This is a nice article that could explain the easy interconversion from circular to linear DNA. The plasmid or transposon that inserts itself in the circle automatically provides the protection against the end replication problem. The nuclear chromosomes have a complete machinery for this ‘task’ which make sit unlikely that their origin is the same. Anyway, they identified the problem and provided a fairly easy solution and may explain the diversity of linear mitochondrial chromosomes. Couldn’t get my hands on the article itself, so the abstract will have to do for now.
Abstract Mitochondrial genomes in organisms from diverse phylogenetic groups vary in both size and molecular form. Although the types of mitochondrial genome appear very dissimilar, several lines of evidence argue that they do not differ radically. This would imply that interconversion between different types of mitochondrial genome might have occurred via relatively simple mechanisms. We exemplify this scenario on patterns accompanying evolution of mitochondrial telomeres. We propose that mitochondrial telomeres are derived from mobile elements (transposons or plasmids) that invaded mitochondria, integrated into circular or polydisperse linear mitochondrial DNAs (mtDNAs) and subsequently enabled precise resolution of the linear genophore. Simply, the selfish elements generated a problem — how to maintain the ends of a linear DNA — and, at the same time, made themselves essential by providing its solution. This scenario implies that insertion or deletion of such resolution elements may represent relatively simple routes for interconversion between different forms of the mitochondrial genome.