Detecting networks of Coadapted genes by genome-scale analysis of DNA polymorphisms: application to the model legume Medicago truncatula – DeCoD

Climate change represents an immediate threat for agriculture and food security worldwide. In this context, crop production must adapt. One strategy is developing new genetic resources able to face abiotic and biotic local conditions (Challenge 5, Research axis 1 of the ANR Program). This can be achieved notably by integrating agronomically interesting genes or Quantitative Trait Loci (QTL) in elite varieties by marker-assisted selection (MAS). QTL mapping is an efficient tool to identify the genetic bases of complex traits, including gene networks through the analysis of epistatic interactions between genes, but it has some flaws such as the cost and time to produce populations segregating for the trait, imprecise mapping and a restriction to the genetic diversity of parental lines often not reflecting genetic variation at the species level. Thanks to the decreasing cost of sequencing, it is now possible to identify genetic variants underlying adaptations at the species level by using population genomics and genotype-phenotype association (Genome-Wide Association Studies - GWAS) approaches, because they are either associated with an attractive trait (resistance to stresses, yield…) or the target of natural selection. Since my recruitment as assistant professor in the LRSV lab in 2010, I have been applying such methods to study the genetic bases of adaptation to root-associated microorganisms in Medicago truncatula, a model legume used to investigate the genomics of plant–microbe interactions (Bonhomme et al. 2014, New Phytol; Bonhomme et al. 2015, Mol Biol Evol). Association mapping is a powerful fine mapping approach, allowing epistatic interactions to be tested yet at the price of computational burden, but precisely phenotyping a lot of individuals remains a heavy task and GWAS results are highly dependent on the measured trait. On the other hand, selection scans focus only on molecular polymorphisms, but currently lack the ability to detect genetic interactions or gene networks without supplementary biological information. Facing the urgency of providing solutions for accelerated and more effective crop genetic improvement (be it for legumes -pea, soybean- or not), selection programs will have to target comprehensive gene networks conferring local adaptation, rather than single major genes. Straightforward genetic methods must be developed to identify such networks. This is the goal of the DeCoD proposal, submitted for funding through the ANR JCJC instrument.
The DeCoD project aims at (i) developing innovative population genomic approaches to identify networks of “coadapted” genes, by exploiting the genetic signature of epistatic selection between genes of the same biological network, (ii) applying these methods to massive DNA polymorphism data (16 millions of Single Nucleotide Polymorphisms – SNPs) already available in Medicago truncatula (http://www.medicagohapmap.org/) to validate, complete and identify new gene networks involved in its adaptation to the environment, and (iii) functionally characterize some candidate genes and simple gene networks involved in Medicago truncatula adaptive response to pathogenic and/or symbiotic micro-organisms, and then to identify orthologous genes and alleles through comparative genetics with crop legumes. The immediate perspective of the project is to accelerate crop legumes genetic improvement through MAS and even targeted genomic selection. The developed methods could be applied to genome-wide DNA polymorphism data obtained in other crop species and, to a larger extent, to domesticated species. The main scientific input of the DeCoD project is that it will support advanced understanding of the genetic bases of adaptation in complex biological systems made of several interacting genes. Because virtually applicable to any species for which genome-wide DNA polymorphisms data are available, this approach may be used in various scientific and technological domains in the coming years.