Canine Genetics Canine Origins Our canine studies are divided into genomics of domestication, morphologic variation, and disease gene mapping. With regard to the first we have published recent, high-profile papers that describe the domestication of dogs (vonHoldt et al., 2013). This builds on previous work, published in Nature, where we presented a refined description of how breeds are related to each other (vonHoldt et al., 2010). Very recently, in a collaboration led by Dr. John Novembre at UCLA, sequencing of wild canids revealed a severe 30-fold reduction in effective population size that likely occurred during the domestication bottleneck, which we now estimate occurred about 15,000 years ago (Freedman, Submitted). These data also show that none of the extant wolf lineages from putative domestication centers are directly ancestral to dogs, implying that a now extinct population of wolves may be the missing link in dog evolution. Finally, the study demonstrates selection on genes affecting brain function, metabolism, and morphology, supporting a major role for regulatory evolution in domestication (Freedman, Submitted). Morphology A majority of our dog papers over the past four years reveal our growing understanding of canine genome organization and its relationship to morphologic variation between breeds. We have, for instance, continued our studies of body size, identifying IGF1R as a major gene controlling very small body size (Hoopes et al., 2012). Continued studies in this area focus on five other loci, identified in a large genome wide association study (GWAS) by us, that appear to be important in body size, and which are currently undergoing fine mapping and statistical analysis. A new avenue of morphologic study is aimed at understanding the genetic underpinning of skull shape variation (Schoenebeck, 2013). Skull size and shape vary dramatically across breeds (Schoenebeck, 2013;Schoenebeck et al., 2012). To quantify the variation, we collected data from 533 museum skulls at 51 landmarks using a microscribe digitizer. The resulting principal components analysis (PCA) showed that the top four PCs account for about 77% of skull variance across breeds. PC1 describes profound changes in rostrum length and angle, palate and zygomatic arch width, and depth of the neurocranium;essentially the continuum of craniofacial features that extend between brachycephalic (short head like a bulldog) and dolichocephalic (long head like a greyhound) skulls. Our GWAS on PC1, done using CanMap data, revealed several loci. The CFA32 locus demonstrated a marked reduction in observed heterozygosity (Ho) and elevated genetic differentiation (FST), which are hallmarks of strong selection. To identify causative variants on CFA32, we analyzed whole genome sequence from 12 dog breeds of widely varying skull shapes. This reduced the dataset thousands of variants to only one that met all prioritization criteria which included evolutionary conservation, strong association with the phenotype, and likely functional impact. The mutation was a F452L missense mutation in the bone morphogenesis protein 3 (BMP3) gene (BMP3F452L). In the resulting paper we show that the mutation is functional by several criteria including, most importantly, studies in zebrafish (Schoenebeck et al., 2012). Knockdown of the gene by morpholinos causes severe hypoplasia of cartilaginous elements that form the viscerocranium and head structures, showing that Bmp3 function in cranial development is both necessary and ancient, and demonstrating a role for this gene in skull shape for the first time. We also hypothesize a role for BMP3 in the craniofacial dysmorphism observed in human 4q21 deletion syndrome. Ongoing studies focus on fine mapping and functional studies of other skull-associated loci, particularly those accounting for the continuum from dolichocephalic to brachycephalic shapes. Canine Cancer The tremendous phenotypic diversity of modern dog breeds represents the end point of a >15,000-year experiment in artificial and natural selection. Most breeds came into existence within the last 300 years, and many are derived from small numbers of founders. Each has undergone strong artificial selection, in which dog fanciers selected for many traits including body size, fur type, color, skull shape, and even behavior, to create novel breeds. The adoption of the breed barrier rule no dog may become a registered member of a breed unless both its dam and sire are registered membersensures a relatively closed genetic pool within each breed. As a result, there is strong phenotypic homogeneity within the 177 breeds recognized by the American Kennel Club (AKC), but between breeds there exists extraordinary phenotypic variation. One consequence of this population structure is breed-associated genetic disease. We have taken advantage of this to identify genes for a large number of canine maladies for which mouse models do not exist. Our recent genetic studies of dog disease have focused almost exclusively on cancer, which we argue is a strong model for human cancer genetics. Squamous cell carcinoma of the digit (SCCD) is a locally aggressive cancer typified by lytic bone lesions, recurrence, and occasional death from metastasis. Standard Poodles are among the breeds with the highest risk of SCCD, however only the dark pigmented standard poodles are susceptible. We conducted a GWAS using on black Standard Poodles, and demonstrated that the Kit Ligand (KITLG) locus is strongly associated with SCCD. After extensive sequencing and haplotype analysis, a 5.7 kb CNV with the signature of a transcriptional enhancer element was identified as the most likely causative mutation (Karyadi, 2013). We found that four cis copies of the CNV element represent the critical threshold for disease risk. Dogs without at least one allele with four copies in cis are not at any risk for disease (Karyadi, 2013). This locus is under strong selective pressure in dogs, likely in response to coat color preferences, as KITLG has been linked to pigmentation in humans, mice and fish. In separate studies we have performed a GWAS on Bernese Mountain Dogs (BMD) to identify a locus associated with histiocytic sarcoma (HS), and found a strongly associated haplotype on CFA11 that spans MTAP and part of CDKN2A (Shearin, 2012) that we continue to follow up. We have also initiated studies of transitional cell carcinoma (TCC) of the bladder and gastric cancer.
Showing the most recent 10 out of 43 publications
