Identification of genes associated with soil carbonate tolerance by BSA-Seq analysis
During meiosis, recombination reasserts the complement of alleles segregating in hybrid progeny [43]. Thus, the study of F2 populations from contrasted phenotype crosses can provide valuable information on genetic mechanism by association [44]. With the purpose of performing a BSA-Seq study, plants from 65 families of the A1xT6 cross and from 28 families of the T6xA1 cross were self-pollinated. F3 progeny plants were cultivated and screened under the same soil and growth conditions to select the extreme pools (Dataset SD5). We realized that the most sensitive families had very low germination rates, or the seedlings died after a few days post germination (NP plants), therefore we were not able to sequence the sensitive pool. Despite this, considering the families with a higher number of flowered individuals and with an elevated silique production, 7 families were selected as the tolerant pool (F3TP) (Figure 4D).
The allele frequency (AF) comparison of the F3TP with the sensitive parental (T6(c-)) enabled the identification of the top 0.5% outlier SNPs from the empirical distribution (Figure 4E). The 5617 SNPs of both tails were associated to 1119 genes, mainly located in chromosome 5. Within this chromosomal group, the higher proportion of SNPs were assigned to the short arm (Figure 4F, Dataset SD6). Salomé et al. [43] found that crossovers and segregation distortion in F2 populations from different A. thaliana accessions were more frequent in chromosomes 1 and 5. Here, 68% of our phenotype-associated divergent outlier genes belong to chr5, suggesting that this chromosome harbors key polymorphisms that facilitate soil carbonate tolerance.
Additionally, we sequenced the whole genome of 6 A1(c+)and 6 T6(c-) individuals collected from their natural habitat. It is well known that amino acid substitutions in a protein can cause a drastic phenotypic effect [45]. The AF comparison between A1(c+) and T6(c-) picked out 977 genes with 3 to 10 non-synonymous variants with high or moderate predicted effect (Dataset SD7). There are many examples of genomic variants whose frequencies are correlated with environmental variables and temporal changes consistent with natural selection in A. thaliana [35, 47]. A1(c+) is locally adapted to carbonated soils, thus it was important to consider the genomic differences between A1(c+) and T6(c-) in order to narrow the BSA results down. The 69 genes in common between these two analyses were selected as the “genomic candidate genes list” (Figure 4G, Dataset SD8).