Multi-population joint genome-wide association study to detect genomic regions associated with litter size in sheep

Introduction: Reproduction is one of the most important economic traits in sheep with within and between-breeds variation. Reproductive traits normally show low to medium heritability and therefore response to conventional selection methods is not satisfactory for these traits. Considering the genetic information of the genetic variants underlying reproduction variability could efficiently increase the selection efficacy. Genome-wide association studies (GWAS) have been used to identify associations between genotypes and phenotypes as well as candidate genes for economically important traits. Statistical power in GWAS is mostly affected by sample size. The low sample size is hence a main obstacle in GWAS. Combining multiple data sets of different studies for joint (mega) GWAS provides an opportunity to increase the sample size required for GWAS. This study was performed to identify genomic regions affecting litter size in sheep using the mega-analysis of GWAS.
Materials and methods: Multi-population joint GWAS was performed using genotypic and phenotypic data of six sheep breeds retrieved from the database. Quality control was performed using the Plink software. The markers or individuals were removed from the further study based on the following criteria: (1) unknown chromosomal or physical location, call rate <0.95, missing genotype frequency >0.05, minor allele frequency (MAF) < 0.05, and a P-value for Hardy–Weinberg equilibrium test less than 10^-6. Before analysis, imputation of missing genotypes for combined data set was implemented by LD-kNNi method. Mega-analysis was performed using a mixed linear model in TASSEL software considering kinship and population structure (top five components of principal component analysis (PCA)) as confounding effects. The quantile–quantile (Q–Q) plot was visualized by plotting the distribution of obtained vs. expected log10 (P-value). The association results along the genome and the significant SNPs were visualized in the Manhattan plot. To account for multiple test problem and identify the genome-wide and chromosome-wide significance level, Bonferroni test was used based on the number of independent SNPs obtained from pairwise linkage disequilibrium analysis. After GWAS analysis, the 300 bp sequence upstream and downstream of the significant SNP was explored to identify the adjacent candidate genes using Ovis aries_v4.0 (UCSC).
Results and discussion: In the present study, we implemented a mega GWAS using six different sheep breed data to identify the genetic mechanisms responsible for litter size in sheep. After quality control, 305 animals and 351,615 SNP markers with a mean MAF of 0.33 were kept for further analysis. The results of the mega-analysis identified one marker on chromosome 21 at the genome-wide level and 10 markers at the chromosome-wide level on chromosomes 1, 2, 3, 14, 17, and 22. The quantile–quantile plot that features the total distribution of the observed P-values (−log10 P-values) of quality passed SNPs vs. the expected values, showed the effective control for confounding effects. Many of the significant SNPs identified in this study were located in or very adjacent to known genes (OPCML, GULP1, RBP4, MMP2, and LPCAT2) that have been already reported for their contribution to fertility and pregnancy success. It has been reported that OPCML is more consistently expressed in cells lining the uterus, oviduct, and rete ovarii. OPCML has been reported as a tumor suppressor protein that is frequently inactivated in epithelial ovarian cancer. It has been reported that the RBP4 gene is expressed during the period of fast elongation of the pig blastocyst which is a crucial period for the survival of the embryos. Also, it has been suggested that RBP4 has the main contribution in uterine and conceptus physiology during the establishment of pregnancy and therefore can be considered as a candidate gene for litter size. MMP2 has an essential function during ovulation and pregnancy through extracellular matrix (ECM) components degradation and therefore enabling cell migration and angiogenesis.
Conclusions: Comparison of the results of this study with previous reports showed that the mega-analysis of GWAS, compared to the meta-analysis already reported for GWAS results, had comparable power in identifying genomic regions influencing litter size in sheep but identified fewer genomic regions than individual GWAS for each breed. No previously reported major genes controlling litter size in sheep were identified using our mega GWAS. The results of our research are suggested for further investigations in identifying causal genetic variants or genomic regions underlying the litter size variation in sheep and can be used to understand the genetic mechanism controlling this trait.