INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Papillomaviruses (PVs) cause a benign hyperproliferation of epithelial cells that sometimes progress to form carcinomas. The model system for studying papillomaviruses has been bovine papillomavirus type 1 (BPV1) because of its ability to infect and transform a variety of rodent cells in tissue culture (41).

Papillomaviruses encode early and late proteins that are involved in regulation of virus gene expression and replication, and assembly of the virion, respectively. E2 is a dimeric multifunctional early protein that is intimately involved in the regulation of gene expression and viral genome replication. The primary structures of papillomavirus E2 proteins are highly conserved. They consist of an N-terminal domain that can act as a transcriptional activator and is also involved in viral DNA replication and interaction with the viral DNA replication protein E1; a central, poorly conserved hinge region; and the C-terminal DNA binding and protein dimerization domains (17). Analysis of the crystal structure of the C terminus of the BPV1 E2 protein bound to its DNA binding site revealed that two -helices, one from each monomer of an E2 dimer, bind in the major groove of the DNA (8, 9). The original studies of the BPV1 E2 protein identified the high-affinity DNA binding sites within the BPV genome to be 12 bp long and to minimally contain the sequence ACC-N₆-GGT (canonical site), with the highest-affinity site being ACCG-N₄-CGGT (2, 15).

The occurrence of BPV1 E2 high-affinity DNA binding sites in the long control region (LCR) of the human papillomaviruses (HPVs) has led investigators to assume that the HPV E2 proteins preferentially bind this site (1, 33, 39, 40, 42).

Many E2 functions are dependent on the relative affinity of the protein for its various DNA binding sites. HPVs that infect the genital mucosal epithelium contain multiple E2 binding sites within their LCRs, both proximal and distal to the transcription initiation site for early-gene expression. The distal, higher-affinity sites apparently act as enhancers, and the proximal, lower-affinity sites act as repressors of early gene expression (17). These sites are thought to be part of a switching mechanism that modulates the levels of early gene expression during the viral life cycle. In addition, it has been shown that E2 can compete for the binding of cellular transcription factors from their neighboring or overlapping sites within the LCR (5, 37, 38).

The papillomavirus proteins E1 and E2 are necessary for HPV genome replication (12, 32, 36). For HPVs, the minimal origin of replication is defined as an E1 (E1BS) and an E2 binding site (E2BS) in close proximity flanked by an A/T-rich region, with additional E2BSs facilitating replication (3, 16, 36). In the absence of an E1BS, two E2BSs near the A/T-rich region can support transient HPV replication (36). Deletion analyses around the BPV1 origin of replication have revealed that the location of E2BSs with respect to the E1BS and A/T-rich region is flexible. Moreover, the affinity of the E2 protein for a particular site directly correlates with its ability to stimulate DNA replication (6, 43, 45).

The primary objective of our studies was to elucidate the highest-affinity binding sites for HPV E2 proteins. To address this question, we employed the nonbiased cyclic amplification and selection of targets (CASTing) technique (48). We identified a unique set of sequences, ACAC-N₅-GGT, that HPV E2 proteins (but not the BPV1 E2 protein) bind with a relative affinity that is indistinguishable from the canonical high-affinity site. Our studies also suggest the existence of preferred nucleotides within the flanking and core (N₅) sequences. Comparisons of the relative affinities and binding complex half-lives were made for the HPV51, HPV-16, and BPV1 E2 proteins with the different DNA binding sites. These novel sites are located within papillomaviruses genomes at locations where E2BSs are typically found.

In order to assess how an E2 protein might utilize the novel E2BSs, we used it in place of the wild-type sites found in the early promoter and replication origin of the better-characterized HPV type 16. HPV16 infects the epithelial cells of the genital mucosa and, like all other high-risk HPVs, is strongly associated with cervical carcinoma. We have designed a single plasmid that allows the assay of both transient transcription and transient replication. This plasmid, pOri16L, was built with a portion of the HPV16 LCR that contains both the origin of replication and the early promoter driving the expression of the firefly luciferase reporter gene. There are three canonical E2BSs within this portion of the LCR (see Fig. 5). In this study, we mutated each of the wild-type E2BSs to either eliminate binding (BS-KO) or create new E2BSs with the novel binding site sequence ACACAAATCGGT. Here we demonstrate that the novel E2BS functionally substitutes for the native E2BSs within this portion of the LCR in both transient-transcription and replication assays. Because bp 3 of the novel E2BSs disrupts the canonical site's palindrome, we also addressed the influence of the orientation of this binding site on E2 function. The results of our experiments with these mutated LCRs demonstrate that the functional role for the ACAC-N₅-GGT sites in E2-mediated replication of and transcription from the HPV genome can be dependent on binding site orientation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

DNA constructs for protein expression. DNA constructs used for expression of proteins in bacteria were made by PCR amplification of portions of the genes encoding the E2 proteins from HPV types 51 and 16 and BPV type 1. The resulting PCR products were digested with BamHI and EcoRI, whose sites are in the primers used for amplification (underlined in the primer sequences listed in Table 1) and then ligated in frame with and C-terminal to the glutathione S-transferase (GST) gene of pGEX-3X (Amersham Pharmacia Biotech, Piscataway, N.J.). These constructs were designed to express GST fusion proteins with either the full-length (fl) or the short C terminus (sct) E2 proteins that contain only the DNA binding and dimerization domain of the respective E2 proteins. The fl constructs GST-₅₁E2fl, and GST-B₁E2fl contain the entire E2 coding sequences. The set constructs GST-₅₁E2sct, GST-₁₆E2sct, and GST-B₁E2sct contain papillomavirus nucleotide sequences 3536 to 3811 from HPV51, 3584 to 3892 from HPV16, and 3457 to 3840 from BPV1, respectively. Nucleotide numbering corresponds to papillomavirus genome sequences listed in the Human Papillomavirus Compendium, HPV database (22). The primers used to create these constructs are listed in Table 1.

Protein purification. Escherichia coli strain BL21/DE3 was used for bacterial expression of recombinant proteins. Proteins were extracted from cultures (500 to 1,000 ml) of bacteria grown in liquid overnight at 25°C without IPTG (isopropylthiogalactoside) induction because induction resulted in partitioning of most of the E2 proteins to the insoluble fraction upon extraction (data not shown). Total-cell extracts were made by sonication, and proteins were purified by their affinity for glutathione-agarose beads (Amersham Pharmacia Biotech). Cleavage of the GST from the E2 protein was performed by addition of factor Xa protease (Roche Molecular Biochemicals, Indianapolis, Ind.) to the glutathione-eluted fraction and incubation with gentle agitation at 4°C for 16 to 24 h in factor Xa cleavage buffer (50 mM Tris-HCl [pH 8.0], 100 mM NaCl, 10 mM MgSO₄, 1 mM CaCl₂, 5 mM dithiothreitol [DTT]). To further purify this protein, it was applied to an S-Sepharose (Amersham Pharmacia Biotech, Piscataway, N.J.) column that was equilibrated with S-Sepharose buffer (20 mM Tris-HCl [pH 8.5], 100 mM NaCl, 5 mM EDTA [pH 8.0]). Protein was eluted with a linear salt gradient (100 mM to 1 M NaCl) in S-Sepharose buffer, and 1-ml fractions were collected and screened for active E2 protein by electrophoretic mobility shift assay (EMSA) and Western blotting (data not shown).

Cloning of PCR-amplified sequences. Sequences amplified by PCR were cloned using the TA cloning kit (Invitrogen Corp., Carlsbad, Calif.) with the pCRII and pCR2.1 vectors according to the manufacturer's instructions.

Sequencing reactions. DNA sequencing was performed using the Sequenase version 2.0 DNA sequencing kit as per the manufacturer's instructions (Amersham/USB) with M13 forward and M13 reverse primers end labeled with [-³²P] ATP using T4 polynucleotide kinase.

CASTing. The CASTing method was performed as previously described (48) with minor modifications. Each oligonucleotide in the degenerate oligonucleotide library (DOL) contains PCR primer binding sites that flank a core region of 20 randomly generated nucleotides. The oligonucleotides used were DOL, 5'-AGACGGATCCATTGCA-N₂₀-CTGTAGGAATTCGGA-3', and the primers used for PCR amplification were N₂₀-B (5'-AGACGGATCCATTGCA-3') and N₂₀-R (5'-TCCGAATTCCTACAG-3'). CASTing was performed by adding about 10 µg of double-stranded DOL to 20 µl of glutathione-agarose beads with approximately 15 to 20 µg of GST fusion protein (GST-HPV₅₁E2fl, GST-His₆, or GST-BPV₁E2fl) captured from dialyzed extracts in a total volume of 100 µl of binding buffer (10 mM Tris-HCl [pH 7.5] 50 mM NaCl, 1 mM EDTA, 4 mM DTT, 250 µg of bovine serum albumin [BSA] per ml, 5% glycerol). This mix was allowed to interact at room temperature for 3 h with gentle agitation followed by three washes with binding buffer. The complexes of glutathione-agarose beads/GST protein/double-stranded DOLs were then resuspended in the PCR buffer mix (30 µl of 10× Taq buffer [Promega], 18 µl of 25 mM MgCl₂, 3 µl of 10 mM deoxynucleotide triphosphate (dNTP) mix 3 µl of Taq DNA polymerase, 3 µl of 500-pmol/µl N₂₀B primer, 3 µl of 500-pmol/µl N₂₀R primer, in a total volume of 300 µl) and amplified according to the following protocol: 95°C for 5 min, then 10 PCR cycles of 94°C for 1 min, 40°C for 1 min, and 72°C for 30 sec, followed by cooling to 4°C, and a 100-µl aliquot was removed and stored on ice. The remaining 200 µl was cycled four more times (14 in total) and cooled to 4°C, and another 100-µl aliquot was removed and stored on ice. The remaining 100 µl was subjected to four more rounds of amplification (18 in total), followed by cooling to 4°C. A 10-µl aliquot (plus 1 µl of loading dye) from each of the PCR amplifications (10, 14, and 18 cycles) was electrophoresed on a 2.5% agarose-1× TBE (90 mM Tris-borate, 1 mM EDTA [pH 8.5]) gel and visualized by ethidium bromide staining and UV transillumination. A portion of the pool obtained from each round of CASTing was amplified using radiolabeled primers and tested for enrichment by EMSA. After nine rounds of CASTing with GST-HPV₅₁E2fl, a protein-probe complex was easily detected by EMSA, and the CASTing procedure was stopped. The amplified pools were cloned into the pCRII vector (Invitrogen), the resulting plasmid DNAs were isolated from individual clones, and their sequences were determined.

Additional plasmid constructs. (i) E2 binding sites. The construct pBPV1E2BS (a gift from Eliot Androphy) is a pUC18 derivative containing the sequence tcgagaACCGAATTCGGTagcc cloned into the polycloning sequence. This clone was used as a positive control in EMSAs that analyzed the products of the CASTing reactions and for screening TA clones (last two lanes of Fig. 1). Additional E2 binding sites used as probes for EMSAs were cloned by annealing two complementary oligonucleotides (H and B clones; see Table 2) that contain the same flanking sequences; +, 5'-CCAGAGTGAATTCCAGA-(12-bp binding site)-TCCCAAGCTTGGCG-3', and , 5'-CGCCAAGCTTGGGA-(complement of 12-bp binding site)-TCTGGAATTCACTCTGG-3'.These were then cleaved with EcoRI and HindIII and ligated into a pUC18 vector between its unique EcoRI and HindIII sites.

(ii) pOri16L. The pOri16L plasmid was constructed by building a portion of the HPV16 LCR and the firefly luciferase gene into a pUC19 plasmid backbone. The portion of the HPV16 LCR from nucleotide positions 7800 to 73 (nucleotide numbering as in reference 22) that includes the HPV16 origin of replication and the P97 early promoter, including its TATA box, was amplified by PCR. The primers used to amplify this portion of the LCR contain novel PstI and BamHI restriction sites to facilitate cloning: 16Ori-upper/PstI primer, 5'- CATGAACTGTCTGCAGGTTAGTCATAC-3', and 16Ori-lower/BamHI primer, 5'- GTGCATAAAGGATCCGCTTTTATAC-3'.

(iii) pOri16L mutants. The binding site knockouts and sequence substitutions were all created by site-directed mutagenesis of pOri16L using either the MORPH kit (5 Prime  3 Prime, Inc., Boulder, Colo.) or the QuickChange site-directed mutagenesis kit (Stratagene Cloning Systems, La Jolla, Calif.) as per the manufacturers' instructions. For plasmids with mutations in more than one E2BS, one site was altered using the pOri16L plasmid as the template and the second site was altered using the partially mutated plasmid as the template. All mutated plasmids were screened by DNA sequence analysis. Upon isolation of mutant clones, origin-containing fragments were removed by digestion with PstI and EcoRI (the EcoRI used for this recloning step is found in the luciferase BamHI cassette) and then ligated into a pOri16L plasmid from which the wild-type LCR was removed by digestion with the same enzymes.

(iv) pCMV Series. The pCMV-E2₁₆ and pCMV-E1₁₆ expression plasmids were kindly provided by Peter Howley (32). They express the full-length HPV16 E1 and HPV16 E2 proteins driven by the cytomegalovirus (CMV) promoter. The pCMV₄-XS plasmid was constructed by digesting pCMV-E1₁₆ with XbaI and SmaI, filling in the resulting overhangs with deoxyribonucleotides using T4 DNA polymerase, and ligating the resulting blunted ends as described above for the pUC19-EX plasmid construct. The identities of all of the above constructs were confirmed by DNA sequence analysis.

EMSA. M13rev primers were end labeled with [-³²P]ATP using T4 polynucleotide kinase. PCR was then performed using this radiolabeled primer plus an unlabeled M13for primer to create a single end-labeled, double-stranded DNA probe for EMSA. The constructs used as templates for making probes were either the CASTing TA clones in the pCRII vector or pH and pB clones (see above for cloning details and Table 2 for binding site sequences) in pUC18. The resulting PCR products were gel purified. EMSAs were performed in binding buffer with purified protein, 1 µg of sonicated salmon sperm DNA (Sigma, St. Louis, Mo.), 3 µg of BSA, and 10⁴ cpm of radiolabeled DNA probe. This binding reaction was incubated at room temperature for 30 min and then loaded directly onto 6 to 8% native polyacrylamide gels containing 0.25×TBE and 2.5% glycerol.

Tissue culture. J2-3T3 cells were cultured in Dulbecco's modified Eagle's medium with 10% bovine calf serum. SCC-13 cells (29) were grown on mitomycin C-treated J2-3T3 cell feeder layers in E medium as described previously (19).

Luciferase expression assays. SCC-13 cells were plated on mitomycin C-treated J2-3T3 feeder layers in 35-mm dishes 24 h before transfection. The plasmids pOri16L, pRL-TK, and pCMV-E2₁₆ or pCMV₄-XS (see the legends to Fig. 6 and 7 for details) were introduced into the SCC-13 cells using LipofectAmine, as per the manufacturer's instructions (Life Technologies). Cells were harvested and assayed at 36 h posttransfection using the dual luciferase kit (Promega, Madison, Wis.), and units of luciferase activity were determined using a Berthold Lumat LB9501 luminometer (Berthold Systems, Inc., Pittsburgh, Pa.). Expression levels were determined from duplicate transfections in three independent experiments.

Transient-replication assays. Transient-replication assays were performed with SCC-13 cells as previously described for HPV31 (12). Briefly, plasmid DNAs (quantities and identities detailed in the legend to Fig. 5) were electroporated into SCC-13 cells as detailed by Hubert et al. (12) and Ustav and Stenland (44). Replication was assayed after DpnI digestion of low-molecular-weight DNA extracted by a modified Hirt protocol (10) followed by gel electrophoresis and Southern blot hybridization.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cyclic amplification and selection of targets. High-affinity DNA binding sites for the HPV E2 protein were identified using the CASTing technique (48). A GST fusion with the full-length HPV type 51 E2 protein (GST-₅₁E2fl) was used to select specific DNA binding sites from a random pool of degenerate double-stranded oligonucleotides (see Materials and Methods for details).

View larger version (67K): [in this window] [in a new window]   FIG. 1.   EMSA analysis of CASTing-enriched probes. Oligonucleotide probes that formed DNA-protein complexes were amplified by PCR and radiolabeled by incorporation of [-32P]dCTP in the reaction mixture. These DNA probes were used in EMSAs with the GST-HPV51E2 protein to form complexes. The original degenerate oligonucleotide pool (N20S) was used as the template for the lane labeled round 0. The next five lanes represent amplified probes from rounds 6, 7, 8, and 9, as labeled above the EMSA. The final two lanes were loaded with complex formed between radiolabeled probe amplified by PCR from the pBPV1-E2BS plasmid, which contains the binding site sequence gagaACCGAATTCGGTagcc as the template with (+) or without () the GST-HPV51E2 protein.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Determination of a binding site consensus from the GST-HPV51E2fl CASTing. PCR amplimers from multiple rounds of the CASTing process were cloned, and the sequence of the insert was determined. Seventy-five sequences from the GST-HPV51E2fl CASTing experiment were isolated, sequenced, and aligned based on homologies within the variable 20 bp of each clone. (A) The numbers represent the frequency with which a nucleotide was found at each position (expressed as a percentage of the total). Listed above the table is the consensus arrived at by determination of the most frequently occurring nucleotide at each position. (B) The frequency of specific sequences within the pool of sequenced clones expressed as a percentage of the total. Note that the first two and last two sequences listed are subsets of one another, respectively.

HPV E2 proteins bind to the novel nonpalindromic site with higher relative affinities than to a canonical BPV E2 site. EMSAs were performed using E2 proteins derived from different mucosal HPVs to ask if binding to the nonpalindromic HPV site was a general characteristic of HPV E2 proteins (Fig. 3). The probes used were the CASTing consensus sequence and a high-affinity canonical E2 binding site. Equal amounts of extract were used for each EMSA reaction; however, because the expression levels of the different E2 proteins varied, meaningful conclusions can only be drawn from comparisons between the relative amounts of the probes shifted by the same E2 protein extract. In addition, the expression levels of the baculoviruses expressing BPV₁E2fl and HPV₁₆E2ct were very low, making the complexes in Fig. 3, lanes 6 and 10, difficult to see in this reproduction of the autoradiograph. Note also that the complexes formed with nuclear extracts may contain more proteins that interact with the probes than just the baculovirus-expressed E2. Thus, the relative mobilities of the various E2 complexes cannot be meaningfully compared. The GST-₅₁E2fl protein has a higher relative affinity for the CASTing consensus site than for the canonical E2 binding site (Fig. 3, lanes 1 and 2). The same result is found with the full-length HPV51 E2 protein purified from a baculovirus expression system (Fig. 3, lanes 3 and 4), demonstrating that recognition of the novel binding site was not a property of either the GST fusion or bacterial expression of the protein. Note that in the case of the GST fusions with full-length E2 proteins, because both the E2 proteins and the GST tag can dimerize independently, there is a pair of shifted complexes that likely represent dimers and tetramers of these proteins (Fig. 3, lanes 1 and 2). When a full-length BPV1 E2 protein was used in the EMSA, there was no detectable shift of the GST-₅₁E2 CASTing consensus site (Fig. 3, lanes 5 and 6), consistent with the fact that this site was not represented within the pool of sequenced GST-BPV₁E2fl CASTing clones (data not shown).

View larger version (67K): [in this window] [in a new window]   FIG. 3.   EMSA of multiple E2 proteins. Two clones from the GST-HPV51E2fl CASTing experiment were used to make radiolabeled probes. They represent the GST-HPV51E2fl CASTing consensus sequence (ACACAAATCGGT, odd numbered lanes) and a high-affinity binding site similar to those previously described for other E2 proteins (ACCGAATATGGT, even numbered lanes). Below each lane number is the percentage of the total probe shifted by each E2 protein. These numbers were calculated from the scanned autoradiogram using ImageQuant software to calculate pixel densities for each band. Lanes 1 and 2 represent complexes formed between these probes and the bacterially expressed GST-HPV51E2fl proteins used for the CASTing experiment. Lanes 3 through 12 all contained nuclear extracts from baculovirus-infected insect cells. Lanes 3 and 4 are binding reactions with the full-length E2 protein of HPV51. Lanes 5 and 6 are binding reactions with full-length E2 protein of BPV1. Lanes 7 and 8 are binding reactions with full-length E2 protein of HPV16. Lanes 9 and 10 are binding reactions with the C-terminal portion of the HPV16E2ct protein. Lanes 11 and 12 contain extracts from insect cells infected with a baculovirus expressing -glucoronidose. Lanes 13 and 14 contain the two probes loaded without exposure to any protein extracts.

Analysis of DNA binding by highly purified E2 proteins. So that a more stringent determination of relative affinities and off-rates could be made, a purification scheme was designed to isolate highly purified ₁₆E2sct, ₅₁E2sct, and GST-BPV₁E2sct. Attempts to cleave the GST portion of the GST-BPV₁E2sct protein with factor Xa resulted in loss of all DNA binding activity (data not shown). Therefore, factor Xa cleavage was not performed on the GST-BPV₁E2sct protein. The GST-E2sct constructs used above were expressed in bacteria and purified as described in Materials and Methods. Fractions were screened for activity by EMSA and examined for the level of purity by sodium dodecyl sulfate (SDS)-PAGE (Fig. 4A). The active, highly purified fractions were pooled and used for all further work defining the relative affinities and off-rates for these proteins with selected DNA binding sites. The E2 proteins used in this study were of comparable length, and this choice was based on the results of Pepinsky et al., who demonstrated a correlation between the length of BPV1 E2 C-terminal constructs and their binding affinity (25).

View larger version (88K): [in this window] [in a new window]   FIG. 4.   Determination of the Krel and T1/2 for the HPV51E2sct protein with various DNA binding sites. (A) Silver-stained 4 to 12% gradient SDS-PAGE of the highly purified HPV51E2sct protein preparation (lane 1). Proteins in lane 2 are prestained molecular size markers. (B) EMSAs to determine the relative affinity of the E2 protein for various DNA probes. Reaction mixtures were made with the HPV51E2sct protein and equal amounts of radiolabeled Bm#3 probe DNA with no competitor DNAs (lanes 0) or with sequential dilutions corresponding to a 2-, 100-, and 500-fold excess of the PCR-amplified unlabeled competitor DNA listed below the EMSAs. Binding was for 30 min before electrophoresis. Gels were dried, exposed to PhosporImager screens, and analyzed using ImageQuant software. (C) EMSAs to determine the rate of E2 protein-DNA dissociation. The HPV51E2sct protein was reacted for 30 min with the probe DNAs identified below each section of the EMSA. Then, a 1,000-fold excess of unlabeled competitor DNA, Bm#3, was added to each reaction mixture. Lanes 0 were loaded immediately, while others were allowed to incubate for 1, 5, 10, 30, or 90 min before electrophoresis. Gels were dried, exposed to PhosphorImager screens, and analyzed using ImageQuant software.

Relative affinities and off-rates for E2BSs. To further confirm the CASTing results, we constructed a novel set of binding site clones with defined nucleotide differences based on the consensus E2BSs determined from the GST-HPV₅₁E2fl and GST-BPV₁E2fl CASTing experiments. Figures 4B and C are representative EMSAs used to define the relative affinities and off-rates for these protein-DNA complexes. Table 2 lists the binding sites utilized and the EMSA results for HPV₁₆E2sct, HPV₅₁E2sct, and GST-BPV₁E2sct. The relative affinities suggested by the frequency with which a sequence was identified in the GST-HPV₅₁E2fl CASTing experiment correlate with the relative affinities of both HPV E2sct proteins in the EMSAs. Changes in the most highly conserved nucleotides, the first and last two base pairs, result in abrogation of binding (Table 2, compare Hwt with Hm#7). HPV E2 proteins prefer A/T base pairs in the core of their recognition sites for (Table 2 compare Hwt with Hm#9 and Bm#3 with Bwt). The HPV E2sct proteins bound both the CASTing consensus and canonical E2BSs with very similar affinities (Table 2, compare Hwt with Bm#3). This result does not directly correlate with the GST-HPV₅₁E2fl CASTing experiment because the canonical E2BS was identified with a much lower frequency than the novel E2BS (see Fig. 2B). A novel palindromic sequence, ACACAAATGTGT (Table 2, Hm#5), was also bound by the HPV E2sct proteins with relative affinities that were lower but still within the same order of magnitude as the CASTing consensus and canonical E2 binding sites (Table 2, compare Hm#5 with Hwt and Bm#3).

Novel DNA binding sites are located throughout HPV genomes. As a first step in addressing if these novel E2BSs may have biological relevance, we searched the papillomavirus genome database (21) for the occurrence of ACAC-N₅-GGT, ACC-N₅-GTGT, and ACAC-N₄-GTGT. Most papillomavirus genomes contain one or more of these sites in regions where E2BSs are commonly found. Table 3 identifies the locations of these sites in selected papillomavirus genomes. The HPV51 genome has two such novel high-affinity sites within the LCR proximal to the putative E1BS. The site ACCGATTTGTGT (Table 3, column LCR [ori/E]) closest to the E1BS is identical to the HPV51 E2 CASTing consensus sequence (Fig. 3). For other HPVs that infect mucosal epithelium (e.g., HPV11 and HPV18), the analogous E2BS is involved in initiation of replication (E2BS 3) (3, 4, 31, 36), suggesting that this novel site may serve a similar function in replication of HPV51.

Construction of the pOri16L reporter plasmids. We next asked if the novel E2BS could functionally substitute for the canonical sites. The E2 protein is involved in the regulation of both papillomavirus gene transcription and genome replication. By substituting the novel E2BS for the canonical site in a system that allows assay of E2-mediated effects, we can assess the functional role of the sites. The well-characterized HPV type 16 genome was used for these experiments. To this end, the pOri16L reporter was constructed; it contains the 3 portion of the HPV16 LCR that includes the DNA origin of replication and the overlapping early promoter structure that contains three canonical E2BSs fused to a luciferase reporter. All three sites are involved in the HPV16 E2 protein-mediated modulation of viral replication and transcription (4, 26, 32).

View larger version (56K): [in this window] [in a new window]   FIG. 5.   pOri16L constructs. A fragment of the LCR that contains three of the four E2BSs from the HPV16 LCR (A) was fused to the luciferase gene. The three E2BSs are labeled and represented by the circles below the LCR (BS1 to BS3). Adjacent to BS2 is an Sp1 site, and adjacent to BS1 is the early-gene TATA box. Also contained within this portion of the LCR is the viral origin of replication (ori). The minimal viral ori is defined by the proximal E1BS and E2BS and their neighboring A/T-rich regions. All of these LCR E2BSs stimulate viral DNA replication. (B) The wild-type and modified sequences for each of the three E2BSs of the pOri16L constructs. The names correlate with either the single-nucleotide-transition mutations or the orientation of the novel E2BS that is created within each mutated pOri16L construct. Within the sequences, lowercase lettering corresponds to flanking nucleotides, uppercase letters correspond to the 12-bp E2BS, and underlined letters correspond to those nucleotides that differ from the wild type for each site. At the right is a schematic of each construct. Open boxes are wild-type E2BSs; an X through the box corresponds to a knocked-out binding site; transition mutations are specified within the E2BS box; and large arrows within a box correspond to the novel E2BS in either the forward or reverse orientation.

Relative affinities and off-rates of the pOri16L E2BSs. Relative affinity and stability studies with the HPV₁₆E2sct protein and the pOri16L construct E2BSs were performed to analyze the effects of the sequence changes on two important biochemical parameters. Table 4 lists the results of these HPV16 E2 binding affinity and complex stability analyses.

10

11

Effects of site substitutions on basal promoter activity. This experiment was designed to determine if the novel E2BS can functionally substitute for the wild-type E2BSs within pOri16L in either orientation. Previous studies have shown that the HPV16 E2 protein affects expression from the early promoter at the level of transcription (26). Therefore, a luciferase cassette driven by the pOri16L constructs was used as a reporter to monitor the effects of the E2 protein on transcription from the mutated HPV16 promoters. These assays were first performed in the absence of the E2 protein to determine what effects the sequence alterations might have on the basal activity of the HPV16 early promoter. The pOri16L wild-type and mutant plasmids were introduced into SCC-13 cells along with a reference plasmid and pRL-TK (see Materials and Methods and Fig. 6 legend for details).

View larger version (38K): [in this window] [in a new window]   FIG. 6.   Basal expression levels from the pOri16L plasmids. SCC-13 cells were transfected with 0.05 µg of pRL-TK and 0.5 µg of pCMV4-XS plus 1 µg of each pOri16L template. Dual luciferase assays were performed on cell extracts prepared at 36 h posttransfection. The luciferase activity for each cell lysate was expressed as the ratio between the firefly and Renilla luciferase activities and then normalized to the pOri16L wild-type (W.T.) levels (100%) for comparisons between experiments. The average results from three independent experiments are plotted for each set of templates. (A) E2BS#3; (B) E2BS#2; (C) E2BS#1; (D) constructs with alterations to more than one E2BS. Error bars correspond to the standard deviation for each data set.

Novel E2BS functionally substitutes for wild-type E2BSs in transient transcription assays. The E2 protein can repress transcription from the HPV early promoter (30). Therefore, an HPV16 E2 protein expression construct was cotransfected into SCC-13 cells along with each of the pOri16L mutants and the reference plasmid pRL-TK to determine whether the novel E2BS could functionally substitute for the wild-type sites in an E2-responsive manner.

View larger version (18K): [in this window] [in a new window]   FIG. 7.   pOri16L repression by the HPV16E2 protein. SCC-13 cells were transfected with 0.05 µg of pRL-TK, 1 µg of each pOri16L template, and either 0.5 µg of pCMV4-XS or 0.5 µg of pCMV-E216. Dual luciferase assays were performed on cell extracts prepared at 36 h posttransfection. The luciferase activity for each cell lysate was expressed as the ratio between the firefly and Renilla luciferase activities. The levels of repression were calculated by dividing the basal promoter activity (transfections with the pCMV4-XS plasmid) by the promoter activity in the presence of the E2 protein. The results from three independent experiments are plotted for each set of templates: (A) E2BS#3; (B) E2BS#2; and (C) E2BS#1. Error bars correspond to the standard deviation for each data set.

Novel E2BS functionally substitutes for the wild-type E2BSs in transient-replication assays. Another major function of the E2 protein in the viral life cycle is the stimulation of DNA replication in conjunction with the papillomavirus E1 protein. To determine if the novel E2BS can functionally substitute for the canonical E2BSs in E2-mediated DNA replication, we used the pOri16L mutants in transient-replication assays. The pOri16L constructs were designed to be analogous to plasmids used in other studies of HPV replication (32, 36). Each of the E2BSs contained in the pOri16L plasmid is known to influence the efficiency of replication. Transient-replication assays were performed as described for HPV31 (12, 28, 32).

View larger version (32K): [in this window] [in a new window]   FIG. 8.   Transient-replication assays with pOri16L mutants. (A) Representative Southern blot prepared by electrophoresis of Hirt DNA digested with both AlwNI and DpnI, transferred to a nylon membrane, and visualized with radiolabeled probe made using pOri16L as the template. In the left lane is one-eighth of the wild-type sample that was digested only with AlwNI. The next three lanes contain dilutions of DNA standards (in picograms) to control for hybridization efficiency between blots. (Because all cells are not efficiently transfected by the electroporation procedure, these standards cannot be used to determine pOri16L copy numbers replicated per cell) The lanes labeled E1/E2, E2, and E1 were transfected with wild-type (W.T.) pOri16L alone or with the addition of only the E1 or E2 expression plasmid respectively. The remaining lanes are labeled with the name of the pOri16L clone (3 µg) that was transfected into SCC-13 cells along with the E1 and E2 expression plasmids (in equal molar amounts; 3 µg of pOri16L plus 3.3 µg of pCMV-E216 plus 3.8 µg of pCMV-E116). Southern blots were exposed to PhosphorImager screens and analyzed using ImageQuant software. Shown are the relative replication activities from three independent transfections of pOri16L templates with mutations to their (B) E2BS#3, (C) E2BS#2, or (D) E2BS#1. Error bars correspond to the standard deviation for each data set.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

CASTing was used to identify a novel family of HPV E2BSs whose members are bound with affinities similar to that of the canonical E2BS. The CASTing results reveal the promiscuous nature of the HPV E2 protein with respect to DNA binding site selection. Our results allude to the potential flexibility of the HPV E2 protein's conformation upon binding to the E2BSs, as these proteins can bind to either the CASTing consensus (ACAC-N₄-CGGT) or canonical E2BS (ACCG-N₄-CGGT) with similar relative affinities if their core and flanking nucleotides are conserved (Table 2, Hwt versus Bm#3). These novel sites are present in HPV genomes at locations where E2BSs are commonly found.

To show that the novel site also has biological significance, we substituted it for each of three of the wild-type canonical sites within the HPV16 LCR (Fig. 5). We demonstrated that in both transient-transcription assays (Fig. 7) and transient-replication assays (Fig. 8), substitution for the wild-type E2BSs with the novel E2BSs had, in some instances, very strong and unpredicted effects on the ability of the HPV16 E2 protein to modulate transcription and replication. In the case of BS3-hE2BS, reporter expression was activated rather than repressed. For BS2, replication was only supported when the site was replaced with the novel hE2BS in one orientation. Thus, binding to this novel, asymmetric site affects the HPV16 E2 protein's activities in an orientation-dependent manner.

Binding properties of HPV E2 proteins. The CASTing experiment (Fig. 2) and relative affinity studies (Tables 2 and 4) reveal that the HPV E2 proteins that we examined have a preference for an A/T-rich 4-bp core. There was a >100-fold difference in binding affinities of the HPV16 E2 protein for an A/T-rich core versus a core that contained only 2 G/C bp (e.g., Table 2, Hwt versus Hm#9).

Structural determinants for E2 binding site preferences. The molecular basis for the differences in the DNA binding activities of these proteins is not readily apparent. Cocrystal structures have been published of only the BPV1 and HPV18 E2 proteins bound to canonical DNA binding sites (9, 13) (Fig. 9). If we make two assumptions, that the affinity for the novel hE2BS site and the specific protein-DNA contact points are conserved between the HPV16 and HPV18 E2 proteins, then the cocrystal structures cannot explain the differences in binding properties between the HPV and BPV E2 proteins. The identity of all but one of the amino acids that contact the DNA is conserved among the E2 proteins used in this study. The single difference is a change from Phe343 in the BPV1 E2 protein to a Tyr in the corresponding site in the HPV E2 proteins (Fig. 9A). As both crystal structures indicate that these residues make comparable DNA contacts, this sequence difference cannot explain the differences in the binding properties of the BPV and HPV E2 proteins. In addition, this aromatic residue makes contact with the T at the 3 end of the DNA binding site (-GGT) that is absolutely conserved in all functional E2BSs examined to date.

View larger version (41K): [in this window] [in a new window]   FIG. 9.   BPV1E2ct protein-DNA contacts. (A) The amino acid sequence of the -helix from the BPV1E2 protein that contacts DNA is aligned with the homologous regions from the E2s of HPV51, -16, and -18. Numbers above the sequence corresponds to the BPV1 E2 protein. Amino acid residues conserved between these proteins are highlighted. Arrows above and below the alignment indicate amino acids that make contact with the bases in the DNA major groove of the E2BS and are presumed for HPV51 and HPV16 E2 (9, 13). (B) Schematic of the E2 protein -helix contacting the DNA in the major groove. The amino acids that make contact with the DNA are capitalized and in boldface. The 5 and 3 ends of the DNA strands are indicated at the top and bottom of each strand. (C and D) DNA double helix with protein-contacting base pairs exposed for the BPV1E2 (C) and HPV18E2 (D) proteins. DNA-contacting amino acid residues are listed along the sides. Contact points are indicated by arrows between the letters representing the amino acids and nucleotides. Those contacts that are mediated by a water molecule are indicated by the waterdrop symbol.

Effect of E2 protein affinity for its binding sites. Steger et al. (35) used in vitro transcription assays to demonstrate that the amount of HPV18 E2 can determine whether it acts as an activator or repressor of transcription from the HPV18 early promoter. When template containing the HPV18 LCR (with four E2BSs) is in excess, addition of the HPV18 E2 protein results in stimulation of transcription. By contrast, when the HPV18 E2 protein is in excess, transcription is repressed. Thus, E2 abundance may act as a switch to differentially control early-gene expression at distinct stages of the viral life cycle.

Mechanisms of E2 protein-mediated transcription repression. All three of the E2BSs within pOri16L facilitate E2-mediated repression of gene expression from this promoter (17). The spacing and relative locations of the E2BSs are well conserved, with respect to the TATA box for the early promoter and with respect to the binding sites of the known cellular transcription factors and the viral E1 protein, among the mucosal HPVs. At two E2BSs, E2BS#1 and E2BS#2, the E2 protein can compete for binding with the cellular transcription factors TFIID and Sp1, respectively (5, 37, 38). These two factors are involved in the stimulation of early-gene expression from this promoter in the absence of the E2 protein.