he Ostrander lab is interested, first, in expanding the vocabulary of growth control related genes. As such, the researchers are building on the extensive genomic infrastructure we developed for the dog to identify genes associated with breed-specific differences in morphology. The team seeks, secondly, to find genes associated with cancer susceptibility in dogs, where breed predilection likely reflects the selective pressures used to develop and enhance specific traits. Their interest in cancer extends beyond dogs to human prostate cancer (PC), where they are working both independently and within consortia to find genes associated with susceptibility and progression to aggressive forms of PC.
Canine Morphology: The Ostrander lab has long championed the canine system (Ostrander NEJM, 2012) for the study of complex traits by building maps (e.g. Guyon et al., PNAS 2003; Hitte et al., Nat Rev Genet 2005), organizing sequencing of the canine genome (Lindblad-Toh et al., Nature 2005), and studying population structure (Parker et al., Science 2004; Genome Res 2007), domestication (VonHoldt et al., Nature 2010), and performance (Mosher et al., PLoS Genet 2007). In doing so they have demonstrated the power of breed structure for mapping traits that are have proven intractable through human studies (e.g. Parker et al., Genome Res 2007). The lab's ability to map the genetic underpinnings of body size (Sutter et al., Science 2007), pelage (Cadieu et al., Science 2009), leg length (Parker et al., Science 2009), skull shape (Schoenebeck et al., PLoS Genetics 2012; Schoenebeck et al., Genetics 2013), and other phenotypes has provided unique insights into the fundamental biology of natural phenotypic variation in mammals, offering a completely new lens for viewing these processes in humans. They have built a large data set for such studies encompasing 85 breeds and over 1000 dogs, all of which they have genotyped with a dense SNP chip. Using that data, their GWAS analyses of canine skull morphologies, for instance, has identified a new role for the BMP3 gene in early skull formation. Next-generation sequencing of multiple dogs has further identified a highly conserved amino acid change that is strongly associated with the bracychephalic skull shape observed in breeds such as the bulldog. They have also focused on body size, identifying eight major size genes in dogs, the first of which was IGF-1 (Sutter et al., Science 2007). Their research suggests that these newly identified genes interact in a non-additive fashion to determine breed size, which ranges from 2-200 pounds (Rimbault et al., Genome Res 2013). In theory, variants at several additional loci fine-tune the system, accounting for the continuum of size observed in modern breeds. The lab's near term aims focus on fine mapping and functional studies for the remaining body size and skull shape loci.
Canine Cancer: Cancer mapping studies in humans often fail due to sample size limitations and locus heterogeneity; factors the Ostrander lab argues can often be overcome in canines (Parker et al., Annu Rev Genet 2010). Dogs generally develop cancer in the same organs and respond to similar therapies as humans. However, predisposition to cancer type is often breed-specific. The lab first demonstrated the utility of dogs for cancer genetics by mapping kidney cystadeno- carcinoma in German Shepherds, prior to its localization in humans (Jonasdottir et al., PNAS 2000). They subsequently showed that mutations in the folliculin gene are causative, as they are in humans (Lingaas et al., Hum Mol Genet 2003).
The group subsequently tackled other cancers. Their GWAS for histocytic carcinoma identifies two loci, the first of which spans CDKN2A, an important cell cycle regulator, while the second region includes several candidates (Shearin et al., CEBP, 2012). RNAseq and ChiPseq data are being done to characterize transcripts and regulatory regions. They have also made progress in studies of squamous cell carcinoma of the digit (SCC), a breed-specific nail bed cancer occurring primarily in poodles, briards and giant schnauzers. They identified KIT ligand (KITLG) as one of the genes associated with the disorder (Karyadi et al., PLoS Genet 2013). Additional work suggests that variant forms of MC1R and MITF are protective. Interestingly, mutations in MC1R, KIT and MITF are associated with some melanomas in humans. The group's hypothesis is that for human acral (palm, soles, or nail bed) melanoma the oncogenic potential of the KIT pathway requires a functional MC1R, and that SCC may serve as a precursor model. Other studies focus on canine gastric and bladder cancer.
Prostate Cancer: PC is a major cause of human morbidity and mortality, and no good naturally occurring animal models exist. Finding genes associated with aggressive disease has been minimally successful (Schaid et al., Hum Genet 2007; FitzGerald et al., CEBP 2011). The Ostrander lab has collaborated with Janet Stanford, Ph.D., from the Fred Hutchinson Cancer Research Center, Seattle, to identify PC loci using two lines of investigation. First, the group has performed multiple genome scans using 304 high-risk families, (Janer et al., Hum Mol Genet 2003; Stanford et al., Hum Mol Genet 2009), resulting in the identification of multiple loci. Family-based association methods identified one locus on chromosome 22q12 which we mapped to a few kilobases (Johanneson et al., Hum Mol Genet 2010). They are also part of two international and highly successful consortia (e.g. Lu et al., Prostate 2011; Haiman et al., Nat Genet 2011) that have used GWAS to enumerate and now validate several new PC risk loci (e.g. Eeles et al., Nat Genet 2009; Kote-Jarai et al., Nat Genet 2011; Eeles et al., Nat Gen 2013).
Their second approach is a hypothesis-driven candidate gene analysis using a population-based cohort of 1,309 PC patients under long-term surveillance and upon which we have published extensively (e.g. Stanford et al., Cancer Res 1997; Kwon et al., CEBP 2011; Kwon et al., CEBP 2012). They have identified and validated five SNPs associated with PC specific mortality (PCSM) (Lin et al., CEBP 2011). Patients with 4-5 (vs. 0-2) of the at-risk genotypes have a 50 percent higher risk of PCSM, and risk increases with number of at-risk genotypes. These results provide the first validation of a genetic predisposition for lethal PC and highlight genes of interest.
Recently, in a third approach, the group began sequencing high-grade prostate tumors to identify drivers of disease. Their goal is to find markers of prostate cancer specific mortality. They are also doing whole-exome sequencing of men from multiple high-risk PC families and aim to coordinate these and other data in men with long-term clinical follow-up to identify prognostic markers of disease.
Comparative Genetics Section Members
Heidi G. Parker, Ph.D. Staff Scientist
Danielle Karyadi, Ph.D. Staff Scientist
Brian Davis, Ph.D., Postdoctoral Fellow
Dayna Dreger, Ph.D., Postdoctoral Fellow
Adrienne Bhatnager, Ph.D. Postdoctoral Fellow
Brennan Decker, M.S., NIH Oxford Cambridge graduate student
Eric Karlins, M.S., Biologist
Erica Chapman, B.A., contractor
Last Updated: June 2, 2015