Selection signature analyses and genome-wide association reveal genomic hotspot regions that reflect differences between breeds of horse with contrasting risk of degenerative suspensory ligament desmitis

Abstract

Degenerative suspensory ligament (SL) desmitis (DSLD) is a progressive idiopathic condition that leads to scarring and rupture of SL fibers in multiple limbs in horses. Prevalence of DSLD is breed-related. Risk is high in the Peruvian Horse (PH), whereas pony and draft breeds have low breed risk. DSLD occurs in families of PHs, but its genetic architecture has not been definitively determined. We investigated contrasts between breeds with differing risk of DSLD and identified associated risk variants and candidate genes. We analyzed 670K single nucleotide polymorphisms (SNPs) from ten breeds, each of which was assigned one of four breed DSLD risk categories: Control (Belgian, Icelandic Horse, Shetland Pony, and Welsh Pony), Low Risk (Lusitano, Arabian), Medium Risk (Standardbred, Thoroughbred, Quarter Horse), and High Risk (Peruvian Horse). SNPs were used for genome-wide association and selection signature analysis using breed assigned risk levels. We found that the PH is a population with low effective population size and our breed contrasts suggest that DSLD is a polygenic disease. Variant frequency exhibited signatures of positive selection across DSLD breed risk groups on chromosomes 7, 18, and 23. Our results suggest DSLD breed risk is associated with disturbances to SL homeostasis where matrix responses to mechanical loading are perturbed through disturbances to aging in tendon (PIN1), mechanotransduction (KANK1, KANK2, JUNB, SEMA7A), collagen synthesis (COL4A1, COL5A2, COL5A3, COL6A5), matrix responses to hypoxia (PRDX2), lipid metabolism (LDLR, VLDLR), and BMP signaling (GREM2). Our results do not suggest that SL proteoglycan turnover is a primary factor in disease pathogenesis.

Genetic architecture and polygenic risk score prediction of degenerative suspensory ligament desmitis (DSLD) in the Peruvian Horse

“…Glycosaminoglycans have a key role in extracellular matrix composition of tendons and disturbed metabolism of the extracellular matrix of tendon; accumulation of aggrecan in SL tissue and disturbance to decorin glycosylation are key features of DSLD ( Kim et al, 2010 ; Plaas et al, 2011 ; Haythorn et al, 2020 ). Mechanotransduction has a key role in tendon and ligament homeostasis and genes that have regulatory effects on mechanotransduction were a key finding in a previous categorical GWAS of DSLD in multiple breeds of horse ( Momen et al, 2022 ).…”

“…The Peruvian Horse is a breed with a small effective population ( Momen et al, 2022 ), enabling detection of significant associations and accurate PRS predictions with a relatively small sample size. In the cross-validation experiment, the models we studied demonstrated moderate predictive performance on both the probit and logit liability scales.…”

“…By performing eigen decomposition of the GRM using the base eigen () function in R ( R Core Team, 2021 ), the eigenvectors and eigen values were obtained and the eigenvectors were normalized. Finally, PCs were computed by multiplying eigenvectors by the square root of the associated eigenvalues ( Bryant & Yarnold, 1995 ; Momen et al, 2022 ). To review the results, we plotted the projection of the individuals on the first two PCs, with colors corresponding to their group assignment.…”

“…DSLD is typically more evident in the pelvic limbs verses the thoracic limbs, although in some Peruvian Horses DSLD develops in all four limbs. Reproduced from Momen et al, 2022 .…”

“…From an evolutionary point of view, selection for a desired phenotype through careful breeding results in an increased frequency of haplotypes containing the gene(s) and functional allele(s) conferring the phenotype at a rate greater than expected under a null model of neutral evolution ( Cutter and Payseur, 2013 ). GWAS and detection of signatures of selection (SOS) are two common genetic analysis approaches for case control association between a disease phenotype and genetic markers, typically single nucleotide polymorphisms (SNPs) ( Patron et al, 2019 ; Momen et al, 2022 ).…”