Genetics of domestication in common bean (P. vulgaris L.): an approach for the analysis of candidate genes via tilling-by-sequencin

Objectives According to Charles Darwin, domestication can be thought as a great model to study convergent evolution. Among crops, common bean (Phaseolus vulgaris L.) represents a unique example of multiple parallel independent domestications: wild common bean is organized in two geographically isolated and genetically differentiated wild gene pools (Mesoamerican and Andean) that diverged from a common ancestral wild population, then independently domesticated in Mexico and in South America nearly 8,000 years ago. These processes resulted in morphological changes (e.g. seed and leaf sizes, seed coat color, growth habit, photoperiodic responses) that distinguish culturally adapted classes of beans. In addition to the interest emerging from its domestication history, common bean also carries a pivotal agronomic value: it is one of the most important grain legumes for human consumption and, as a legume, it also has a role in sustainable agriculture owing to its ability to fix atmospheric nitrogen. To gain deeper insights into the genetic basis of common bean domestication we used an available TILLING population, obtained by chemical EMS-mutagenesis of the Mesoamerican genotype BAT 93 (1), and developed a TILLING-by-sequencing pipeline starting as a method for the identification of mutants of candidate domestication genes. Materials and Methods DNA was extracted from seeds of 1728 M4 individuals of an EMS-mutagenized population developed in the Mesoamerican genotype BAT93 [1]. Samples were combined using a 3D pooling system (Figure1) and sequenced by Illumina at High Coverage (50x) after enriching target regions using custom capture probes [2, 3]. Probes were designed to cover a total of 719 genes of interest (approx. 491 Mb), based on the presence of: i) one or more signals of domestication or ii) differential expression between the Andean genotype and Mesoamerican genotype, and iii) known involvement in the phenomenon of shattering, in seed traits, in the cytokinin hormonal pathway and in phytic acid biosynthesis and metabolism. Results Sequencing data were aligned on the BAT 93 reference genome. Variant calling from 3D pools was performed using four variant callers and low frequency variant calling was performed by combining different algorithms (NGSEP v4.1.0, GATK Mutect2 v4.2.0, VarScan v2.4.2, CRISPR v0.2): 144 variants were identified on 128 candidate genes. Data validation on the same DNA samples was performed via PCR and Sanger re-sequencing: 84 variants were confirmed (65,6%) and were carried by a total of 78 lines, along 103 of the screened genes. Among these variants, 8 had a high inferred impact (gain of stop codon or frameshift), 50 had a moderate inferred impact (non-synonymous variants) while the remaining did not produce any amino acid change. Selected M4 (35) lines carrying high (8) and medium impact (27) mutations are being genotyped and reproduced to obtain M5 homozygous individuals; phenotypical analysis of selected lines will be also carried out. Conclusions The TILLING by Seq approach adopted here (three-dimensional DNA pooling strategy coupled with a High Coverage Illumina sequencing after enriching target regions using custom capture probes) allowed us to identify 53 target genes carrying variants with high/medium impact and whose phenotypic effects are currently under evaluation on 31 of them. Considered the relevance of these results and to provide the scientific community a structured platform to look for mutants in common bean, the development of a new TILLING population in the Meccearly dwarf borlotto type variety (Blumen) has recently started and 3516 M1 plants are currently under reproduction.