The Papillomavirus E2 Regulatory Proteins

Cycle The papillomaviruses are small DNA viruses that induce be- nign proliferative squamous epithelial and fibroepithelial lesions (warts and papillomas) in their natural hosts. They have been isolated from a variety of animal species, and over 60 human papillomavirus types have been described so far (reviewed in Ref. 1). The papillomavirus life cycle is closely linked to keratinocyte differentiation. In the basal epithelial cells and dermal fibroblasts of a fibropapilloma the viral genome is maintained as a multicopy episome, and only “early” viral genes are expressed. Vegetative DNA replication, late gene expression, and virion production are restricted to differentiated squamous epithelial cells. Bovine papillomavirus type 1 (BPV-1)’ has served as the prototype for the genetic analysis of the papillomavirus functions. BPV-1 is able to transform certain rodent cell lines; within these cells the viral genome is maintained as an extrachromosomal element, and only “early” gene products are expressed. This latent, non-productive infection is thought to he analogous to the virus-host interaction characteristic of the infected basal epithe- lial cells or dermal fibroblasts of a fibropapilloma. Transforma-tion of rodent cells by BPV-1 has enabled functions important for viral transformation, replication, and transcriptional regulation to be mapped. Such studies have revealed that products encoded by the BPV-1 E5 and E6 genes are required for full transformation, E l products are necessary for viral DNA replication, and E2 polypeptides function both in replication and transcriptional regulation (reviewed in Refs. 2 and


Papillomavirus Life Cycle
The papillomaviruses are small DNA viruses that induce benign proliferative squamous epithelial and fibroepithelial lesions (warts and papillomas) in their natural hosts. They have been isolated from a variety of animal species, and over 60 human papillomavirus types have been described so far (reviewed in Ref. 1). The papillomavirus life cycle is closely linked to keratinocyte differentiation. In the basal epithelial cells and dermal fibroblasts of a fibropapilloma the viral genome is maintained as a multicopy episome, and only "early" viral genes are expressed. Vegetative DNA replication, late gene expression, and virion production are restricted to differentiated squamous epithelial cells.
Bovine papillomavirus type 1 (BPV-1)' has served as the prototype for the genetic analysis of the papillomavirus functions.
BPV-1 is able to transform certain rodent cell lines; within these cells the viral genome is maintained as an extrachromosomal element, and only "early" gene products are expressed. This latent, non-productive infection is thought to he analogous to the virus-host interaction characteristic of the infected basal epithelial cells or dermal fibroblasts of a fibropapilloma. Transformation of rodent cells by BPV-1 has enabled functions important for viral transformation, replication, and transcriptional regulation to be mapped. Such studies have revealed that products encoded by the BPV-1 E5 and E6 genes are required for full transformation, E l products are necessary for viral DNA replication, and E2 polypeptides function both in replication and transcriptional regulation (reviewed in Refs. 2 and 3).
The genomic organization of the human papillomaviruses (HPVs) is very similar to BPV-1 (see Fig. 1). Extensive studies have shown that the E6 and E7 proteins from those HPV types that have been associated with cancer are able to immortalize human squamous epithelial cells. The HPV E2 genes also encode transcriptional regulatory protein(s). Little is known specifically about HPV DNA replication, hut, by analogy with BPV-1, it is likely that proteins encoded by the E l and E2 ORFs are also important for this function (reviewed in Ref. 4).

E2 Polypeptides and Transcriptional Regulation
Mutations in the BPV-1 E2 ORF are pleiotropic, disrupting transformation, replication, and transcriptional regulation (5-9). Genetic analysis of BPV-1 demonstrated that the E2 ORF encodes a transactivator which interacts directly with specific enhancer elements within the long control region (LCR) resulting in increased transcription from viral promoters P7940, Ps9, P2443, and P:IORO (10-17). The E2 ORF also encodes two transcriptional repressors that function through the same enhancer elements hut inhibit E2-dependent transactivation (18-20). One of these repressors, E2-TR, is expressed from the P30R0 promoter and translated from an internal methionine at amino acid 162 of the E2 ORF. Another repressor, E8/E2, is encoded by a spliced message that encodes 11 amino acids from the E8 ORF linked to the C-  (21). These correspond to the E2, E2-TR, and E8/E2 proteins, respectively (19). The relative abundance of repressor over transactivator proteins in transformed cells (1 E2:lO E2-TR3 E8/E2) is thought to result in a low level of viral gene expression. As described below, in certain circumstances the full-length E2 transactivator is also able to repress transcription (22,23).
Less is known about the HPV E2 Polypeptides. All of the HPVs so far sequenced encode an E2 polypeptide homologous to the transactivator. However, although cDNAs have been isolated which could potentially encode spliced or shorter E2 species, such polypeptides have not been specifically identified. The distribution of E2 DNA-binding sites in the HPV genomes differs from the pattern observed in fibropapillomaviruses such as BPV-1 ( Fig. 1). In the cancer-associated human papillomaviruses HPV-16 and HPV-18, the major promoter (Pg7 and Ploo, respectively) is negatively regulated by E2 (24)(25)(26). This repression is mediated through two E2-binding sites immediately adjacent to the TATA box of the promoters (26, 27). Occupation of these sites by E2 is thought to interfere with assembly of the transcription preinitiation complex (see Fig. 3A and below). E2 regulation of the E6 promoter of HPV-11 also provides evidence for both positive and negative transcriptional modulation (28,29).
In the majority of HPV-containing cervical carcinomas the viral genome is often integrated such that the E l and/or E2 ORFs are disrupted. It has been postulated that in the absence of these polypeptides, the E6 and E7 oncoproteins are expressed in a deregulated manner which in turn may contribute to tumor development. In support of this model, it has been shown that mutation of the promoter proximal E2-binding sites in the HPV-16 genome results in a slight hut reproducible increase in the ability of the viral DNA to immortalize human keratinocytes. Moreover, mutations which disrupt the HPV-16 E l or E2 ORF cause a further augmentation of the immortalization potential, suggesting that these ORFs play additional roles in regulating the levels or activities of the viral oncoproteins.'

Structural and Functional Domains of the E2 Polypeptide
A comparison of the predicted amino acid sequences of a number of papillomavirus E2 ORFs reveals that these polypeptides probably consist of three domains (30,31). Two of the domains, consisting of about 220 amino acids at the N terminus and 90-100 amino acids at the C terminus, are relatively well conserved among papillomaviruses. However, the internal region varies both in length and amino acid sequence (Fig. 2).
The N-terminal region, which is unique to the full-length transactivator protein, encodes the activation domain (32)(33)(34). This domain is large compared with many other transactivators, and this may reflect the multifunctional role of E2 in transcription and replication. The activation domain contains two regions within the first 85 amino acids that can be predicted to form acidic amphipathic helices. These structures have been shown to he important for the activation function of a number of eukaryotic transactivators (35). Deletion analysis of BPV-1 E2 demonstrated that the region containing the potential amphipathic helices is necessary hut not sufficient for transactivation (34). DNA-hound dimers of the BPV-1 transactivator protein can interact in vitro to form DNA loops, and this association requires the N-terminal domain of E2 (36). In addition, as described below, the transactivator can associate with the BPV-1 E l polypeptide and is necessary for viral DNA replication (37)(38)(39). It is likely that sequences in the N-terminal domain are required because the E2 repressors cannot interact with E l or support viral replication.
The DNA binding function is encoded by the C-terminal 85 amino acids of the BPV-1 E2 ORF, and this domain is present in all three E2 polypeptides (40)(41)(42). The E2 proteins bind to a single DNA-binding site as dimers, and dimerization is mediated through the DNA-binding domain (34,43,44). This C-terminal domain does not contain any obvious consensus DNA binding motifs. Some E2 proteins contain several leucine residues spaced at seven-amino acid intervals, similar to the leucine zipper dimerization structure; others contain hydrophobic residues at most of these positions. However, this region is not predicted to form an a-helix, and we have therefore designated these residues the hydrophobic heptad repeats. Overlapping the repeats is a conserved stretch of basic amino acids which likely interacts directly with the E2 DNA-binding site (Fig. 2). E2 has been demonstrated to show some cooperativity in DNA binding, and this property requires some unidentified sequences outside of the miniial DNA-binding domain (42).
The non-conserved internal region of the E2 polypeptide is designated the hinge region. It does not seem to be required for transcriptional transactivation in transient transfection assays, and may simply provide a flexible link between the DNA-binding and activation domains (33,34). This region overlaps with the E3 and E4 ORFs, and it is possible that selective pressure on these overlapping ORFs has resulted in sequence divergence of the E2 ORF. However, as described below, this region contains two major serine phosphorylation sites of the BPV-1 E2 proteins (45).

E2 DNA-binding Sites
The E2 proteins bind specifically to a 12-base pair palindromic sequence, ACCNsGGT, which is repeated several times in the papillomavirus genomes (11, 12, 46). BPV-1 has 12 sites that correspond to this consensus and an additional five closely related sequences that are also bound by E2 (47). The majority of these sites are located in the LCR, but there are several situated close to the promoters in the early region (Fig. 1). The affinity of the BPV-1 E2 protein for the various DNA-binding sites ranges over several orders of magnitude. The higher affinity sites are clustered in the LCR in regions that have been designated E2-dependent enhancer elements E2REl and E2REz (13).
A low level of E2-dependent activation can be obtained with one E2 binding motif in Saccharomyces cereuisiae and in mammalian cells. However, two E2 DNA-binding sites cooperate to constitute a strong enhancer (48)(49)(50)(51)(52). In enhancer elements such as EPRE, multiple binding sites cooperate to form a highly responsive element (53). Some cooperativity in DNA binding has been observed in vitro with the BPV-1 E2 protein, but this is not sufficient to explain the strong synergism seen in uiuo, suggesting that additional mechanisms are probably involved (36,42,48).
E2 transactivates the PZM and Pam promoters located in the early region, and activation depends on the E2-dependent enhancers in the LCR (15-17). It was postulated that the E2-binding sites adjacent to these promoters might cooperate with those in the LCR to bring about activation. DNA-bound E2 transactivator molecules have been demonstrated to associate and form DNA loops, providing a model as to how activation could occur at a distance (36). Alternatively, binding of E2 to these sites could repress expression from the promoters in a manner analogous to the situation with the HPV-16 Pg, and HPV-18 Plo5 promoters. However, mutation of the E2 site adjacent to the PZ4*3 promoter (which has the lowest affinity of all the BPV-1 sites) has no effect on E2 regulation of this promoter (17, 54), and mutation of the two E2 sites adjacent to the PW promoter results in only a minimal decrease in E2-dependent promoter activity (17). Nevertheless, it is possible that these sites could have an important regulatory role under different circumstances. For example, the DNA binding specificity of E2 for the various motifs could change during the cell cycle, with the state of cellular differentiation, or at a particular stage of the viral life cycle. The number and distribution of the E2-binding sites differ in the human papillomaviruses as compared with the animal papillomaviruses. In the genital papillomaviruses HPV-16 and HPV-18, only four Ea-binding sites have been identified, all of which are located in the LCR (Fig. 1). As noted above, two of these sites are situated immediately upstream from the respective TATA boxes of the E6/E7 viral promoters (designated Ps, in HPV-16 and P1% in HPV-B), and binding of E2 to these sites represses rather than activates transcription from these promoters (24)(25)(26)(27). Mutation of these proximal E2-binding sites so that they no longer bind E2 results in E2-mediated activation of transcription of the Pg7 or Plos promoters through interaction with the remaining upstream E2-binding sites (26, 27). These results indicate that E2 modulation of HPV transcription can involve both positive and negative regulation. The final outcome may be a function of the different binding affinities of E2 for each of the sites (27, 55). Alternatively, E2 may positively regulate some other, as yet unidentified, HPV promoters.
The precise contact points of the E2 proteins on the ACCNsGGT motif have been determined (41,55). The protein contacts the guanosine residues on either strand of the motif, which are predicted to be present on the same face of the DNA helix. The nucleotides in the non-conserved core and immediately outside the binding site determine the affinity for the E2 protein.
Binding of both BPV-1 and HPV-16 E2 proteins has been shown to bend DNA, and the HPV-16 E2 preference for an A-T rich core sequence may reflect the flexibility of such a sequence (55,56).

Mechanisms of E2 !lYanacriptional Regulation
As mentioned above, there are several ways in which the E2 proteins can either activate or repress transcription, depending on the position of the E2-binding sites and the nature of the E2 polypeptides. These are shown diagrammatically in Fig. 3. In one case, such as for the BPV-1 PTgd0 and Pas promoters, binding of the full-length E2 transactivator to a number of sites located upstream from the promoter activates transcription (13,14). E2 can also stimulate transcription from heterologous promoters which contain several E2-binding sites located at some distance either upstream or downstream from the promoter (51). In these cases, the E2 sites behave like classical conditional enhancer elements. However, the E2 transactivator can also repress transcription when the E2 motifs are situated close to the transcriptional start site, overlapping the binding sites for essential cellular transcription factors. In the case of HPV-16 (and HPV-18), two E2 sites are located just upstream from the TATA box of the Pg7 promoter, and binding of E2 to these sites may sterically hinder the binding of the cellular factor, TFIID, and/or interfere with the assembly of the transcriptional initiation complex (26, 27, 57). There is also an essential Spl-binding site located just upstream from the E2 motifs, but it is not known whether occupation of the E2 sites interferes with binding of Spl to this region of the promoter (58). In a similar situation, an E2-binding site located downstream from the BPV-1 PT1& promoter overlaps an Spl-binding site. This Spl site is essential for P7185 activity, and binding of E2 polypeptides can repress expression from this promoter, probably by interfering with Spl binding (22,23). Finally, as described above, the BPV-1 PZu3 promoter has a low affinity E2-binding site located just upstream from the TATA box, but this site does not appear to play a role in the regulation of this promoter. This E2 site is adjacent to an Spl site which is critical for both basal and E2-dependent promoter activity, and E2 transactivation of this promoter depends on the E2 sites located 2.5 kilobases upstream in the LCR (17, 54). The mechanism by which E2 stimulates transcription has not yet been elucidated, but it is likely that activation results from a direct or indirect interaction of E2 with some component of the basic transcriptional machinery. E2 is unable to activate minimal promoters containing only a TATA box but requires additional promoter proximal elements which bind factors such as Spl (59). There are Spl sites located close to many papillomavirus promoters, and several of these sites have been shown to be essential for promoter activity (17, 23,54,58). Many of these sites are adjacent to or overlap E2 binding motifs and may play a role in E2-mediated regulation. Spl can interact directly with the fulllength E2 polypeptide and can enhance E2 binding to weak affinity DNA sites (17). Binding of Spl and E2 to sites spaced widely apart on a DNA fragment results in the formation of DNA loops in vitro and provides a conceivable mechanism by which the LCR E2-binding sites can activate transcription from promoters located far downstream in the early region, such as P2413 and P3080. However, it seems unlikely that the interaction between E2 and Spl is exclusive because E2 is able to synergize with other promoter proximal elements (59).
Modulation of papillomaviral transcription also depends on the nature and type of interaction of the E2 polypeptides (Fig.  3B). There are several mechanisms by which the E2 repressor proteins could inhibit transcriptional activation mediated by the full-length transactivator. The repressors could bind and titrate out essential cellular factors required for activation. Alternatively, the repressors could block E2 transactivation by binding competitively to the ACCNsGGT DNA-binding site or by forming potentially inactive heterodimers with the transactivator species. Since all three E2 regulatory factors share the C-terminal domain required for DNA binding and dimerization, the latter two models are likely mechanisms for E2 repression (Fig. 3B).

E2 Polypeptides and Viral DNA Replication
It has recently been shown conclusively that the full-length products of the E l and E2 ORFs are both necessary and sufficient for BPV-1 transient DNA replication (39). The E l polypeptide has some homology with SV40 and polyomavirus T antigens and binds ATP and binds DNA nonspecifically (38,60, 62). In addition, the full-length E l and E2 polypeptides form a complex capable of binding to E2 DNA sites, suggesting that E2 might direct the E l replication protein to the viral origin of replication (37,38). In support of this hypothesis, the minimal origin required for transient replication has been mapped to a 105-base pair fragment spanning the HpuI restriction site at nucleotide 1 of the BPV-1 genome:' This fragment also contains E2-binding sites 11 and 12. However, subsequent studies have shown that the E l polypeptide does in fact bind specifically to a region spanning the HpaI site (63) and that the adjacent E2 sites are not required for transient DNA replication." The role of E2 in replication is therefore less clear. It may function by stabilizing the interaction of E l with DNA or by altering the chromatin structure around the origin of replication thus allowing greater accessibility of cellular replication factors. In support of the former hypothesis, it has recently been shown that cellular extracts supplemented with purified BPV-1 E l and E2 proteins can support replication of exogenously added papillomavirus DNA. The E2 transactivator stimulates binding of the E l replication protein to the origin of replication and thereby activates DNA replication (64).
Only the full-length E2 protein is able to complex with El, but it is possible that the E2 repressors could regulate replication by competitively binding to E2 sites adjacent to the replication origin or by complexing with the E2 transactivator. In support of this model, a BPV-1 mutant that is unable to express the E2-TR repressor replicates at a high copy number (65,66).

Regulation of E2 Function
The abundance and activities of the E2 polypeptides can be regulated at several levels. As described above, the promoters from which the E2 gene products are thought to be expressed are themselves E2-responsive and therefore could potentially be autoregulated by the E2 proteins (15-17). In addition, the functions of the E2 transactivator can be modulated by association or competition with the shorter repressor species.
The E2 proteins may also be regulated by post-translational modifications. The E2 polypeptides are phosphorylated primarily on two serine residues at amino acids 298 and 301 in a region of the hinge immediately adjacent to the DNA-binding domain and present in all three E2 polypeptide species (45). Substitution of serine 301 with an alanine residue results in a virus that replicates to a much higher copy number than that of the wild-type virus (67). This suggests that E2 phosphorylation may play a direct role in viral DNA replication. In support of this hypothesis, E l binds preferentially to the underphosphorylated form of E2, suggesting that phosphorylation may modulate the El-E2 protein interaction (68). Alternatively, E2 phosphorylation may modulate replication indirectly by regulating viral gene expression.
While it seems likely that an E1-E2 protein complex is required for viral DNA replication, there is little data on the effect of this complex on the transcriptional properties of the E2 transactivator. It has been postulated that the E l ORF encodes a transcriptional repressor because BPV-1 E l mutants exhibit increased viral transcription (69,70); however, the mechanism by which this phenotype arises has not yet been elucidated.