Source: UNIV OF MINNESOTA submitted to NRP
GENETIC DIVERSITY AND SELECTION IN THE DOMESTIC HORSE
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
AFRI COMPETITIVE GRANT
Reporting Frequency
Annual
Accession No.
0227973
Grant No.
2012-67015-19432
Cumulative Award Amt.
$499,481.00
Proposal No.
2011-03216
Multistate No.
(N/A)
Project Start Date
Apr 1, 2012
Project End Date
Jun 30, 2016
Grant Year
2012
Program Code
[A1201]- Animal Health and Production and Animal Products: Animal Breeding, Genetics, and Genomics
Recipient Organization
UNIV OF MINNESOTA
(N/A)
ST PAUL,MN 55108
Performing Department
Veterinary Population Medicine
Non Technical Summary
The horse has a 102 billion dollar annual impact on the US economy. Approximately 60% of the 9.2 million horses in the US are used for work, performance or breeding, while the remainder are used primarily for recreation. Like other domestic species, the modern horse was developed through human-directed breeding and recent intense selection, leading to tremendous variation in traits across breeds. While selective breeding has shaped the equine genome and led to the fixation of desirable traits, intense selection and inbreeding have likely limited genetic diversity and increased the incidence of undesirable traits (and genetic disease) within breeds. Little is known about the impact of breed formation and artificial selection on population structure and diversity of the domestic horse; and the specific genotypes that underlie the traits under selection are largely unknown. Our goal is to improve the health and performance of the horse by identifying genes responsible for peak athletic performance as well as those responsible for disease, allowing horse breeders to select for traits of interest while limiting the impact of deleterious genetics. We will evaluate the effects of selective breeding on the genetic diversity and structure of domestic horse populations, and identify regions of the genome undergoing selection genes targeted by selection for economically important and health-related traits.
Animal Health Component
80%
Research Effort Categories
Basic
20%
Applied
80%
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
3033810108080%
3043810108020%
Knowledge Area
303 - Genetic Improvement of Animals; 304 - Animal Genome;

Subject Of Investigation
3810 - Horses, ponies, and mules;

Field Of Science
1080 - Genetics;
Goals / Objectives
Our long-term goal is to improve horse health and performance by identifying genes/alleles responsible for performance traits as well as for disease traits, allowing breeders to select for specific traits while limiting the impact of deleterious alleles. The goals of this proposal are to evaluate the effects of selective breeding on the genetic diversity and structure of domestic horse populations, identify genomic regions undergoing selection, and identify putative functional alleles targeted by selection. We have three objectives: In objective 1 we will quantify genetic diversity within and among horse populations, define breed relationships, and identify ancestry informative markers in for use in the domestic horse. In objective 2 we will use genome-wide approaches to identify loci undergoing selection using both FST- and haplotype-based statistics, to detect both recent population-specific, and partial selective sweeps. In objective 3 we will use haplotype analysis and targeted resequencing to identify putative functional variants that underlie selection at targeted loci. Regions prioritized for investigation will be those which have strong evidence for selection and are compelling candidate regions for biologically or economically important phenotypes (i.e., gait, performance, athletic ability, etc). Completion of this project will provide the largest and most comprehensive data on genetic diversity in the modern horse. Further, we will have systematically identified the regions of the genome that have been influenced by selective breeding during the history of the domestic horse and, in some instances, the alleles underlying desirable traits. The identification of these regions and alleles will allow unprecedented insight into the biology behind particular phenotypes, and for the first time enable us to start untangling the genetic basis of enhanced performance in certain populations and disciplines. These data will allow insight into the overall effect of current breeding practices, and identification specific alleles will allow breeding strategies that to focus on desirable traits while avoiding disease alleles. These results are key to making educated breeding choices that will lead to healthier horses and a robust equine industry in future generations.
Project Methods
In objective 1 we will use genotype data from greater than 50,000 nuclear SNPs in 2,150 horses representing 45 globally-distributed and phenotypically diverse breeds. This data will be used to quantify genetic diversity and linkage disequilibrium within breeds, to decipher the relationships between domestic horse populations (principal components analysis, FST, analysis of molecular variance, genetic assignment and admixture analysis, haplotype diversity and haplotype sharing) and to identify ancestry informative markers. In objective 2, we will identify putative genomic regions undergoing selection using di and iHS statistics. Di is a population-differentiation based method, which compares allele frequencies between a breed, or group of breeds with a shared phenotype, to all other breeds in the analysis. This approach tests the hypothesis that intense selection in a population for the same phenotypic characteristic will result in a detectable difference in allele frequencies at neutral markers linked to the functional allele exceeding that expected from drift alone. iHS, a haplotype-based method, is useful in circumstances where there is strong, very recent selection in favor of alleles that have not yet reached fixation (partial selective sweeps), particularly where these alleles and selective pressures occur across populations. Based on preliminary data we anticipate that these approaches will identify multiple regions undergoing selection, and that marker-based haplotype analysis will allow us to narrow regions to approximately 0.5 to 6 Mb. In objective 3 we will identify potential functional alleles underlying these signatures of selection. To minimize cost, and maximize the number of loci that can be examined, we propose a target-enrichment strategy with sequence capture on pooled samples, followed by next-generation sequencing, allowing for discovery of variants within each region of interest. We will be prioritizing loci in objective 3 based on potential impact on health, disease and performance, thus we anticipate that discovered alleles will be valuable to the industry. To quantify the importance of these alleles, allele frequencies will be determined larger sample populations in subsequent studies. We expect these results to be used by the industry to select for or against particular functional alleles.

Progress 04/01/12 to 06/30/16

Outputs
Target Audience:Horse owners, breeders, equine veterinarians and equine genetic researchers Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The MSTN work was conducted primarily by a post-doctoral fellow, Dr Jessica Petersen. Dr Petersen is now a tenure-track Assistant Professor at the University of Nebraska-Lincoln. Work on allele discovery and follow-up genotyping for gait was performed by PhD student Annette McCoy DVM, MS, DACVS), and by veterinary summer scholar Samantha Beeson. Dr. McCoy is now an assistant professor at the University of Illinois and Ms. Beeson is now a DVM/PhD dual degree student at the University of Minnesota. Post-doctoral fellow Felipe Avila has been working on refinement of signatures of selection using 2 million array data across breeds for multiple loci. The ECA6 size and metabolic trait locus was performed by Dr Avila and graduate student Dr Elaine Norton (DVM, MS, DACVIM). And, the SNP array design and genotyping array work was completed by postdoctoral fellow Robert Schaefer. How have the results been disseminated to communities of interest?Data has been presented at the Plant and Animal Genome Meeting (2012, 2013, 2014, 2015, 2016), the International Society for Animal Science Meeting, International Society of Animal Genetics (2012), the European Federation of Animal Sciences (2012), and the Havemeyer Equine Genomics Workshop (2013, 2015), The Alberta Horse Owners Conference (2016), and the Colorado State Ranch Owners Forum (2016), and the American College of Veterinary Internal Medicine Forum (2016). the following additional manuscripts are also in preperation: Manuscripts submitted or in prep. Felipe Avila, Robert Schaefer, James R. Mickelson, Samantha Beeson, Rebecca K. Splan, Molly E. McCue. The Genetic Underpinnings of Diverse Athletic Performance in the American Quarter Horse Felipe Avila, Robert Schaefer, James R. Mickelson, Molly E. McCue. Bred for Speed: Genomic Signatures of Selection in Racehorses. Robert Schaefer, James R. Mickelson, and Molly E. McCue. Next Generation Genomic Tools for the Horse: The MNEc670k SNP chip - designing a high density genotyping array designed for genotype imputatiion. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Functional effects of the MSTN locus. The frequency of a SINE insertion allele in the promoter region of the MSTN gene is approximately 0.90 in the American Quarter Horse. We have confirmed the association between MSTN promoter SINE insertion genotype and increased type 2B fiber proportion. We have also genotyped multiple different breeds for the major MSTN variants and detailed the haplotype distribution across breeds that have implications for interpreting potential functional effects of this locus on racing performance. We have also demonstrated that the SINE insertion is associated with altered metabolic phenotype in Quarter Horses. Identification of genomic regions associated with alternative gait in Standardbred horses. A recently described functional mutation in DMRT3, thought to be permissive for the ability to perform alternative gaits in the horse, is nearly fixed in Standardbred horses despite the fact that not all Standardbreds naturally pace. We have worked to identify the regions associated with pace versus trot in the Standardbred horse using a combination of GWAS with 542 Standardbred horses, whole genome sequencing of 18 individual horses, and pooled whole genome sequencing. Using these methods we identified a handful of regions across the genome that appear to segregate between pacers and trotters and have discovered variants for the design of a custom Sequenom genotyping assay with 303 putative functional alleles and 100 ancestry informative markers. ~700 horses 428 trotters, 231 pacers) were genotyped using this Sequenom assay. Association analysis using mixed regression model (GEMMA) identified 156 SNPs that were statistically significant Random forest analysis resulted in an out-of-box error rate of 1.01%. The SNPs identified as most important in correctly classifying individuals as pacers or trotters were located within the same loci identified by the GEMMA mixed model analysis. 10-fold cross-validation of the data using linear discriminate analysis (LDA) resulted in an estimated misclassification error rate of 0.01 (1%). Conditional inference models were created and evaluated; a model with as few as 7 SNPs could accurately predict an individual as a pacer or trotter with 98.6% overall accuracy (97.1% sensitivity, 99.3% specificity with "pacer" set as the "positive" condition). Design of 2M and 670k SNP arrays. Considerable effort was invested to create high-density SNP arrays for use by the equine genomics community; these arrays would also greatly improve our own abilities to identify signatures of selection. Whole genome sequences (6-12X coverage) from 166 horses from 32 breeds were used to identify > 23M SNPs with the Genome Analysis Toolkit (GATK) UnifiedGenotyper and Samtools mpileup. 2,001,826 high quality SNPs were included on a 2M SNP array. 347 horses representing 20 breeds (3-22 per breed), including ≥ 18 horses from the 11 target breeds representing diversity in the horse were genotyped on these 2M SNP arrays. After QC based on genotyping in these 347 horses, a set of 1,846,988 SNPs (~1.8M) were retained for further analysis. SNPs were selected for a 670K array with the goal of maximizing imputation accuracy. SNPs tagging the underlying haplotype structure (tagSNPs) were identified using FastTagger. Both across breed (inter-breed) as well as breed-specific (intra-breed) tagSNPs were identified. TagSNPs from each approach were collapsed using multiPopTagSelect. This resulted in 355,903 inter-breed tag SNPs and 206,822 intra-breed tag SNPs present in at least 5 breed-specific tagSNP lists. 13,993 SNPs identified as tag SNPS in ≥ 3 out of the 4 most diverse breeds were included to improve imputation efficiency in these breeds. 16,398 SNPs were added to 50 kb windows with < 8 SNPs, 7,394 SNPs were included within the MHC, and 70,295 SNPs were included from the previous 54K/65K arrays, resulting in a total of 670,840 SNPs on the 670K array. Identification of signatures of selection in racing horses and subpopulations of Quarter Horses. We focused on identifying genomic signatures of selection in the domestic horse using data available form the new 2M SNP high-density genotyping array, which allowed for better delineation of fine scale haplotype structure in regions identified as genomic signatures of selection. Signatures of selection for dense genotyping data were identified in 9 target breeds chosen to represent the major breed groupings of domestic horses as defined by PCA, FST, and a distance-based NJ tree. Each of these breeds has high diversity, large census population size, or is the subject of active genomic research. We next worked to identify the genomic regions undergoing positive selection in three breeds of racehorses, and to identify candidate genes within these regions that contribute to the racing phenotype. For that, 546 Standardbreds, 519 Thoroughbreds, and 24 elite racing Quarter Horses were genotyped for 2 million DNA markers across the genome, and their genetic profiles were compared to those of 520 individuals from 13 non-racing breeds listed above. Genomic regions harboring signatures of selection in the racehorses were identified using di, runs of homozygosity, hapFLK, and hapQTL. A total of 125 regions of interest (ROI), distributed across all 32 equine chromosomes, were identified as undergoing positive selection in these individualsInteresting findings included genes associated with exercise-related physiology, such as the CASQ2 gene (located on chromosome 5) that is involved in skeletal muscle contraction; and THSD4 (chromosome 1) and FSTL4 (chromosome 14), associated with lung and heart function, respectively. Lastly, we sought to characterize the major genetic differences that contribute to athletic performance traits within six performance groups of the American Quarter Horse that have distinct genetic profiles, namely racing, halter, cutting, working cow, reining and western pleasure. We investigated the genetic profile of 144 elite individuals (24 from each of the six subpopulations) across the 2 M SNP markers. di, hapFLK, and hapQTL were used to pinpoint regions of the genome undergoing positive selection in each subpopulation. A total of 152 genomic regions were found to be undergoing positive selection in the study cohort. These regions were identified across all 31 chromosomes; on 5 of these (chromosomes 1, 10, 18, 20 and 28), at least one ROI was shared among all subpopulations. Identification of genomic regions associated with height and metabolic traits. We have identified a region in the genome that appears to control small stature in ponies and also may explain metabolic differences between horses and ponies. It has been previously shown that pony breeds have higher fasting insulin concentrations and are more insulin resistant than horses. In humans and mice, genetic alleles resulting in small skeletal size have also been associated with alterations in insulin and glucose metabolism. We hypothesized that region(s) of the genome associated with height also play a role in metabolic traits in ponies. Genome-wide association performed for height and fasting insulin in 232 WP and 286 Morgan horses identified a region on chromosome 6 to be significantly associated with height and fasting insulin in the WP, but not in Morgans. The HMGA2 gene, contained within this region, was identified as a potential candidate gene, due to a recently identified mutation associated with height in Shetland ponies. We then analyzed 250 WP and 60 Morgans for this HMGA2 mutation and identified a frequency of 0.802 and 0.375, respectively, for the mutated form of the gene (which confers shorter stature to carriers). Further, positive correlations were also identified for the additive effect of the mutated form of the gene on height and baseline insulin levels in both WP and Morgans.

Publications

  • Type: Journal Articles Status: Published Year Published: 2015 Citation: Petersen JL, Mickelson JR, Valberg SJ, McCue ME. Genome-wide SNP data shows little differentiation between the Appaloosa and other American stock horse breeds. Animal Genetics Article first published online: 22 MAY 2015 DOI: 10.1111/age.12301. 46.5 (2015): 585-586.
  • Type: Journal Articles Status: Published Year Published: 2014 Citation: Petersen JL, Valberg SJ, Mickelson JR, McCue ME. Haplotype diversity in the equine myostatin gene with focus on variants associated with race distance propensity and muscle fiber type proportions. Animal Genetics published online: 26 August 2014 DOI: 10.1111/age.12205. 2014 Dec; 45(6):827-35 PMID:25160752.
  • Type: Journal Articles Status: Published Year Published: 2014 Citation: McCoy AM, Schaefer R, Petersen JL, Slamka MA, Morrell PL, Mickelson JR, Valberg SJ, McCue ME. Evidence of Positive Selection for the GYS1 mutation in Equine Polysaccharide Storage Myopathy. Journal of Heredity 105:163-172, 2014. PMCID: PMC3920812.
  • Type: Journal Articles Status: Published Year Published: 2014 Citation: Petersen JL, Cleary K, Mickelson JR, McCue ME. The American Quarter Horse: Population Structure and relationship to the Thoroughbred. Journal of Heredity 105:148-162, 2014. PMCID: PMC3920813.
  • Type: Journal Articles Status: Published Year Published: 2013 Citation: Petersen JL, Mickelson JR, Cothran EG, Rendahl AK, Andersson LS, Axelsson J, Bailey E, Bannasch D, Binns MM, Borges AS, Brama P, de C�mara Machado A, Distl O, Felicetti M, Fox-Clipsham L, Graves KT , Gu�rin G, Haase B, Hasegawa T, Hemmann K, Hill EW, Leeb T, Lindgren G, Lohi H, Lopes MS, McGivney BA, Mikko S, Orr N, Penedo MCT, Piercy RJ, Raekallio M, Rieder S, R�ed KH, Silvestrelli M, Swinburne J, Tozaki T, Vaudin M, Wade CM, McCue ME. Genetic diversity in the modern horse illustrated from genome-wide SNP data. PLoS One 2013: 8(1): e54997 DOI:10.1371/journal.pone.0054997.
  • Type: Journal Articles Status: Published Year Published: 2013 Citation: Petersen JL, Mickelson JR, Rendahl AK, Valberg SJ, Andersson LS, Axelsson J, Bailey E, Bannasch D, Binns MM, Borges AS, Brama P, de C�mara Machado A, Capomaccio S, Cappelli K, Cothran EG, Distl O, Fox-Clipsham L, Graves KT , Gu�rin G, Haase B, Hasegawa T, Hemmann K, Hill EW, Leeb T, Lindgren G, Lohi H, Lopes MS, McGivney BA, Mikko S, Orr N, Penedo MCT, Piercy RJ, Raekallio M, Rieder S, R�ed KH, Swinburne J, Tozaki T, Vaudin M, Wade CM, McCue ME. Genome-wide analysis reveals selection for important traits in domestic horse breeds. PLoS Genetics 2013: 9(1): e1003211, DOI:10.1371/journal.pgen.1003211.
  • Type: Journal Articles Status: Published Year Published: 2012 Citation: 24. Andersson LS, Schwochow D, Larhammar M, Rubin CJ, Arnason T, Wellbring L, Hj�lm G, Imsland F, Petersen JL, McCue ME, Mickelson JR, Cothran G, Roepstorff L, Mikko S, Kullander K, Lindgren G and Andersson L. Mutations in DMRT3 affect locomotion in horses and spinal circuit function in mice. Nature 2012: 488(7413):642-6. DOI: 10.1038/nature11399.


Progress 04/01/15 to 03/31/16

Outputs
Target Audience:Equine genomics researchers, horse owners, breeders and breed organizations. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The work over the past year was conducted by several students in the laboratory. Felipe Avila PhD, is a post-doctoral fellow hired in September 2014. Dr. Avila has been working on refinement of signatures of selection using 2 million array data across breeds for multiple loci. SNP array design and genotyping array work was completed by Dr. Robert Schaefer a new post-doctoral fellow in the laboratory. How have the results been disseminated to communities of interest?Data has been presented at the Plant and Animal Genome Meeting (2012, 2013, 2014, 2015, 2016), the International Society for Animal Science Meeting, International Society of Animal Genetics (2012), the European Federation of Animal Sciences (2012), and the Havemeyer Equine Genomics Workshop (2013, 2015), The Alberta Horse Owners Conference (2016), and the Colorado State Ranch Owners Forum (2016). And it will also be reported at the American College of Veterinary Internal Medicine Forum (2016). What do you plan to do during the next reporting period to accomplish the goals?Between the end of this reporting period (March 2016 and the end of the award (June 2016) we will work to prepare manuscripts (3) detailing the signatures of selection outcomes.

Impacts
What was accomplished under these goals? During the year 4 funding we focused on refining genomic signatures of selection in the domestic horse using data available form the new 2 million SNP high-density genotyping array, which allowed for better delineation of fine scale haplotype structure in regions identified as genomic signatures of selection (objective 3). Signatures of selection for dense genotyping data were identified in 9 target breeds chosen to represent the major breed groupings of domestic horses as defined by PCA, FST, and a distance-based NJ tree (Petersen et al 2013 PLoS One). Each of these breeds has high diversity, large census population size, or is the subject of active genomic research. The 9 breeds with this high-density SNP genotype data included: Lusitano, n=21; Icelandic, n=18; Belgian, n=21; Morgan, n=43; Thoroughbred n=24; Quarter Horse n=77; Standardbred, n=22; Arabian, n=21; and Welsh Pony, n=41. We have worked to identify the regions associated with pace versus trot in the Standardbred horse using a combination of GWAS, whole genome sequencing of individual horses, and pooled whole genome sequencing. Using these methods we identified a handful of regions across the genome that appear to segregate between pacers and trotters and have discovered variants for the design of a custom Sequenom genotyping assay with 303 putative functional alleles and 100 ancestry informative markers. ~700 horses were genotyped using this Sequenom assay. Association analysis using mixed regression model (GEMMA) identified 156 SNPs that were statistically significant as determined by a Bonferroni correction (p<2.06x10-4). Within the top hits were multiple SNPs clustered together in 6 loci on 4 chromosomes. Random forest analysis resulted in an out-of-box error rate of 1.01%. The SNPs identified as most important in correctly classifying individuals as pacers or trotters were located within the same loci identified by the GEMMA mixed model analysis. 10-fold cross-validation of the data using linear discriminate analysis (LDA) resulted in an estimated misclassification error rate of 0.01 (1%). Conditional inference models were created and evaluated; a model with as few as 7 SNPs could accurately predict an individual as a pacer or trotter with 98.6% overall accuracy (97.1% sensitivity, 99.3% specificity with "pacer" set as the "positive" condition). We are currently testing these markers in an independent validation population of Standardbred horses and in other gaited breeds including Tennessee Walking Horses (TWH), Paso Finos, Peruvian Pasos, and Missouri Fox Trotters, among others. We continue to work on the discovery of putative functional variants that underlie differences in body size. We have identified a region in the genome that appears to control small stature in ponies and also may explain metabolic differences between horses and ponies. We are currently sequencing two candidate genes in this region. We have also phenotyped an across breed population of ~1500 horses for body size, and a population of ~900 horses for metabolic traits to allow for follow-up of putative functional variants. Other regions that are being actively investigated are a putative region underlying race performance that is shared by Thoroughbreds, Racing Quarter Horses and Standardbreds, as well as two regions on other chromosomes that potentially represent chromosomal rearrangements with high frequency in three different breeds. We have phenotyped and collected DNA samples form nearly 2000 Standardbred horses to follow-up on racing performance and gait projects.

Publications

  • Type: Journal Articles Status: Published Year Published: 2015 Citation: Petersen JL, Mickelson JR, Valberg SJ, McCue ME. Genome-wide SNP data shows little differentiation between the Appaloosa and other American stock horse breeds. Animal Genetics Article first published online: 22 MAY 2015 DOI: 10.1111/age.12301. 46.5 (2015): 585-586.


Progress 04/01/14 to 03/31/15

Outputs
Target Audience: Nothing Reported Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The work over the past year was conducted by several students in the laboratory. Work on allele discovery and follow-up genotyping for gait has been performed by a PhD student (Annette McCoy DVM, MS, DACVS) and by a veterinary student summer scholar (Samantha Beeson). Dr. McCoy (DVM, PhD, DACVS) is now an assistant professor at the University of Illinois. Ms. Beeson is now a DVM/PhD dual degree student and began her full time PhD training in September 2014. Felipe Avila PhD, is a new post-doctoral fellow hired in September 2014. Dr. Avila has been working on refinement of signatures of selection using 2 million array data across breeds for multiple loci. ECA11 size locus has also been the focus of a laboratory rotation for a PhD student, Dr. Kate O'Conor (DVM, MS, DACVIM). SNP genotyping array work was done by Mr. Robert Schaefer (PhD expected August 2015). Training was also provided for an undergraduate student (Nicole Tate) who has now transitioned to a full time laboratory technician position. Ms. Tate will by applying to graduate programs in Fall 2015. How have the results been disseminated to communities of interest?Data has been presented at the Plant and Animal Genome Meeting (2015), Five abstracts related to this funding have been accepted for presentation at the Havemeyer Equine Genomics Workshop (2015). What do you plan to do during the next reporting period to accomplish the goals?During year four we will continue our work to test putative alleles for association with size, gait and performance, and we will also use the 2 M SNP genotypes collected across the 11 breeds to look for additional signatures of selection across the genome and to refine previously identified haplotypic windows. Gait loci will be investigated in additional breeds as described above. Putative function variants for size will be investigated in ~900 horses from multiple breeds.

Impacts
What was accomplished under these goals? We finished the development of 2 high density genotyping arrays (~1.8 million SNPs and ~670,000 SNPs), which will allow for better delineation of fine scale haplotype structure across breeds (objective 1), and also followed-up regions identified as genomic signatures of selection (objective 3). SNPs for array construction. Briefly, 3-30x paired-end Illumina HiSeq 100 bp whole genome sequences from 166 horses representing 32 breeds were mapped to an extended version of the EquCab 2 genome. Variants were called using both the GATK UnifiedGenotyper and Samtools mpileup. > 23 million variants were called by both tools. After initial quality control filtering based on Affymetrix initial design criteria, ~10.6M SNPs were further filtered based on genome distribution, variant calling group, and linkage disequilibrium (LD). LD was calculated for all SNPs pairs within 10kb. When r2 was > 0.90, the SNP with representation in draft and pony breeds. 5,352,650 newly discovered SNPs, as well as ~74,000 VIP SNPs from previous 54K/65K arrays were submitted to Affymetrix for probe design. Based on design scores, ~37 SNPs were chosen per 50kb window, including all VIP SNPs from previous generation equine SNP arrays unless probe design was not possible (70,295 SNPs), and all SNPs in the equine MHC. In total, 2,076,397 SNPs were included. 2 M SNP array genotyping. 9 target breeds representing the major breed groupings of domestic horses as defined by PCA, FST, and a distance-based NJ tree (Petersen et al 2013 PLoS One)., chosen for high diversity, large census population size, or is the subject of active genomic research were included in genotyping: Lusitano, n=21; Icelandic, n=18; Belgian, n=21; Morgan, n=43; Thoroughbred n=24; Quarter Horse n=77; Standardbred, n=22; Arabian, n=21; and Welsh Pony, n=41. Samples included 6 trios. Additional individuals from 11 other breeds were genotyped (n=3-22 per breed): total of 347 horses from 20 breeds. Based on genotyping results SNPs were categorized based on conversion rates and genotyping quality using Affymetrix's QC criteria. 1,846,993 of the 2,076,397 SNPs passed quality control. SNP selection for the 670K commercial array. 2 M SNP array genotypes from the 347 horses were combined with genotypes from 175 horses in which whole genome sequences were available; resulting in a total of 522 horses with genotypes at 1,846,993 SNPs. SNPs were selected for the 670K array with the goal of maximizing imputation accuracy. SNPs tagging the underlying haplotype structure (tag SNPs) were identified using FastTagger. To identify inter-population tag SNPs, FastTagger was run on the entire dataset using a minor allele frequency of ≥ 0.01 and an r-squared of ≤ 0.99; resulting in the identification of 355,903 inter-population tag SNPs. To identify intra-population tagSNPs, samples were split into 15 breed groups. FastTagger was run with parameters to differentiate tagSNPs within each population on a resolution of ≥ 0.10 minor allele frequency and r-squared ≤ 0.90. Tag SNP sets from each population were collapsed using the program multiPopTagSelect which assesses overlap between tag SNP sets and to identify the subset of SNPs which minimally spans the populations. 206,822 intra-breed tag SNPs that were present in at least 5 breed-specific tag SNP lists were included in the final array design. An additional 13,993 SNPs identified as tag SNPS in ≥ 3 out of the 4 most diverse breeds were included to improve imputation efficiency in these breeds. An additional 16,398 SNPs were added within 50 kb windows with < 8 SNPs, 7,394 SNPs were included within the MHC region, and 70,295 SNPS were included from previous generations arrays, resulting in a total of 670,840 SNPs. The imputation accuracy from 670K to ~1.8M genotypes for the breeds represented by at least 10 individuals ranged from ~95% to 98.5%. Genomic loci that determine gait. Using GWAS, whole genome sequencing of individual horses, and pooled whole genome sequencing,we identified a handful of regions across the genome that appear to segregate between pacers and trotters and have discovered variants for the design of a custom Sequenom genotyping assay with 303 putative functional alleles and 100 ancestry informative markers. ~700 Standardbred horses were genotyped using this Sequenom assay. Association analysis using a mixed regression model identified 156 SNPs that were statistically significant as determined by a Bonferroni correction (p<2.06x10-4). Within the top hits were multiple SNPs clustered together in 6 loci on 4 chromosomes. Random forest analysis resulted in an out-of-box error rate of 1.01%. The SNPs identified as most important in correctly classifying individuals as pacers or trotters were located within the same loci identified by the GEMMA mixed model analysis. 10-fold cross-validation of the data using linear discriminate analysis (LDA) resulted in an estimated misclassification error rate of 0.01 (1%). Conditional inference models were created and evaluated; a model with as few as 7 SNPs could accurately predict an individual as a pacer or trotter with 98.6% overall accuracy (97.1% sensitivity, 99.3% specificity). In the coming year we will test these markers in other gaited breeds including Tennessee Walking Horses, Paso Finos, Peruvian Pasos, and Missouri Fox Trotters, etc, to determine if any of these putative gait-modifying variants and loci are shared across breeds. Genomic loci that determine body size. We continue to work on the discovery of putative functional variants on ECA11 that underlie differences in body size. We have been unable to successfully Sanger sequence through all the candidate genes in this region, and have generated whole genome sequences form several draft horses, Welsh ponies, Shetland Ponies and Miniature horses to allow for de novo assembly in the region. Further, using 1.8 million array genotyping data we have performed fine scale haplotype analysis in this region on ECA 11 narrowing the putative candidate genes and refining the association. Variant discovery in several biologically relevant candidate genes is underway. We have also phenotyped an across breed population of ~900 horses for body size, to allow for follow-up of putative functional variants. Functional effects of a MSTN allele. We have followed-up our initial work associating variants in the myostatin (MSTN) gene with muscle fiber type in gluteal muscle biopsies of Quarter Horses in both Thoroughbred and Belgian horses and confirmed the association between MSTN promoter SINE insertion genotype and increased type 2B fiber proportion. We have also genotyped multiple other breeds for the major MSTN variants and have a manuscript detailing the haplotype distribution across breeds that have implications for interpreting potential functional effects of this locus on racing performance (Petersen et al., 2014, Anim Genet 45, 827). We have also demonstrated that the SINE insertion is associated with altered metabolic phenotype in Quarter Horses. The SINE is significantly associated with girth to height ratio (a measure of adiposity), fasting and post oral sugar test insulin, and fasting leptin and adiponectin concentrations. With each additional copy of the SINE allele, horses have less apparent adiposity, and lower fasting and OST insulin, and leptin concentrations. In contrast, serum adiponectin concentrations were positively associated with the SINE, with adiponectin concentrations increasing with each additional SINE allele. Serum triglyceride concentrations also tend to be lower with additional copies of the SINE. Other loci. Other regions that are being actively investigated are a putative region underlying race performance that is shared by Thoroughbreds, Racing Quarter Horses and Standardbreds, as well as two regions on other chromosomes that potentially represent chromosomal rearrangements with high frequency in three different breeds.

Publications

  • Type: Journal Articles Status: Published Year Published: 2014 Citation: McCoy AM, Schaefer R, Petersen JL, Slamka MA, Morrell PL, Mickelson JR, Valberg SJ, McCue ME. Evidence of Positive Selection for the GYS1 mutation in Equine Polysaccharide Storage Myopathy. Journal of Heredity 105:163-172, 2014. PMCID: PMC3920812.
  • Type: Journal Articles Status: Published Year Published: 2014 Citation: Petersen JL, Valberg SJ, Mickelson JR, McCue ME. Haplotype diversity in the equine myostatin gene with focus on variants associated with race distance propensity and muscle fiber type proportions Animal Genetics published online: 26 August 2014 DOI: 10.1111/age.12205. PMID:25160752.
  • Type: Journal Articles Status: Accepted Year Published: 2015 Citation: Petersen JL, Mickelson JR, Valberg SJ, McCue ME. Genome-wide SNP data shows little differentiation between the Appaloosa and other American stock horse breeds. Animal Genetics accepted for publication


Progress 04/01/13 to 03/31/14

Outputs
Target Audience: Target audiences include equine genome research community (SNP array development); breeds/breed groups (38 total) in which diversity and signatures of selection data were produced, and students who recieved training and veterinarians and horse owners. Changes/Problems: We have moved to design and use of a new high density array (2 million SNPs) for fine scale haplotype analysis instead of using the 65,000 SNP array. What opportunities for training and professional development has the project provided? This work was conducted primarily by a post-doctoral fellow, Dr Jessica Petersen. Dr Petersen is now a tenure-track Assistant Professor at the University of Nebraska-Lincoln. Work on allele discovery and follow-up genotyping for gait has been performed by a PhD student (Annette McCoy DVM) and by a veterinary summer scholar (Samantha Beeson). We are in the process of hiring a new post-doctoral researcher, with a tentative start date of May 2014. And we anticipate the Ms Beeson will enter the graduate program as a DVM/PhD dual degree student to continue work on related project goals. How have the results been disseminated to communities of interest? Data has been presented at the Plant and Animal Genome Meeting (2012, 2013, 2014), the International Society for Animal Science Meeting, International Society of Animal Genetics (2012), the European Federation of Animal Sciences (2012) and the Havemeyer Equine Genomics Workshop (2013). What do you plan to do during the next reporting period to accomplish the goals? The 2M SNP array data will be used to investigate haplotype sharing and fine scale haplotype structure across breeds that are part of our signatures of selection study. During year three we will continue our work to test putative alleles for association with size, gait and performance, and we will also use the 2 M SNP genotypes collected across the 11 breeds to look for additional signatures of selection across the genome and to refine previously identified haplotypic windows. Side benefits of our development of this 2 M SNP chip construction, genotyping and haplotype resource effort include: the capability to perform imputation to enhance ongoing GWAS analyses for many traits by the international equine genomics community; and the selection of 670K SNPs to be used on a very affordable commercially-available genotyping array.

Impacts
What was accomplished under these goals? During the second year of funding, we have focused on the development of a new high-density SNP genotyping array, which will allow for better delineation of fine scale haplotype structure across breeds (objective 1), and also on following-up regions identified as genomic signatures of selection (objective 3). Briefly, we have capitalized on whole genome sequences form 166 horses representing 32 breeds to discover 31 million novel equine SNPs. From these SNPs we have identified ~5.3 million SNPS that are of the highest quality and represented across multiple breeds and breed groups. These 5.3 millions SNPs have been submitted for design QC for the Affymetrix axiom platform and will be used to design a 2 million (2M) SNP test array. 384 horses from 11 different breeds will be genotyped on this 2M array and these genotypes will be used to investigate haplotype sharing and fine scale haplotype structure across breeds. We have continued to follow-up on genomic signatures of selection for gait, size and other performance phenotypes. Briefly for the gait work, we have worked to identify the regions associated with pace versus trot in the Standardbred horse using a combination of GWAS, whole genome sequencing of individual horses, and pooled whole genome sequencing. Using these methods we identified a handful of regions across the genome that appear to segregate between pacers and trotters and have discovered variants for the design of a custom Sequenom genotyping assay. We have recently completed genotyping for ~450 putative functional alleles and 100 ancestry informative markers in a cohort of 500 horses, and are in the process of analyzing this data to identify those alleles that are most likely underlying pace versus trotting ability in Standardbreds. We are also actively exploring genomic signatures of selection in other gaited breeds including Tennessee Walking Horses (TWH), Paso Finos and Peruvian Pasos, including analysis of capture and resequencing data in the TWH. We are also in the process of collecting DNA samples from TWH for which quantitative scoring for gait has been performed. We continue to work on the discovery of putative functional variants on ECA11 that underlie differences in body size. We have been unable to successfully Sanger sequence through all the candidate genes in this region, and have generated whole genome sequences form several draft horses to allow for de novo assembly in the region. Other regions that are being actively investigated are a putative region underlying race performance that is shared by Thoroughbreds, Racing Quarter Horses and Standardbreds, as well as two regions on other chromosomes that potentially represent chromosomal rearrangements with high frequency in three different breeds. We have followed-up our initial work associating variants in the myostatin (MSTN) gene with muscle fiber type in gluteal muscle biopsies of Quarter Horses in both Thoroughbred and Belgian horses and confirmed the association between MSTN promoter SINE insertion genotype and increased type 2B fiber proportion. We have also genotyped multiple different breeds for the major MSTN variants and are have a manuscript detailing the haplotype distribution across breeds that have implications for interpreting potential functional effects of this locus on racing performance in preparation.

Publications

  • Type: Journal Articles Status: Published Year Published: 2014 Citation: McCoy AM, Schaefer R, Petersen JL, Slamka MA, Morrell PL, Mickelson JR, Valberg SJ, McCue ME. Evidence of Positive Selection for the GYS1 mutation in Equine Polysaccharide Storage Myopathy. Journal of Heredity 2014: 105 (2): 163-172, published online: November 9, 2013; DOI: 10.1093/jhered/est075.
  • Type: Journal Articles Status: Published Year Published: 2014 Citation: Petersen JL, Cleary K, Mickelson JR, McCue ME. The American Quarter Horse: population structure and relationship to the Thoroughbred. Journal of Heredity 2014: 105(2): 148-162, published online: November 23, 2013 DOI: 10.1093/jhered/est079.
  • Type: Journal Articles Status: Published Year Published: 2014 Citation: McCoy AM, McCue ME. Validation of imputation between equine genotyping arrays. Animal Genetics 2014: 45 (1): 153, published online: October 27, 2013 DOI: 10.1111/age.12093.


Progress 04/01/12 to 03/31/13

Outputs
Target Audience: During the first year of funding, we have completed the bulk of the work outlined in objective 1. Namely we have used a genome-wide set of autosomal SNPs and 814 horses from 36 breeds to provide the first detailed description of equine breed diversity. FST calculations, parsimony, and distance analysis demonstrated relationships among the breeds that largely reflect geographic origins and known breed histories. Low levels of population divergence were observed between breeds that are relatively early on in the process of breed development, and between those with high levels of within-breed diversity, whether due to large population size, ongoing outcrossing, or large within-breed phenotypic diversity. Populations with low within-breed diversity included those which have experienced population bottlenecks, have been under intense selective pressure, or are closed populations with long breed histories. Cluster analysis was performed with an increase in an overall mean ln at K=35, however a true value of K was not obvious from the data. These data were published in early 2013. Ancestry informative markers (AIMs) have been identified across breeds with nearly all the diversity present and captured by PCA being captured by as few as 1000 markers. A manuscript detailing the AIMs and the new algorithm developed for AIMs discovery is in preparation. These results provide new insights into the relationships among and the diversity within breeds of horses. In addition these results will facilitate future genome-wide association studies and investigations into genomic targets of selection. The Fst-based approached to identifying signatures of selection in Objective 2 has been completed. 744 individuals from 33 breeds, and a 54,000 SNP genotyping array, breed-specific targets of selection were identified using an FST-based statistic calculated in 500-kb windows across the genome. A 5.5-Mb region of ECA18, in which the myostatin (MSTN) gene was centered, contained the highest signature of selection in both the Paint and Quarter Horse. Gene sequencing and histological analysis of gluteal muscle biopsies showed a promoter variant and intronic SNP of MSTN were each significantly associated with higher Type 2B and lower Type 1 muscle fiber proportions in the Quarter Horse, demonstrating a functional consequence of selection at this locus. Signatures of selection on ECA23 in all gaited breeds in the sample led to the identification of a shared, 186-kb haplotype including two doublesex related mab transcription factor genes (DMRT2 and 3). The recent identification of a DMRT3 mutation within this haplotype, which appears necessary for the ability to perform alternative gaits, provides further evidence for selection at this locus. Finally, putative loci for the determination of size were identified in the draft breeds and the Miniature horse on ECA11, as well as when signatures of selection surrounding candidate genes at other loci were examined. This work provides further evidence of the importance of MSTN in racing breeds, provides strong evidence for selection upon gait and size, and illustrates the potential for population-based techniques to find genomic regions driving important phenotypes in the modern horse. Progress on objective 3 includes the design of a Agilent capture array to interrogate 7 genomic regions in 4 breeds, and sequencing of pooled DNA for target and non-target horses for each breed. Due to the falling costs of next generation sequencing, and the poor capture at the end of bait tiling regions, additional next generation sequencing efforts will focus on whole genome sequencing for variant discovery. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? This work is being conducted primarily by a post-doctoral fellow, Dr Jessica Petersen. Follow-up Sanger sequencing and genotyping is being performed by two undergraduate students. Evaluation of whole genome sequences for variant discovery in regions of interest for one trait, gait (pace vs trot) is being performed by a veterinary summer scholar. How have the results been disseminated to communities of interest? Data has been presented at the Plant and Animal Genome Meeting, the International Society for Animal Science Meeting and the Havemeyer equine Genomics Workshop. What do you plan to do during the next reporting period to accomplish the goals? During year two, we will focus on identification of functional alleles underlying loci identified by the FST-based statistic, and testing these alleles for genetic association in larger phenotyped populations of horses. Primary targets are the modifying locus affecting pace versus trot in the Standardbred, and the putative size locus on ECA11.

Impacts
What was accomplished under these goals? During the first year of funding, we have completed the bulk of the work outlined in objective 1. Namely we have used a genome-wide set of autosomal SNPs and 814 horses from 36 breeds to provide the first detailed description of equine breed diversity. FST calculations, parsimony, and distance analysis demonstrated relationships among the breeds that largely reflect geographic origins and known breed histories. Low levels of population divergence were observed between breeds that are relatively early on in the process of breed development, and between those with high levels of within-breed diversity, whether due to large population size, ongoing outcrossing, or large within-breed phenotypic diversity. Populations with low within-breed diversity included those which have experienced population bottlenecks, have been under intense selective pressure, or are closed populations with long breed histories. Cluster analysis was performed with an increase in an overall mean ln at K=35, however a true value of K was not obvious from the data. These data were published in early 2013. Ancestry informative markers (AIMs) have been identified across breeds with nearly all the diversity present and captured by PCA being captured by as few as 1000 markers. A manuscript detailing the AIMs and the new algorithm developed for AIMs discovery is in preparation. These results provide new insights into the relationships among and the diversity within breeds of horses. In addition these results will facilitate future genome-wide association studies and investigations into genomic targets of selection. The Fst-based approached to identifying signatures of selection in Objective 2 has been completed. 744 individuals from 33 breeds, and a 54,000 SNP genotyping array, breed-specific targets of selection were identified using an FST-based statistic calculated in 500-kb windows across the genome. A 5.5-Mb region of ECA18, in which the myostatin (MSTN) gene was centered, contained the highest signature of selection in both the Paint and Quarter Horse. Gene sequencing and histological analysis of gluteal muscle biopsies showed a promoter variant and intronic SNP of MSTN were each significantly associated with higher Type 2B and lower Type 1 muscle fiber proportions in the Quarter Horse, demonstrating a functional consequence of selection at this locus. Signatures of selection on ECA23 in all gaited breeds in the sample led to the identification of a shared, 186-kb haplotype including two doublesex related mab transcription factor genes (DMRT2 and 3). The recent identification of a DMRT3 mutation within this haplotype, which appears necessary for the ability to perform alternative gaits, provides further evidence for selection at this locus. Finally, putative loci for the determination of size were identified in the draft breeds and the Miniature horse on ECA11, as well as when signatures of selection surrounding candidate genes at other loci were examined. This work provides further evidence of the importance of MSTN in racing breeds, provides strong evidence for selection upon gait and size, and illustrates the potential for population-based techniques to find genomic regions driving important phenotypes in the modern horse. Progress on objective 3 includes the design of a Agilent capture array to interrogate 7 genomic regions in 4 breeds, and sequencing of pooled DNA for target and non-target horses for each breed. Due to the falling costs of next generation sequencing, and the poor capture at the end of bait tiling regions, additional next generation sequencing efforts will focus on whole genome sequencing for variant discovery.

Publications

  • Type: Journal Articles Status: Published Year Published: 2013 Citation: Petersen, JL, Mickelson, JR, Rendahl, AK, Valberg, SJ, Andersson, LS, Axelsson, J, Bailey, E, Bannasch, D, Binns, MM, Borges, AS, Brama, P, de C�mara Machado, A, Capomaccio, S, Cappelli, K, Cothran, EG, Distl, O, Fox-Clipsham, L, Graves, KT , Gu�rin, G, Haase, B, Hasegawa, T, Hemmann, K, Hill, EW, Leeb, T, Lindgren, G, Lohi, H, Lopes, MS, McGivney, BA, Mikko, S, Orr, N, Penedo, MCT, Piercy, RJ, Raekallio, M, Rieder, S, R�ed, KH, Swinburne, J, Tozaki, T, Vaudin, M, Wade, CM, McCue, ME. Genome-wide analysis reveals selection for important traits in domestic horse breeds. PLoS Genetics 9(1): e1003211, doi:10.1371/journal.pgen.1003211
  • Type: Journal Articles Status: Published Year Published: 2012 Citation: 11. Andersson, LS, Schwochow, D, Larhammar, M, Rubin, CJ, Arnason, T, Wellbring, L, Hj�lm, G, Imsland, F, Petersen, JL, McCue, ME, Mickelson, JR, Cothran, G, Roepstorff, L, Mikko, S, Kullander, K, Lindgren, G and Andersson, L. Mutations in DMRT3 affect locomotion in horses and spinal circuit function in mice. Nature 2012 Aug 30; 488(7413):642-6. doi: 10.1038/nature11399
  • Type: Journal Articles Status: Published Year Published: 2013 Citation: Petersen, JL, Mickelson, JR, Cothran, EG, AK, Andersson, LS, Axelsson, J, Bailey, E, Bannasch, D, Binns, MM, Borges, AS, Brama, P, de C�mara Machado, A, Distl, O, Felicetti, M, Fox-Clipsham, L, Graves, KT , Gu�rin, G, Haase, B, Hasegawa, T, Hemmann, K, Hill, EW, Leeb, T, Lindgren, G, Lohi, H, Lopes, MS, McGivney, BA, Mikko, S, Orr, N, Penedo, MCT, Piercy, RJ, Raekallio, M, Rieder, S, R�ed, KH, Silvestrelli, M, Swinburne, J, Tozaki, T, Vaudin, M, Wade, CM, McCue, ME. Genetic diversity in the modern horse illustrated from genome-wide SNP data. PLoS One 8(1): e54997. doi:10.1371/journal.pone.0054997.