A core group of four open reading frames (ORFs) is present in all known papillomaviruses (PVs): the E1 and E2 replication/transcription proteins and the L1 and L2 structural proteins. Because they are involved in processes that are essential to PV propagation, the sequences of these proteins are well-conserved. However, sequencing of novel subtypes for human papillomaviruses (HPV) 54 (AE9) and 82 (AE2/IS39), coupled to analysis of four other closely related genital HPV pairs, indicated that E2 has a higher dN/dS ratio than E1, L1 or L2. The elevated ratio is not homogeneous across the length of the ORF, but instead varies with respect to E2's three domains. The E2 hinge region is of particular interest, because its hypervariability (dN/dS>1) differs markedly from the two domains that it joins: the transcription-activation domain and the DNA-binding domain. Deciphering whether the hinge region's high rate of non-synonymous change is the result of positive Darwinian selection or relaxed constraint depends on the evolutionary behaviour of E4, an ORF that overlaps E2. The E2 hinge region is contained within E4 and non-synonymous changes in the hinge are associated with a disproportionate amount of synonymous change in E4, a case of simultaneous positive and purifying selection in overlapping reading frames. Modular rates of selection among E2 domains are a likely consequence of the presence of an embedded E4. E4 appears to be positioned in a part of the HPV genome that can tolerate non-synonymous change and purifying selection of E4 may be indicative of its functional importance.
INTRODUCTIONHuman papillomaviruses (HPVs) include a group of closedcircular, double-stranded DNA viruses established as the aetiological agents of cervical neoplasia and invasive cervical cancer Muñoz et al., 2003). Over 100 putative HPVs have been described, more than 90 of which have been cloned and officially designated (de Villiers et al., 2004). Despite an array of types, the development of cervical carcinoma is restricted to a subset of high-risk viruses, of which HPV16 and HPV18 together account for nearly two-thirds of observed cases (Muñoz et al., 2003). Classification of HPV types into categories, such as 'cancer-causing' and 'non-cancer-causing' or cutaneous and mucosal, is based on both epidemiological and phylogenetic studies (Chan et al., 1995;Pfister & Fuchs, 1994;Van Ranst et al., 1992, 1993. Most oncogenic types share a common ancestor and most common ancestors radiate within an ecological niche, such as the skin or mucosa.The typical HPV genome contains six open reading frames (ORFs) encoding the structural proteins (L1 and L2), proteins that mediate the viral life cycle (E1 and E2) and proteins that regulate host-cell DNA replication and transformation (E6 and E7) (Knipe et al., 2001). E6 and E7 bind the cell-cycle regulators p53 and pRb, respectively. Among highly cancerous types, these oncogenes degrade p53 and pRb, resulting in host-cell immortalization and proliferation (reviewed by Fehrmann & Laimins, 2003). Some p...