Retroviridae [ssRNA(+) viruses replicating via a longer-than-genome-length dsDNA intermediate], Hepadnaviridae, caulimoviruses and badnaviruses [family Caulimoviridae, gapped circular dsDNA viruses replicating via longer-than-genome-length RNA intermediates] all share the unlikely attribute of the use of an enzyme complex consisting of a RNA-dependent DNA polymerase/RNAse H in order to replicate. They share this attribute with several retrotransposons, which are eukaryotic transposable cellular elements with striking similarities with retroviruses [such as the yeast Ty element, the mammalian LINE-1 elements, and the Drosophila copia element]; and with retroposons, which are eukaryotic elements which transpose via RNA intermediates, but share no obvious genomic similarity with any viruses other than the reverse transcriptase.
Bacteria such as E coli also have reverse- transcribing transposons -known as retrons - but these are very different to any of the eukaryotic types while preserving similarities in certain of the essential reverse transcriptase sequence motifs.
All of these elements are collectively known as RETROELEMENTS; the fact that the reverse transcriptases of all of them have some amino acid identity suggests a common evolutionary origin.
Several reviewers have pointed out that just such an enzyme as reverse transcriptase would have been necessary for the transition from what is widely believed to have been an RNA world - that is, where all the extant organsisms had RNA genomes - to the present world in which all cellular organisms have DNA genomes.
Viruses with RNA genomes which use RNA-dependent RNA polymerases for their replication may be the only remnants of that pre-DNA era; however, cellular elements and viruses which use reverse transcriptase may share a common origin as cell-derived "modules" coding for a reverse transcriptase, which evolved to become retrons and retroposons and retrotransposons. Addition of structural proteins may have allowed evolution of retroviruses.
The evolution of the DNA retroviruses - Hepadnaviridae, caulimo- and badnaviruses - is more obscure; it appears as though these arose from retrotransposon-like sequences, but this probably occurred near the origin of of these types of element as they are so diverse in sequence and genome organisation.
It is believed that retrotransposons may contribute substantially to the evolution of their hosts. Evidence for this has been obtained by studying human LINE-1s (Long Interspersed Nuclear Elements) - a group of retrotransposable elements which make up approximately 15 % of the human genome. The vast majority of LINE-1s are no longer retrotransposition competent and it is believed that in humans only between 30 and 60 full length LINE-1s are currently active. There is strong evidence from sequences in the sequence databases to suggest that active LINE-1s play an important role in "exon shuffling" (belived to be the major mechanism of macro-evolution whereby entirely new genes are created by reshuffling the components of older genes). The most compelling evidence that LINE-1s do facilitate exon shuffling, however, is the experimental demonstration that they are not only able to move large amounts of non-LINE-1 exonic DNA but also insert this DNA into unrelated expressed genes to obtain chimeras which encode active hybrid gene products.
References:
Chapter 7: "Evolution by transposition" (pp. 172-203) in "Fundamental of Molecular Evolution" by Wen-Hsiung Li and Dan Graur; Sinauer Associates, Inc., Sunderland, Mass., 1991
Moran, J. V., R. J. DeBerardinis, and H. H. Kazazian. 1999. Exon shuffling by L1 Retrotransposition. Science 283:1530-1534.