NEW YORK (GenomeWeb Daily News) – Bone cancer risk loci in dogs vary by breed, but a number overlap in pathways linked to bone formation and growth, researchers from Uppsala University and the Broad Institute reported in Genome Biology yesterday.
The researchers, led by Kerstin Linblad-Toh, conducted genome-wide association studies in three dog breeds — greyhounds, Rottweilers, and Irish wolfhounds — that are highly susceptible to bone cancer, and found 33 inherited risk loci linked to the disease.
"Our results show that the pathways involved in bone formation and growth are important for the disease," said Lindblad-Toh, a professor at Uppsala and the director of vertebrate genome biology at the Broad, in a statement.
Osteosarcoma is an aggressive cancer that, in people, typically strikes children and adolescents and has a high mortality rate. Though bone cancer is more common in dogs, especially large breeds, and the dogs tend to be relatively older, canine osteosarcoma follows a similar progression to human osteosarcoma: tumors form at the ends of long bones and metastasize easily, often to the lungs, the researchers noted.
"Because of the great similarities between bone cancer in dogs and humans, we believe that our findings may contribute to an increased understanding of how bone cancer develops in humans," she added.
In this study, Linblad-Toh and her colleagues genotyped 334 greyhounds, 166 Rottweilers, and 174 Irish wolfhounds using the Illumina canine HD array. They noted that the three breeds were visibly discrete populations and that the 267 racing greyhounds clustered near, though apart from, the AKC greyhounds.
From this, the researchers uncovered 33 regions associated with bone cancer in these dogs, though they saw no overlap in regions between breeds. These 33 loci account for some 50 percent to 80 percent of the disease risk in the breeds, the researchers noted.
They tested those top 33 regions for fixation in the other breeds, finding that the risk allele associated with the top greyhound locus is fixed in Rottweilers and Irish wolfhounds.
By sequencing the locus in eight greyhound cases and seven controls, the researchers homed in on a 15 kilobase non-coding region near CDKN2A, CDKN2B, and CDKN2B-AS1/ANRIL. CDKN2A and CDKN2B, both tumor suppressor genes, regulate the cell cycle, while CDKN2B-AS1/ANRIL is an anti-sense non-coding gene. Altered levels of CDKN2A have been linked to hematopoetic stem cell senescence and development.
Additionally, the researchers pointed out that the risk haplotype at 11q16 is syntenic to a non-coding regulatory region in the human chromosome 9p21, which is deleted in between 5 percent to 21 percent of human osteosarcoma. They also noted that 9p21 is one of the most complex regulatory regions in the human genome.
That led the researchers to suggest "that variant(s) carried on the risk haplotype disrupts enhancer elements upstream of the CDKN2A/B locus, thereby altering expression of one or more genes in the region."
In a human osteosarcoma cell line, Linblad-Toh and her colleagues searched for enhancer activity linked to the CDKN2A/B locus using renilla/firefly luciferase assays. One fragment, containing a top SNP, increased expression by more than 30 fold. In mammals, that SNP is usually a C or a T, but the risk allele A is present in 87 percent of greyhound cases and 68 percent of greyhound controls. Further analysis indicated that PAX5, which regulates B-cell and osteoblast differentiation as well as bone formation, could bind the non-risk C allele, but not the A risk allele. That risk allele was also found at high frequency in Rottweilers and Irish wolfhounds as well as in Leonbergers and Great Pyrenees, two other breeds with high rates of osteosarcoma.
Through pathway analysis, the researchers determined that a number of the 33 regions associated with osteosarcoma in the three dog breeds are linked to bone differentiation and growth, and that the regions are interconnected.
For example, they reported that OTX2, an orthodenticle homeobox protein which was linked to bone cancer in greyhounds, is connected to six other risk loci, including the osteoblast differentiation regulators BMPER and VWC2, and EN1, which modulates osteoblast differentiation and proliferation.
Since GWAS can't detect alleles that are fixed in populations, the researchers examined those regions in each of the breeds, and noted that a number of the fixed regions included genes implicated in bone development and bone cancer. Then by comparing these regions of reduced variability in these dogs to some 28 other breeds, the researchers found seven gene sets that were inflated, including the KIT, p53, and PDGFRB pathways.
Using array CGH, the researchers also examined the frequency and distribution of copy number variation in 20 dog bone cancer tumor samples. The profiles, the researchers reported, were highly similar to each other as well as to CNV profiles of human osteosarcoma. In particular, the tumor samples tended to gain MYC and RUNX2 while losing RB1 and CDKN2A/B.
A recent human GWAS of osteosarcoma, published in Nature Genetics in June, identified two loci, one in the GRM4 gene and one in a gene desert, linked to the disease.
While the canine risk alleles Linblad-Toh and her colleagues uncovered don't include GRM4, they do include other glutamate receptors.
The importance of glutamate receptors in both canine and human osteosarcoma suggested to the researchers that glutamate signaling could be a potential therapeutic target. They noted, though, that as that pathway has diverse physiological functions, such as in learning and memory, targeting it with a therapeutic could be difficult.