In the present study, we described a new canine retinopathy. The clinical form described here is clearly characteristic of a progressive retinal atrophy (PRA): the onset of symptoms occurred at around three years of age, and peripheral vision was then progressively lost over the next years. The appearance of the fundus was compatible with PRA with distinct peripheral pigmentary changes, optic pallor, and attenuated retinal vessels. ERG responses were suggestive of the disease in two-year-old dogs. The predominant symptoms of night blindness, peripheral field defects, and absence of light sensitivity strongly suggest that this is a form of PRA [25, 26].
All the clinical signs that we observed (menace responses, pupillary light reflexes, fundus examination) were evocating a PRA:
focal lesions were evoluting from one eye to both in cases of unilateral lesions at the first examination;
either focal or diffuse bilateral lesions were evolving towards a generalized disease of he fundus in re-examined dogs, with PRA compatible signs: tapetal hyperreflexion, vascular attenuation, palor and enlargement of the optic disc;
vacuolar subcapsular equatorial cortical cataracts observed in some dogs with severe fundic lesions were described in other canine breeds as secondary to PRA [6, 27];
the most important was midriasis, the most incomplete, unstable and the slower was the pupillary light reflex, with disappearing of the menace response in cases with severe fundic lesions.
If only focal isolated lesions were observed, there was no modification of the ERG. In cases of multiple or generalized fundic lesions, the hypovolted traces (sometimes non recordable in blue stimulation) were in favour of a progressive rod and cone degeneration.
We never found either ophthalmoscopic, functional or angiographic signs which correlated with CPRA: this disease was first described in several breeds, considered as recessive autosomal in the Briard shepherd dog, dominant autosomal with incomplete penetrance in the Labrador Retriever, the Shetland sheepdog [25–27], dominant autosomal with incomplete penetrance  or polygenic  in the Border Collie. Second, it appears that environmental factors were involved in CPRA: experimental Vitamin E deficiency exactly mimicked the same fundic lesions in the dog . Nevertheless, in Border Collies, the results of eye examination scheme in the United Kingdom remained in favor of an hereditary origin: in 1965, 12% of the dogs were affected of CPRA; in 1989, only 2% were exhibiting ophthalmoscopic changes characteristic of the disease [25, 26]. All the findings of FA were demonstrating that the primary changes in the fundus were not related to a pigment epithelium dystrophy: we never found the specific lesions of fluorescence mottled masking by lipofuscinic like pigment deposits . We showed that this PRA segregates according to an X-linked mode of transmission. This is compatible with the status of the few affected females identified. Indeed, the six females found affected, after clinical examination, were diagnosed between 18 months and 3 years old and clinical symptoms were very similar to those observed in males. Moreover, all females presented similar clinical features. For three of them, we do not have any genealogical data and thus we do not know the status of the parents. For the other three affected females, which are in the pedigree, they were all born from an affected father and an obligate carrier female. Considering the high level of inbreeding in the breed, we assume that affected females come from obligate carriers and affected males, supporting a fully penetrant recessive inheritance mode. Genetic tests carried out on subsets of dogs excluded the involvement of known mutations.
We excluded the XLPRA1 and XLPRA2 mutations as the possible causal mutation for this X-linked PRA. However, owing to differences in the clinical signs and age-of-onset observed with XLPRA2, the second ORF15 mutation in RPGR could likely be excluded . The CEA mutation is frequent in the Collie lineage, but occurs at only low frequency in the Border Collie breed in France (Chaudieu, personal communication). We nonetheless tested for the presence of the CEA mutation that could co-segregate. As expected, the dogs analyzed did not carry the mutation, consistent with the low prevalence of CEA in the Border Collie. Finally, we also tested for the presence of the mutation for PRCD, which has been involved in many dog breeds displaying PRA-prcd, while these breeds have no evident common origin . Neither CEA, nor prcd were found to co-segregate in affected Border Collies.
In human, retinitis pigmentosa occurs and presents clinical features similar to canine PRA. Human cases experience a decrease in peripheral and night vision in their twenties, then night blindness and diminution of the visual field in their thirties, but visual acuity and color vision may remain normal until advanced stages of the disease . Clinically, X-linked RP are the most severe forms of RP  but only 2 genes have yet been identified on the X chromosome. Human families are often too small for the identification of the causal loci by linkage studies and families cannot be grouped for genetic linkage analyses because of possible genetic heterogeneity. Indeed, several causal genes have already been identified in inbred families . As dog pedigrees are highly inbred, it may be a useful model to identify new candidate genes for X linked RP.
The pedigree used for the segregation analysis is currently being completed for genetic linkage analysis to search for the genetic locus involved in the disease. The XLPRA1 and XLPRA2 mutations are not involved in this XLPRA, but we cannot exclude the involvement of other mutation in the RPGR gene. Indeed, this is the only gene involved in canine XLPRA with two mutations presently described in exon 15. In human, RPGR is frequently and differently mutated in XLRP. Other human candidates, such as the XLRP gene (RP2) or loci (RP6, 23, 24, 34) [14–19] may be analysed. We expect linkage analysis to be more efficient to test the involvement of genes and loci, as gene sequencing cannot unequivocally exclude the involvement of a gene if no mutation is found especially as some regions are usually unscreened, such as introns, 5' and 3' UTR or other regulatory regions.