ear abnormalities in 3 of 4 ( Adgrv1, Cep250, Clrn1 ). The results suggest that longitudinal study of mouse models of USH (as in the published literature) is more sensitive than the high-throughput snapshot screening used by the IMPC. We also concluded that mouse models of USH, when studied longitudinally, recapitulate the human findings associated with hearing (14/15, 93.3%), but the retinal abnormalities were not modeled as faithfully (9/13, 69.2%). All data are found in Table 2 , A.

A similar analysis was performed on mouse lines for the other (non-USH) inherited DSI-confirmed genes ( Table 2 , B). There were 39 known DSI genes exclusive of USH. The IMPC has generated and phenotyped the retina of KO/HET/hemizygous mice for 22 of 39 confirmed DSI genes: Alms1, Bbip1, Bbs1, Bbs4, Bbs5, Bbs7, Bbs9, Bbs10, Bbs12, Cep290, Cfap418, Chd7, Col11a1, Col2a1, Col4a3, Col4a4, Col4a5, Col9a2, Ift172, Ift74, Mks1 , and Snai2 . Two of these 22 lines had retinal phenotypes ( Bbs5 and Ift72 , 9.1%). The IMPC has also reported on the hearing ability of 14 KO/HET/hemizygous mouse lines of the 39 DSI genes: Alms1, Bbs1, Bbs4, Bbs7, Bbs10, Cep290, Cfap418, Col4a3, Col4a4, Col4a5, Col9a2, Ift172, Ift74 , and Snai2 . Only Col9a2 (1/14, 7.1%) had abnormal hearing.