Distinct Roles of Two Binding Sites for the Bovine Papillomavirus (BPV) E2 Transactivator on BPV DNA Replication

Abstract

The modulation of DNA replication by transcription factors was examined by using bovine papillomavirus type 1 (BPV). BPV replication in vivo requires two viral proteins: E1, an origin-binding protein, and E2, a transcriptional transactivator. In the origin, E1 interacts with a central region flanked by two binding sites for E2 (BS11 and BS12), of which only BS12 has been reported to be essential for replication in vivo. Using chemical interference and electrophoretic mobility shift assays, we found that the binding of E2 to each site stimulates the formation of distinct E1-origin complexes. A high-mobility C1 complex is formed by using critical E2 contacts to BS12 and E1 contacts to the dyad symmetry element. In contrast, interaction of E2 with the BS11 element on the other origin flank promotes the formation of the lower-mobility C3 complex. C3 is a novel species that resembles C2, a previously identified complex that is replication active and formed by E1 alone. The binding of E1 greatly differs in the C1 and C3 complexes, with E1 in the C1 complex limited to the origin dyad symmetry region and E1 in the C3 complex encompassing the region from the proximal edge of BS11 through the distal edge of BS12. We found that the presence of both E2-binding sites is necessary for wild-type replication activity in vivo, as well as for maximal production of the C3 complex. These results show that in the normal viral context, BS11 and BS12 play separate but synergetic roles in the initiation of viral DNA replication that are dependent on their location within the origin. Our data suggest a model in which the binding of E2 to each site sequentially stimulates the formation of distinct E1-origin complexes, leading to the replication-competent complex.

Transcription factors have recently been recognized to play important modulatory roles during the initiation of eukaryotic DNA replication. In most cases, these factors act not by regulating neighboring transcription units but, rather, by directly interacting with proteins bound to an origin of replication or with the DNA itself (43). The mechanisms by which transcription factors regulate DNA replication have been most clearly defined by using viral systems such as bovine papillomavirus (BPV) (29), adenovirus (19, 30), and simian virus 40 (SV40) (7), although chromosomal examples exist as well (see, e.g., reference 11). In these systems, transcription factors act through various means, including the recruitment of replication proteins to the origin, the modulation of the activity of bound replication proteins, and the disruption of the local nucleosome structure, allowing replication factors access to the DNA.

In the BPV model system, in vivo replication of DNA containing the BPV origin requires two viral proteins designated E1 and E2 (41), although only E1 is essential in vitro (36, 45). E1 is a DNA helicase (36, 46) and DNA-binding protein (29) that recognizes a dyad symmetry element within the viral origin of replication (16, 42, 45). Origin binding by E1 is cooperative with E2 (24, 32, 37, 45), and this cooperativity is mediated by physical interaction between E1 and E2 (1–3, 22, 29). In the presence of a single-stranded-DNA-binding protein such as human replication protein A (hRPA), E1 unwinds the DNA outward from the origin, with its DNA helicase functioning in the 3′→5′ direction (36, 46).

The region encompassing the origin contains a dyad symmetry element adjacent to an AT-rich domain. This central region is flanked by two binding sites for E2 termed BS11 and BS12, adjacent to the AT-rich and dyad elements, respectively. Mutational analysis to determine the minimal origin sequence has indicated that the central region and BS12 are required for replication in vivo (42). Consistent with the need only for E1 to support replication in vitro, a smaller region lacking most of BS12 can suffice when cell-free systems are used (20). The two viral proteins form various complexes over the origin region as detected by electrophoretic mobility shift assays. E1 in the absence of E2 forms a relatively slowly migrating complex (C2 in the terminology of reference 23), using critical contacts within the dyad element (16, 17, 23, 34). The binding of E1 causes the ATP-dependent induction of structural changes to the viral origin (13). On an origin containing the central region and BS12, the addition of E2 to E1 can give rise to two distinct complexes: a fast-migrating complex containing a lower oligomeric form of E1 (C1 in the terminology of reference 23; a similar complex was observed by the Stenlund laboratory [32, 33]), and an apparent C2 complex (lacking E2 [23, 32–34]). By using an origin that contained both E2-binding sites, it was shown that an increase in the ratio of E1 to E2 caused an extension of the E1 footprint from the dyad region into the AT-rich region and stimulated distortion of the origin (13). The C1 and C2 complexes have different functional properties; the C2 complex is competent for replication, while C1 is inactive (23; see also reference 34).

We have used chemical interference and transient-replication assays to examine the role of E2 and E2-binding sites in viral replication. We find that each flanking E2-binding site plays distinct and important roles during the initiation of BPV DNA replication. E2 binding to BS12 serves to recruit E1 to the origin. In contrast, the interaction of E2 with BS11 stabilizes the binding of E1 across the central origin and BS12 regions, yielding a novel complex that we term C3. We propose that in this final initiation complex, E1 recognizes the origin in a structure similar to that formed by the SV40 T antigen on its cognate origin, using the central palindromic element to produce a complex with twofold dyad symmetry encompassing the AT-rich, dyad, and BS12 regions.

MATERIALS AND METHODS

E1 and E2 proteins.

The GST-E1 (4) and E2 (22) proteins were overexpressed in Sf9 insect cells by using recombinant baculovirus. The E2 protein was purified as described by Seo et al. (37). E1 protein was purified by a modified procedure of that described by Bonne-Andrea et al. (4). Infected cells were thawed in 5 volumes of hypotonic buffer (20 mM Tris-HCl [pH 8.0], 5 mM KCl, 1 mM MgCl₂, 1 mM dithiothreitol [DTT], 0.1 mM phenylmethylsulfonyl fluoride, proteinase inhibitors [0.05 mM EGTA, 20 μg of aprotinin per liter, 20 μg of leupeptin per liter, 10 μg of antipain per liter]) and lysed by Dounce homogenization with 20 strokes with a type B pestle. The lysate was centrifuged for 15 min at 10,000 rpm in a Beckman SS-34 rotor, and the pelleted nuclei washed with 20 mM Tris-HCl (pH 8.0)–10% (wt/vol) sucrose–1 mM EDTA. Nuclei were resuspended in NR buffer (20 mM Tris-HCl [pH 8.0], 50 mM MgSO₄, 500 mM NaCl, 0.5% [vol/vol] Nonidet P-40, 5 mM DTT, 0.1 mM phenylmethylsulfonyl fluoride, proteinase inhibitors) and incubated on ice for 30 min. The nuclei were pelleted as above, and the supernatant was mixed with glutathione-Sepharose beads (previously equilibrated in NR buffer) and nutated for 1 h at 4°C. The beads were washed with 50 bead volumes of NR buffer: three washes with NR buffer containing 1 M NaCl, and three washes with XPa cleavage buffer (50 mM Tris-HCl [pH 8.0], 10 mM MgSO₄, 100 mM NaCl, 10% [vol/vol] glycerol, 1 mM CaCl₂, 5 mM DTT). E1 was then cleaved from the beads by incubation with biotinylated XPa (Boehringer Mannheim) for 4 h at 4°C. The beads were briefly centrifuged, and the supernatant containing the liberated E1 was removed. Streptavidin beads (Boehringer Mannheim) were added to remove the biotinylated XPa.

BPV constructs.

The BPV DNA containing the mutated origin was generated by PCR with the Stratagene Quick Change kit. The following oligonucleotides were used for mutagenesis (mutated bases underlined): ΔBS12, 5′-GTTGTTAACAATAATCACGTTCTCACGTACTTTTCAAGCGGGAAAAAATAGCC (top primer) and 5′-GGCTATTTTTTCCCGCTTGAAAAGTACGTGAGAACGTGATTATTGTTAACAAC (bottom primer); and ΔBS11, 5′-GCAGCATTATATTTTAAGCTCGTTCAAACGTACAAGTAAAGACTATGTATTTTTTCC (top primer) and 5′ GGAAAAAATACATAGTCTTTACTTGTACGTTTGAACGAGCTTAAAATATAATGCTGC (bottom primer). The top strand is defined as that containing the run of T’s between BPV positions 7925 and 7930. The template plasmid used to prepare the ΔBS12 and ΔBS11 mutants was pXS (in which the BPV XbaI-SmaI fragment from nucleotides [nt] 6132 to 945, was inserted into a vector derived from pML-1 [21]). The template plasmid for the ΔBS12ΔBS11 double mutant was pXSΔBS12. Mutations were verified by sequencing. To generate the full-length viral DNA containing the mutated origins, the pXS plasmids were digested with MluI and MunI and the origin-containing fragment was inserted into the MluI-MunI site of pSS3 (28).

BPV origin-containing DNA fragments.

The BPV origin-containing DNA fragments (∼120 bp) were generated by PCR amplification of pKSO (45), pXSΔBS12, or pXSΔBS11 (to prepare the wild-type, ΔBS12, or ΔBS11 origin, respectively). One of the two origin-flanking primers was 5′-³²P labeled with T4 polynucleotide kinase (Boehringer Mannheim) to a specific activity of approximately 1 × 10⁶ to 2 × 10⁶ cpm/pmol.

Electrophoretic mobility shift assays.

To prepare E1-origin and E1-E2-origin complexes, reaction mixtures (30 μl) containing 25 mM potassium phosphate (pH 7.5), 0.1 M potassium glutamate, 7 mM MgCl₂, 1 mM EDTA, 0.5 mM DTT, 4 mM ATP, 10% glycerol, 300 ng of pBluescript KS+ (as a nonspecific competitor), 200 fmol of the origin-containing fragment, and E1 alone or E1 and E2 (as indicated) were incubated for 15 min at 37°C. Glutaraldehyde (final concentration, 0.1%) was added, and the reaction mixtures were incubated for an additional 5 min. The resulting complexes were separated by electrophoresis through a native 5% polyacrylamide gel (acrylamide/bisacrylamide ratio, 29:1) and visualized by autoradiography.

Interference assays.

Chemical modification of the origin-containing DNA fragment was performed as described previously (35). After the separation of the protein-DNA complexes by a gel retardation assay (see above), gel slices containing the complexes were excised, crushed, and soaked overnight in gel elution buffer (0.5 M ammonium acetate, 0.1% sodium dodecyl sulfate, 1 mM EDTA). The eluted DNA was precipitated with ethanol and then resuspended in TE (10 mM Tris-HCl [pH 8.0], 1 mM EDTA). The DNA was extracted with phenol-chloroform (1:1, vol/vol) and precipitated with ethanol. The modified DNA was then cleaved as described previously (35). Assignation of interfering or stimulatory modifications was determined by careful comparison of the results of multiple independent interference experiments.

Transient-replication assays.

Transient-replication assays were performed as described by Mendoza et al. (28). The BPV genomic DNA was released from the vector by digestion with BamHI. Linearized DNA was purified by phenol-chloroform (1:1, vol/vol) extraction followed by ethanol precipitation and resuspension in TE. C127 cells growing in the log phase were trypsinized, pelleted, washed, and resuspended in Dulbecco’s modified Eagle’s medium (DMEM)–5 mM N,N-bis(2-hydroxyethyl)-2-aminoethanesulfonic acid (BES) (pH 7.2) at a concentration of 2 × 10⁷ cells/ml. The cell suspension (0.25 ml) was mixed with 2 μg of input DNA, 0.5 μg of pSS3 linearized with SalI (containing the wild-type viral DNA not released from the vector, to serve as an internal standard), and 50 μg of sheared salmon sperm DNA and transferred to a electroporation cuvette (0.4-cm gap; Bio-Rad). Electroporation was performed at 270 V and 960 μF in a Bio-Rad gene pulser. The cell material was then transferred to 10 ml of DMEM containing 10% fetal bovine serum, and 1 ml was added to each 10-cm plate containing 9 ml of DMEM with 10% fetal bovine serum. For each time point, low-molecular-weight DNA was extracted from two plates of cells by the method of Hirt (15). The isolated viral DNA was restricted with MunI to linearize the viral genome and DpnI to remove unreplicated DNA. DNA was detected by Southern blot analysis with nick-translated pSS3 as a probe. The replication activity in vivo was quantitated by excision of the bands and counting in a scintillation counter.

RESULTS

Key origin contacts used by E1 and E2 identified by interference assays.

Three distinct complexes can be detected when the BPV E1 protein alone or E1 and E2 proteins are incubated with DNA fragments containing the wild-type origin (the central region flanked by BS11 and BS12) in an electrophoretic mobility shift assay (Fig. 1). In the presence of E1 alone, a relatively slow-migrating C2 complex is detected (lane 1) (23, 34). When both E1 and E2 are added, a novel complex (that we define as C3) that migrates slightly more slowly than C2 is generated, in addition to the quickly migrating C1 complex (lane 2 [23; see also references 32 and 33]).

FIG. 1.

Complexes formed on the wild-type BPV origin by E1 and E2. E1 alone (100 ng) (lane 1) or E1 (100 ng) plus E2 (60 ng) (lane 2) were incubated for 15 min at 37°C with a ³²P-labeled DNA fragment containing the wild-type origin. The complexes were cross-linked by treatment with glutaraldehyde, separated by electrophoresis through a native 5% polyacrylamide gel, and visualized by autoradiography. The locations of the C1, C2, and C3 complexes are indicated.

To characterize these different complexes and explore their function in viral DNA replication, we determined the critical contacts on DNA used by the E1 and E2 proteins to form each complex. These contacts were determined by an interference assay. In this approach, 5′-³²P-labeled DNA fragments containing the BPV origin were modified at the base or phosphate positions to an extent of less than one ‘hit’ per molecule. The modified substrate was incubated with E1 or with E1 and E2, and the resulting protein-DNA complexes were separated by nondenaturing gel electrophoresis. The free and bound DNA fragments were excised and cleaved at the modified sites in a chemical cleavage reaction. The modification pattern was then determined by subjecting the cleavage products to denaturing gel electrophoresis. Bands corresponding to modifications that interfere with complex formation are underrepresented in the bound fraction and overrepresented in the free fraction, compared to the initial substrate.

Differential requirements for E2 binding in the C1 and C3 complexes.

Key protein contacts with purines and pyrimidines were examined by the “missing-contact” approach of Brunelle and Schlief (6). Purine contacts were determined by using a DNA fragment modified with formic acid, leading to partial depurination (Fig. 2) (6, 27). Conversely, protein contacts with pyrimidines were examined by using a DNA fragment treated with hydrazine, a reagent that causes destruction of the pyrimidine base (Fig. 3) (6, 27). Each reagent leaves the sugar-phosphate backbone intact. The effect of each modification was examined on a DNA fragment that was 5′-³²P labeled on either the top (Fig. 2A and 3A) or bottom (Fig. 2B and 3B) strand.

FIG. 2.

Interference of the formation of E1- and E1-E2-origin complexes by partial origin depurination. Duplex DNA fragments, 5′-³²P labeled on either the top (A) or bottom (B) strand, were subjected to partial depurination by treatment with formic acid. The DNA fragments were then incubated with E1 alone (300 ng) (lanes 2 and 3) or E1 (75 ng) plus E2 (60 ng) (lanes 4 to 6). Each reaction mixture was then subjected to native gel electrophoresis to isolate the C2 complex (lane 3), or the C1 (lane 5) and C3 (lane 6) complexes from the corresponding free (unbound) DNA (lanes 2 and 4, respectively). After separation, the free DNA pools, the DNA molecules within each complex, and the initial origin substrate DNA (lane 1) were isolated and chemically cleaved at the sites of modification. The cleavage products were separated by electrophoresis through a denaturing 8% polyacrylamide gel and visualized by autoradiography.

FIG. 3.

Interference of the formation of E1- and E1-E2-origin complexes by partial origin depyrimidation. Duplex DNA fragments, 5′-³²P labeled on either the top (A) or bottom (B) strand, were subjected to partial depyrimidation by treatment with hydrazine. The DNA fragments were then incubated with E1 alone (300 ng) (lanes 2 and 3) or E1 (75 ng) plus E2 (60 ng) (lanes 4 to 6). Each reaction mixture was then subjected to native gel electrophoresis to isolate the C2 complex (lane 3) or the C1 (lane 5) and C3 (lane 6) complexes from the corresponding free (unbound) DNA (lanes 2 and 4, respectively). After separation, the free DNA pools, the DNA molecules within each complex, and the initial origin substrate DNA (lane 1) were isolated and chemically cleaved at the sites of modification. The cleavage products were separated by electrophoresis through a denaturing 8% polyacrylamide gel and visualized by autoradiography.

Comparing the cleavage pattern of the DNA within the C2 complex (Fig. 2 and 3, lanes 3) to unbound DNA (Fig. 2 and 3, lanes 2) reveals numerous top- and bottom-strand bases within the dyad region whose modification inhibited C2 complex formation (i.e., whose intensity was lower in lane 3 than in lane 2). Inhibition of complex formation was caused by modification of all top-strand bases from nt 7940 to 16 and bottom-strand bases from nt 7943 to 14 (compiled below in Fig. 5C), although the loss of bases encompassing positions 7940 to 10 was observed to have a greater effect on complex formation. Outside of this region, the loss of two top-strand purines within the BS12 element at nt 19 and 22 and three bottom-strand purines within the AT-rich element (nt 7925 to 7927) were also seen to have significant effects on C2 complex formation (Fig. 2, lanes 3). Although the effects of modification of these purines were modest, they were consistently observed in our depurination studies. Thus, we conclude that E1 binding in the C2 complex is stabilized primarily by contacts with the dyad element, although the interaction of E1 with bases in BS12 and the AT-rich element play a supporting role.

FIG. 5.

Compilation of depurination, depyrimidation, and phosphate ethylation interference data for E1 and E2 binding to the BPV origin. Maps of modifications that interfere with C1 (A), C3 (B), and C2 (C) complex formation are shown. Phosphates whose modification strongly inhibits complex formation are indicated by solid triangles; weakly interfering phosphates are shown by open triangles. Bases whose removal (i.e., by depurination and depyrimidation) reduces complex formation are indicated by solid circles above (bottom strand) or below (top strand) the affected base. We also include the results from previous footprinting analyses (13). The top- and bottom-strand regions protected from DNase I cleavage by each complex are indicated by solid boxes above and below the sequence, respectively. Thymines hyperreactive to KMnO₄ (a probe of DNA structure) are indicated by ovals. The previous data for the C1 and C3 complexes was taken from complexes formed at low (50 ng) and high (400 ng) E1 levels. (D) Helix map of phosphates and purines in the nt 7930 region whose modification interferes with C3 complex formation. Phosphates whose ethylation inhibits complex formation are indicated by solid circles, while critical purines are indicated by open circles. The top and bottom strands are indicated. The dashed line in the center of the map distinguishes the two helical sides.

In the presence of E2, removal of most purine (Fig. 2) or pyrimidine (Fig. 3) bases within the dyad region affected both C1 (lanes 5) and C3 (lanes 6) complex formation (compare with lanes 4). Bases within BS12 also affected the formation of both complexes but in a differential manner, as described below. Modification of bases in the dyad region were observed to have a smaller effect on C1 and C3 formation than on C2 formation (compare lanes 5 and 6 with lanes 3). This result suggests that E2-mediated stabilization of E1 within the C1 and C3 complexes causes E1 binding to be less dependent on contacts with the dyad element (see also references 32 and 34).

Comparing the C1 and C3 complexes, we observed clear differences in the effect of modification of each flanking E2 binding site. Formation of the C1 complex was dependent upon contacts within BS12, as indicated by the effect of modification of top-strand bases at nt 15 to 18 and 24 to 27 and of bottom-strand bases at nt 15 to 18, 21, and 24 to 26. These data are in agreement with previous results showing the importance of E2 binding to BS12 for the formation of the C1 and similar fast-migrating complexes (24, 32, 34). Concerning the C3 complex, modification of BS12 bases needed for C1 complex formation had relatively modest effects on C3 complex formation. In contrast, modification of the E2 BS11 had more severe consequences. Top-strand bases from nt 7897 to 7910 and bottom-strand bases from nt 7896 to 7900 and nt 7904 to 7909 were found to be critical for C3 complex formation. We noted that when the interference pattern of DNA from the C3 complex was compared to that of unbound DNA, the intensity of certain bands in the BS11 region was reduced >90% in densitometric analysis (data not shown). Because base modification in this region had no effect on C2 complex formation, these data demonstrate that the recovered C3 complex was not contaminated by significant levels (>10%) of C2, which could potentially complicate the analysis of our interference results. In summary, because we have previously shown that E2 binds the BS11 region when incubated with E1 (13), our data indicate that E2 binding to BS11 is critical for C3 complex formation.

It was observed that modifications within BS11 stimulate the formation of the C1 complex, seen as an increase in the intensity of bands within the BS11 region compared to those for the unbound and substrate DNA (compare lanes 1, 4, and 5 in Fig. 2 and 3). The use of DNA substrates modified with dimethyl sulfate for methylation interference also indicates that modification of BS11 stimulates C1 complex formation (data not shown). These data argue that the C3 complex can occur in a pathway that uses the C1 complex as an intermediate.

Overlap of AT-region phosphate contacts with site of primary origin distortion.

The contacts made by E1 and E2 to the sugar-phosphate backbone were examined (Fig. 4). The DNA substrate was ethylated at phosphate positions with N-nitroso-N-ethylurea, generating phosphotriesters (38). Similar to that observed for the missing-base assays, E1 interaction with the backbone of the dyad region was critical for C2 complex formation (Fig. 4, lanes 3; compiled in Fig. 5C).

FIG. 4.

Interference of the formation of E1- and E1-E2-origin complexes by ethylation of the origin phosphates. Duplex DNA fragments, 5′-³²P labeled on either the top (A) or bottom (B) strand, were ethylated on a small fraction of DNA phosphates. The DNA fragments were then incubated with E1 alone (300 ng) (lanes 2 and 3) or E1 (75 ng) plus E2 (60 ng) (lanes 4 to 6). Each reaction mixture was then subjected to native gel electrophoresis to isolate the C2 complex (lane 3) or the C1 (lane 5) and C3 (lane 6) complexes from the corresponding free (unbound) DNA (lanes 2 and 4, respectively). After separation, the free DNA pools, the DNA molecules within each complex, and the initial origin substrate DNA (lane 1) were isolated and chemically cleaved at the sites of modification. The cleavage products were separated by electrophoresis through a denaturing 8% polyacrylamide gel and visualized by autoradiography. Phosphates are numbered according to the base position on the 5′ side.

Compared to the C2 complex, C1 had a smaller number of phosphate contacts in the dyad region (Fig. 4, lanes 5 [top-strand phosphates at nt 7940 to 7942] and lanes 3 and 4 [bottom-strand phosphates at nt 7947 and 9 to 11]). E1 also appeared to utilize top-strand phosphate contacts at nt 12 to 15, located between the dyad element and BS12, because they were seen to play a role in C2 complex formation. These contacts, when plotted on a helix map, appear on one face of the helix (data not shown). C1 complex formation also utilized six phosphate contacts within the BS12 (top-strand phosphates at nt 23 to 25; bottom-strand phosphates at nt 18 to 20). The C3 complex had a similar pattern of contacts in the dyad region to that observed for C2 (Fig. 4, lanes 6). The most notable feature of the C3 complex is the deleterious effect of BS11 modification (top-strand phosphates at nt 7896, 7897, 7905, 7906, and 7909; bottom-strand phosphates at nt 7898 to 7901, 7909, and 7910). Thus, as was seen for the base interference studies, the C1 and C3 complexes had differential requirements for the flanking E2-binding sites.

The interference data for the C1, C3, and C2 complexes were compiled (Fig. 5A, B, and C, respectively). Included in each compilation were the results of previous DNase I footprinting and KMnO₄ modification studies, the latter indicating the sites of ATP-dependent DNA distortion induced by E1 (13). We noted that the patch of phosphate and purine contacts in the nt 7930 region for the C3 complex overlapped the primary site of DNA distortion (centered at nt 7932). When these phosphate and base contacts were mapped on a DNA helix, they were found to be located on the same face of the helix (Fig. 5D).

Each E2-binding site stimulates formation of a different complex.

Our data suggest that, contrary to published data (40), both BS11 and BS12 play key roles during the initiation of BPV DNA replication. We therefore examined complex formation on BPV origins in which one of the two E2-binding sites was mutated to prevent E2 binding (Fig. 6). Complex formation was tested by a gel retardation assay in the presence of low levels (6.25 to 25 ng) of E1 to more closely mimic the expression levels in infected cells. Binding to the wild-type origin, as well as to mutant origins lacking BS11 (ΔBS11) or BS12 (ΔBS12), was tested. In the absence of E2, E1 formed the C2 complex on the wild-type (lanes 1 to 3) and ΔBS11 (lanes 7 to 9) origins, but only at the highest levels of E1 (25 ng). In contrast, C2 formed to a lower degree on the ΔBS12 origin at 25 ng of E1 (lanes 4 to 6). This result again indicates that sequences within the BS12 element stabilize E1 binding to the BPV origin.

FIG. 6.

Effect of E2 binding-site mutation on complex formation by E1 and E2. DNA fragments (³²P labeled) containing the wild-type (WT) ΔBS12, or ΔBS11 mutant origin were incubated with increasing levels of E1 (6.25, 12.5, and 25 ng) in the absence or presence of E2 (15 ng; as indicated). Complexes were cross-linked with glutaraldehyde, separated by electrophoresis through a native 5% polyacrylamide gel, and visualized by autoradiography. The locations of the C1, C2, and C3 complexes are indicated. Note that the amounts of E1 are less than that used in the experiment in Fig. 1 (100 ng), accounting for the reduced level of C3 complex on the wild-type origin (lanes 10 to 12) in this experiment.

The presence of E2 moderately stimulated C3 complex formation on the wild-type origin and allowed significant C1 complex formation at all levels of E1 (Fig. 6, lanes 10 to 12). A C1 complex was not detected on ΔBS12 (lanes 13 to 15), even using 10-fold-higher levels of E1 or E2 (data not shown). In constrast, C3 complex formation was observed on ΔBS12 using 25 ng of E1. On ΔBS11, C1 complex formation was seen at a level similar to that on the wild-type origin, while C3 formation was not observed. In summary, BS12 was required for C1 complex formation, BS11 was necessary for formation of the C3 complex, and both BS11 and BS12 were required for maximal stimulation of C3 complex formation.

Both E2-binding sites are required for wild-type levels of replication in vivo.

The requirement for the E2-binding sites in viral DNA replication was tested in vivo using a transient replication assay. To reproduce the levels of E1 and E2 occurring during viral infection, mutant origins were constructed in the context of viral DNA. A plasmid containing the BPV genome (pSS3 [28]; designated the wild type) was mutated to remove one or both of the E2-binding sites from the origin (designated ΔBS11, ΔBS12, and ΔBS11ΔBS12, lacking BS11, BS12, and both elements, respectively). As a negative control, an origin was used with a 15-bp insertion in the dyad symmetry element (designated LI 15C), previously shown to inactivate the viral origin (28). Each of these constructs was linearized with BamHI to liberate the viral genome from the vector, and the linearized DNA was transfected into C127 cells by electroporation. The test plasmids were cotransfected with wild-type viral DNA, not released from the vector, which served as an internal control. After 72 and 96 h, the viral DNA was isolated and linearized and the unreplicated DNA was destroyed by digestion with DpnI.

The ΔBS11ΔBS12 and LI 15C viral constructs failed to replicate (Fig. 7). At each time point, the ΔBS11 and ΔBS12 origins were defective compared to the wild type, since ΔBS11 and ΔBS12 replicated to 30 and 20% of the wild-type level, respectively. Similar effects on replication were observed when these experiments were repeated by varying the ratio of the wild-type-to-mutant origin template (1:10 and 1:1; data not shown), indicating that the inhibition is not a result of altered expression of E1 and E2. The presence of both E2-binding sites therefore resulted in a synergistic response of viral DNA replication. The inability of the mutant origins to replicate at wild-type levels indicates that both flanking E2-binding sites are important for BPV replication under physiological levels of E1 and E2 expression.

FIG. 7.

Mutation of either BS11 or BS12 is deleterious for transient BPV DNA replication in vivo. (A) Schematic showing the origins that were tested for replication activity. From top to bottom, these origins are the wild-type origin, LI 15C (containing a 15-bp insertion in the dyad symmetry element which inactivates the origin [28]), ΔBS12 and ΔBS11 (lacking the BS12 and BS11 elements, respectively), and ΔBS12ΔBS11 (lacking both E2-binding sites). (B) The viral DNA molecules (2 μg) were released from the vector and transfected into murine C127 cells by electroporation. As a control, the test plasmids were cotransfected with wild-type viral DNA (0.5 μg) not released from the vector. After 72 and 96 h, the viral DNA was isolated by the method of Hirt (15) and treated with MunI to linearize the viral genome and with DpnI to digest unreplicated DNA. DNA was detected by Southern blot analysis with nick-translated pSS3 as a probe. (C) The replication activity in vivo was quantitated by excising bands corresponding to linearized viral DNA and counting in a scintillation counter. Replication activity was normalized with respect to the replication activity of the wild-type origin at 96 h.

DISCUSSION

Our data indicate that the BPV E2 transactivator enables viral replication by facilitating discrete steps during formation of the viral initiation complex. These data lead to a model in which the initiation complex is formed in a pathway that entails the sequential interaction of E2 with two binding sites flanking the central origin region, each stabilizing a distinct E1-origin complex. In the first step, E2 binding to the BS12 element is required to form a replication-incompetent complex which contains a low oligomeric form of E1 bound to the dyad symmetry element within the central origin region. Additional E1 monomers bind and extend this complex outward, forming a replication-active species, in a step in which E2 binding to BS11 becomes paramount and E2 bound to BS12 is displaced.

A previous study concluded that only one of the two E2-binding sites (BS12) flanking the central origin region was required to support efficient viral DNA replication in vivo (40). In contrast, we found that both BS11 and BS12 play distinct roles of similar importance, indicated by the comparable reduction of transient replication caused by mutation of either of these two sites. The likely cause for this difference is that Ustav et al. (40) overexpressed E1 and E2 while our replication assays expressed E1 and E2 in the context of the viral genome. Since E1 is detected at extremely low levels in infected cells (31, 39), our data suggest that E1 overexpression inadequately reproduces replication conditions during infection.

The initiation of viral replication appears to involve the formation of the C1 recognition complex, although this complex per se is not active in replication. Stenlund and colleagues have made compelling arguments that a similar fast-migrating complex is critical by virtue of increasing the specificity of E1 for origin sequences (32, 33). Indeed we found that at low levels of E1, the C1 complex recruits E1 more effectively to the origin than does the C3 complex. These workers also found that a single E2-binding site engineered at different positions within the dyad-proximal flank of the origin can support complex formation by E1 and E2 (32). Using origins that contain BS11 but lack BS12, we were unable to observe a rapidly migrating complex similar to C1, even at very high levels of E1 or E2 (data not shown). In that the E1 recognition element has general twofold symmetry (see, e.g., reference 16), it seems unlikely that C1-like complexes can form only by using an E2-binding site located on the BS12 side of the origin. Two non-mutually exclusive causes appear more reasonable. First, the distance between E2 bound to BS11 and E1 bound to the dyad element may be too great to allow stable complex formation. Second, since the transition of C1 to C3 correlates with an extension of the E1 footprint into the AT-rich region (13), physical interaction between E2 binding to BS11 and the adjacent E1 may greatly favor the formation of C3 compared to a complex containing E1 bound only to the dyad region.

The C1 and C3 complexes are distinctly different by various criteria. Most obvious is the relative importance of the BS11 and BS12 elements for C3 and C1 complex formation, respectively, but other notable differences exist. Formation of C3 requires a larger number of phosphate contacts, particularly in the dyad region, supporting the hypothesis that E1 is in a higher oligomeric state in the C3 (and C2) complex compared to that in C1 and similar complexes (23, 33). C3 formation also utilizes contacts with purines in the AT-rich region (this work) and, from our previous DNA-probing analysis of the C3 complex in solution (13), results in protection of the minimal core origin sequence from nuclease attack. In contrast, E1 and E2 in the C1 complex had significant interactions only with the right half of the origin including the dyad region. Finally, our prior footprinting study of the C3 and C1 complexes in solution indicated that the C3 complex was competent to induce distortion in the origin structure while C1 was lacking in this ability (13).

In contrast to the dissimilarity between C3 and C1, the overall disposition of E1 in the C2 and C3 complexes appears similar. The pattern of base and phosphate interference for the C2 and C3 complexes is alike (Fig. 5), as is the ability of these complexes to distort origin structure (13). The mobility of the C3 complex by a gel retardation assay was only slightly reduced compared to that of C2, reflective of the additional molecular weight provided by E2 binding to BS11, indicating that the E1 oligomeric state in the two complexes is similar. Because of the great similarity of the C2 and C3 complexes, our data argues that C3, like C2, is replication competent. It is clear that E2 binding to BS11 significantly stabilizes C3 at low E1 levels (Fig. 6), indicating that the main role of E2 binding to BS11 is to stabilize this replication-active complex. The low expression levels of E1 in infected cells lead us to suggest that the normal replication-active complex in BPV-infected cells is C3 rather than C2.

Various data indicate that the C3 complex forms by using a favored pathway with C1 as an intermediate. First, our mobility shift assays show that formation of the C3 complex is stimulated by the presence of BS12, which is critical for C1 formation. Conversely, the BS11 element does not stimulate formation of the C1 complex. Second, modification of BS12 phosphates which are important for C1 complex formation also inhibits C3 formation, although to a lesser degree. This observation would be expected for a pathway in which C1 precedes C3. Third, modifications within BS11 that prevent C3 complex formation were found to stimulate the formation of C1. In other words, preventing E2 from binding to BS11 led to an increase in C1 complex formation. This result would also be predicted by a model in which C1 converts to C3. These observations strongly support the hypothesis that the C1 complex initially forms and then is transformed into the C3 complex.

We note, however, that if C1 complex formation were essential for the subsequent formation of the final initiation complex, BS12 modification would be expected to have similar effects on C1 and C3 complex formation, a result we did not observe. We therefore suggest that replication-active complexes can form independent of C1, although this pathway is conditional on the stabilization of E1 binding to the origin by the E2-BS11 complex. Similarly, we can conclude that BS11 is not essential for replication if E2 binding to BS12 is possible. Thus, formation of an active viral replication complex can occur by multiple pathways. Because the viral DNA lacking either BS12 or BS11 replicates to levels 20 to 30% of that in the wild type, the pathway involving both the C1 and C3 complexes appears to be favored.

We found that removal of nearly any base in the dyad element is similarly deleterious for C1 and C3 complex formation. This is particularly surprising with respect to C1, since the phosphate contacts fall predominantly on one helical face (data not shown), a result observed previously for a C1-like complex (33). We find it doubtful that E1 could form important contacts with nearly every dyad base yet use primarily one helical face of the DNA for binding. A more probable explanation derives from observations that DNA molecules containing an abasic residue have structural perturbations (see, e.g., reference 9). Loss of bases in the origin would therefore be expected to alter both the conformation and dynamic properties of the DNA, resulting in destabilization of the E1-origin complex.

We previously observed that E1 binding to the origin induced ATP-dependent structural distortions within the AT-rich region and, to a lesser degree, within the dyad element and BS12 (Fig. 5B) (13). In the present study, we found six phosphates and three bases in the nt 7930 region whose modification inhibited C3 complex formation. The base contacts were important for both C2 and C3 formation, showing that E1 uses limited sequence recognition of the AT-rich region. The nt 7930 region base and phosphate contacts, which fall on one helical face of the DNA, overlap the primary site of DNA distortion (Fig. 5B) (13), suggesting that these contacts are used to induce the structural transitions. These phosphate contacts are apparently more critical for C3 formation because their modification did not noticeably disrupt C2 complex formation. E2 was shown previously to lower the amount of E1 required to distort this region (13). These data suggest that E2 bound to BS11 causes E1 to more closely approach the DNA in the nt 7930 region. This effect may be a more general property of E1-E2 interactions, because ethylation of top-strand phosphates at nt 12 to 15 (adjacent to BS12) had a greater effect on C1 formation than on C2 formation. The ability of E2 to increase the interfering properties of phosphate ethylation may be due to an E2-induced conformational change within E1 that alters the interaction of E1 with DNA. Since E2 is known to bend DNA fragments (14), a related effect may be that E2-mediated DNA bending at each E2-binding site both heightens the deleterious effects of phosphate modification between the dyad and each E2 binding site and facilitates the E1-mediated DNA distortion in the nt 7930 region.

Our previous DNase I footprinting analysis of the wild-type viral origin showed that in the presence of E2, an increase in the concentration of E1 caused an extension of protection into the region between the dyad element and BS11 (13). The boundary of the footprint at low E1 concentrations (corresponding to the C1 complex [Fig. 5A]) overlapped the region of primary structural distortion. From this data, we suggested that increases in E1 concentration caused an additional lobe of E1 to bind adjacent to that E1 situated over the dyad element (13). Since we observed few key phosphate contacts in the AT-rich region other than those in the nt 7930 region, our results do not support the suggestion that two independent lobes of E1 bind to the dyad and AT-rich regions. Instead, they indicate that E1 binding to the AT-rich region is an extension of the E1 bound over the dyad element. This larger E1 structure would therefore be responsible for the induction of structural changes within the viral origin.

Clues to the mechanism of formation of the C3 complex by E1 and E2 can likely be obtained from comparison with the SV40 T antigen. E1 is homologous to T antigen, particularly in their C-terminal regions (8), which, for T antigen, contain the ATPase and other elements critical for DNA helicase activity (12). The two proteins have similar activities including the ability to bind and unwind the origin in the presence of a single-stranded-DNA-binding protein (10, 36, 44, 46). Each protein can bind the origin in both low and high oligomeric forms by using a central palindromic structure (23, 25, 26, 32, 33). In the presence of ATP, structural changes are induced on each side of the central palindrome (5, 13). Analysis of the T-antigen–SV40 origin complex by scanning transmission electron microscopy and DNA probing has shown that T antigen is bound as a double hexamer, with twofold symmetry around the central palindrome (26, 35).

The similar characteristics of E1 and T antigen, combined with the results of previous binding studies, lead to the following model of E1 binding to the BPV origin (Fig. 8). E1 initially binds the origin in a low oligomeric state, using symmetrical contacts with the dyad symmetry element, and is stabilized by E2 bound to BS12 (13, 23, 32, 33) (see above). The binding of additional E1 monomers enlarges the complex, both toward the proximal edge of the BS11 element and 10 bp beyond the distal edge of the BS12 element, as observed in the DNase I footprint of the C2 complex (Fig. 5C). A key feature of this model is that the twofold symmetry around the central palindrome is maintained. The rightward extension of the complex displaces the E2 bound to the BS12 element but is stabilized by (and stabilizes) E2 binding to BS11. Recent data by Berg and Stenlund (2) has shown that E2 has two distinct domains that can interact with E1, the DNA-binding domain and the transactivating domain, and our data indicates that each domain may be differentially used in the C1 complex and the C3 complex. In the presence of ATP, this higher oligomeric C3 complex induces structural transitions, primarily within the nt 7930 region, but also to significant levels within BS12. Similar to T antigen, the final complex would have two helicase entities, one bound to the left half of the dyad element and the AT-rich region and the other bound to the right half of the dyad element and the BS12 region. Since hRPA can denature DNA under conditions supporting DNA replication (18), the interaction of hRPA with the distorted DNA region(s) leads to origin denaturation and unwinding of the DNA by the DNA helicase activity of E1.

FIG. 8.

Model of E1 and E2 binding to the BPV origin to form a replication initiation complex. See the text for details. The light zigzag lines indicate regions of DNA distortion induced by E1 in the AT-rich and BS12 regions.

Although transcriptional transactivators have been found to modulate DNA replication by a diversity of mechanisms, our data indicates that E2 is perhaps unique in its ability to use multiple binding sites to sequentially activate viral replication. Other papillomaviruses have multiple E2-binding sites in origin-proximal regions, and it would not be surprising to find the differential usage by E2 of these sites during the initiation of replication.