The presence of one or two blue eyes, without large white patches on the coat, is among the aesthetic traits that have recently become very popular in domestic cats. Cat breeders have selected this trait to create a breed (Abitbol, Couronné, et al. 2024) and various breeding lines (Abitbol, Dufaure de Citres, et al. 2024). Since the recent publication of the first feline variant of the PAX3 (Paired Box 3) gene associated with a phenotype of blue eyes and minimal white spotting (DBE or Dominant Blue Eyes; Abitbol, Couronné, et al. 2024), three further variants of this gene have been reported in cats (Abitbol, Dufaure de Citres, et al. 2024; Rudd Garces et al. 2024; Abitbol et al. 2025). All variants are characterised by an autosomal dominant inheritance with variable expressivity and incomplete penetrance for both minimal white spotting and blue eyes, and pleiotropy (deafness and lethality). However, these four variants do not account for all the different feline breeds and lines that exhibit DBE (Abitbol, Dufaure de Citres, et al. 2024).
To initiate the search for the other DBE variants, we collected DNA and phenotypic data from 71 DBE cats from various lineages (for a description of the lineages, see Abitbol, Dufaure de Citres, et al. 2024), and from 69 control cats from various breeds and lines (Appendix S1). We present the study of two DBE breeding lines established from two founding females, Nadeya and Marusya (Figure 1a,b).
[图:PAX3 variants associated to DBE in two lineages of domestic cat. (a) DBE phenotype in the founding female of the Nadeya ]

The Ermine Trace Nadeya line was founded in Russia in 2016 using a purebred British shorthair female named Nadeya (Figure 1a), born to two registered British shorthair parents: a black silver shaded colourpoint male with no white and blue eyes due to colourpoint, and a black silver shaded mother with no white and green eyes. This line was not outcrossed and is maintained within the British genetic background. The DBE phenotype in the line is characterised by heterochromia or two blue eyes, minimal to no white spotting and the presence of latent cats (see Abitbol, Dufaure de Citres, et al. 2024). Three deaf cats were recorded among tens of DBE kittens that were produced. Pedigree data confirmed the dominant inheritance of DBE (Figure 1c). Repeated matings between DBE cats never produced abnormal kittens. The proportion of DBE and non‐DBE kittens in litters suggested that homozygotes were lethal in utero (Table S1). Forty‐one cats from the Nadeya line (including 20 DBE and 21 controls) were genotyped using the Illumina Feline 63 k SNP array (Appendix S1), among which 60 611 SNPs yielded usable results (minor allele frequency MAF > 5%, genotyping rate > 95%). All 41 cats had genotyping rates > 95%, and all were conserved for the analysis. Following case control analysis using a dominant model, the 25 highest significant associations were identified for 25 SNPs, among which 11 markers were located on chromosome C1 (Appendix S1, Figure 1e, Table S2). The 11 SNPs from chromosome C1 were located between position 203 566 121 bp and position 212 616 786 bp according to Felis_catus 9.0 reference genome (Table S2) and between position 202 431 315 bp and position 211 450 929 bp according to the new Felis.catus_Fca126_mat1.0 reference genome and defined the candidate region. This region contains a strong candidate gene, PAX3 (Paired Box 3, chromosome C1: 205 766 080 bp −205 862 770 bp, Felis.catus_Fca126_mat1.0 reference genome). We had previously sequenced the exons and intron‐exon boundaries of PAX3 for a DBE‐Nadeya cat but were unable to identify any coding variant (Abitbol, Dufaure de Citres, et al. 2024). We thus analysed the whole genome sequence of a Nadeya‐DBE cat obtained from paired‐end reads (2 × 150 bp) collected from a shotgun DNA fragment library (achieving genome‐wide coverage of 30X) to screen for non‐coding variants in PAX3, PAX3 regulatory sequences and across the candidate region (Table S3). We specifically searched for insertions and deletions (structural variants) as we had previously identified non‐coding variants associated to DBE in two feline lines (Abitbol, Couronné, et al. 2024; Abitbol, Dufaure de Citres, et al. 2024), and conserved elements located in the vicinity of PAX3 have been identified to be critical for PAX3 expression (Degenhardt et al. 2010; Moore et al. 2013). Two structural variants located in the candidate region drew our attention (Table S3). To confirm and characterise these variants, we analysed the whole genome data of a second Nadeya‐DBE cat obtained from 30 Gigabases PacBio single‐molecule real‐time sequences (Appendix S1). We identified the two variants as a complex deletions‐inversion NC_058375.1:g.[205097674_207411974del;207411975_211290497inv;211290498_211387174del] starting at position 205 097 674 bp, spanning 6 289 500 bp (Felis.catus_Fca126_mat1.0 reference genome) and encompassing PAX3 (Figure 1f,g).
We then genotyped 39 DBE cats and 47 non‐DBE control cats from the Nadeya line and individuals from other DBE lines (see Abitbol, Dufaure de Citres, et al. 2024) for the complex variant using primers drawn across the break points (Table 1, Appendix S1, Figure 1g). None of the cats that did not belong to the Nadeya lineage carried the variant (Table 1). All 39 DBE cats related to Nadeya were heterozygous for the variant. Seven cats whose DBE status was unknown due to their colourpoint coat pattern were genotyped. Three of them were heterozygous and four were wild‐type. The 47 non‐DBE British cats were wild‐type (Table 1). None of the cats were homozygous for this variant, strengthening the hypothesis that the homozygous genotype is lethal. Nadeya, the purebred British founder, was also heterozygous for the variant. Segregation of the variant in the line is consistent with dominant inheritance and does not appear to be associated with increased risk of deafness according to breeder's data. We observed a perfect genotype–phenotype correlation (Table 1, Figure 1c). The NC_058375.1:g.[205097674_207411974del;207411975_211290497inv;211290498_211387174del] variant consisted of a central inversion starting at position 207 411 975 bp and ending at position 211 290 497 bp, flanked by two deletions. The first deletion starts at position 205 097 674 bp and ends at position 207 411 974 bp. It includes PAX3 and 13 described genes (Felis.catus_Fca126_mat1.0 reference genome; Figure 1f,g; Table S4). Among these 13 genes, FARSB (Phenylalanyl‐TRNA Synthetase Subunit Beta) and MRPL44 (Mitochondrial Ribosomal Protein L44) have been associated with another phenotype that was not found in the cats studied here (Table S4). In both cases, deleterious variants were described as autosomal recessive (Table S4). The second deletion starts at position 211 290 498 bp and ends at position 211 387 174 bp (Figure 1f,g). No gene is described in that region (Felis.catus_Fca126_mat1.0 reference genome). As no deleterious phenotypes have been reported in DBE cats from the Nadeya line, it is likely that the loss of a functional allele for the 13 genes included in the proximal deletion is not responsible for haploinsufficiency, unlike the PAX3 gene, for which phenotypes have been associated with dominant alleles (Abitbol, Couronné, et al. 2024; Abitbol, Dufaure de Citres, et al. 2024; Abitbol et al. 2025; Rudd Garces et al. 2024; Boudjadi et al. 2018). In utero lethality in homozygotes is likely linked to the absence of a functional PAX3, as has been observed in other homozygous loss‐of‐function variants in PAX3 (Rudd Garces et al. 2024; Boudjadi et al. 2018). However, the involvement of one or more of the other genes, either individually or in combination, cannot be ruled out. Based on the perfect genotype–phenotype correlation and the nature of the identified variant, which is likely to lead to a PAX3 haploinsufficiency, we propose that this NC_058375.1:g.[205097674_207411974del;207411975_211290497inv;211290498_211387174del] variant represents the DBE NAD (Nadeya Dominant Blue Eyes) allele in the domestic cat.
TABLE 1. Genotypes for the DBE NAD and DBE MARU alleles.
| Phenotype | Lineage/Breed | Number | DBE NAD * | DBE MARU ** |
|---|---|---|---|---|
| DBE | Nadeya | 39 | Heterozygous | 3 C/C among 3 tested b |
| Unknown (colourpoint cats) | Nadeya | 7 | 3 heterozygous 4 wild‐type | |
| Control | Nadeya/British | 47 | Wild‐type | |
| DBE | Marusya | 14 | 2 tested b 2 wild‐type | 14 C/T |
| Dead kitten a | Marusya | 1 | Wild‐type | T/T |
| Control | Marusya | 1 | C/C | |
| DBE | Igor Azur Dream | 3 | Wild‐type | 1 C/C among 1 tested |
| DBE | Malvina | 7 | 4 tested 4 wild‐type | 4 C/C among 4 tested |
| DBE | Celestial | 3 | Wild‐type | 2 C/C among 2 tested |
| DBE | Rociri Elvis | 1 | Wild‐type | C/C |
| DBE | Pillowtalk‐Foreigner | 1 | Wild‐type | C/C |
| DBE | Agostino | 2 | Wild‐type | 1 C/C among 1 tested |
| Control | Bengal | 1 | Wild‐type | C/C |
| Control | Birman | 1 | Wild‐type | C/C |
| Control | Chartreux | 2 | Wild‐type | C/C |
| Control | Domestic shorthair | 1 | Wild‐type | C/C |
| Control | Devon Rex | 1 | Wild‐type | C/C |
| Control | Donskoy | 1 | Wild‐type | C/C |
| Control | Sphynx | 3 | Wild‐type | C/C |
| Control | Siamese | 2 | Wild‐type | C/C |
| Control | Siberian | 2 | 1 tested 1 wild‐type | C/C |
| Total | 140 | |||
| Control | 99 Lives dataset | 348 | C/C |
The Cyrridwen Marusya line was founded in Russia in 2019 using a female outbred domestic shorthair named Marusya, with no known ancestry, and Exotic shorthair, Persian and British cats (Figure 1b). No data is available regarding the hearing status of DBE cats in this line.
As with the Nadeya line, the DBE phenotype is characterised by heterochromia or two blue eyes, minimal to no white spotting and the presence of latent cats. Pedigree data confirmed the dominant inheritance of DBE in the line (Figure 1d). A single DBE × DBE mating gave birth to an abnormal kitten that died at birth and was supposed to be homozygous for the DBE variant segregating in the line. A candidate gene approach conducted by Sanger sequencing of PAX3 exons and intron‐exon boundaries (Appendix S1) in two DBE cats from this line revealed a nonsense variant in the fifth exon of PAX3 (Table S5). This XM_019838731.3:c.784C>T variant was predicted to produce a truncated PAX3 protein: XP_019694290.1:p.(Arg262*) that would lack 46% of the amino acids of the wild‐type protein (Figure S1). We then genotyped 16 cats from the Marusya family for this XM_019838731.3:c.784C>T variant (Appendix S1). All 14 DBE cats that were born to a DBE parent and a non‐DBE parent were heterozygous for the variant (Table 1, Figure 1d). The non‐DBE cat from the family was wild‐type, Marusya the founder female was heterozygous (Figure 1d) and the abnormal dead kitten born to two DBE parents was homozygous for the variant (Table 1). We searched for the XM_019838731.3:c.784C>T in the genomes of the 99 Lives cat whole genome sequencing project (Buckley et al. 2020; Lyons et al. 2021) but did not detect the variant (Table 1). Finally, 10 DBE cats from other lines were wild‐type (Table 1). We observed a perfect genotype–phenotype correlation for XM_019838731.3:c.784C>T in cats from the Marusya line and segregation of this nonsense variant is consistent with dominant inheritance (Figure 1d). We propose that this XM_019838731.3:c.784C>T variant of PAX3 represents the DBE MARU (Marusya Dominant Blue Eyes) allele in the domestic cat. Along with the DBE NAD allele it expands the panel of PAX3 alleles in cats.
The identification of these two variants will make it possible to offer genotyping tools to cat breeders, but further research is required to fully characterise the complex DBE NAD allele and to identify the other feline DBE variants.
Author Contributions
Tess Eiselé: investigation, data curation, resources. Marie Abitbol: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, project administration, data curation, supervision, resources. Elizaveta Lipovenko: investigation, data curation, resources.
Funding
This work was supported by LOOF‐Agria, DBEcat.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
[内容请查看原文]
Acknowledgements
Authors wish to thank the 99 Lives Cat Genome Consortium for sharing feline genome sequence information. Authors also thank breeders and owners for providing samples, pedigree data and pictures, especially Sylvie Mitteaux‐Martin, Florence Lalliard and Nataliya Zalesskaia. This project was funded by the LOOF‐Agria fund (Livre Officiel des Origines Felines, the French feline studbook: www.loof.asso.fr). Feline DNA samples are part of the Feli‐DNA biobank, which is part of the CRB‐Anim infrastructure (ANR‐11‐INBS‐0003, in the framework of the ‘Investing for the Future’ program (PIA)). Open access publication funding provided by COUPERIN CY26.
Data Availability Statement
SNP genotyping data were deposited at OSF (https://osf.io/nex5f/overview). Illumina short‐read whole genome sequence of a DBE‐Nadeya cat was submitted to SRA. Accession numbers are (BioProject ID: PRJNA1073398, BioSample ID: SAMN61108699). PacBio long‐read whole genome sequence of a second DBE‐Nadeya cat was submitted to SRA. Accession numbers are (BioProject ID: PRJNA1073398, BioSample ID: SAMN61111837).
References
[内容请查看原文]
Associated Data
Supplementary Materials
[内容请查看原文]
Data Availability Statement
SNP genotyping data were deposited at OSF (https://osf.io/nex5f/overview). Illumina short‐read whole genome sequence of a DBE‐Nadeya cat was submitted to SRA. Accession numbers are (BioProject ID: PRJNA1073398, BioSample ID: SAMN61108699). PacBio long‐read whole genome sequence of a second DBE‐Nadeya cat was submitted to SRA. Accession numbers are (BioProject ID: PRJNA1073398, BioSample ID: SAMN61111837).
兽研