Genome-wide scans for signatures of selection in Mangalarga Marchador horses using high-throughput SNP genotyping

Abstract

Background: The Mangalarga Marchador horse (MM) is one of the breeds shaped over generations by local adaptations and specific preferences of Brazilian breeders to morphology, functionality, and locomotion. The animals genetically have “batida” or “picada” natural gait trait, which is a trademark of the breed. The movement biomechanics of this breed promote stability during the execution, comfort, and softness of the ride. The detection of signatures of selection in genomic regions provides insights into the evolutionary process, enabling discoveries regarding complex phenotypic traits. In this research, we focused on the identification of genomic regions affected by different selection pressures, mainly highlighting recent selection, as well as understanding the candidate genes and functional pathways associated with the signatures of selection in the MM genome. A broader discussion of genes and traits of importance in this breed, especially traits related to the type and quality of the gait, temperament, conformation, and locomotor system, was also provided.Results: Three different methods were used to search for signals of selection: Tajima’s D (TD), the integrated haplotype score (iHS), and runs of homozygosity (ROH). The samples were composed of males (n=62) and females (n=130) that were initially chosen considering well-defined phenotypes for gait: picada (n=86) and batida (n=106). All horses were genotyped using a 670k Axiom ® Equine Genotyping Array​ (Axiom MNEC670). In total, 169 pruned candidate genes harboring important biological processes were found, highlighting the following: anterior/posterior pattern for the set of genes (GLI3, HOXC9, HOXC6, HOXC5, HOXC4, HOXC13, HOXC11, and HOXC10); limb morphogenesis, skeletal system, proximal/distal pattern formation, JUN kinase activity (CCL19 and MAP3K6); and muscle stretch response (MAPK14). Other candidate genes were associated with energy metabolism, bronchodilator response, NADH regeneration, reproduction, keratinization, and the immunological system.Conclusions: Collectively, our findings revealed for the first time some pieces of evidence for selection signals acting on athletic performance, gait type, and energy to muscle activity in the MM breed, with the particular involvement of the HOX family of genes. Gene network analysis showed no relationship between the signals observed in this population and the DMRT3 gene, a relevant gene associated with gait trait, however candidate genes that were located close (~28Mb) to DMRT3 were identified.

Comprehensive analysis of runs of homozygosity and heterozygosity in Holstein cattle on the basis of medium and high density SNP panels and large population sample

“…For ROH detection, the following criteria of identification were applied: a minimum number of 30 consecutive homozygous SNPs; a minimal ROH length set to 1 Mb; a maximum distance between SNPs equal to 1 Mb; a maximum of 1 SNP with a heterozygous genotype; and a maximum of 1 SNP with a missing genotype. For RO-Het, since there are no standard criteria for their analysis and only a few studies concerning ROHet exist, we followed the criteria proposed by Biscarini et al (2020), Santos et al (2021) and Li et al (2022): a minimum number of 15 consecutive homozygous SNPs; a minimal ROH length set to 1 Mb; a maximum distance between SNPs equal to 1 Mb; a maximum of 3 SNPs with homozygous genotypes; and a maximum of 2 SNPs with missing genotypes. The identified ROH were assigned to five length categories: >1 Mb, >2 Mb, >4 Mb, >8 Mb and above 16 Mb.…”

“…It should be noted that the results of ROH identification can be strongly biased by the density of the microarray used, the parameters and the method of ROH identification. Recently, the most common method of ROH identification based on a sliding window was established to provide some form of analytical bias, which is why in this study, a consecutive approach that resolves around checking homozygosity status for each adjacent SNP was used (Santos et al, 2021). Compared to the sliding window method, this approach, however, is characterized by the identification of a smaller number of long ROH, especially in the case of high-density SNP panels, due to breaks in long ROH caused by a limited number of allowed heterozygotes.…”

“…However, compared to ROH, more errors are generally allowed in ROHet; in most publications, up to three homozygous SNPs for short regions are allowed. ROHet can be associated with heterozygous clusters (Williams et al, 2016), and their analysis can be used to identify balancing selection events or loci of lethal recessive mutations (Biscarini et al, 2020;Santos et al, 2021). Only a few reports on ROHet identification can be found.…”

“…Only a few reports on ROHet identification can be found. The numbers of identified ROHet per individual in livestock vary from 9.9 for cattle (Biscarini et al, 2020) and 28.3 for sheep (Tsartsianidou et al, 2021) to 57.8 for turkeys (Marras et al, 2018) and up to 52.2 for horses (Santos et al, 2021). The numbers have differed greatly due to the methods of ROHet identification and the density of the SNP panel used.…”

“…Natural or artificial selection acts as a strong pressure in shaping ROH distribution and characteristics across the genome, allowing for tracking of recent and ancient selection history by assess-ing the size and frequency of ROH regions (Howrigan et al, 2011;Mastrangelo et al, 2016;Peripolli et al, 2017). Several studies have pinpointed that ROH hotspots are associated with positive selection events in various animal species (Szmatoła et al, 2019;Biscarini et al, 2020;Santos et al, 2021;Li et al, 2022;Szmatoła et al 2022) that in most scenarios increase homozygosity in certain genomic regions (Makino et al, 2018). Additionally, ROH can be easily utilized for estimating inbreeding via assessment of the ROH-based inbreeding coefficient (F ROH ), calculated as the ratio of the total sum of ROH per individual and the autosomal genome's length (McQuillan et al, 2008).…”