Introduction

Progressive retinal atrophy (PRA) is a major health concern in purebred dogs with greater than 100 affected breeds reported to date. It is characterized by a steady deterioration of bilateral retinal function over time, commonly resulting in blindness. There is currently no effective treatment for PRA. It is important to find mutations in the genes underlying PRA so that breeders can identify carriers in order to make informed breeding decisions. With the development of genetic tests, it should be possible to greatly reduce the incidence of PRA in purebred dogs.

Although it has become possible to sequence complete genomes, it is still relatively expensive to use this method to find mutations that cause PRA. Currently, a more practical and far less expensive way to search for culprit genes is to use the candidate gene approach. A common way to test candidate genes is to directly sequence them. This is also relatively expensive and time consuming, but has the advantage of requiring only a relatively small sample of affected dogs. On the other hand, pedigree-based linkage analysis requires the collection of large and nearly complete pedigrees, which can be difficult to do and is often very time consuming. We have developed an alternative approach that is fast, efficient, and cost effective and also only requires a few affected dogs from any breed to test the hypothesis that a given gene is the culprit. To work effectively, this method requires the identification of highly variable markers that flank candidate genes. We report here a panel of these markers for thirty genes that underlie recessively inherited retinitis pigmentosa or Leber congenital amaurosis (RP or LCA respectively, the human equivalents of canine PRA), and we provide an example of how candidate genes can be quickly eliminated as being the culprit, leaving only a few genes to be studied more carefully.

Methods

Thirty genes were chosen as candidates for recessively inherited PRA because of their proven association with the PRA-like diseases, retinitis pigmentosa and Leber congenital amaurosis, in humans. The UCSC Genome Browser repeat track was used to locate two unique microsatellite (MS) markers flanking the ends of each candidate gene for development of the MS genotyping panel. Microsatellite markers were designed within 1 megabase (Mb) of the gene and a majority of markers could be designed within 0.25 Mb. Primer3 was used to design the primers. Polymerase chain reaction (PCR) products were analyzed via ABI PRISM 3130 Genetic Analyzer. Four breeds of dogs (in each breed an N of 3 to 5dogs was used) were used to check the variability of the markers. Two breeds in which the mutation that causes PRA is known were used to validate the panel. Both heterozygosity and the number of alleles per marker are ways to measure variability. Heterozygosity was measured by dividing the number of heterozygotes by the total number of dogs run on the panel. The number of alleles was counted for each marker. The observation of heterozygosity among affected dogs in both markers provides strong evidence that a particular gene is unlikely to underlie PRA in a particular breed.

Results

Two highly variable markers for each of the thirty candidate genes were designed for the PRA panel. All but a few primer sets could be run using the same PCR conditions, enabling the easy use of 96 well plates for amplification. The product sizes have a wide range that allowed us to combine three different markers into each lane of the ABI PRISM 3130 Genetic Analyzer, drastically reducing the cost and time for each sample.

Marker heterozygosity ranged from 0.2 to 1.0 with the average heterozygosity above 0.6. The panel was tested in dogs where the mutant gene was already known. The correct gene was in the handful of non-excluded genes at the end of the screening.

Discussion

This MS marker panel will be useful for expediting candidate gene searches for recessively inherited PRA in purebred dogs. It is important to point out that most innocent candidate genes were excluded by using, only five affected dogs. The affected dogs do not need to be closely related, as in pedigree based linkage analysis. The probability of falsely excluding the true culprit gene using this approach is quite low (almost always less than 1/10,000). Future work will focus on developing additional markers for all genes that underlie recessively inherited RP (and, by extension, PRA). In the past the genetic heterogeneity of PRA has led to criticisms of the candidate gene approach. However there will be a finite number of candidate genes and as investigations of retinal dystrophies in humans, dogs and other species continue we will have a more complete list of potential candidate genes to screen. We expect that the use of this panel of markers and future extended versions of it will help to reduce the burden of PRA in the breeds that make up one of our favorite companion animals.