Genome-wide CG hypomethylation of the <i>Arabidopsis</i> ecotype Cvi linked to structural variation and RNAi at the <i>VIM4</i> – <i>VIM2</i> locus

The Cape Verde Island (Cvi)-0 ecotype, a relict lineage of Arabidopsis thaliana , is characterized by exceptionally low CG methylation levels genome-wide. Numerous population-level and molecular studies have highlighted links between this epigenomic feature, phenotypic traits and environmental factors; however, the molecular mechanism underlying Cvi-0’s distinctive methylation profile remains unknown. Using multiomics comparisons between Cvi-0 and Col-0, we identified Cvi-specific small-RNAs derived from an inverted repeat (IR) formed by two inverted-orientation DNA methylation pathway genes, VARIANT IN METHYLATION 2 and 4 ( VIM2 and VIM4 ). An intergenic 2.8-kb deletion brings these genes closer in Cvi-0, enabling read-through transcription and formation of chimeric transcripts. Targeted 5′ RACE and WGBS showed that the resulting siRNAs mediate both posttranscriptional gene silencing and local DNA hypermethylation in cis and trans, reducing the expression of the VIM1 gene, which is required for methyltransferase activity in A.thaliana . Comparative analysis across A. thaliana ecotypes and related species revealed substantial structural polymorphism at the VIM4 – VIM2 IR locus, likely shaped by nonallelic homologous recombination—facilitated by local palindromic structures, intronic repeats, and transposable element activity. Notably, the Cvi-0 VIM4 – VIM2 IR configuration was unique, suggesting it is a rare, lineage-specific natural variant. Taken together, natural variation at the VIM4 – VIM2 palindrome locus, uniquely found in Cvi-0, appears to be the causal variant driving siRNA biogenesis and down-regulation of the methyltransferase pathway. This distinctive regulatory mechanism is likely to have contributed to long-term reduction of CG methylation in Cvi-0, potentially maintained over millennia due to its geographical isolation from the mainland.