CCDC66 frameshift variant associated with a new form of early-onset progressive retinal atrophy in Portuguese Water Dogs

Phototransduction in vertebrate retinas is initiated by highly polarized rod and cone photoreceptors composed of outer and inner segments, a nuclear cell body, and synaptic terminals. The connecting cilium links outer and inner segments and is crucial for photoreceptor integrity and maintenance. Defects in photoreceptor development or ciliary structure/function cause progressive retinal degeneration leading to blindness, many of which are inherited ciliopathies where retinal dystrophy may occur in isolation or as part of syndromes. Over 300 genes/loci are implicated in human inherited retinal diseases, with more than 30 genes causing analogous progressive retinal atrophy (PRA) in dogs. Canine models are valuable for translational therapies and mechanistic insights. The Portuguese Water Dog (PWD) breed, shaped by founder effects and bottlenecks, has limited genetic variation and an engaged breeding community, facilitating genetic studies. PRA in dogs mirrors human retinitis pigmentosa and is typically autosomal recessive, though dominant and X-linked forms exist. Multiple PRA-causing variants can segregate within a breed due to historical breeding practices; examples include Golden Retrievers (PRCD, TTC8, SLC4A3), Irish Setters (PDE6B, C2ORF71), and Miniature Schnauzers (RPGR, HIVEP3-PPT1). Prior to this study, PWDs were known to harbor prcd-PRA due to PRCD, a late-onset PRA recognized at 3–6 years or older and shared across many breeds. Here, the authors report a new, earlier-onset PRA in PWDs not associated with PRCD and identify a causative variant in CCDC66, a gene linked to ciliary function and retinal degeneration in mouse and another dog breed.

The paper contextualizes retinal degenerations as largely genetic, highlighting the role of ciliary dysfunction in photoreceptor disease. It notes >300 human IRD genes (RetNet) and >30 canine genes for PRA. Prior canine studies show both locus and allelic heterogeneity across and within breeds. PRCD-associated prcd-PRA is widespread with variable onset by breed. Other genes implicated in canine PRA include TTC8, SLC4A3, PDE6B, C2ORF71, RPGR, and a previous CCDC66 insertion (c.521_522insA) causing PRA in Schapendoes. In mice, Ccdc66 knockout leads to early photoreceptor degeneration and olfactory defects, supporting a ciliopathy-like mechanism. CCDC66 participates in ciliogenesis and centriolar satellite distribution and interacts with CEP72, CEP290, PCM1. Debate exists regarding CCDC66 in human RDs, but animal data support its relevance. The study leverages mapping-plus-sequencing strategies shown effective for recessive trait gene discovery.

Study design combined clinical phenotyping, pedigree analysis, SNP-chip genotyping, GWAS, linkage analysis, homozygosity mapping, whole-genome sequencing (WGS) of a case-parent trio, variant filtering and annotation, population genotyping, transcript and protein analyses, and in vitro functional assays. - Ethical approval: University of Pennsylvania IACUC (#806301); ARVO guidelines followed. - Phenotyping: Four EOPRA-affected PWD full-siblings (2M, 2F) from a small consanguineous pedigree and 78 related/unrelated controls were examined by board-certified veterinary ophthalmologists via indirect ophthalmoscopy and biomicroscopy after dilation; selected fundus photography performed. One affected dog underwent ERG at age 6 years. Additional normal controls (n=20) provided records. Four external PRA cases with early onset (2–3y) were documented by other specialists. - SNP genotyping: 23 PWDs (4 affected, 19 unaffected) genotyped on Illumina CanineHD BeadChip (173,661 SNPs). Quality control removed markers/individuals with call rate <90%, MAF <5%, and Hardy-Weinberg outliers as appropriate. - GWAS: Conducted in GenABEL (R) with mixed-model to address stratification; after excluding 103,281 non-informative markers due to relatedness, 70,380 SNPs analyzed. Suggestive peaks observed on CFA20 (p=1.4×10^-4) and CFA11 (p=5.9×10^-4); top 100 markers on CFA20. - Linkage: Parametric multipoint linkage (MERLIN) on nuclear family (2 parents, 4 affected, 2 unaffected siblings) under autosomal recessive, full penetrance, allele frequency 0.68. Positive LOD on CFA20 with maximum LOD=2.501 across a 20.63 Mb interval (CFA20: 29.33–49.96 Mb). Smaller positive LODs on CFA2, CFA12. - Homozygosity mapping: Using PLINK, searched for IBD homozygous segments (>1 Mb) shared by all 4 cases; identified a shared region on CFA20 from 21,911,990 to 48,461,664 bp. - Defining candidate interval: Integrated mapping defined a 26.55 Mb critical region on CFA20 (CFA20:21,904,057–48,467,358; CanFam3.1) for variant discovery. - WGS: Case-parent trio sequenced (HiSeq2500, 2×100 bp, ~32.6× coverage total; 1,637,397,795 reads). Alignment: BWA v0.5.9, samtools v1.1; duplicates marked (Picard). Variant calling: GATK UnifiedGenotyper; annotation: SnpEff v5.0. Structural variants: Delly2 v0.8.5 (no disease-associated SV detected). Variants filtered against DBVDC, DogSD, and EVA; also checked idog; used bcftools for filtering. - Variant prioritization: From 33,205 variants private to trio, 158 fit autosomal recessive model (hom in case, het in parents). Protein-changing candidates in interval: (1) CCDC66 c.2262_c.2263insA (CFA20:g.33,717,704_33,717,705insT; frameshift p.Val747SerfsTer8) and (2) RNF123 c.2755C>T (CFA20:g.39,560,171G>A; p.Arg919Cys; PolyPhen-2 possibly damaging, 0.81). An apparent second CCDC66 frameshift at CFA20:33,745,456_33,745,457insT was determined to represent an error in CanFam3.1 reference; confirmed as wild type T in multiple genomes by WGS and Sanger. - Validation and genotyping: PCR and Sanger sequencing confirmed CCDC66 insertion; genotyped CCDC66 (n=102 PWDs) and RNF123 (n=132 PWDs). PRCD variant genotyped as needed. - Population screening: Screened 616 additional PWDs of unknown phenotype for CCDC66 variant; also 644 dogs from other breeds. - Transcript analyses: Canine retinal RNA-seq (8 samples) analyzed with StringTie to define CCDC66 isoforms (8 transcripts; I and II >90% total). Transcripts translated (Expasy) and aligned (Clustal Omega) across species; motif/domain prediction with Pfam, InterPro, MotifScan, PSIPRED. - In vitro expression and localization: Cloned full-length canine CCDC66 (wild type and mutant c.2262_c.2263insA) into Myc-tagged expression vectors (pKMyc). Transfected COS-1 cells; Western blot probed with anti-Myc and anti-β-actin. Immunocytochemistry assessed subcellular localization (anti-Myc, Hoechst). - Retinal localization (IHC): Immunolabeled normal canine retinal cryosections with anti-CCDC66, co-labeled with Goa to mark ON-bipolar cell layer; epifluorescence imaging. - Data availability: WGS ENA accessions ERS4045175–77.

- Clinical phenotype: Four PWD full-siblings exhibited early-onset PRA (2–3 years). Signs progressed from nyctalopia to day vision impairment. Fundus showed peripapillary ring hyperreflectivity (peripapillary conus) progressing to broader peripapillary hyperreflectivity, generalized tapetal hyperreflectivity, diffuse vascular attenuation, optic disc pallor, and multifocal non-tapetal depigmentation. One dog had undetectable scotopic and photopic ERGs at 6 years. No extra-ocular abnormalities reported. - Genetic mapping: GWAS suggested CFA20 (p=1.4×10^-4) with top markers clustering on CFA20; linkage analysis yielded max LOD 2.501 on CFA20 across 20.63 Mb; homozygosity mapping identified a shared segment on CFA20 (∼21.9–48.46 Mb). An integrated 26.55 Mb candidate interval (CFA20:21,904,057–48,467,358) was defined. - Causal variant: WGS of a case-parent trio and filtering identified a 1 bp insertion in CCDC66 (CFA20:g.33,717,704_33,717,705insT; c.2262_c.2263insA) causing a frameshift and premature truncation (p.Val747SerfsTer8). Sanger sequencing confirmed the variant. - Co-segregation and population data: All four affected siblings were homozygous for CCDC66 c.2262_c.2263insA; none of the clinically normal PWDs were homozygous mutant. Among clinically assessed PWDs (n=102): 73 wt/wt, 25 wt/A, 4 A/A; A-allele frequency 16.2%. In 616 PWDs of unknown phenotype: 602 wt/wt, 14 wt/A, 0 A/A; A-allele frequency 1.1%. Total PWDs screened (n=718): 675 wt/wt, 39 wt/A, 4 A/A; A-allele frequency 3.3%. Other breeds (n=644) showed no carriers. The RNF123 c.2755C>T variant was in LD with CCDC66 in the family; in an extended normal PWD panel (n=124), no homozygous mutants were found. - Specificity relative to prcd-PRA: Three external PWD cases with early clinical onset were homozygous for PRCD and lacked the CCDC66 haplotype, likely representing early prcd-PRA. A fourth early-onset case lacked both PRCD and CCDC66 variants, indicating additional genetic heterogeneity. - Protein impact and localization: The CCDC66 insertion truncates isoform I from 951 aa to 755 aa (79.4% length of WT), deleting predicted C-terminal coiled-coil and leucine zipper motifs. Western blot in COS-1 cells showed expression of full-length Myc-CCDC66 (~115 kDa) and truncated mutant (~92 kDa). Immunocytochemistry revealed wild type CCDC66 localized to cytoplasm with punctate perinuclear labeling, whereas mutant CCDC66 mislocalized predominantly to the nucleus. - Retinal localization: In normal canine retina, CCDC66 immunolabeling was intense at the photoreceptor connecting cilia with lighter outer segment labeling and faint OPL labeling consistent with photoreceptor axon terminals. - Conclusion of causality: Perfect co-segregation in the pedigree, absence of homozygous mutants among normals, prior association of CCDC66 with retinal degeneration in dogs/mice, and mutant mislocalization support CCDC66 c.2262_c.2263insA as the causal variant for EOPRA in PWDs.

The study identifies a truncating frameshift in CCDC66 as the likely cause of a newly recognized early-onset PRA (EOPRA) in PWDs. Multiple independent mapping strategies converged on CFA20, and WGS pinpointed a single-base insertion in CCDC66 that co-segregates perfectly with disease. Functional assays demonstrated that the mutant protein is expressed but truncated and mislocalized to the nucleus, implying disruption of its normal ciliary/microtubule-associated roles. CCDC66’s localization to photoreceptor connecting cilia in normal canine retina, together with prior evidence from Ccdc66-/- mice (early photoreceptor degeneration and ciliogenesis defects) and a canine CCDC66 mutation in Schapendoes, reinforce a ciliary pathomechanism. While an RNF123 missense variant in LD was also observed, absence of homozygous mutants among normal dogs and lack of prior association with retinal phenotypes argue against RNF123 as causal. The findings demonstrate locus heterogeneity in PWD PRA (CCDC66-EOPRA versus PRCD prcd-PRA) and suggest allelic heterogeneity across breeds (different CCDC66 mutations). The distinct peripapillary conus phenotype and earlier onset (2–3 years) distinguish EOPRA from typical prcd-PRA in PWDs (3–6+ years). Given debates on human CCDC66 pathogenicity, these canine data underscore CCDC66 as a relevant candidate gene for human ciliopathy-related retinal dystrophies and highlight the need to consider tissue-specific isoforms and domain functions, especially in the conserved C-terminus affected by the PWD variant.

A combined GWAS, linkage, homozygosity mapping, and WGS approach identified a 1 bp insertion in CCDC66 (c.2262_c.2263insA; p.Val747SerfsTer8) as the cause of a new early-onset PRA in Portuguese Water Dogs. The variant truncates the C-terminal portion of CCDC66, producing a stable but mislocalized protein. Results support a critical role for CCDC66 in photoreceptor ciliary function and retinal viability. This establishes CCDC66 as the second PRA gene in PWDs after PRCD, demonstrating locus heterogeneity within the breed and allelic heterogeneity across breeds. Future work should include: (1) functional dissection of the CCDC66 C-terminus and interacting partners in photoreceptors; (2) in vivo analyses in affected canine tissues/animal models; (3) expanded genetic screening of human RD cohorts, particularly suspected ciliopathies; and (4) assessment of potential breed-specific modifiers influencing age of onset and severity.

- Affected retinal tissues were not available, limiting in vivo assessment of transcript stability, native protein expression, and retinal ultrastructure. - Small pedigree and related sample set led to population stratification and limited GWAS power; linkage evidence (LOD 2.5) is suggestive but not definitive. - RNF123 missense variant in LD could not be fully excluded by population screening alone, though evidence weighs against causality. - In vitro overexpression in COS-1 cells may not fully recapitulate photoreceptor-specific trafficking and interactions. - RNA-seq read length (2×100 bp) constrained comprehensive isoform reconstruction; minor transcripts may be underrepresented.