STUDY OF GENOMIC STRUCTURE AND SIGNATURES OF RECENT POSITIVE SELECTION IN CATTLE | STUDY OF GENOMIC STRUCTURE AND SIGNATURES OF RECENT POSITIVE SELECTION IN CATTLE

The knowledge of the extent and pattern of linkage disequilibrium (LD) is necessary for estimating the number of SNPs required for implementing association mapping studies as well as describing genomic structure of the bovine genome as a whole. In the first work of this study we used Illumina Bovine SNP50K BeadChip genotypes in a sample of German Holstein¨CFriesian cattle and developed a second generation of LD map statistics which has four times higher resolution compared to the maps available so far. These results revealed a lower level of LD for SNP pairs at distances ¡Ü100 Kb than previously thought. The level of LD obtained in this study indicated that a denser SNP map would be beneficial to capture the LD information required for whole-genome fine mapping and genomic selection and to completely assess the pattern of LD across the genome. Effective population size (Ne) was estimated based on the direct estimates of recombination rates from haplotype data and showed a persistent decline in about 100 individuals at the current generations. The impact of allele frequency in analyzing genome-wide LD was also explored in this part. Our observation revealed that minimizing the allele frequency difference between SNPs, reduces the influence of frequency on r2 estimates and provides a useful metric for analyzing LD. The larger block size in Holstein cattle observed in this study indicates substantially greater LD in cattle than in human populations. The second task of this thesis involved our attempts to find traces of decades of intensive artificial selection for traits of economically importance in modern cattle. In the first experiment we employed the recently described Extended Haplotype Homozygosity (EHH) test for tagging the genome wide footprints of positive selection in Holstein-Friesian cattle. This test uses the characteristics of haplotypes to detect selection by measuring the decay of haplotype homozygosity within a single population. To formally assess the significance of these results, we compared the combination of frequency and the Relative Extended Haplotype Homozygosity (REHH) value of each core haplotype with equally frequent haplotypes across the genome. A subset of the putative regions showing the highest significance genome-wide was mapped. Regarding the fact that problems arising from multiple testing may have affected the results we performed a further validation by aligning the 12 regions of extreme REHH to the bovine genome (Btau 4.0) to verify any coincidence of the preliminary signals observed with important genomic regions. We found co-location of a panel of genes such as FABP3, CLPN3, SPERT, HTR2A5, ABCE1, BMP4 and PTGER2 and some others with putative regions. This panel represents a broad range of economically important traits such as milk yield and composition as well as reproductive and behavioral traits. We also reported high values of LD and a slower decay of haplotype homozygosity for some candidate regions harboring major genes related to dairy quality. The results of this study provided a genome wide map of selection footprints in Holstein genome. In further experiments we exploited the variation among populations to explore the signatures of past selection. In this sense, we developed a new Bayesian approach for detecting differentiated loci based on FST and applied it to a set of geographically separated populations with identical or diverse breeding goals. This algorithm was able to deal with a large battery of marker information. Clustering the genome-wide estimates of FST values between Holstein and Brown Swiss versus Angus and Piedemontese breeds using Akaike¡¯s criterion recognized two groups, one representing putatively neutral loci, and the other possibly corresponding the genomic regions affected by selection. We examined the potential of FST analysis in detecting selection signals by testing some candidate major genes in our data set. The results revealed FST values larger than expected (P < 10%) for regions harboring the Casein cluster, GHR, STS, LP and IGF-1 genes which are supposed to be targets for artificial selection. However, we were not able to propose strong candidate genes on the basis of the gene content in the vicinity of extreme signals. As an explanation, we theorized that selection may work on genes that were not considered the primary targets of selection so far. Consistent with the previous reports our results mostly revealed gene deserts in the location of extreme peaks, which may reflect selection acting on uncharacterized regulatory region or simply fixation of non-coding DNA by genetic drift in the absence of any selection. Thus, these results in combination with the observations on human population data suggest that non-coding regions have been an important substrate for adaptive evolution. In a parallel analysis the integrated Haplotype Homozygosity Score (|iHS|) a derivation of EHH test, was applied for tracing on-going sweeps. After estimating |iHS| for each locus, we defined regions of the genome that may contain targets of positive selection as windows in the extreme of empirical distribution. This criterion resulted in 94 significant windows (P ¡Ü 0.05). These results revealed significant enrichments for genes such as SPATA17, MGAT1, PGRMC2 and SRD5A2 in the region of clustered signals which belong to the number of functional categories relevant to reproduction including gamete generation, embryo development and spermatogenesis and genes in these categories may provide strong candidates for selection for fertility traits. Another interesting observation is the presence of the genes like Actinin, Collagen and fibroblast activation protein as well as the gene responsible for developing the cartilage rudiments in the positively selected regions of beef cattle. These results suggest that selection for muscle related phenotypes play a major role in the shaping the beef cattle. These results generally are consistent with the previous reports and begin to suggest general themes about the types of genes that have been targets of positive selection in cattle genome. Overall, based on the results of this study we conclude that high-resolution genome scans using dense markers are capable to identify outlier regions that potentially contain genes contributing to within and inter-breed phenotypic variation. Our results may be of future interest for identifying signatures of recent positive artificial selection between the cattle breeds or as additional evidence for any polymorphisms that show associations with beef or milk traits.