The Lhasa Apso PRA Project
in the University of Cambridge 

reproduced by kind permission of Jesus Aguirre-Hernandez


Figures Related to the Disease

By April 2001, 37 Lhasa Apsos affected with progressive retinal atrophy (PRA), and distributed in six countries, had been reported to ILIAC. In addition to these, several unreported individuals are known.

In some countries, such as Sweden, affected individuals have died, so the number of countries with living affected individuals has decreased.

PRA is a recessive disease. This means that affected animals have two copies of the mutant gene, while carriers have one normal copy and a mutant one. Each of the versions of a gene, in this case the normal and the mutant, is called an allele.

From the number of eye-tested individuals, both affected and non-affected, it is possible to estimate the frequency of the mutant allele in the population, as well as the proportion of carriers. We know that approximately 2% of all eye-tested individuals are affected. From this we can deduce that in the countries where PRA has been detected, 25% of the population is composed of carriers.


Locating a Disease-Gene

When searching for a gene associated with a disease the first problem that arises is where to start looking for it: dogs may have around 40 000 genes (the precise figure being unknown), and they are distributed among 37 pairs of canine chromosomes plus two sex chromosomes (X and Y) that contain enough genetic material to span at least 1 meter. In principle, the gene underlying a disease may be anywhere in that genome, This illustrates the formidable task faced when searching for them. However, human ingenuity has managed to devise several ways of finding those genes.


In principle, there are three methods for finding the gene involved in PRA in Lhasa Apsos, and we have been working with the three of them:

· the candidate-gene approach,

· linkage analysis, and

· linkage disequilibrium.


1. The Candidate-Gene Approach

This involves searching for mutations in genes which can reasonably assumed to be involved in PRA. Usually the genes that are studied are chosen because they are known to be involved in similar diseases in other species, such as retinitis pigmentosa in humans, or because they are known to be expressed in the retina, due to studies using model animals.

This method has allowed dog geneticists to identify all but one of the canine genes currently known to be involved in diseases. In our laboratory, it has enabled us to identify two PRA genes known to date: PDEB in the Irish setter and PDEA in the Cardigan Welsh corgi.

We have performed some experiments with the candidate-gene approach during the last few months in order to find the PRA-gene in Lhasa Apsos. However, we have not found any abnormalities in the genes we have screened. This approach is relatively inefficient, since it requires testing many genes until the mutated one is found. The ratio of success to failure is very low. This situation has prompted us to try other methods.


2. Linkage Analysis

This is a more powerful procedure. It has been used in very few studies since it requires a map of the canine genome, which has only been available since very recently. This map is a set of sequences with precise locations in the genome, and each one different from all the others. These sequences are called markers. Markers vary between individuals, reflecting the genetic differences between them. As with genes, each variant of a marker is called an allele.

In linkage analysis, the pattern of inheritance of markers distributed along the genome, in known locations, is analysed in a multigenerational pedigree. If one of these marker sequences shows a pattern of inheritance resembling the one we expect for the allele causing the disease, then we may reasonably conclude that that marker is close to the gene involved in the disease. The only reason (apart from chance or coincidence) why the pattern of inheritance of one such sequence may match the pattern expected for the disease allele, is because they are physically close, and thus are inherited together.

We have done experiments using this procedure. Results show that the method cannot be applied usefully to the Lhasa Apsos available to us. The reasons for this is are twofold.

Firstly, the affected Lhasa Apsos are not clustered together in a few families. Instead of this, there is generally only one affected individual per family. Affected siblings are extremely uncommon, as well as affected individuals in consecutive generations of the same family. Given this distribution of affected animals, it is not possible to infer the inheritance of the disease allele in these families. This means that when the patterns of inheritance of markers throughout the genome are analysed, there is no pattern to which to compare them (since it is impossible to deduct the pattern of inheritance of the disease allele).

Secondly, even in the best families we have for study, we are missing crucial individuals who would contribute information. In a late-onset disease, such as PRA in Lhasa Apsos, it is impossible to get grandparents, and we have also had problems tracing siblings of affected individuals. Each missing individual in a family reduces our power to analyse it.


3. Linkage Disequilibrium (LD)

This is the method to which we are shifting. It requires, as the preceding one, a map of the canine genome.

We expect that alleles in the disease locus (locus is the location in the genome of a gene or any other sequence) will be distributed differently in affected and non-affected individuals: all affected individuals will have two copies of the disease allele, and no copies of any other allele, while non-affected individuals may have different combinations of alleles for that gene, with the only restriction that they will not have two copies of the disease allele. In LD, therefore, we study the distribution of alleles for many markers, among affected and non-affected animals, until we find one with a distribution resembling the one described above for the disease locus. If a marker sequence shows the distribution we expect for the disease locus, it must be because they are close to each other, and hence they are inherited together. The only other explanation would be coincidence, but there are ways to rule this out.

LD is based on the idea that the mutation that gave rise to the disease allele arose only once in the population and since then it has been transmitted to successive generations. This means that all individuals having the mutant allele are descendants of a single ancestral individual in which the mutation took place, and they are all relatives. When these related individuals are bred, both parents may transmit the mutant allele to their offspring which will have the disease. Therefore, affected individuals will have two copies of the same mutant gene that originated in an ancestor common to both parents.

As mentioned, LD works by studying sequences along the entire canine genome to see if any one of them is identical in all affected individuals, as would happen with the mutant allele. In very close relatives, a very large proportion of their genomes will be identical, since it will have been inherited from a very close ancestor common to both of them. For more distant relatives, the proportion of their genomes that is identical will be smaller. The larger the number of generations that separate two relatives from a common ancestor, the smaller the proportion of their genomes that will be identical. As generations pass, less and less of the genome will be identical for distant relatives. Less and less of the genome will have been received from the same distant ancestor, and more and more of the genome of these distant relatives will be due to ancestors not shared by them. This means that as generations pass, the length of the chromosome fragments that are identical for these relatives becomes smaller. In order for alleles of a marker to show the distribution we expect for the disease allele, both, the marker and the disease locus, must reside in the chromosome fragment that is identical in all affected individuals and that has remained intact since their common ancestor. The more distant two or more affected individuals are from their common ancestor, the smaller the chromosome fragment identical in both of them; this means that the marker and the disease locus must be very close to each other if they both are to be in that same chromosome fragment.

We have just finished the analysis of the pattern of inbreeding in the families of affected Lhasa Apsos for which we have blood samples. Except for one family, the parents of all affected individuals share a set of three ancestors. This does not mean that the mutation arose in any one of them. In fact, the presence of that single family not sharing any of these three ancestors suggests that the mutation arose in a still more distant ancestor, common to this and to the rest of the families. It has not been possible to identify this ancient ancestor.

A maximum of 8 generations separate the affected individuals from those three common ancestors. With this information, it is possible to estimate the average length of the chromosome fragments identical for all affected individuals. This length may then be compared to the average spacing of the markers in the canine genome. If this spacing is smaller than the estimated length of the chromosome fragments conserved intact since the mutation arose in the common ancestor, then there will be at least one marker within this conserved fragment. This is crucial, since only markers located in the chromosome fragments that are identical for all affected individuals, as well as to the common ancestor, will show the same distribution pattern as the disease locus. Our results show that average spacing between markers in the current canine genome map is approximately the same as the length of the chromosome fragments identical for all affected individuals. This means that there may be one marker close enough to the disease locus, such that both, the marker and the disease locus, belong to a chromosome fragment that has remained intact since it was originated in the ancestor common to all parents of affected individuals.

The fact that the average length of identical chromosome fragments in all affected individuals (fragments that have remained intact despite the several generations that separate all of them from their common ancestor) is approximately the same as the average distance between markers on the current canine genome map, means that it may be possible to find a marker with the same distribution as the disease allele, although the chances are slim. If the average spacing of the markers on the canine genome were smaller (e.g., a more dense set of markers) then the chances of finding such pattern would increase, since there would be more than one marker within the identical conserved chromosome fragment, and so more opportunities for finding the distribution pattern we are looking for.


Given this situation, a first attempt will be made to find a marker with the same distribution pattern as the disease allele, using the set of available markers on the canine genome map.

In parallel to this work, in the course of the following months a more dense set of markers will be developed, and then they will be used to detect that pattern.



Use the 'Back' button on your browser to return to the previous page.