Schultz, Nichol2018-07-262018-07-262016-01https://hdl.handle.net/11299/198365University of Minnesota Ph.D. dissertation. January 2016. Major: Comparative and Molecular Biosciences. Advisors: James Mickelson, Molly McCue. 1 computer file (PDF); xxiv, 349 pages.Equine metabolic syndrome (EMS) is a clustering of clinical signs associated with increased risk of laminitis, a potentially life-threatening condition of the foot. Similar to human metabolic syndrome (MetS), generalized and/or regional adiposity, hyperinsulinemia, insulin resistance and dyslipidemia, are reported components of EMS. However, there is ongoing debate regarding the definition of EMS, its etiology and pathogenesis, and the mechanisms linking EMS to its secondary consequences. Conflicting reports regarding EMS reflect the limitations of prior EMS studies, and that EMS is likely a complex, multifactorial condition similar to MetS. The primary objectives of this thesis were to characterize metabolic variation and EMS across horse and pony breeds and to identify candidate genes for EMS risk. Chapter 2 details the largest-ever epidemiological investigation of EMS in which 11 metabolic traits were measured in >600 horses and ponies from 166 farms. The use of multivariate, multilevel regression modeling allowed, for the first time, quantification of the relative importance of environmental (farm, dietary composition, exercise, etc.) and individual (age, breed, sex etc.) factors on these metabolic traits, while accounting for the often strong correlation between the trait measures. Age, sex, breed, obesity, prior laminitis status, and time of year were all strongly associated with one or more metabolic traits. Despite strong associations, these factors only explained 9.6% to 36.3% of the variation across these 11 traits, thus the majority of the variability in these measures remained unexplained. Unexplained variation at the farm level after accounting for diet, exercise, and sampling time of year, suggests that additional unmeasured environmental factors explain the similarity in metabolic measures between horses sampled from the same farm. Similarly, unexplained variation at the individual level suggests that unmeasured individual characteristics, for example genetics, are responsible for a large proportion of individual trait variation. Differences in the incretin response may also contribute to individual trait variation. The incretin response, defined as the difference in insulinemic responses between an oral and intravenous glucose challenge, is controlled by intestinal secretion of peptides, such as GLP-1, that stimulate pancreatic insulin secretion. While the incretin response has been hypothesized to play a role in the EMS pathogenesis, this hypothesis has not been adequately tested. In Chapter 3, the glycemic, insulinemic, and total and active GLP-1 responses to an oral sugar challenge, and the activity of DPP4, the major protease that breaks down GLP-1, were characterized. The use of a longitudinal analysis, rather than the traditional area under the curve analysis, allowed for more power to detect differences in these responses, including variation due to breed, obesity, and prior laminitis status. Unexplained individual level variation and breed differences in metabolic phenotypes support the hypothesis that there is an underlying genetic susceptibility to EMS. The final objective of this thesis was to identify candidate genes associated with EMS. MetS is a highly polygenic syndrome where numerous candidate genes have been identified. Whereas MetS associated variants are typically of small effect size; it was hypothesized that in EMS a small number of moderate to large effect loci contribute to variation in metabolic traits due to the fact that horse populations do not randomly mate and experience substantial selection pressure. 286 Morgan horses were genotyped on the Illumina SNP50 chip and imputed up to >800,000 SNPs to perform a genome wide association study (GWAS) to identify candidate genes for EMS. Additive genetic variance estimated from a genomic relationship matrix calculated from genotyped SNPs (“chip heritability”) indicated that the 11 measured metabolic traits were moderately heritable. Yet initial genome-wide scans using standard linear mixed models failed to detect significant associations. In Chapter 4, an improved linear mixed model for mapping polygenic traits in a population with familial relationships similar to that in many equine GWAS was developed and validated. The model incorporates a Bayesian variable selection method to rank SNPs and a stepwise feature selection process to determine the optimal SNPs to model the random polygenic effect, while including a random effect for each sampled herd or “familial cluster”. The method was validated using the QTL-MAS 2010 dataset, and Morgan horse and Welsh pony height datasets, and demonstrated increased power while controlling the false positive rate. Using this improved linear mixed model, 76 suggestive and 17 genome-wide significant candidate loci were identified for the 11 metabolic traits in the 286 Morgan horse cohort. Candidate genes had a substantial overlap with MetS candidate genes such as VEGFA, NRXN3, GRIK2, and TRIB2. Other interesting candidate genes included ISL, which encodes insulin enhancer protein that is thought to play an important role in regulating insulin gene expression; and AHR which encodes the aryl hydrocarbon receptor, a ligand activated transcription factor known to bind endocrine disrupting chemicals such as polycyclic aromatic hydrocarbons and dioxins. AHR is an interesting candidate gene given the potential role of endocrine disrupting chemical in the pathophysiology of MetS, and unexplained sources of farm level variation in Chapter 2. A unifying theme of Chapters 2-5 was the similarities between EMS and MetS, and the complex phenotypic and genetic architecture in both species. The use of advanced statistical modeling approaches allowed for a more complete understanding of the metabolic phenotypic variation in Chapters 2 and 3, and for the identification of many associated genetic loci in Chapter 5. The shared candidate genes for metabolic syndrome in humans and horses suggests similar underlying pathophysiological mechanisms and provides opportunity for exploring similar preventative and therapeutic management strategies.enequine metabolic syndromeGWASincretinlaminitismetabolic syndromemultilevel modelThesis or Dissertation