Coevolution of persistently infecting small DNA viruses and their hosts linked to host-interactive regulatory domains.

Although most RNA viral genomes (and related cellular retroposons) can evolve at rates a millionfold greater than that of their host genomes, some of the small DNA viruses (polyomaviruses and papillomaviruses) appear to evolve at much slower rates. These DNA viruses generally cause host species-specific inapparent primary infections followed by life-long, benign persistent infections. Using global progressive sequence alignments for kidney-specific Polyomaviridae (mouse, hamster, primate, human), we have constructed parsimonious evolutionary trees for the viral capsid proteins (VP1, VP2/VP3) and the large tumor (T) antigen. We show that these three coding sequences can yield phylogenetic trees similar to each other and to that of their host species. Such virus-host "co-speciation" appears incongruent with some prevailing views of viral evolution, and we suggest that inapparent persistent infections may link virus and most host evolution. Similarity analysis identified three specific regions of polyoma regulatory gene products (T antigens) as highly conserved, and two of these regions correspond to binding sites for host regulatory proteins (p53, the retinoblastoma gene product p105, and the related protein p107). The p53 site overlaps with a conserved ATPase domain and the retinoblastoma site corresponds to conserved region 1 of E1A protein of adenovirus type 5. We examined the local conservation of these binding sequences and show that the conserved retinoblastoma binding domain is characteristic and inclusive of the entire polyomavirus family, but the conserved p53-like binding domain is characteristic and inclusive of three entire families of small DNA viruses: polyomaviruses, papillomaviruses, and parvoviruses. The evolution of small-DNA-virus families may thus be tightly linked to host evolution and speciation by interaction with a subset of host regulatory proteins.