The Multi-allelic Genetic Architecture of a Variance-heterogeneity Locus for Molybdenum Accumulation Acts as a Source of Unexplained Additive Genetic Variance

Abstract

Most biological traits are regulated by both genetic and environmental factors. Individual loci contributing to the phenotypic diversity in a population are generally identified by their contributions to the trait mean.Genome-wide association (GWA) analyses can also detect loci based on variance differences between genotypes and several hypotheses have been proposed regarding the possible genetic mechanisms leading to such signals. Little is, however, known about what causes them and whether this genetic varianceheterogeneity reflects mechanisms of importance in natural populations. Previously, we identified a varianceheterogeneity GWA (vGWA) signal for leaf molybdenum concentrations in Arabidopsis thaliana. Here, finemapping of this association to a ~78 kb Linkage Disequilibrium (LD)-block reveals that it emerges from the independent effects of three genetic polymorphisms on the high-variance associated version of this LDblock. By revealing the genetic architecture underlying this vGWA signal, we uncovered the molecular source of a significant amount of hidden additive genetic variation ("missing heritability"). Two of the three polymorphisms on the high-variance LD-block are promoter variants for Molybdate transporter 1 (MOT1), and the third a variant located ~25 kb downstream of this gene. A fourth independent association was also detected ~600 kb upstream of the LD-block. Testing of T-DNA knockout alleles for genes in the associated regions suggest AT2G25660 (unknown function) and AT2G26975 (Copper Transporter 6; COPT6) as the strongest candidates for the associations outside MOT1. Our results show that multi-allelic genetic architectures within a single LD-block can lead to a variance-heterogeneity between genotypes in natural populations. Further they provide novel insights into the genetic regulation of ion homeostasis in A. thaliana, and empirically confirm that variance-heterogeneity based GWA methods are a valuable tool to detect novel associations of biological importance in natural populations.

On the relationship between high-order linkage disequilibrium and epistasis

“…To extrapolate these 256 findings to the genome level, the proportions of all evaluated predictor pairs that 257 displayed these patterns were calculated and then multiplied with the total number of 258 possible cis-cis/cis-trans/trans-trans pairs in the genome (Table S1). an earlier study of this dataset (Forsberg et al 2015), it was revealed that a large 270 . CC-BY 4.0 International license It is made available under a (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.…”

“…How dependent is the prevalence of high order LD 307 and cis vs trans predictors on the population size? We also present an empirical 308 example where high order LD exists between a cis-trans predictor pair with 309 significant statistical epistasis and a locus displaying a strong genetic variance 310 heterogeneity due to independent contributions by multiple linked polymorphisms 311 (Forsberg et al 2015). This illustrates how complex inheritance patterns of individual 312 loci, something usually not explored in GWAS data, further complicates the 313 interpretation of detected statistical epistatic signals.…”

“…This linkage block was originally identified due to its 273 large marginal, variance heterogeneity effect in the population (Shen et al 2012). It is 274 known that statistical epistasis and genetic variance heterogeneity can emerge from 275 similar genetic architectures (Forsberg et al 2015), and this population was therefore 276 selected for further evaluations of whether high order LD between the genotyped 277 SNPs and these hidden polymorphisms could lead to statistical epistasis in a two locus 278 association analysis. We performed an exhaustive, two-dimensional genome scan for 279 -10 ) was defined as a Bonferroni corrected nominal 5% significance 287 threshold.…”

“…(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. (Forsberg et al 2015) to dissect a locus with a highly 382 significant variance heterogeneity association for leaf molybdenum concentration 383 (Shen et al 2012) to the contributions of four independent associations in an extended 384 LD block on chromosome 2. Several of these associations were found to structural 385 variants that were poorly tagged by the SNP markers (Forsberg et al 2015).…”

“…Our 386 pairwise genome wide scans for pairs of epistatic loci identified 396 significant SNP 387 pairs. For 290 pairs both markers were located in the narrow region on chromosome 2 388 that was earlier dissected in detail (Forsberg et al 2015). All these are examples of 389 cis-cis predictor pairs.…”