The Genetic Basis of Coat Variation in Dogs

    By Kevin E. Noonan --

The genetic age, characterized by numerous genome projects (not the least of which is the Human Genome Project) has focused the public's attention on the influence of our genes on our characteristics.  This has led to questions about whether there is a "gene" for risk-taking, or homosexuality, or aggressiveness.  Indeed, the pervasiveness of the "gene" question for complex traits has produced something of a backlash, illustrated by Richard Lewontin's books, Not in Our Genes and It Ain't Necessarily So, making the reasonable point that genetic reductionism has its limitations.

But genes are certainly implicated in a number of characteristics and traits, and one of the consequences of studying genomes has been the elucidation of these relationships.  In 2007, a report in Science disclosed that body size in domesticated dogs is associated with a particular single nucleotide polymorphism (SNP) in the insulin-like growth factor I gene (IGF-1) (see "From Toy Poodle to Rottweiler:  Why Is Fido So Small (or Large)?").  Last week, in Science, a group from the National Institutes of Health report on another canine phenotype, coat variation, and the genes that are associated with variations in several coat related traits (Cadieu et al., 2009, "Coat Variation in the Domestic Dog Is Governed by Variants in Three Genes," Science DOI: 10.1126/science.1177808).

The report, from Elaine Ostrander's lab at the NIH and researchers at the Veterinary Genetics Laboratory at the University of California, Davis, the Biology Department at the University of Utah, the Department of Ecology and Evolutionary Biology at UCLA, Cornell University's Department of Biological Statistics and Computational Biology, the Faculte de Medicine in Rennes, France, and Affymetrix, involves genome-wide association studies (GWAS) for three coat variations:  the presence or absence of "furnishings" (eyebrows and mustaches as arise in wire-haired dogs), coat length, and coat curliness.  The study involved more than 1,000 dogs from 80 breeds of domestic dogs, and resulted in the identification of three genes associated with variations in coat.  These genes are RSPO2, encoding R-spondin-2; FGF5, encoding fibroblast growth factor 5; and KRT71, encoding keratin-71.  These genes, and specifically SNPs associated with each of these genes, were found to be associated with the presence or absence of furnishings (RSPO2), length (FGF5), and curliness (KRT71).

Unlike the situation with body size, these three genes form a combinatorial group of genes that can be independently segregated (i.e., they all reside on different chromosomes) and can thus create a number of different coat variations depending on the different variant alleles inherited by particular breeds of dog (since coat characteristics are frequently important components of the breed standard for different dog breeds).  Three datasets were used in these experiments:  the first comprised DNA from 96 dachshunds varying as wire-haired (with furnishings), smooth and long-haired coats; 76 Portuguese waterdogs, varying in curl type; and 903 dogs form 80 breeds representing "a wide variety of [coat] phenotypes."  The initial results of polymorphisms segregating with the different coat traits were validated using a larger panel of at least 661 dogs from 106 breeds that included the appropriate controls.

From the dachshund (at right) studies was identified the association of RSPO2 with furnishings as a locus residing on canine chromosome 13 (CFA13).  As reported in the paper, "[a] 718 Kb homozygous haplotype in all dogs fixed with furnishings was located within both the original 3.4Mb haplotype observed in the dachshund-only GWAS, and a 2.8 Mb haplotype identified in crossover analysis within the dachshund pedigree."  From this chromosomal location the RSPO2 gene was identified:

Fine-mapping allowed us to reduce the homozygous region to 238 Kb spanning only the R-spondin-2 (RSPO2) gene, excluding the 5'UTR and the first exon (Fig. 1D; fig. S1 and table S3).  RSPO2 is an excellent candidate for a hair growth phenotype as it is known to synergize with Wnt to activate β-catenin . . . , and Wnt signaling is required for the establishment of the hair follicles . . . .  Moreover, the Wnt/β-catenin pathway is involved in the development of hair follicle tumors or pilomatricomas . . . , which occur most frequently in breeds that have furnishings . . . .  Recent studies have shown that a mutation in the EDAR gene, also involved in the Wnt pathway, is responsible for a course East-Asian hair type found in humans . . ., with some similarity to canine wirehair.

The polymorphism associated with the furnishings trait in dachshunds and confirmed in 704 dogs of varying phenotype is a 167 bp insertion within the 3' untranslated region (UTR) of the RSPO2 gene; "297/298 dogs with furnishings were either homozygous (268) or heterozygous (29) for the insertion, while all 406 dogs lacking the trait were homozygous [for 3' UTR lacking the insertion]."  The authors also report a functional consequence of the insertion, wherein there is a 3-fold increase in RSPO2 expression in muzzle skin from dogs with furnishings, suggesting that the insertion in the 3' UTR influences mRNA stability.

For the association of FGF5 with hair length, the same type of GWAS were performed, locating a polymorphism on canine chromosome 32 (CFA32) at the FGF5 locus.  Here, the polymorphism affected the coding sequence of the protein, resulting in a Cys --> Phe change at a conserved residue (95) in exon 1 of the FGF5 gene (corresponding to a G --> T transversion mutation).  This association was not unexpected, as prior work had shown that the FGF5 gene was associated with a long-haired or "fluffy" phenotype in Welsh corgis.  Long hair was associated with the TT genotype, consistent with a genetically recessive mode of inheritance.  The T allele was found in 91% of all long-haired dogs tested and in only 3.9% of short-haired dogs (with medium-haired dogs having an intermediate 30% frequency level for this allele).  Curiously, Afghan hounds that have particularly long hair do not have the Cys95Phe mutation.

Finally, for the curly coat variation GWAS were performed with Portuguese water dogs (PWD) (at left, First Dog Bo).  These studies revealed a SNP on canine chromosome 27 (CFA27) located in the keratin 71 (KRT71) genetic locus.  As reported in the paper, "Non-curly haired dogs carried the CC genotype [and] curly coated dogs had the TT phenotype," and all three genotypes (CC, CT and TT) were found in PWDs, a breed that expressed curl hair to varying extents.  This SNP is also located in the coding sequence (in the second exon) of the gene, causing an Arg --> Trp change at residue 151.  These findings were also rationalized by the authors:

Keratins are obvious candidates for hair growth . . . and mutations in KRT71 have been described in curly coated mice . . . .  The mutation described in our study is within the second exon of the gene and may affect either or both of two protein domains; a coiled-coil and a prefoldin domain.  Conceivably, sequence alterations in these domains could affect cellular targeting, receptor binding or proper folding of the protein after translation.

These three mutations were sufficient to account for the coat phenotype in 95% of the dogs tested, a total of 622 dogs representing 108 of the 160 dog breeds recognized by the American Kennel Club.  The Supplemental Materials contain a combinatorial diagram of the different variations at the three loci that give rise to at least 7 different coat types (representing the majority of coat types in modern breeds):

* The TT genotype was never found in this combination, even though it would likely also display a short phenotype.

The authors set forth the following conclusions.  First, the "ancestral state" of all three genes (Cys95 in FGF5, Arg151 in KRT71, and absence of an insertion in the 3' UTR in RSPO2) is found in short-haired dog breeds (as well as grey wolves, the ancestors of all modern dog breeds).  Wire-haired breeds (which all have furnishings) have the insertion in the RSPO2 gene.  Curly wire hair is found in dogs having both the RSPO2 insertion and the KRT71 mutation, while longer-haired dog breeds have the FGF5 mutation.  Dogs carrying both the FGF5 mutation and the RSPO2 insertion have long, soft coats and furnishings, whereas dogs with the FGF5 and KRT71 mutations have long, curly coats.  Finally, dogs bearing all three of the variants have long, curly coats, and furnishings.

Elucidation of this genetic architecture of observable phenotypic traits is possible because of the large number of different dog breeds and the great variability, on one hand, and the consistent stability, on the other hand, of genetic transmission in these breeds.  This study suggests that these analytical methods can (and will) be applied to assess the genetic bases of other traits in dogs, and the extension of these results to homologous phenotypes in other mammals including humans.

Photograph of mixed breed terrier (above) by Chris Barber, from the Wikipedia Commons under the Creative Commons license.