Combining different pedigrees to fine-map QTL in the pig

Pig domestication started around 10,000 years ago during the Neolithic age, independently in Europe and China, and most current pig breeds originate from these two areas. Among the 560 pig breeds that have been recorded over the world in 2007 by the FAO, only few of them have been intensively selected for production. Domestication and, more recently, pig breeders have relied on naturally occurring mutations to select individuals exhibiting favourable traits related to reproduction, growth, fatness, resistance to diseases and behaviour. In order to identify these mutations underlying the phenotypic variations of these traits, a number of QTL detection programs was set up, and thousands of QTLs have been detected in the 2000s. However, for only a few of them, fine-mapping has resulted in the identification of the causal polymorphism. In chapter 1, the general introduction provides an overview of QTL detection in pig in relation to the molecular tools that are available for pig geneticists and to the different mapping strategies that can be used. Major limitations to QTL fine-mapping in pig (but also valid for other livestock species) concern the number of individuals, the number of informative genetic markers and the ability to detect non-additive QTLs. To increase the statistical power by increasing the number of individuals, a combined linkage analysis is presented in chapter 2. To carry out this work, two pig F2 pedigrees comprising about a thousand individuals each and based on similar breeds (Large White and/or Landrace crossed with Meishan individuals) were combined. Both pedigrees had been developed in the late 1990s at INRA and WUR. Common QTLs segregating on SSC2, SSC4 and SSC6 were confirmed in the combined analysis, but QTLs that were specific to one pedigree disappeared or were detected at a lower significance threshold. Despite the limited benefits in term of the number of QTLs, the increase in the number of individuals, enabled us to separate two linked QTLs that were previously detected as a single one. False-positive QTLs were also detected as well as new QTLs characterised by a low frequency and/or a small effect. In addition, both pedigrees could be compared regarding the imprinting status at the IGF2-intron3-G3072A substitution, segregating on SSC2. The mutation was segregating within the European founders used in both pedigrees, Meishan individuals being all homozygous for the wild allele (G). This analysis, presented in chapter 3, shows that the structure of the pedigree (number of F1 individuals and size of half-sib families), the number of F1 heterozygous females at the IGF2 locus and the segregation of another QTL at a distance of 40 cM from the IGF2 locus influence the ability to detect imprinting at the IGF2 mutation. This spurious maternal effect can lead to incorrect conclusions regarding the imprinting status of the IGF2 mutation, with maternal effects being detected whereas they do not exist. In order to fine-map the second QTL segregating on pig chromosome 2, a backcross design was set up. Sires that were finally progeny tested were all homozygous for a Meishan haplotype in the IGF2 region, so the phenotypic variation could not be due to the IGF2-intron3-G3072A mutation. Results from the progeny-testing presented in chapter 4 confirmed that a QTL underlying fatness traits was segregating on the short arm of SSC2. However, the size of the QTL interval could not be reduced because of epistatic interactions. These epistatic effects could be detected because full-sibs with Identical-by-descent haplotypes in the QTL regions were progeny-tested. This particular design could be analysed without the strong assumptions of the line-cross models (according to which QTL alleles are fixed within breeds), so interactions could be detected. The re-analysis of one of the two F2 pedigrees confirmed that a region on SSC13 interacts with the QTL segregating in SSC2, but other candidate regions still need to be considered. The combined analysis of different pedigrees finally gives few benefits regarding the number of new QTLs that were detected. However, these combined analyses enabled to successfully consider non-additive effects such as imprinting and epistasis. During the work described in this thesis, a major technological advance occurred for pig geneticists, with the commercialisation of the Illumina PorcineSNP60 Beadchip. With this tool, the number of genotypes that can be included in a QTL analysis tremendously increased. In order to properly use this new type of information, the order of the SNPs along the genome must be reliable. In chapter 5, the first high-density genetic map of the pig is presented. This genetic map was computed using information from in silico and RH mapping of the SNPs in combination with recombination rates between them and finally comprised 38,599 SNPs. Four pig pedigrees based on different breeds were analysed separately, and the analysis of the recombination rate along the pig genome highlighted that the more recombinant regions tend to cluster around the telomeres irrespective of the location of the centromere. Two of the four analysed pedigrees comprised enough male and female meiosis to construct sex-specific maps. Major sex-differences in recombination were observed with a higher recombination rate in the females only within GC-rich regions, with females exhibiting a much stronger correlation between recombination rate and specific sequence features. This new information will be of major importance when dealing with QTL fine-mapping and pig genome evolution. Finally, in the general discussion presented in chapter 6, arguments toward further fine-mapping of QTLs in pig are given despite the increasing interest in genomic selection. Despite major improvements that have been made in the development of high-density SNP chips, efforts are still needed to overcome the biases linked to the design of the chips. In parallel to the development of high-density genotyping tools, few improvements were made regarding phenotyping. In this final chapter, various programs dedicated to the description of highly precise phenotypes and to the development of homogenous phenotyping practices are presented. Such programs, in combination with the development of appropriate genotyping tools, will facilitate the detection of causal variants. These efforts that are still necessary are not only required for pig but also to most livestock species for which QTL fine-mapping is still needed.