Genetic diversity in candidate genes for developmental traits and cell wall characteristics in perennial ryegrass (Lolium perenne)

In outbreeding species in which linkage disequilibrium (LD) spans short distances, the use of markers in candidate genes is an appropriate approach for association mapping (AM) (Zhu et al. 2008). Several studies have successfully used this approach in Lolium perenne. For example, Skøt et al. (2007) examined the association between an alkaline invertase candidate gene (LpcAI) and carbohydrate content, and between the photoperiod control gene LpHD1 and flowering time; Skøt et al. (2011) investigated the association between polymorphisms in the L. perenne ortholog of the Arabidopsis FT gene (LpFT3
) and flowering time; Auzanneau et al. (2007) examined the association between
L. perenne Gibberelic Acid Insensitive (LpGAI) and organ growth. Recently, Yu et al. (2013) examined the association between polymorphisms in a set of 14 genes and drought tolerance. However, these studies focused on a relatively small number of genes, which is, at least partially, due to the relatively large population sizes typical of AM studies and the lack of cost-efficient genotyping technologies for highly polymorphic genomes. These studies used either amplicon resequencing across the population (Auzanneau et al. 2007; Yu et al. 2013), or identified alleles in a representative discovery panel followed by screening the AM population with the subset of polymorphisms that discrimi-
nated those alleles using TaqMan assays (Skøt et al. 2007) or PCR-fragment length polymorphisms (Skøt et al. 2011) as genotyping tool. While analysis of a limited number of candidate genes can be appropriate for the dissection of traits with a clearly defined genetic control and easy to pin-point candidate genes, it represents a clear limitation when dealing with complex traits. Indeed, even when a significant association is found between polymorphisms in one candidate gene and the investigated trait, it is difficult to rule out that other genes might also have an effect, or that (part of) the phenotypic diversity observed is due to interactions between multiple genes. Furthermore, most high-throughput genotyping platforms only allow screening of a priori known SNPs, so that only a subset of the potentially relevant genetic diversity is considered in the AM study. As a consequence, the development of versatile, cost-efficient methods to screen a large number of polymorphisms in multiple candidate genes for a diverse set of L. perenne genotypes is required. Targeted resequencing of a panel of several hundred candidate genes using next generation sequencing (NGS) provides the possibility to simultaneously discover and genotype a priori unknown and/or low-abundant polymorphisms across a large population. We have investigated whether probe capture enrichment followed by NGS is a feasible approach for targeted resequencing in the highly heterozygous species
L. perenne, for which high SNP densities of about five SNPs per 100 bp have been
reported (Yu et al. 2013; Ruttink et al. 2013). In this approach, oligonucleotide baits are designed to bind regions of interest (i.e. in candidate genes), which are enriched by in-solution sequence capture before sequencing (Davey et al. 2011).
Our final objective is to conduct candidate gene AM of plant developmental traits and cell wall characteristics in a population of about 700 genotypes including breeding materials and wild accessions. Therefore, we identified candidate genes putatively involved in the regulation of plant growth and development, plant architecture, meristem activity, induction of flowering, cell wall biogenesis, and phytohormone biosynthesis, signaling, and response, and selected 539 L. perenne candidate genes for genotyping. During probe design, we excluded target regions containing repeat sequences or with high sequence similarity to paralogous genes in the reference genome sequence, which can affect target-probe hybridization during the enrichment step and cause ambiguous read mapping after sequencing (Uitdewilligen et al. 2013). Here, we evaluate the performance of this newly developed probe set for the screening of 539
L. perenne genes, discuss the effect of read depth (RD) on target coverage, SNP and InDel density, and calling accuracy and evaluate LD patterns in this set of genes. The analysis is based on a set of 95 L. perenne genotypes from 17 accessions originating from France and the inbred line used for genome sequencing (Byrne et al. 2014).