Transcriptional activity among high and low risk human papillomavirus E2 proteins correlates with E2 DNA binding.

The full-length E2 protein, encoded by human papillomaviruses (HPVs), is a sequence-specific transcription factor found in all HPVs, including cancer-causing high risk HPV types 16 and 18 and wart-inducing low risk HPV types 6 and 11. To investigate whether E2 proteins encoded by high risk HPVs may function differentially from E2 proteins encoded by low risk HPVs and animal papillomaviruses, we conducted comparative DNA-binding and transcription studies using electrophoretic mobility shift assays and cell-free transcription systems reconstituted with purified general transcription factors, cofactor, RNA polymerase II, and with E2 proteins encoded by HPV-16, HPV-18, HPV-11, and bovine papillomavirus type 1 (BPV-1). We found that although different types of E2 proteins all exhibited transactivation and repression activities, depending on the sequence context of the E2-binding sites, HPV-16 E2 shows stronger transcription activity and greater DNA-binding affinity than those displayed by the other E2 proteins. Surprisingly, HPV-18 E2 behaves more similarly to BPV-1 E2 than HPV-16 E2 in its functional properties. Our studies thus categorize HPV-18 E2 and BPV-1 E2 in the same protein family, a finding consistent with the available E2 structural data that separate the closely related HPV-16 and HPV-18 E2 proteins but classify together the more divergent BPV-1 and HPV-18 E2 proteins.

These are classified as high risk HPVs. In contrast, HPV types that are rarely found in cancers but are associated with genital warts, such as HPV-6 and HPV-11, are considered low risk HPVs (1)(2)(3). Because the genomic structures of HPVs are highly conserved, it is important to determine the functional differences among individual HPV gene products which lead to etiologically high and low risk phenotypes. Previous comparative studies have primarily focused on HPV-encoded E6-and E7-transforming proteins. These studies found that E6 and E7, from high risk HPVs, lead to cellular transformation much more readily than low risk E6 and E7 proteins (for review, see Refs. 4 and 5). In both high and low risk HPVs, expression of E6 and E7 is transcriptionally regulated via the E6 promoter by many cellular and viral proteins. The full-length viral E2 protein is a sequence-specific transcription factor that functions as an activator or repressor to regulate tightly the E6 promoter through four consensus E2-binding sites (E2-BSs), ACCGN 4 CGGT (6, 7), whose locations within the upstream regulatory region (URR) are highly conserved among genital HPVs. Efficient activation of the E6 promoter requires binding of E2 protein to the promoter-distal E2-BSs in conjunction with binding of cellular factors to the enhancer elements also located within the URR (8 -11). Activation directly mediated through E2 protein may occur via different mechanisms. First, E2 may recruit the general transcription machinery to the promoter through direct interactions with TATA-binding protein (TBP), TBP-associated factors in TFIID, TFIIB, as well as RNA polymerase II (12)(13)(14)(15)(16)(17). Second, E2 interacts with nuclear factors Sp1, Gps2/AMF-1, or TopBP1, which may then serve as coregulators for transcription (18 -20). Third, E2 may potentiate transcription by remodeling DNA structure at the chromatin level (21). Indeed, studies have shown E2 interacts with proteins (CREB-binding protein, p300, and p300/CBP-associated factor) possessing histone acetyltransferase activity (22)(23)(24). It is likely that different types of E2 proteins function differentially in regulating preinitiation complex assembly and show varied cooperativity with cellular enhancer-binding factors that recognize constitutive, inducible, or cell type-specific enhancer elements located in the URR, as well as with general or genespecific transcription cofactors that modulate E2 activity on HPV chromatin.
Although low levels of E2 occupy promoter-distal binding sites, it is postulated that transcriptional repression occurs with high levels of E2 leading to occupancy at promoter-proximal E2-BSs thus displacing cellular factors critical for E6 promoter activity (9,10,(25)(26)(27). We have shown previously that E2 protein can also actively repress transcription by preventing preinitiation complex formation (28). This active repression was demonstrated to be independent of cellular enhancer-binding factors, Sp1, and general cofactors such as TBP-associated factors, TFIIA, mediator and PC4 (28). Conversely, cellular proteins may bind to DNA and hinder access of E2 protein, thus inhibiting E2 function (29 -31). Papillomaviruses may also use its own proteins to modulate E2 activation function. These include various forms of E2 repressor proteins created through alternative splicing to generate gene products that contain the same DNA-binding domain linked to different N-terminal regions (32)(33)(34) or the use of viral E1 protein to abrogate E2 transactivation function through direct proteinprotein interaction (35,36). Because E2 only weakly activates (ϳ2-fold) but strongly represses (up to 100-fold) the native HPV promoter in both transfected cells (9) and in currently available cell-free transcription systems (for review, see Ref. 28), our initial effort was focused on dissecting the mechanisms of transcriptional repression by various E2 proteins on their homologous E6 promoter (28).
Interaction between viral E1 and E2 proteins also regulates papillomavirus DNA replication and viral episome maintenance (37)(38)(39)(40). In addition to binding to viral DNA, E2 ensures proper segregation of viral genomes during cellular replication which is independent of the E2 DNA-binding domain (41)(42)(43). Analogous to abrogation of E2 transactivation function, it has been suggested that E1 also regulates E2 binding to mitotic chromosomes (44). Besides regulating viral transcription, replication, and viral episome maintenance, E2 has been implicated in several cellular processes relevant to carcinogenesis. E2 transfected into HPV DNA-containing cell lines can lead to growth arrest, apoptosis, or abrogation of mitotic checkpoints (45)(46)(47). E2 repression of E6 and E7 oncogene expression leading to reactivation of p53 and pRb tumor suppressor pathways has been the proposed mechanism for growth arrest in HPVpositive cell lines (48,49). Surprisingly, growth arrest and repression of E6 and E7 expression require an intact E2 transactivation domain (50 -52). Expression of E2 protein in a transformed cell line devoid of HPV DNA also displayed growth arrest through an undefined mechanism (45,51). E2-triggered apoptosis can occur through both p53-independent (53) and p53-dependent pathways (54). Similar to E2-induced growth arrest, E2 caused apoptosis in HPV-negative cell lines through unknown pathways (54).
Although many functions have been attributed to papillomaviral E2 proteins, very few studies have been conducted to compare directly the transcription activities among HPV and BPV-1 E2 proteins (11,55,56). Experiments directly comparing high risk HPV-16 and HPV-18 E2 proteins with low risk HPV-6 E2 protein by transient transfection assays found that high risk E2 proteins have higher activation activities than low risk E2 protein, but this activity did not correlate with either higher steady-state levels of protein in vivo or with DNAbinding properties of the two classes of E2 proteins (56). Therefore, it was hypothesized that high risk E2 proteins were intrinsically better transcriptional activators than low risk E2 proteins, which in turn regulate the levels of E6 and E7 oncoproteins in transformed cells and thereby control the processes leading to carcinogenesis. In a more recent study, high risk E2 was found to have higher affinity for transcriptional coactivator CREB-binding protein than that of low risk E2 (23). These studies on E2 transcriptional activity were assayed primarily in transiently transfected cells. Therefore, interpretation of the results may be complicated by interference of other cellular factors, potentially masking the true activity of E2 protein.
To address the differences in transcriptional activities between high and low risk HPV E2 proteins, we employed in vitro transcription systems reconstituted either with individually purified recombinant general transcription factors (TFIIB, TFIIE, and TFIIF), epitope-tagged protein complexes (TFIID, TFIIH, and pol II) and general cofactor PC4, or with a preassembled RNA polymerase II (pol II) holoenzyme complex and TBP (17,28). Using this highly purified in vitro transcription system along with an HPV responsive G-less cassette template containing two E2-BSs (17), high risk HPV-16 E2 protein was found to possess greater transactivation activity than those exhibited by E2 proteins encoded by HPV-18, HPV-11, and BPV-1. For repression assays, we employed an in vitro transcription system reconstituted with pol II holoenzyme and TBP with HPV template containing either homologous or heterologous URR linked to a G-less cassette. HPV-16 E2 (16E2) was determined to be the strongest transcriptional repressor, followed by BPV-1 E2 (BE2), HPV-18 E2 (18E2) and HPV-11 E2 (11E2). The transcriptional activities of these E2 proteins correlated well with their corresponding binding affinities to E2-BSs derived from the promoter-proximal regions of naturally occurring HPV-11, HPV-16, and HPV 18 E6 promoters, as well as BPV-1. Although previous in vivo studies revealed little correlation between E2 transcriptional activities and DNAbinding affinities, our in vitro studies indicate that higher transcriptional activation and repression activities of various E2 proteins correlate directly with their relative DNA-binding affinities for E2-BSs.
In Vitro Transcription-The transcription repression assays were performed in a two-component transcription system reconstituted with TFIID-deficient pol II holoenzyme and TBP using 50 ng of p11URR-GLess, p16URR-GLess, p18URR-GLess, or p7862-70(23M)GLess/I Ϫ G-less cassette template, 50 ng of pML⌬53, 1 ng of TBP, and 3 l (ϳ 90 ng) of pol II holoenzyme, in the absence or presence of 11E2, 16E2, 18E2, or BE2 protein as specified in the figures. Reactions were then conducted and analyzed as described previously (28,60).
The E2 transactivation assay was carried out in a highly purified in vitro transcription system in a 30-l reaction containing 50 ng of pG 5 MLT, 50 ng of p2E2(IR)⌬53, 10 ng of TFIIB, 20 ng of TFIID, 2.5 ng each of TFIIE␣ and TFIIE␤, 2 ng of TFIIF, 22.5 ng of TFIIH, and 5 ng of pol II, in the presence or absence of 150 ng of PC4 and an increasing amount of 11E2, 16E2, 18E2, or BE2 as specified, following the two-step incubation protocol (17). Reactions were then performed and analyzed as described previously (17,60). Transcription signals were quantified by PhosphorImager analysis (Molecular Dynamics).
EMSA was conducted in a 10-l reaction containing ϳ0.5 fmol of 32 P-labeled DNA probe in the presence of 10 mM HEPES-Na (pH 7.9), 5 mM dithiothreitol, 0.2 mM EDTA, 4 mM MgCl 2 , 10% glycerol, 0.1 mg/ml bovine serum albumin, 10 ng/l poly(dI-dC), and increasing amounts of respective purified E2 proteins (adjusted to 2 l in BC300). Reactions were incubated at 30°C for 1 h and then loaded onto a 5% polyacrylamide gel in 0.5ϫ TBE and run at 175 V for ϳ2.5 h at 4°C. The gels were then dried and exposed to x-ray films. The signals were quantified by PhosphorImager analysis.
Calculation of Equilibrium Binding Constants-The intensity of the remaining probe in each lane after EMSA was quantified using Image-Quant (Molecular Dynamics) without () or with (⑀ for each lane) an increasing amount of 11E2, 16E2, 18E2, or BE2 protein. Fractional occupancy () was then determined by Equation 1.
Binding isotherms were obtained by plotting fractional occupancy as a function of protein concentrations and analyzed by nonlinear least squares analyses. The equilibrium binding constant, K d , was determined by analysis of the titration curves against Equation 2, where ϭ fractional occupancy, [protein] ϭ protein concentration, and K d ϭ the equilibrium binding constant (62). Constructions of various G-less cassettes driven by HPV-11 URR (p11URR-GLess), HPV-16 URR (p16URR-GLess), or HPV-18 URR (p18URR-GLess) were as described under "Experimental Procedures." Each template contains the naturally occurring HPV URR and the TATA box of the E6 promoter preceding a 377-nucleotide G-less cassette. Nucleotide numbering for boundaries of E2-BSs is based upon previous studies (9, 28) and HPV nucleotide accession numbers AF125673 (for HPV-16) and X05015 (for HPV-18). proteins in bacteria. All E2 proteins share similar structural features with an N-terminal activation domain linked to a C-terminal DNA-binding/dimerization domain by a flexible hinge region (Fig. 1A). These bacterially expressed E2 proteins were purified to near homogeneity and clearly migrated at positions corresponding to their predicted molecular sizes (Fig.  1B). The amounts of various purified E2 proteins were then normalized by Western blotting with anti-FLAG monoclonal antibody, which recognizes the same epitope sequence introduced at the N terminus of individual E2 proteins (data not shown). For functional studies, we also created three transcription templates, containing enhancer elements and the E6 promoter derived from each type of HPV, by cloning respective URR into a G-less cassette of 377 nucleotides (Fig. 1C). We demonstrated previously that correct initiation from the HPV E6 promoter whose transcription is regulated by different amounts of E2 proteins, similar to the in vivo observations, was recapitulated in our cell-free transcription systems using the G-less cassette templates (28).

Transcriptional Repression Mediated by High and Low Risk
When examined in a cell-free transcription system reconstituted with TBP and human pol II holoenzyme, which provides pol II and general transcription factors TFIIB, TFIIE, TFIIF, and TFIIH (60,61), transcription from these URR-driven G-less templates was inhibited by BE2, 11E2, 16E2, and 18E2, in a dose-dependent manner (Fig. 2). Repression of the E6 promoter could be observed on both homologous and heterologous URR-driven templates, irrespective of the types of E2 proteins used (compare left panels of Fig. 2, A-C). In contrast, the internal control template pML⌬53, which contains the adenovirus major late core promoter linked to a G-less cassette of ϳ280 nucleotides, did not respond to increasing amounts of E2 proteins, indicating a URR-specific repression mediated by E2 proteins. We had illustrated in our previous studies (28) that repression of the E6 promoter from HPV-11 URR-containing G-less cassette templates was mediated mainly through E2 binding to the promoter-proximal #3 and #4 E2-BSs, which are immediately adjacent to the TATA box of the E6 promoter (see Fig. 1C).
These comparative studies also revealed several interesting findings. First, 16E2 is apparently a better repressor compared with 18E2, 11E2, and BE2, because 10 ng of 16E2 almost completely inhibited HPV transcription, whereas 10 ng of the other E2 proteins only led to less than 50% of inhibition (compare Fig. 2, A-C, lanes 3, 8, 13, and 18, and the diagrams shown on the right of each panel). Obviously, ϳ50 ng (i.e. a 5-fold higher amount) of 18E2, 11E2, and BE2 is required to reach a level of repression similar to that achieved by 10 ng of 16E2 (see the diagrams in Fig. 2, A-C). Again, stronger repression of the E6 promoter by 16E2, relative to the other E2 proteins, was observed on both homologous (Fig. 2B) and heterologous (Fig.  2, A and C) HPV templates. Second, animal papillomavirus E2 protein (BE2) could also inhibit HPV E6 promoters as efficiently as HPV E2 proteins (Fig. 2), consistent with many previous studies using the BPV-1 E2 expression plasmid to study HPV promoter regulation by transfection assays (11,55,56). Third, the repression activity of 18E2 is more similar to that exhibited by 11E2 and BE2, but not to its closely related family member (i.e. high risk 16E2). This observation could be best explained by the available C-terminal DNA-binding/ dimerization domain structures of 16E2, 18E2, and BE2 (see "Discussion").
Transcriptional Activation Mediated by High and Low Risk HPV and BPV-1 E2 Proteins-To compare the transactivation activity of E2 proteins encoded by HPV and BPV-1, we employed a highly purified in vitro transcription system reconstituted with only recombinant general transcription factors (TFIIB, TFIIE, and TFIIF), general cofactor PC4, and epitopetagged multiprotein complexes (TFIID, TFIIH, and pol II). It has been shown previously (17) that 11E2 can activate transcription from p2E2(IR)⌬53 DNA template, which contains two copies of the HPV-11 #2 E2-BS linked to a major late core promoter-driven G-less cassette of ϳ280 nucleotides, but not from an internal control template containing five copies of Gal4-binding sites connected to a similar major late core promoter construct with a longer G-less cassette (Fig. 3A, bottom  drawings). When we compared HPV and BPV-1 E2 proteins directly in this reconstituted transcription system, we found that 16E2 was the strongest activator achieving a maximum level of activation at only 5 ng of protein compared with the other E2 proteins (Fig. 3A, compare lanes 5, 13, 23, and 31, and  Fig. 3B). BE2 and 18E2 displayed similar levels of activation (Fig. 3A, lane 7 versus lane 34, 11.6-and 13.7-fold activation, respectively), but BE2 reached maximum activation at a lower dose-dependent amount than 18E2 (Fig. 3A, compare lanes 7 and 34, 50 ng of BE2 versus 150 ng of 18E2). 11E2 attained its maximum level of activation at the same dosage as BE2 (Fig.  3A, lane 15, 50 ng of 11E2), but with only 7.4-fold activation. After the maximum level of activation is achieved for each E2 protein, a general decline in fold activation, representing transcriptional squelching, is observed (Fig. 3B). It is clear from both experiments for activation and repression that 16E2 is the dominant activator and repressor compared with 18E2, 11E2, and BE2 in these comparative studies.
DNA-binding Activities of High and Low Risk HPV and BPV-1 E2 Proteins-Because E2 is a sequence-specific DNAbinding protein, we speculated that the transcription activity of E2 protein may correlate with its DNA-binding activity, thereby accounting for the functional differences in transactivation and repression between 16E2 and the other E2 proteins. To explore this possibility, we performed EMSA using purified E2 proteins and DNA probes derived from promoter-proximal #3 and #4 E2-BSs of respective HPV E6 promoters and a URR region containing #11 and #12 E2-BSs from BPV-1 (63). As shown in Fig. 4A, two protein⅐DNA complexes (C1 and C2) were detected with each E2 protein when a DNA probe containing E2-BSs #3 and #4 of HPV-11 was used for EMSA. The C1 complex represents an E2 dimer binding to one E2-BS first appearing at low concentrations of E2, whereas the C2 complex is formed between two E2 dimers binding to both E2-BSs and observed at increasing concentrations of E2 protein. At higher concentrations of 16E2 and 18E2, we also observed additional protein⅐DNA complexes migrating slower than the C2 complex (Fig. 4A, lanes 30 and 40), presumably generated by oligomerization of E2⅐DNA complexes. An appearance of a band (indicated by an asterisk) below the C1 complex of BE2 (Fig. 4A,  lanes 11-20) is likely formed between minor degradation products of BE2 and the DNA probe. Consistent with BE2 being the largest E2 protein in this assay, BE2⅐DNA complexes migrated slower than HPV E2⅐DNA complexes. Similar patterns of E2⅐DNA complexes were also detected with DNA probes derived from HPV-16 (Fig. 4B) and HPV-18 (Fig. 4C). When BPV-1 probe was used, 16E2 and 18E2 showed migration patterns similar to those observed with HPV probes (Fig. 4D, lanes  21-40). The C1 and C2 complexes formed with BE2 apparently dissociated during electrophoresis, causing two clusters of large smears on the gel (Fig. 4D, lanes 11-20). The affinity of 11E2 for the BPV probe is extremely low because no clear complexes could be detected even with the use of 200 ng of protein (Fig. 4D, lanes 1-10), which accidentally led to the formation of large protein⅐DNA aggregates unable to enter the gel. It is apparent from these EMSAs that the C1 complex was formed at much lower concentrations of 16E2 and 18E2 than 11E2 and BE2 (compare lanes 21-40 with lanes 1-20 of Fig. 4, A-C), indicating that high risk E2 has higher affinity for E2-BSs than 11E2 and BE2.
To compare and quantify accurately the binding affinities among various E2 proteins for each DNA probe, we calculated the equilibrium binding constant (K d ) estimated from three or more gel shift assays by measuring the reduction of free probe ( Table I). Calculations of the equilibrium binding constant clearly indicate that 16E2 has the highest affinity to both homologous and heterologous HPV E6 promoter-proximal E2-BSs, followed in order by 18E2, BE2, and 11E2 (see Table I

. Transcriptional repression mediated by high and low risk HPV and BPV-1 E2 proteins.
A, transcriptional repression of the HPV-11 E6 promoter mediated by different E2 proteins. In vitro transcription was performed in a two-component system comprising pol II holoenzyme and TBP using p11URR-GLess template, which contains the HPV-11 URR spanning nucleotides 7072-7933/1-70, and the internal control template pML⌬53, in the absence (Ϫ) or presence of increasing amounts (2, 10, 50, and 200 ng) of BE2 (lanes 1-5, respectively), 11E2 (lanes 6 -10), 16E2 (lanes 11-15), or 18E2 (lanes 16 -20). Protein concentrations of different E2 proteins were normalized by Western blotting with anti-FLAG M2 monoclonal antibody (Sigma). The line graph shows the relative transcription intensity of HPV-11 template with different amounts of E2 protein. Relative intensity is defined as the signal intensity, quantified by PhosphorImager analysis, from p11URR-GLess relative to that from the same DNA template performed in the absence of E2 (i.e. the first lane of each reaction set). B, transcriptional repression of the HPV-16 E6 promoter mediated by different E2 proteins. In vitro transcription was performed as described in A, except that a transcriptional template p16URR-GLess containing the HPV-16 E6 promoter spanning HPV-16 nucleotides 7007-7904/1-72 was used. The line graph shows relative transcriptional intensity for the HPV-16 E6 promoter at increasing concentrations for each E2 protein titration. C, transcriptional repression of the HPV-18 E6 promoter mediated by different E2 proteins. In vitro transcription reactions were performed as described in A, except that a transcriptional template p18URR-GLess containing the HPV-18 E6 promoter spanning HPV-18 nucleotides 6929 -7857/1-81 was used. The line graph shows relative transcriptional intensity for each HPV-18 E6 promoter signal derived from the titration of each E2 protein.
probe, and similar comparisons for HPV-16 and HPV-18 probes). In the case of BPV-1 DNA-derived probe containing two E2-BSs, homologous BE2 bound with the highest affinity (K d ϭ 1.079) followed by 16E2 (K d ϭ 1.567), 18E2 (K d ϭ 8.422), and 11E2 (K d ϭ 25.360), consistent with results published previously (11). This indicates that there may be some species specificity between the DNA-binding activity of BPV and HPV E2 proteins, which allows BE2 to bind better to BPV-1 DNA and HPV E2 to bind better to HPV DNA. This phenomenon likely reflects inherent variation between HPV and BPV E2-BS sequences with HPV preferring A/T spacer nucleotides (ACCGN 4 CGGT, see Fig. 1C), whereas BPV contains more G/C spacer nucleotides within their respective E2-BSs. Differences among E2 recognition of DNA (see "Discussion") and decreased stability of BE2 protein may also contribute to the variation in results from EMSA performed with HPV DNA compared with BPV-1 DNA (7, 64).
Correlation of Transcriptional Repression Activity with DNAbinding Activity between 18E2 and 11E2-There is a clear distinction between repression activity of 16E2 and both 18E2 and 11E2 proteins (see Fig. 2). This activity correlated well with the binding affinities among HPV E2 proteins to DNA probes derived from the HPV E6 promoter-proximal region, in that 16E2 had the highest affinity followed by 18E2 and then 11E2 (Tables I and II). If there is indeed a direct correlation between binding affinities and E2 repressor activity, 18E2 should be a better transcriptional repressor than 11E2 because it has a higher affinity for promoter-proximal E2-BSs than 11E2 (Table I and Fig. 4). However, this was not clearly observed (Fig. 2). If the mechanism of transcriptional repression at the E6 promoter is caused by disruption of preinitiation complex formation through promoter-proximal DNA binding (9,10,25,28), then one possibility that may account for this lack of difference is that there were not enough titration points to measure the differences in activity between 18E2 and 11E2 adequately. The other possibility is that promoter-distal E2-BSs may sequester E2 protein away from promoterproximal E2-BSs.
To address these issues, we performed an in vitro transcription assay using the two-component transcription system with an HPV-11 URR-containing G-less cassette template where E2-BSs #2 and #3 were mutated (28) while carefully titrating both 11E2 and 18E2 (Fig. 6A). At low concentrations of 11E2 and 18E2 (e.g. 1 ng), the activity from the HPV-11 E6 promoter is almost reduced by half with 18E2, whereas 11E2 showed negligible reduction in E6 promoter activity (Fig. 6A, compare lane 15 with lane 5). Plotting relative intensity determined by PhosphorImager analysis revealed that there is a significant difference between the repression activities of 18E2 and 11E2  at low amounts of E2 proteins, which indeed correlates with their relative DNA-binding affinities for E2-BSs (Fig. 6B).

DISCUSSION
In this report, we described the first comparative analysis of transcription and DNA-binding activities of the full-length high risk and low risk HPV and BPV-1 E2 proteins in well defined cell-free environments devoid of other cellular proteins whose presence in the cell may complicate the studies of intrinsic activities of E2 proteins. We uncovered a direct correlation of transcriptional activity with the DNA-binding activity in which high risk 16E2 is the strongest transcriptional activator/ repressor, compared with 18E2, 11E2, and BE2, and shows the highest affinity for both homologous and heterologous HPV E6 promoter-proximal E2-BSs. Our studies thus provide unequivocal comparison of functional properties of E2 proteins encoded by high risk and low risk HPVs as well as BPV-1, which play an important role in virus-induced pathogenesis in both humans and animals.
High Risk E2 Proteins Display More Transcriptional Activity than Low Risk HPV and BPV-1 E2 Proteins-The full-length E2 protein can function as a transcriptional repressor to inhibit the expression of virus-encoded gene products mainly through promoter-proximal E2-BSs (25,27,28,(67)(68)(69). We found that, similar to high versus low risk HPV E6 and E7 oncoproteins, there are also distinct differences in the repression activities between high and low risk HPV E2 proteins. Using an in vitro transcription system reconstituted with human pol II holoenzyme and TBP, we defined a strict order of transcriptional repression activity with 16E2 acting as the strongest repressor, followed by 18E2 and BE2, and lastly 11E2 (Figs. 2 and 6). The potency of transcriptional repression activity exhibited by different E2 proteins strongly correlates with their DNA-binding affinities for E2-BSs ( Fig. 4 and Table I). 16E2 bound with the highest affinity for the E6 promoter-proximal DNA fragments derived from HPV-11, HPV-16, and HPV-18 and was followed closely, again, by 18E2, BE2, and finally 11E2. The strong DNA-binding activity observed with 16E2 and 18E2 even allows these high risk E2 proteins to bind to DNA probes containing half-site mutations in both promoter-proximal #3 and #4 E2-BSs (Fig. 5), whose binding could not occur with low risk 11E2 protein.
Although comparison of transactivation activity among some high and low risk HPV E2 proteins has been described in previous studies using either transient transfection or in vitro transcription performed with HeLa nuclear extract (11,55,56), the presence of numerous unidentified cellular proteins in their assays often complicates the interpretation. Therefore we have developed a highly purified E2-dependent in vitro transactivation system (17) comprised only of essential recombinant general transcription factors (TFIIB, TFIIE, and TFIIF), recombinant general cofactor PC4, and epitope-tagged multiprotein complexes (TFIID, TFIIH, and pol II). Using this reconstituted cell-free transcription system, we found that 16E2 is the most responsive and strongest transactivator compared with 18E2, BE2, and 11E2 (Fig. 3), suggesting that 16E2 may be more effective at recruiting components of the general transcription machinery to the promoter region. This interesting possibility remains to be investigated further.
Significance for the Order of Transcriptional Activity between High and Low Risk HPV E2 Proteins-In our studies, we found a strict order of both transcriptional activation and repression activity among various E2 proteins. High risk 16E2 displays the highest activation and repression activity, followed by high risk 18E2 and then low risk 11E2. In a recent epidemiological FIG. 6. Transcriptional repression of the HPV-11 E6 promoter mediated by 11E2 and 18E2. A, transcriptional repression of the HPV-11 E6 promoter by different amounts of E2 proteins. In vitro transcription was performed in a twocomponent system as described in Fig.  2A, except that p7862-70(23M)GLess/I Ϫ (28), a transcriptional template containing HPV-11 URR spanning nucleotides 7862-7933/1-70 with mutations at #2 and #3 E2-BSs, was used. B, line graph of relative intensity for 11E2 and 18E2 proteins. Relative intensity is defined as the signal intensity, quantified by Phospho-rImager, from p7862-70(23M)GLess/I Ϫ relative to that from the same DNA template performed in the absence of E2 (i.e. the first lane of each reaction set). study, 99.7% of 1,000 cases of invasive cervical cancer were HPV DNA-positive with HPV-16 DNA (53%) being the most prevalent followed by HPV-18 DNA (15%) (2,3). Previous studies have also shown that high risk HPV E6 and E7 are more active in inducing cellular transformation than low risk E6 and E7 (for review, see Refs. 4 and 5). A common hallmark of cervical cancer is viral integration, which often disrupts the E2 open reading frame, leading to the loss of E6 promoter regulation and thus increased the expression of HPV E6 and E7 oncoproteins. We speculate that HPV-16 developed a more transcriptionally active E2 protein to regulate its highly active E6 and E7 oncoproteins tightly and govern the viral life cycle more stringently than low risk HPVs, which express less potent E6 and E7 proteins. Once regulation of the E6 promoter by high risk E2 proteins is lost because of viral integration, cancer may then develop.
Differences in DNA Target Site Discrimination by Papillomavirus E2 Proteins-Although previous studies of binding site affinities between HPV and BPV-1 E2 proteins revealed little variation (11,56), we found a distinct order of E2 binding to both wild-type and mutated HPV E6 promoter-proximal E2BSs among 16E2, 18E2, BE2, and 11E2 (Figs. 4 and 5 and Tables I and II). One possible explanation for the variation between our results and previously published studies might be attributed to the DNA probes from which the equilibrium binding constants were calculated. Although we used a much larger DNA probe containing two E2-BSs derived from the natural HPV URRs, others have used much smaller DNA probes containing only one or two E2-BSs with minimal flanking regions (11,56). Smaller DNA probes would naturally have more inherent DNA flexibility than larger DNA probes, which would favor BE2 binding to more flexible DNA, whereas HPV E2 has a predisposition to bind to pre-bent DNA (7). Variations in Mg 2ϩ salt concentrations between DNA-binding experiments may also play a role in the differences in calculating equilibrium binding constants because there is evidence that Mg 2ϩ enhances HPV E2 recognition of specific E2-BSs (70). Furthermore, the fournucleotide spacers within our E2-BSs also vary from previous DNA probes (11), consistent with previous studies identifying that HPV E2 favors spacer nucleotides that are A/T-rich, specifically AATT, whereas BPV E2 does not display an ability to discriminate between spacer nucleotides (7,71). This may explain our results in which 16E2 favors binding to DNA probes derived from HPV URR, whereas BE2 binds better to BPVderived E2-BSs that are less A/T-rich ( Fig. 4 and Table I), a conclusion in agreement with previous studies (11).
It should not be surprising that E2 proteins vary in binding affinities, considering that there are distinct differences in the quaternary structures of the DNA-binding domains between 16E2 and 18E2 (7). These differences are critical because they dictate the spatial arrangement of side chains presented to the major grooves for DNA sequence recognition (64). Furthermore, structural and biochemical studies of E2 proteins from BPV-1 and HPV-16 have suggested they use different mechanisms to discriminate their DNA targets (71) even though they are highly homologous proteins. Recent E2⅐DNA-binding studies have also reported that there are significant differences in the binding affinities between the DNA-binding domains of HPV-16 and BPV-1 (66). In fact, BE2 shares more structural similarities to 18E2 than 16E2, whereas 16E2 is structurally more similar to HPV-31 E2 protein (7). This may explain why the transcriptional and DNA-binding activity of 18E2 is more equivalent to BE2 than to its closely related human homolog 16E2. Although BE2 and 18E2 are structurally similar, and they both induce the same global DNA deformation upon binding, their mechanisms of deformation are different. BE2 has a positive charge localized near the C terminus of the recognition helix, whereas 18E2 has a positive charge in the center of its DNA interaction surface, and 16E2 is even less charged all along the DNA interaction surface (7). This would lead to differences in E2⅐DNA contacts, with 16E2 making fewer contacts with the DNA phosphate backbone than either BE2 or 18E2 (7), which would be consistent with increased DNA-binding specificity. Although the DNA-binding domain sequences of high risk 16E2 and 18E2 share 54% identity and 77% homology, they still differ in DNA binding. Thus it is likely that the DNA-binding domain of low risk 11E2 may vary further in structure from both 16E2 and 18E2, thus contributing to variations in DNA-binding affinity among high and low risk HPV E2 proteins. This remains to be elucidated once the structure of 11E2 becomes available.