Concerted evolution rapidly eliminates sequence variation in rDNA coding regions but not in intergenic spacers in Nicotiana tabacum allotetraploid

Nicotiana tabacum (tobacco) is a natural allotetraploid that formed from two diploid progenitors (N. sylvestris—S-genome, N. tomentosiformis—T-genome) within past 0.2 million years. Previous classical studies have shown that its 35S rDNA has been largely homogenised towards T-genome-like homeologs. However, the degree of conversion at single nucleotide resolution remains unknown. Here, we analysed intragenomic variation of rDNA at high resolution in natural tobacco, synthetic tobacco and the progenitors employing genomic, molecular and cytogenetic methods. In synthetic tobacco, we identified 13 highly (≥10% units) polymorphic sites in the 18S-5.8S-26S coding region. In contrast, only a single polymorphic site was detected in natural tobacco, indicating that gene conversion has removed most of the polymorphisms over shallow evolutionary times. However, the non-coding 26S-18S intergenic spacer (IGS) was highly polymorphic in both natural (57 polymorphic sites) and synthetic tobacco (128 polymorphic sites). In natural tobacco, most (64%) IGS polymorphisms were inherited from the N. tomentosiformis progenitor, while 36% appeared de novo indicating rapid rates of sequence divergence of IGS. FISH revealed that the T-genome-like units (harbouring N. tomentosiformis-type IGS) occurred on all four loci in tobacco variety 095-55, including those loci derived from N. sylvestris progenitor, while the variety SR-1 retained 1–2 S-genome loci unconverted and transcriptionally silenced. We discuss potential caveats associated with experimental and in silico approaches used for determination of rDNA polymorphisms. We also hypothesise that polyploidy-associated gene conversion may eliminate mutated and non-functional genes that have accumulated in progenitor genomes, thereby contributing to success of polyploidy species.