Developing RAD markers as a resource for plant breeding

In order to improve the ability of plant breeding programmes to deliver the agricultural increases mandated by a growing population and changing climate, new techniques must be developed for rapid discovery and genotyping of genetic markers. The RAD (Restriction-site Associated DNA) sequencing (RADSeq) technique developed by Professor Eric Johnson of the University of Oregon, generates tens of thousands of genetic 'tags' from genomic DNA. Due to the capabilities of modern DNA sequencers, it is possible to sequence each of these tags many times and thus reliably spot genetic differences between two individuals. The capacity of second-generation sequencers is such that tags from multiple individuals can be pooled within a single sequencing run whilst still maintaining a high enough coverage of each tag to identify genetic differences. Tags from each individual can be identified by adding a unique 'molecular identifier' to the DNA prior to sequencing. By carefully selecting the right number of tags to be generated, it becomes possible to screen enough individuals within a single run to cover an entire genetic mapping population. RAD sequencing therefore combines the discovery, genotyping and mapping of genetic markers into a single step. Furthermore, if the phenotype of the samples is known, the data can be used to identify markers which segregate along with the phenotype, assisting in gene mapping and potentially gene identification. To date, RADSeq has primarily been used in animal or microbial systems. We propose to apply the RADSeq technique to a model cereal species, Lolium perenne (perennial ryegrass), in order to determine the applicability of this technique to improving plant breeding efforts. As a test case, we will use an existing mapping population designed to identify the two genetic loci controlling a self-incompatibility system in Lolium (ryegrass). We will perform RADSeq in the parents of this population at high coverage using two tag densities. We will then screen pooled mapping population progeny from each of four segregating genotypes (two per locus) in order to identify RADSeq markers which appear unique to each genotype. Finally, we will use the RADSeq marker information to construct a genetic map for this population and confirm the bioinformatic identification of a small subset of genetic markers using conventional genotyping. The proposed work will enable us to determine how well the RADSeq technique performs as a method for rapid marker discovery and genotyping in crops, using one of the most difficult examples - a highly heterozygous, outbreeding species. If RADSeq performs well under these conditions, it should easily be applicable to other crop systems. The usefulness of RADSeq as a tool for mapping of genetic loci will also be assessed by attempting to map polymorphisms associated with the self-incompatiblity loci of grasses. Identifying these genes is of high importance to grass breeders as they would allow greater control of mating during breeding programmes.

RAD sequencing (RADSeq) is a high-throughput technique for marker discovery/genotyping. The technique generates short sequence reads adjacent to restriction enzyme sites, which can then serve as 'tags' to identify restriction polymorphisms between samples (similar to AFLP) or be mined to identify the presence of SNPs. By choosing a restriction enzyme which cuts infrequently within the study genome, the complexity of even large genomes can be reduced to a level where high-throughput marker discovery/genotyping can be performed using next-generation sequencing. A restriction enzyme is selected to produce 50-250,000 fragments per digest. Given that one lane of a Solexa flow-cell produces ~12 million reads, individual tags will be sequenced multiple times, making it possible to score presence/absence in the same manner as standard AFLP. However, RADSeq possesses the additional benefit of sampling sequence diversity around restriction sites, allowing for SNP discovery. Multiple individuals can be pooled within a single lane (using a unique 5bp molecular identifier for each sample) and the required coverage of tags needed for SNP genotyping will still be achieved. It is here that the power of next-generation sequencing becomes apparent: a whole flow-cell could potentially be used to assay an entire mapping population. We therefore aim to assess the feasibility of adapting the RADSeq technique to marker discovery/genotyping and mapping in crops, using the forage grass Lolium perenne as a model.We will initially test a series of restriction enzymes on the parental genotypes to produce two tag densities and identify parent-specific RADSeq markers, a subset of which will be validated using another genotyping technique. We will then conduct genotyping of a mapping population of 60 segregants and use this data to 1) construct a genetic map of RADSeq markers and 2) identify marker/trait associations for two genetic loci involved in self-incompatibility.

1. Development of optimal wet-lab and bioinformatics protocols for implementation of RAD sequencing for NERC science 2a. Specific users this work might be of interest to and how they will benefit This proposal is aimed at deliveing much-improved technology for a wide range of BBSRC science researchers. The proposal details how we will prove the utility of and improve the accessibility of RADSeq technology. In particular we will develop robust wet-lab technologies and easy-to-use bioinformatics tools for RADSeq data generation and analysis. 2b. Techniques, methods or activities with which you will engage with this group We have already started a RADSeq users group within the UK and held two meetings. These have been well received, and well attended, and we intend to continue to promote RADSeq through this avenue. We also have a RADSeq wiki site, and one key goal is to develop this into a forum where RADSeq users (and newcomers) can find the information they need for starting, executing and delivering RADSeq analyses. 3a. Wider user interest The RADSeq technioque will be of interest we believe to a wide range of BBSRC-area science practitioners, many of whom may have decided against genetic analyses because of the perceived failings of existing technologies. 3b. Methods of disseminating information to this group We will enthusuiastically promote the RADSeq technology through presentation at research meetings, in publications and in online for a where RADSeq could be of interest to the community. 4. Milestones and measures of success In this 18 month grant, a key milestone will be the completion of the analyses of the differential utility of the different enzymes in the target plant genome. These will inform researchers planning to use RADSeq in the future; we are particularly interested in the performance of methylation sensitive enzymes in this regard. The delivery of optimised bioinformatic analytic tools will be the second major milestone. We will judge success in this segment by the take-up of the software by other users, and from the feedback we receive in its use. In the latter part of the project we will liaise closely with the staff of the GenePool, to ensure efficient transfer of the technology in to the GenePool portfolio. 5. Summary of resources We have requested funding for attendance of project staff at key UK science meetings. The RADSeq wiki is hosted (for free) on the University of Edinburgh computing services infrastructure.