Norton, Elaine2019-12-162019-12-162019-10https://hdl.handle.net/11299/209197University of Minnesota Ph.D. dissertation. October 2019. Major: Comparative and Molecular Biosciences. Advisors: Molly McCue, James Mickelson. 1 computer file (PDF); xvii, 335 pages.Laminitis is a painful, debilitating disease of the hoof, often resulting in these horses being humanely euthanized due to uncontrolled pain. The most commonly cited cause of this life-threatening disease is a clustering of clinical signs resulting from a metabolically efficient phenotype, termed equine metabolic syndrome (EMS). While EMS is a commonly diagnosed syndrome, knowledge of the underlining pathophysiology is lacking and recommendations for diagnostic criteria are vague and inconsistent. EMS is thought to be complex disease, and identification of its underlying genetic risk factors and key gene-by-environment interactions will improve our understanding of EMS pathophysiology and allow for early detection of high-risk individuals and intervention prior to the onset of laminitis. We hypothesized that major genetic risk factors leading to EMS and laminitis susceptibility are shared across breeds of horses, and that differences in the severity and secondary features of the EMS phenotype between breeds, or between individuals within a breed, are the result of modifying genetic risk alleles with variable frequencies between breeds. To test these hypotheses, my PhD thesis has consisted of using phenotype and genotype data on 286 Morgan horses and 264 Welsh ponies, two high risk breeds for EMS. Phenotype data collected on all horses included: signalment, medical history, laminitis status, environmental management (feed, supplements, turnout and exercise regimen), and morphometric measurements (body condition score (BCS), wither height, and neck and girth circumference). After an eight hour fast, an oral sugar test (OST) was performed using 0.15mg/kg Karo lite corn syrup. Biochemical measurements included baseline insulin, glucose, non-esterified fatty acids (NEFA), triglycerides (TG), adiponectin, leptin and ACTH; and measurements 75 minutes after the OST included insulin (INS-OST) and glucose (GLU-OST). For inclusion in the study, each farm had to have at least one control and one horse with clinical signs consistent with EMS under the same management. Single nucleotide polymorphism (SNP) genotyping was performed on all horses. Haplotype phasing and genotype imputation up to two million SNPs was performed on horses genotyped on lower density arrays using Beagle software. Quality control on the imputed data was performed using the Plink software package. After genotype pruning, 1,428,337 and 1,158,831 SNPs remained for subsequent analysis in the Welsh ponies and Morgan horses, respectively. In chapter 2, SNP genotype data from the Welsh ponies and Morgan horses were used to estimate the heritability of the nine EMS biochemical measurements. Heritability (h2SNP) was estimated using a restricted maximum likelihood statistic with the inclusion of genetic relationship matrix, which was corrected for linkage disequilibrium (regions of the genome which are not independent as they are inherited together). The confounders of age, sex and season were included in the model based on the Akaike information criteria. In the Welsh ponies, seven of the nine biochemical traits had h2SNP estimates with p-values that exceeded the Holm-Bonferroni corrected cut-off as follows: triglycerides (0.31), glucose (0.41), NEFA (0.43), INS-OST (0.44), adiponectin (0.49), leptin (0.55), and insulin (0.81). Six of the nine EMS traits in the Morgans had h2SNP estimates with p-values that exceeded the Holm-Bonferroni cutoff as follows: INS-OST (0.36), leptin (0.49), GLU-OST (0.57), insulin (0.59), NEFA (0.68), and adiponectin (0.91). Insufficient population size and high trait variability may have limited power to obtain statistically significant h2SNP estimates for ACTH (both breeds), glucose and triglycerides in Morgans and GLU-OST in Welsh ponies. These data provide the first concrete evidence of a genetic contribution to key phenotypes associated with EMS and demonstrate that continued research for identification of the genetic risk factors for EMS phenotypes within and across breeds is warranted. Although heritability estimates provide valuable insight on the genetic contribution to a trait, they do not provide information on the number of contributing genes, specific genes involved, or where these genes are located within the genome. Genome wide association analyses (GWA) use SNP genotype data to identify those key regions of the genome that are associated with a trait. The objectives of chapter 3 were to (i) perform within breed GWA to identify significant contributing loci in Welsh ponies and Morgans, and (ii) use a meta-analysis approach to identify shared and unique loci between both breeds. For each trait, within breed GWA were performed from the imputed SNP genotype data using custom code for an improved mixed linear regression analysis. Prior to analysis, traits were adjusted to account for known covariates, with sex and age included as fixed effects and farm as a random effect. GWA meta-analysis was performed with a random effects model using the Morgans and Welsh pony GWA summary data from the 688,471 SNPs that were shared between breeds. To define the boundaries of the region, a pairwise comparison of linkage disequilibrium (LD) was calculated for all SNPs within the region. A custom code was used to identify regions where LD for all SNPs dropped below the LD threshold of 0.3 and spanned at least 100kb both 5' and 3' to the widest peak of LD within the window, which was used to define the boundaries of the ROI. An LD-region was identified as shared if it was within the boundaries of another LD-region and prioritized as described above for the fixed regions. Regions were prioritized based on whether they were identified as shared between breeds on meta-analysis (high priority), shared across traits (medium priority), or found in a single breed but not shared across traits (low priority). Prioritization resulted in 56 high, 26 medium, and 7 low priority genomic regions for a total of 1853 candidate genes in the Welsh ponies, and 39 high, 8 medium and 9 low priority regions for a total of 1167 candidate genes in the Morgan horses. Meta-analysis identified 65 of these regions that were shared across breeds. These data demonstrate that EMS is a polygenic trait with both across breed and breed specific genetic variants. In chapter 4, we utilized imputed whole-genome sequencing (WGS) and linear regression analysis in order to fine-map selected high priority LD-ROI in both the Morgan horses and Welsh ponies. LD-ROI were fine-mapped if they contained at least 5 SNPs with one SNP exceeding the threshold for genome-wide significance. Five fine-mapped regions from each breed were further interrogated for predicted impact using variant annotation. Protein-coding genes containing non-coding or coding variants within the fine-mapping region were then further prioritized based on known function and biological evidence in other species utilizing the PubMed search engine. A total of 19 positional candidate genes were identified as having biological evidence for a role in EMS. These data provide support for the process of fine-mapping GWA ROI by increasing marker density and using biological evidence across species to further prioritize candidate genes. In chapter 5, a missense mutation in the first exon of HMGA2 was identified as a putative functional allele for height and EMS phenotypes in Welsh ponies. It is well recognized that ponies (short horses) are at high risk for developing EMS; and in humans shorter individuals have an increased risk of developing cardiovascular disease, type II diabetes and metabolic syndrome. We hypothesized that genetic loci affecting height in ponies have pleiotropic effects on metabolic pathways and increase the susceptibility to EMS. Pearson’s correlation coefficient identified an inverse relationship between height and baseline insulin (-.26) in the Welsh ponies. Genomic signature of selection analysis was performed using a di statistic and identified a ~1.3 megabase region on chromosome 6, that was also identified on GWA. Haplotype analysis using HapQTL confirmed that there was a shared ancestral haplotype between height and insulin. This region contributed ~40% of the heritability for height and ~20% of the heritability for insulin. HMGA2 was identified as a candidate gene, and sequencing identified a single a c.83G>A variant (p.G28E) in HMGA2, previously described in other small stature horse breeds. In our cohort of ponies, the A allele had a frequency of .76, was strongly correlated with height (-.75) and was low to moderately correlated with metabolic traits including: insulin (.32), insulin after an oral sugar test (.25), non-esterified fatty acids (.19) and triglyceride (.22) concentrations. For this allele, model analysis suggested an additive mode of inheritance with height and a recessive mode of inheritance with the metabolic traits. This was the first gene identified as having a pleotropic effect for EMS. In conclusion, the results of my thesis are major steps forward in understanding the genetic contributions of EMS in two high risk breeds. Future directions include the continued identification of the specific genes and alleles contributing to EMS and could include prioritization of the positional candidate genes identified in aim 2 via (1) identification of biological candidate genes based on known gene function and evidence from other species; (2) use of whole genome sequencing and linear regression analysis to fine map regions; (3) use of intermediate phenotypes such as metabolomics or transcriptomics to identify shared regions; or (4) network analysis for identification of genes within similar, relevant pathways.enequine metabolic syndromegeneticsgenome-wide associationhorselaminitisrisk factorsThesis or Dissertation