Complex evolutionary events at a tandem cluster of Arabidopsis thaliana genes resulting in a single-locus genetic incompatibility.

Non-additive interactions between genomes have important implications, not only for practical applications such as breeding, but also for understanding evolution. In extreme cases, genes from different genomic backgrounds may be incompatible and compromise normal development or physiology. Of particular interest are non-additive interactions of alleles at the same locus. For example, overdominant behavior of alleles, with respect to plant fitness, has been proposed as an important component of hybrid vigor, while underdominance may lead to reproductive isolation. Despite their importance, only a few cases of genetic over- or underdominance affecting plant growth or fitness are understood at the level of individual genes. Moreover, the relationship between biochemical and fitness effects may be complex: genetic overdominance, that is, increased or novel activity of a gene may lead to evolutionary underdominance expressed as hybrid weakness. Here, we describe a non-additive interaction between alleles at the Arabidopsis thaliana OAK (OUTGROWTH-ASSOCIATED PROTEIN KINASE) gene. OAK alleles from two different accessions interact in F(1) hybrids to cause a variety of aberrant growth phenotypes that depend on a recently acquired promoter with a novel expression pattern. The OAK gene, which is located in a highly variable tandem array encoding closely related receptor-like kinases, is found in one third of A. thaliana accessions, but not in the reference accession Col-0. Besides recruitment of exons from nearby genes as promoter sequences, key events in OAK evolution include gene duplication and divergence of a potential ligand-binding domain. OAK kinase activity is required for the aberrant phenotypes, indicating it is not recognition of an aberrant protein, but rather a true gain of function, or overdominance for gene activity, that leads to this underdominance for fitness. Our work provides insights into how tandem arrays, which are particularly prone to frequent, complex rearrangements, can produce genetic novelty.

The authors have declared that no competing interests exist.