Why do we selectively breed dogs

genetics of our "best friend"

These genome-wide association studies are much more informative in the dog. While humans need around 1 million SNPs distributed over the genome, only around 15,000 to 30,000 SNPs are necessary to cover the entire genome for the analysis of a certain characteristic of a dog breed. The reason for this is the much more pronounced uniformity of the genome within a dog breed. This breed was established less than 200 years ago with a few dogs, principally only one pair of dogs. Only about 50 to 200 generations have passed since then. Accordingly, this genome is only recombined by a few recombination or cross-over events. Large portions or fragments of the genome have not yet been recombined at all, that is, the identical SNP arrangement exists for these fragments in all dogs from this dog breed (Fig. 3). The technical term for the identical SNP arrangement within these fragments is called linkage disequilibrium. The arrangement of the SNPs on a single fragment is always identical for dogs from one breed, because they have not yet been separated from one another by a cross-over. In dogs, these fragments are approximately 1 million base pairs long. In humans, around 10,000 generations have passed since the emergence of modern humans 200,000 years ago. Accordingly, these fragments are much smaller, they are around 25,000 base pairs long. Therefore, a much larger number of SNPs and much more complex calculation methods are required for the investigation in humans. It has been shown that this disadvantage in humans cannot be compensated for by a larger computational effort. The situation is completely different with the dog. For simple muddling traits, you only need 10 to 20 dogs and a chip with around 15,000 to 30,000 SNPs to map the trait in the genome with absolute certainty. With this first step, the trait is relatively roughly assigned in the dog's genome. In a second step, animals of other dog breeds that happen to show this characteristic are examined for an exact fine mapping. That sounds surprising at first because it is believed that this mutation only occurs in the specific breed of dog. But you have to know that it was not new mutations that led to today's breeds, but that the selective breeding of variants already existing in the dog species led to the uniform breed. The very different breeds of dogs reflect the natural variation in the starting population of the dogs. The peripheral isolates, which have often been examined, e.g. in Iceland or Finland, represent a comparable situation in humans. However, they are still very heterogeneous and much less informative than dog breeds. For the second step in the context of gene mapping, you now need a new chip on which the gene region recognized in the first step is now represented with many closely spaced SNP markers.

Figure 3: Schematic representation of the coupling imbalance in humans (above), dogs in general (middle) and within a dog breed (below). The horizontally drawn rod represents a DNA molecule. The colored marking shows a fragment with a conserved SNP arrangement (coupling imbalance). This is generally smaller in dogs than in humans, since more generations have passed and the wolf as a starting species is genetically more heterogeneous than the human population was 200,000 years ago. The members of a dog breed are still very little mixed up with cross-over events. They have the identical SNP arrangement for large fragments and it is sufficient to test a single SNP for around 1.0 million base pairs (below).

If you examine different breeds of dogs in this second step, you go back to the origins of dogs 15,000 years ago and the number of generations since then is correspondingly large. This means that the fragments with the coupling imbalance become very small, they are actually even smaller than in humans (~ 10,000 base pairs, Fig. 3). In any case, you get an assignment to a gene, or to a very manageable DNA sequence, which can be sequenced and compared between dogs with different characteristics. This two-stage approach of genome-wide association analysis has proven to be very effective. Table 1 shows examples of the genes or loci identified with this method for certain diseases in dogs. The second part of the table lists tumor diseases for which there is also a breed-specific predisposition. In general, these breed-specific risks show what a large part the family and thus genetics contribute to the etiology of diseases. This also applies to humans.

Mapping opens up new perspectives

In the meantime one has also experience with the mapping of complex features in dogs. For example, as previously described for humans, body size was also examined. In a first step, 463 dogs of different sizes of one breed were examined and in the second step, after the region had been determined in the first step, only 17 dogs from different breeds were examined. With IGF1, insulin like growth factor 1, a gene locus has been determined that explains 15% of the genetic variation in body size in dogs. The corresponding effort in humans has already been mentioned. It should be added here that for the mapping of complex diseases not only the genomic prerequisites in dogs are much more favorable than in humans, but also the complexity of a trait or a disease in closed populations or isolates is much lower and therefore the people involved Genes can be recognized much better with these methods. It should not be concealed that the great advantage of GWA studies is the complete impartiality with which one approaches the understanding of diseases with this method. All genes are checked from the start. This can lead to the surprising discovery of new mechanisms for the development of diseases that no one has thought of before. E.g. in hairless dogs it is an ectodermal dysplasia with simultaneous tooth anomalies (Fig. 4). This disorder in dogs has many parallels with anhidrotic ectodermal dysplasia in humans, for which mutations in genes of the ectodysplasin signaling pathway are known. A GWA study in dogs now also makes the transcription factor FOXI3 responsible for this phenotype, which points in a completely new direction (working group Tosso Leeb, Bern). The 180 loci that contribute to height were mentioned earlier. This also includes structural proteins, enzymes and signaling pathways in the metabolism that have never been considered before.

Figure 4 (a): An example of the hairless dogs (a) composed
with a dog with normal coat (b).

Figure 4 (b): An example of the hairless dogs (a together
with a dog with normal coat (b).

Figure 4 (c): In (c) the tooth anomalies in the hairless
Dogs shown. This canine ectodermal dysplasia has many
Parallels with anhidrotic ectodermal dysplasia in
People up.

New therapy options

The dog also offers advantages for therapeutic approaches because of its physiology, which is much more similar to humans (size, age, same environment). For example, the mouse with a mutation in the Duchenne muscular dystrophy gene does not suffer from this disease in any way. Either it does not live long enough (about 2 years) or the much shorter muscle cells can better compensate for the defect in the muscle sarcolemma. Corresponding genetic defects have been discovered in the Golden Retriever in dogs. The young male dogs are visibly affected by the age of 4 to 6 months and die by around one year. A surprising variant of gene therapy was tested on this model, using a culture of mesangioblasts that grow out of a muscle biopsy en masse in vitro. Mesangioblasts are mesodermal progenitor cells. These mesangioblasts are transfected in vitro with a modified dystrophin gene and these gene-treated cells are injected into the dog's blood vessel system. These me-normal precursor cells migrate independently from the capillary system and thus find their own way into the muscle, into which they migrate, in order to then differentiate into muscle cells in this environment.
In fact, the already visibly sick young dog becomes a joyful comrade again for a certain period of time. The treatment of Leber's congenital amaurosis, a blindness in early childhood, is much more advanced. One form concerns the vitamin A metabolism (RPE65) and occurs in a genetically identical form in dogs. With a subretinal injection of a recombinant wild-type RPE65 gene, one can achieve visual improvement that lasts for up to three years (previously). In particular, it can be proven in dogs that not only a purely physical improvement in vision is achieved, because the dog's behavior is also normalized. It is precisely this latter aspect that makes the dog model particularly attractive. Typical diseases of the elderly, such as behavioral disorders, dementia, but also tumor diseases, which are difficult to investigate in humans solely on the basis of time, run about 5 times faster in dogs and can therefore be worked on in the time span of a research career.

Insight into the genetics of behavior

We got a little insight into the really very complex genetics of behavior from a close relative of the dog, namely the fox, more precisely the Siberian silver fox. This fox is bred for its beautiful fur, but it causes great problems due to its aggression (bite). As a result, about 30 generations were bred for tame behavior and received a fox that you can cuddle with on your arm, but this fox no longer has the beautiful fur and, to the great surprise, the shape of the head has also changed. One learns from this that the genes responsible for behavior also fulfill many other vital functions during early development and during metabolism.

Final remark
I would like to conclude with a remark for animal lovers. For these genetic examinations you only need a single blood sample of a few milliliters from the individual dogs, which does not stress the dogs. In the long term, animal and human health will benefit from these comparative studies.
I thank Prof. Dr. Tosso Leeb, Institute for Genetics, University of Bern, for many suggestions and for reviewing the manuscript.