DNA binding and bending by the human papillomavirus type 16 E2 protein. Recognition of an extended binding site.

The human papillomavirus (HPV) 16 E2 protein (hE2) binds to four sites present upstream of the P97 promoter and regulates transcription of the viral E6 and E7 oncogenes. We have determined the relative binding constants for the interaction of the full-length hE2 protein with these sites. Our results show that hE2 binds tightly to site 4, less tightly to sites 1 and 2, and weakly to site 3. Similar results have previously been obtained using a C-terminal fragment of the hE2 protein suggesting that the C-terminal domain is the sole determinant of DNA binding affinity and specificity. Using circular permutation assays we show that binding of the hE2 protein induces the formation of a significant DNA bend and that the hE2-induced DNA bend angle is the same at both tight and weak hE2-binding sites. An alignment of the four hE2-binding sites from the HPV 16 genome suggests that this protein recognizes an extended binding site when compared with the bovine papillomavirus E2 protein. Here we show that the hE2 protein binds tightly to sites containing an A:T or a G:C base pair at position 7 of its binding site but weakly to sites containing either C:G or T:A at this position. Using site-directed mutagenesis we demonstrate that an arginine at position 304 of the hE2 protein is responsible for the recognition of specific base pairs at this position.

The recognition of specific DNA sequences by transcription factors is often the first step in the regulation of gene expression. An understanding of how these proteins recognize their target sequences, and the effect that this has on DNA conformation, is central to the question of how genes are controlled. We are studying the human papillomavirus E2 protein, a sequence-specific DNA-binding protein involved in the regulation of viral gene expression and DNA replication. Papillomaviruses infect epithelial cells and induce the formation of benign hyperproliferative lesions or warts. Over 70 distinct types of human papillomavirus (HPV) 1 have been described. Some of these viral types produce lesions that have the potential to undergo malignant transformation. HPV 16 and HPV 18, for example, are thought to play a primary role in the development of cervical cancer (for a review, see Ref. 1). The products of the viral E6 and E7 genes form complexes with the cellular tumor suppressor proteins p53 and Rb, respectively. These interactions bring about a change in cell growth rate and promote cell immortalization (reviewed in Ref. 2). In HPV 16, transcription of the E6 and E7 genes is under the control of a single promoter (P97) that lies immediately upstream of the E6 gene (3). The activity of the P97 promoter is regulated by a variety of cellular transcription factors and by the viral E2 protein (4 -6).
Much early work concentrated on the bovine papillomavirus (BPV) E2 protein. The BPV E2 protein (bE2) binds as a dimer to 12 inverted repeats (consensus sequence 5Ј-ACCGN 4 CGGT-3Ј) present upstream of the BPV early genes and activates transcription (7,8). Binding of bE2 protein to DNA is cooperative (9,10), and once bound, bE2 dimers can associate further to form DNA loops (11). The bE2 protein contains an N-terminal transcription activation domain and a C-terminal DNA-binding/dimerization domain separated by a flexible hinge (12). The structure of the bE2 C-terminal domain bound to a specific DNA sequence has been determined by x-ray crystallography (13). The two subunits of the bE2 dimer each form half of a ␤-barrel over which the DNA is bent, allowing the interaction of two ␣-helices with the exposed edges of the base pairs in two successive major grooves of the DNA.
The HPV 16 E2 protein (hE2) and the BPV E2 protein share a high degree of sequence similarity at the protein level. The amino acids involved in the recognition of specific DNA sequences, for example, are conserved between bE2 and hE2; the single exception being a bE2 phenylalanine (Phe-343) that is replaced by a tyrosine in the hE2 protein (Tyr-303). The hE2 protein binds to four sites present upstream of the HPV 16 P97 promoter (14). The precise function of the hE2 protein in the regulation of this promoter is the subject of some controversy. Over-expression of the hE2 protein has been shown to repress the activity of some P97 promoter fragments linked to reporter genes (15,16). However, in similar experiments over-expression of hE2 was found to activate transcription from other promoter derivatives (4,5,17,18). This confusion over the role of the hE2 protein is probably brought about by differences in the contribution of each of the four hE2-binding sites toward P97 promoter activity. The organization of the P97 promoter is shown diagrammatically in Fig. 1. Two hE2-binding sites (separated by only 3 base pairs) are located immediately upstream of the P97 TATA box and are flanked on the 5Ј side by a binding site for the cellular transcription factor Sp1. The binding of hE2 (or bE2) to these sites has been shown to block the binding of Sp1 and TBP (the TATA box binding factor) and to repress the promoter activity of DNA fragments carrying these sequences * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. (16,19,20). The two remaining hE2-binding sites are located around 150 and 550 base pairs upstream of the P97 transcription start point. The binding of hE2 to sites in this "enhancer configuration" has been shown to activate transcription from artificial promoter constructs (21,17).
To understand the role of the hE2 protein in the regulation of the P97 promoter it is important to know the relative affinities of the hE2-binding sites present within this promoter and to characterize fully the DNA binding activity of this protein. To this end we have purified the full-length hE2 as a GST fusion protein and determined the relative binding constants of the four hE2-binding sites present in the HPV 16 genome. Using gel retardation assays we show that the DNA bend angle induced by the binding of hE2 is the same at both tight-and weak-binding sites. We also demonstrate that unlike bE2, the hE2 protein binds preferentially to an extended recognition sequence that contains a purine at the Ϫ7 position and we use site-directed mutagenesis to show that this specific interaction is mediated by arginine 304 in the hE2 DNA-binding domain.

EXPERIMENTAL PROCEDURES
Plasmids Used in This Study-The HPV 16 E2 gene was amplified by PCR from HPV 16 DNA using the oligonucleotide primers: 5Ј-GACTG-GATCCATGGAGACTCTTTGCCAACG-3Ј and 5Ј-GACTGAATTCCAT-CATATAGACATAAATCCAG-3Ј. These primers placed BamHI and EcoRI restriction sites at the 5Ј and 3Ј ends of the hE2 coding sequence, respectively. The PCR product was cloned between the BamHI and EcoRI restriction sites of the prokaryotic expression vector pGEX-2T (Pharmacia), creating an in-frame GST-hE2 fusion (pGEX-hE2). The R304A mutation of hE2 was made by PCR. The primers 5Ј-GCTTTTT-AAATGCATATCTTAAAC-3Ј and 5Ј-GGCTGGCAAGCCACGTTTGG-3Ј bind to hE2 and pGEX sequences, respectively, and were used to amplify the 5Ј end of the hE2 gene. In a separate reaction, the primers 5Ј-GTTTAAGATATGCATTTAAAAAGC-3Ј and 5Ј-TCAGTCACGG-TACGGCCG-3Ј, which also bind to hE2 and pGEX sequences, respectively, were used to amplify the 3Ј end of the hE2 gene. The underlined bases mismatch the hE2 sequence and introduce the R304A mutation. The 5Ј end and 3Ј end PCR products were purified from agarose gels, mixed, and re-amplified using the pGEX-specific primers. This generated a full-length hE2 gene containing the R304A mutation. This PCR product was digested with BamHI and EcoRI, and cloned into pGEX-2T as described above. The resulting pGEX-hE2 and pGEX-hE2 R304A plasmids were sequenced using a panel of E2-specific sequencing primers to check for the occurrence of any point mutations generated by the Taq polymerase.
Protein Purification-Escherichia coli XL-1 blue cells containing either pGEX-hE2 or pGEX-hE2 R304A were grown at 37°C in Terrific broth supplemented with 0.01% ampicillin to an OD 600 of 0.5. Fusion protein expression was then induced by the addition of 0.15 mM isopropyl-1-thio-␤-D-galactopyranoside and the cells were grown at 28°C overnight. Bacteria were harvested by centrifugation, resuspended in phosphate-buffered saline, and sonicated at 4°C. The cell lysate was cleared by centrifugation (15,000 ϫ g for 30 min at 4°C) then incubated with 0.1% DNase I for 30 min at 4°C. After re-centrifugation, the supernatant was loaded onto a glutathione-Sepharose 4B column preequilibrated in 50 mM Tris, pH 8.0. The column was washed with 10 column volumes of 50 mM Tris, pH 8.0, 500 mM NaCl, and then equilibrated in buffer 1 (50 mM triethanolamine buffer, 50 mM KCl, 20 mM MgCl 2 , pH 7.5). The column was then washed with 20 column volumes of 5 mM ATP in buffer 1 to remove co-purifying GroEL (Cpn 60). After re-equilibration in 50 mM Tris, pH 8.0, the fusion proteins were eluted using 50 mM reduced glutathione in 50 mM Tris, pH 8.0. The eluted proteins were dialyzed against 50 mM Tris, pH 8.0, overnight at 4°C, then stored in 10% glycerol, 200 mM NaCl at Ϫ70°C. Circular dichroism spectra of the wild type GST-E2 and R304A GST-E2 proteins (0.2 mg/ml in 25 mM phosphate buffer, pH 7.9) were obtained using a JY CD6 CD spectrometer with 2-mm path length cells. Spectra were collected at 1-nm increments, using a 20-s integration time.
Circular Permutation Assays-The oligonucleotides shown below carry hE2-binding sites 3 and 4, respectively, and were cloned into pBend3 (26) between the XbaI and SalI restriction sites.
Recombinant clones were digested with the restriction enzymes Bg-lII, XhoI, SmaI, KpnI, and BamHI to generate a series of fragments in which the position of the E2-binding site varies evenly from one end of the fragment to the other (shown in Fig. 4a). These fragments were dephosphorylated using calf alkaline phosphatase then 5Ј-end labeled with [␥-32 P]ATP using T4 polynucleotide kinase. Unincorporated label was removed using Sephadex G-50 columns (Stratagene) and the labeled oligonucleotides (10,000 cpm) were incubated with purified GST-hE2 in the binding buffer described above. Free and bound DNA was resolved and visualized as described in the previous section. Each experiment was repeated at least three times and the DNA bend angle (␣) was calculated using the empirical relationship, where m equals the mobility of the complex with hE2 bound in the middle of the fragment, and e equals the mobility of the complex with hE2 bound near the end of the fragment (22).

Purification of a GST-hE2 Fusion Protein-
The full-length hE2 protein is poorly expressed in bacteria, shows low solubility, and is unstable (data not shown). To avoid these problems we cloned the HPV 16 E2 gene into the GST expression vector pGEX-2T and used a modification of the standard GST fusion purification procedure (23) to obtain pure GST-hE2. The use of a GST-hE2 fusion improves expression, solubility, and stability of hE2 and also allows rapid purification. However, GST-hE2 was found to co-purify with a bacterial protein of around 60 kDa (Fig. 2, lane 2). N-terminal peptide sequencing showed that the co-purifying protein was GroEL; also known as Cpn 60 (data not shown). The GroEL chaperonin is overexpressed by bacteria on cell stress and frequently co-purifies with recombi- nant fusion proteins. The binding of ATP to GroEL has been shown to induce a conformational change from a tight-binding form of the protein to a weak-binding form (24). To remove the co-purifying GroEL we included an ATP washing step in the purification procedure (25). GST-hE2/GroEL, bound to a glutathione-Sepharose column, was washed with 5 mM ATP in column buffer. GST-hE2 free of GroEL could then be eluted from column using reduced glutathione (Fig. 2, lane 3).
Binding of hE2 to the HPV 16 Sites-We used gel retardation assays to study the binding of hE2 to the four hE2-binding sites found in the HPV 16 genome. Increasing amounts of purified GST-hE2 were added to labeled oligonucleotides carrying hE2 sites 1-4 (shown in Fig. 3a). Free and bound oligonucleotides were then separated by polyacrylamide gel electrophoresis and visualized by autoradiography. Binding to hE2 site 1 was repeated on each gel as an internal control. Fig. 3b shows the results of a typical experiment; in this case, the binding of GST-hE2 to hE2 site 1 (Fig. 3b, lanes 1-7) was compared with the binding GST-hE2 to hE2 site 2 (Fig. 3b, lanes 8 -14). The binding of GST-hE2 results in the formation of two retarded bands (indicated by arrowheads in Fig. 3) probably as a consequence of the dimerization of GST. The quantity of free and bound DNA (both complexes) was determined using a Phospho-rImager and these values are shown graphically in Fig. 3c. The resulting weak binding curves ([DNA] Ͻ K d ) were used to calculate the relative dissociation constant and the change in ⌬G (or ⌬⌬G) for each of the hE2-binding sites (Table I). These data indicate that, compared with hE2 site 1, hE2 binds more tightly to site 4, marginally more tightly to site 2, and more weakly to hE2 site 3.
DNA Bending by the hE2 Protein-In the x-ray crystallography-derived structure of the bE2 DNA-binding domain-DNA complex, the DNA is bent around the protein through an angle of around 50° (13). As the HPV 16 P97 promoter contains four hE2-binding sites, any DNA bending caused by the binding of hE2 would be expected to have consequences for promoter architecture. We used gel retardation assays to investigate the DNA bending properties of the full-length hE2 protein at both tight and weak hE2-binding sites. hE2 sites 3 and 4, the weakest and tightest of the HPV 16 hE2 sites, respectively, were cloned into the DNA-bending vector pBend3 (26). Restriction digestion of the resulting pBend3-hE2(3) and pBend3-hE2(4) plasmids generated a set of circularly permuted fragments in which the position of the E2-binding site is varied (see Fig. 4a). Fig. 4, b and c, show the results of gel retardation assays performed using this set of fragments carrying E2 site 3 and E2 site 4, respectively. In each case the retarded bands show different relative mobility depending on the position of the E2-binding site within the DNA fragment. These differences in mobility indicate that the binding of hE2 induces a significant bend in the DNA. Each gel retardation assay was repeated at least three times and the angle of the induced bend calculated as described under "Experimental Procedures." The binding of  1-7) or hE2 site 2 (lanes 8 -14) were incubated with increasing amounts of purified hE2 (40, 80, 160, 320, 640, and 1050 nM, respectively). Free and bound DNA (hE2 c ) were separated on a 6% polyacrylamide gel and visualized by autoradiography. c, the graph shows the binding of hE2 to sites 1-4 assayed as described above. At each protein concentration the amount of free and bound DNA was determined using a PhosphorImager and the percentage of oligonucleotide bound ([bound oligonucleotide/(bound oligonucleotide ϩ free oligonucleotide)] ϫ 100) is plotted against the amount of hE2 added. Although site 3 shows poor binding and does not reach saturation, each of the binding curves extrapolates to the same end point (within experimental error). The experiments were repeated a minimum of four times and weak binding curves were fitted to the data to determine the dissociation constants shown in Table I. Binding to site 1 was repeated on each gel as an internal control. q, site 1; E, site 2; ç, site 3; ⌬, site 4. hE2 to site 3 (the weak-binding site) and site 4 (the tightbinding site) induced DNA bends of 61.4°Ϯ 3.0°and 61.2°Ϯ 1.8°, respectively. Thus, hE2 induces very similar, if not identical, DNA bends at both weak-and tight-binding sites. It is interesting to note that the experimentally determined hE2induced DNA bend angle is in close agreement with the bE2induced DNA bend angle (50°) measured from the crystal structure of the bE2-DNA complex (13).
hE2 Recognizes an Extended Binding Site-An alignment of the four hE2-binding sites present within the HPV 16 P97 promoter produces a consensus sequence that differs from that obtained when the bE2-binding sites are aligned (Fig. 5a).
While the consensus bE2 site shows no base pair preference at the Ϫ7 position, or the symmetrically related ϩ7 position, the HPV 16 hE2 sites all contain an adenine at Ϫ7 and a thymine at ϩ7. To investigate the significance of this difference we assayed the binding of hE2 to a series of otherwise identical oligonucleotides containing A:T, T:A, G:C, C:G, or A:U at the Ϫ7/ϩ7 position (shown in Fig. 5b). The binding of hE2 to each of these sites was assayed exactly as described in Fig. 3 and is shown graphically in Fig. 5c. The effect of different bases at the Ϫ7/ϩ7 position on the relative dissociation constant is shown in Table II. These data show that changing the Ϫ7/ϩ7 position from A:T to either T:A or C:G significantly reduces the binding of hE2 (8-fold and 4-fold reductions in binding, respectively). In contrast, changing the Ϫ7/ϩ7 position from A:T to G:C has little or no effect. Given that the structure of the bE2-DNA complex derived by x-ray crystallography indicates that the bE2 protein makes specific contacts with base pairs in the major groove, these data suggest that hE2 contacts the N-7 atoms of adenine and guanine when either A:T or G:C are at the Ϫ7/ϩ7 position of the hE2-binding site. Changing the Ϫ7/ϩ7 position from A:T to A:U resulted in only a slight reduction in the binding of hE2. This would also seem to indicate that the hE2 protein makes specific contact with the purine at Ϫ7/ϩ7. Taken together these data suggest that the hE2 protein forms a specific contact with a purine at the Ϫ7/ϩ7 position and that this contact stabilizes the binding of hE2 to the HPV 16 E2 sites.
Arg-304 of hE2 Contacts the Ϫ7/ϩ7 Position-Although the structure of the bE2-DNA complex has been determined by x-ray crystallography, no structural information is available on the hE2-DNA interaction. The bE2 protein makes no direct contacts with the G:C base pair at the Ϫ7/ϩ7 position of the bE2-binding site. However, Arg-344 in bE2 appears to form a water-mediated hydrogen bond with the guanine at this position as well as side chain hydrogen bonds to the phosphate groups of the Ϫ7 and Ϫ8 base pairs (13). Since the amino acids involved in sequence-recognition appear to be conserved between bE2 and hE2, we reasoned that Arg-304 of hE2 (the equivalent of Arg-344 in bE2) might be responsible for the recognition of a purine at the Ϫ7/ϩ7 position of the hE2binding site.
Using site-directed mutagenesis we changed Arg-304 of hE2 to alanine. hE2 R304A was expressed in bacteria as a GST fusion protein and purified exactly as described for the wild-  4. DNA bending by hE2. a, the series of circularly permuted DNA fragments used to assay the DNA bending activity of hE2. Fragments 1-5 were produced by digestion of pBend3-hE2(3) and pBend3-hE2(4) with BglII, XhoI, SmaI, KpnI, and BamHI, B, X, S, K, and Ba, respectively. Each fragment contains the hE2-binding site (filled rectangle) at a different position relative to the ends of the molecule. b, the graph shows the relative mobility of the protein-DNA complexes formed when hE2 binds to the DNA fragments described above carrying hE2 site 3 (the weakest hE2-binding site). c, the graph shows relative mobility of the protein-DNA complexes formed when hE2 binds DNA fragments carrying hE2 site 4 (the tightest hE2-binding site). type protein. Circular dichroism (CD) was used to test whether the presence of the R304A mutation altered the folding or dimerization of hE2. The CD spectra for the wild-type and mutant proteins (shown in Fig. 6a) are very similar, suggesting that this mutation has little or no effect on these properties. The R304A protein did show reduced affinity for DNA compared with the wild-type protein (data not shown) and consequently, binding to the position 7 variants was assayed in the absence of competitor DNA. Fig. 6b shows the results of a gel retardation assay in which increasing amounts of hE2 R304A were added to labeled oligonucleotides carrying E2-binding sites with either A:T or T:A at position Ϫ7/ϩ7. As can be seen from the Figure, hE2 R304A binds equally well to both sites indicating that this mutant has lost the ability to discriminate between A:T and T:A at this position. These data imply that arginine 304 of the hE2 protein makes a specific contact with the N-7 of purine bases at the Ϫ7/ϩ7 position of the hE2binding site.

DISCUSSION
The DNA-binding domain of the bE2 protein is almost identical in structure to the core domain of the Epstein-Barr virus origin binding protein, EBNA1 (27). However, unlike the bE2 C-terminal domain which forms multiple contacts with the base pairs in the E2-binding site, the core domain of the EBNA1 protein makes no direct interactions with the DNA (28). The sequence-specific contacts between EBNA1 and its FIG. 6. Binding of hE2 R304A to oligonucleotides 7A:T and 7T:A. a, circular dichroism was used to test whether the presence of the R304A mutation affected the folding or dimerization of hE2. b, the graph shows the results of a gel retardation assay in which increasing amounts of hE2 R304A (33, 66, 100, 300, and 600 nM, respectively) were added to labeled oligonucleotides carrying either A:T or T:A at position 7 of the hE2-binding site. Free and bound DNA was separated on a 6% polyacrylamide gel and quantified using a PhosphorImager.  binding site appear to be formed by two "flanking domains" that lie outside the E2 homology region. Each flanking domain consists of an ␣-helix and an extended polypeptide chain that form major and minor groove contacts with the DNA (28). Sanders and Maitland (29) have assayed the binding of the isolated hE2 C terminus to the four hE2-binding sites present within the HPV 16 P97 promoter (29). As amino acids outside this C-terminal region could form important contacts with the DNA we have used the full-length protein to determine relative dissociation constants for the interaction of hE2 with these sites. The values obtained for the full-length protein are similar to those which have previously been obtained using the isolated C-terminal domain (29). This suggests that, unlike the EBNA1 core domain, the hE2 C-terminal domain is the sole determinant of DNA binding affinity and specificity.
The data obtained using both the full-length hE2 protein, and the isolated C-terminal domain, indicate that hE2 binds most tightly to E2 site 4, then sites 1 and 2, and most weakly to site 3. The weak binding to site 3 is unsurprising given that unlike sites 1, 2, and 4, this site is not a perfect match to the "consensus" binding site (see Fig. 5a). The tight binding of hE2 to site 4 suggests that at low E2 concentrations this site would be occupied first. E2 site 4 is located around 550 base pairs upstream of the P97 promoter in an "enhancer configuration." As the binding of hE2 to sites in this configuration has been shown to activate transcription from downstream promoters (21,17), the binding of hE2 to E2 site 4 would be expected to activate transcription from the P97 promoter. At higher E2 concentrations E2 sites 1 and 2 would become occupied, possibly displacing Sp1 and the TATA box-binding protein (20). The binding of hE2 to these promoter proximal sites would thus be expected to result in transcriptional repression. The P97 promoter directs transcription of the viral early genes, including E2. Our binding data suggest that the hE2 protein might positively and negatively regulate transcription of its own gene at low and high E2 concentrations, respectively. This hypothesis is consistent with the results of previous studies which have shown that although hE2 activates transcription from the P97 promoter, increased expression of the hE2 protein results in transcriptional repression (18). Similar concentration-dependent effects of HPV 8 E2 have been observed at the HPV 8 late gene promoter (30).
The binding of many transcription factors has been shown to induce DNA bending and, in some cases, these protein-induced DNA bends have important consequences for the regulation of gene expression. The DNA bend induced by E2F, for example, is important for the transcriptional activity of the E2F1 promoter (31). The DNA-binding domain of the bE2 protein has previously been shown to induce DNA bending in an enhancer fragment containing three bE2-binding sites (32). Here we have shown that binding of the full-length hE2 protein introduces a significant bend in the hE2-binding site (around 60°). As the P97 promoter contains four binding sites for hE2, the binding of this protein probably has dramatic effects on the architecture of this DNA. This conformational change might be of importance in the regulation of P97 promoter activity and, as hE2 interacts with the HPV origin recognition protein E1 (33), the regulation of viral DNA replication. Interestingly, the degree of DNA bending induced by the binding of the hE2 protein is the same at both tight and weak hE2-binding sites. In this respect hE2 appears to be similar to the 434 repressor protein which has also been shown to induce identical DNA bends at both tight-and weak-binding sites (34). In contrast, the degree of DNA bending induced by the cAMP receptor protein (CRP or CAP) has been shown depend on the strength of the protein-DNA interaction (35).
An alignment of the four E2-binding sites present in the HPV 16 P97 promoter produces a consensus sequence which differs from that obtained when the E2-binding sites from the BPV genome are aligned. All four E2-binding sites from HPV 16 contain an A:T base pair at the Ϫ7 position and a T:A base pair at the symmetrically related ϩ7 position (shown in Fig. 5a). We have shown that mutations at these positions can significantly reduce the binding of hE2. These results confirm and extend the previous observation that a change from A:T to C:G at this position weakens the binding of the hE2 C-terminal domain (29). In contrast, the consensus bE2-binding site shows no preference for A:T at position 7 and a change from A:T to C:G at this position has been shown to result in a slight increase in the binding of bE2 (8).
Having shown that the hE2 protein discriminates between base pairs at position 7 of its binding site, we set out to determine the amino acid responsible for this effect. Inspection of the bE2-DNA complex showed that arginine 344 (equivalent to arginine 304 in hE2) makes a water-mediated hydrogen bond to the N-7 and O-6 of the guanine base at the Ϫ7 position of the bE2-binding site used in the co-crystal (13). We reasoned that arginine 304 in hE2 might make a contact (either direct or water-mediated) with the N-7 of adenine or guanine present at the position 7 of the hE2-binding site. To test this hypothesis we mutated arginine 304 of hE2 to alanine (R304A) and assayed the binding of the mutant protein to sites containing either an A:T or a T:A at position 7. Unlike the wild-type protein, which showed an 8-fold preference for A:T over T:A at this position, the E2 R304A protein bound equally to both sites, albeit with reduced affinity. This loss of specificity suggests that arginine 304 of hE2 makes a specific contact with position 7 of the hE2-binding site. Given the flexibility of surface arginine residues it would seem possible that bE2 R344 could make direct contact with a purine at position 7 of the bE2-binding site. However, the lack of sequence-specificity at position 7 of the bE2 site suggests that this contact does not occur. The recognition of the base pair at position 7 of the hE2-binding site might not be solely attributable to the Arg-304 side chain but might also involve other contacts that are disrupted when this amino acid is changed to alanine. The lack of a preferred base pair at position 7 of the bE2 consensus site would also seem to indicate that the water-mediated contact between bE2 Arg-344 and the bE2 site makes little contribution to the specificity of the bE2-DNA interaction, however, the interactions with backbone phosphates made by this residue might make an important contribution toward the affinity of bE2 for DNA.
There is a growing body of evidence which suggests that the bE2 and hE2 proteins have significantly different biological properties. For example, the hE2 protein has been shown to activate transcription from the HPV 16 P97 promoter whereas, under exactly the same conditions, the bE2 protein has been shown to repress P97 promoter activity (18). Our results highlight a further difference between these proteins; in comparison to bE2, the hE2 protein recognizes an extended binding site. These functional differences between bE2 and hE2 suggest that caution must be exercised when extrapolating from one system to the other.