Characterization of recombinant HPV6 and 11 E1 helicases: effect of ATP on the interaction of E1 with E2 and mapping of a minimal helicase domain.

To better characterize the enzymatic activities required for human papillomavirus (HPV) DNA replication, the E1 helicases of HPV types 6 and 11 were produced using a baculovirus expression system. The purified wild type proteins and a version of HPV11 E1 lacking the N-terminal 71 amino acids, which was better expressed, were found to be hexameric over a wide range of concentrations and to have helicase and ATPase activities with relatively low values for K(m)(ATP) of 12 microm for HPV6 E1 and 6 microm for HPV11 E1. Interestingly, the value of K(m)(ATP) was increased 7-fold in the presence of the E2 transactivation domain. In turn, ATP was found to perturb the co-operative binding of E1 and E2 to DNA. Mutant and truncated versions of in vitro translated E1 were used to identify a minimal ATPase domain composed of the C-terminal 297 amino acids. This fragment was expressed, purified, and found to be fully active in ATP hydrolysis, single-stranded DNA binding, and unwinding assays, despite lacking the minimal origin-binding domain.


From the Department of Biological Sciences, Boehringer Ingelheim (Canada) Ltd., Laval, Quebec H7S 2G5, Canada
To better characterize the enzymatic activities required for human papillomavirus (HPV) DNA replication, the E1 helicases of HPV types 6 and 11 were produced using a baculovirus expression system. The purified wild type proteins and a version of HPV11 E1 lacking the N-terminal 71 amino acids, which was better expressed, were found to be hexameric over a wide range of concentrations and to have helicase and ATPase activities with relatively low values for K m (ATP) of 12 M for HPV6 E1 and 6 M for HPV11 E1. Interestingly, the value of K m (ATP) was increased 7-fold in the presence of the E2 transactivation domain. In turn, ATP was found to perturb the co-operative binding of E1 and E2 to DNA. Mutant and truncated versions of in vitro translated E1 were used to identify a minimal ATPase domain composed of the C-terminal 297 amino acids. This fragment was expressed, purified, and found to be fully active in ATP hydrolysis, single-stranded DNA binding, and unwinding assays, despite lacking the minimal origin-binding domain.
The papillomaviruses (PVs) 1 are small, nonenveloped DNA viruses that infect and replicate in the cutaneous or mucosal epithelia of humans and other mammals. There are over 100 types of human papillomavirus (HPV), which cause conditions ranging from plantar warts (HPV1) and genital warts (HPV6 and -11) to cervical cancer (HPV16, -18, and -31). HPV6 and -11 are also responsible for laryngeal papillomatosis, a rare but very serious infection of the respiratory tract (1). Antiviral agents capable of specifically inhibiting PV replication could play an important role in the treatment of these diseases, but none exist at this time.
Despite their host and tissue specificity, all of the PV types share a common genomic organization. The closed, circular genome of ϳ8000 base pairs codes for only 10 proteins: eight early proteins termed E1-E8 and two late proteins, L1 and L2, that make up the viral coat (2). E1 and E2 are the only viral proteins required for HPV DNA replication (3,4). E2 is a sequence-specific DNA-binding protein that serves to regulate both transcription and DNA replication. E1, a DNA helicase, is the only PV protein that possesses enzymatic activity (5) and is also the most highly conserved of the PV proteins. For these reasons, E1 has been considered as the most attractive molecular target for the development of antiviral agents (6).
During the initiation of PV DNA replication, E2 serves as a specificity factor to enhance binding of E1 monomers to the origin, which by themselves bind to double-stranded DNA with little sequence specificity (4,(7)(8)(9)(10)(11). E2 dimers bind with high specificity to DNA sequences within the origin and recruit to it E1 monomers, through direct protein-protein interactions (12,13). Upon binding to DNA, E1 monomers assemble into hexamers (14,15), but the E1-E2 interaction is not maintained (16,17). Thus, the role of E2 is to catalyze the assembly of hexameric E1 complexes specifically at the origin. ATP hydrolysis is also required for E1 hexamers to distort the origin and encircle single strands of DNA (18). Using in vitro translated HPV11 proteins, we have previously shown that ATP binding stimulates the E2-dependent association of E1 with the origin (19). ATP binding does not affect the E2 binding step or the E1-E2 interaction but rather a subsequent process, either oligomerization and binding of E1 to DNA or the stability of the assembled E1-origin DNA complex (19,20). ATP hydrolysis is not required in this case, since the nonhydrolyzable analog ATP␥S and ADP have similar effects. For BPV, it has been demonstrated that ATP hydrolysis is required for displacement of E2 from the origin during assembly of E1 multimeric complexes (17). Either concurrent with or after formation of E1 oligomers, the host polymerase ␣ primase (21)(22)(23)(24)(25) and possibly also replication protein A (22,26) bind to E1, and a replication complex is formed together with additional host factors.
We and others have identified subdomains of E1 as well as conserved amino acids that are required for a number of its DNA replication activities. We have recently shown that amino acids 166 -649 of E1 efficiently support DNA replication in a cell-free system and thus must encode for all necessary binding and enzymatic activities required in vitro (25). In vivo, additional sequences spanning amino acids 82-128 (HPV11 E1) are required; these contain an extended nuclear localization sequence (27,28) as well as a cyclin-binding motif and cyclin-dependent kinase phosphorylation sites (29 -31). A minimal DNA binding/origin recognition sequence was mapped to approximately amino acids 140 -300 for BPV E1 (32-35) (amino acids 185-345 for HPV11 E1), although the entire C terminus of the protein (amino acids 186 -649) is required to observe stable binding of HPV11 E1 to the origin (20,36). An HPV11 E1 truncation consisting of the C-terminal amino acids 353-649 retains the ability to bind to E2 or the p70 subunit of polymerase ␣ primase and to oligomerize into hexamers (19,20,25). Within this C-terminal fragment of E1 are four conserved re-gions termed A, B, C, and D (see Fig. 1), which are highly similar to the functionally related T antigens of SV40 and polyomaviruses (37). The conserved region A lies in a minimal E1 oligomerization sequence located within amino acids 353-416 (see Fig. 1). Three motifs characteristic of superfamily 3 of NTP-binding proteins are located between amino acids 478 and 525 of HPV11 E1 (see Fig. 1) (38). Two of these motifs (motifs a and b) correspond to the Walker A and B boxes involved in binding the substrate nucleotide triphosphate-magnesium complex (39), while the exact function of the third motif (motif c) remains unclear. Mutations of conserved residues in all three motifs have been shown to significantly decrease the ATPase activity of the superfamily 3 helicase Rep68, encoded by the adeno-associated virus (40). Mutation of a highly conserved lysine in motif a (Lys 484 in HPV11 E1) has been shown to abrogate the ATPase activity of BPV E1 (41) and HPV11 E1 (42), as also observed for many other ATPases (43).
Most studies on the enzymatic activities of purified E1 have been carried out using the BPV enzyme, purified from baculovirus-infected insect cells (5,41,44,45) or Escherichia coli (15,46). Although the HPV E1 proteins are 40 -50% identical in primary sequence to BPV E1 and presumably have very similar structures and mechanisms of action, it has proven very difficult to isolate sufficient quantities of purified HPV enzymes to characterize their enzymatic activities. Very low levels of expression, compounded by proteolysis and contamination with other proteins, have commonly been observed. Of the few reports on the isolation of HPV E1 (42,(47)(48)(49), in only one case was sufficient material obtained for a basic characterization of helicase activity and biophysical properties (42). Kuo et al. (50) have reported the isolation of epitope-tagged HPV11 E1 that is functional for in vitro DNA replication assays, but the enzymatic activities of their E1 preparations have not been described.
In this report, we present our characterization of the enzymatic activities of HPV11 and -6 E1 helicases. We have made extensive use of an N-terminal truncated form of E1, composed of amino acids 72-649, which we have shown previously is as active as wild type E1 in supporting HPV DNA replication in vitro. We report here that this truncated E1 is expressed in significantly greater yield than full-length E1 and has improved biophysical properties. We show that, in contrast to previous reports (41,42,45,49), our purified E1 preparations have a relatively low value of K m (ATP), well below the physiological concentration of ATP. We show that high concentrations of the E2 transactivation domain raise the apparent value of K m (ATP) and also demonstrate that the binding of ATP-Mg to E1 weakens its interaction with E2. The perturbation of E1-E2 binding by physiological concentrations of ATP suggests a mechanism by which some of the energy gained by tight binding of ATP is used to destabilize the E1-E2 complex, which forms transiently during the initiation of viral DNA replication. We also show that E1-(353-649), which encodes motifs for enzymatic activity but lacks the previously identified minimal origin-DNA binding region, is fully active in ATPase, helicase, and ssDNA binding assays. Thus, while the N-terminal half of E1 is necessary for replication from the HPV origin, all sequences necessary for proper folding, oligomerization, and enzymatic catalysis are within the C terminus.

Construction of Recombinant Baculoviruses-
The complete E1 open reading frame from HPV11 was amplified from plasmid pCR3-E1 (19). The forward primer was 5Ј-CGC GGA TCC AGG ATG CAT CAC CAT CAC CAT CAC GCG GAC GAT TCA CGT ACA GAA AAT GAG-3Ј, and the reverse was GG CTG AAT TCA TAA AGT TCT AAC AAC T-3Ј, inserting 6 histidines immediately after the initial methionine. Similarly, E1-(72-649) and E1-(353-649) were amplified from plasmid pTM1-E1 (19) with the same reverse primer but with the following forward primers, respectively: 5Ј-CGC GGA TCC AGG ATG CAT CAC CAT CAC CAT CAC GCG GAT GCT CAT TAT GCG ACT GTG CAG GAC-3Ј and 5Ј-CGC GGA TCC AGG ATG CAT CAC CAT CAC CAT CAC GAC AGT CAA TTT AAA TTA ACT GAA ATG GTG C-3Ј. To express the corresponding catalytically inactive mutant E1 proteins, similar PCR amplifications were performed on the appropriate mutant E1 template (19). The PCR products were then cut with BamHI and EcoRI, subcloned between the BamHI and EcoRI sites of pFASTBAC1 (Life Technologies, Inc.), and transposed into the baculovirus genome according to the manufacturer's instructions. E1-containing recombinant baculovirus genomes were transfected into SF9 insect cells, and baculovirus-containing supernatants were amplified for large scale protein production. The gene for HPV6 E1 was amplified by PCR from a clinical condyloma sample and subcloned into pCR3.1. PCR primers used were 5Ј-CAA CGA TGG CGG ACG ATT CAG GTA CAG-3Ј (forward, nt 827-853) and 5Ј-TGC TTC GGA CAC CT-3Ј (reverse, nt 2923-2910). The baculovirus expression construct was made by the same procedure as for HPV11 E1, using the same forward and reverse primers for PCR. The DNA sequence matched that reported for HPV6a FIG. 1. Domain structure of the HPV11 E1 helicase. The 649-amino acid-long HPV11 E1 helicase is diagrammed as a rectangle with black boxes labeled A, B, C, and D indicating regions of E1 that are conserved in the large T antigens of SV40 and polyomaviruses (37). The region of HPV11 E1 that corresponds to the minimal DNA-binding domain mapped in BPV E1- (32)(33)(34)(35) is also indicated as a black box. The locations of conserved motifs a, b, and c that are common to members of superfamily 3 NTP-binding proteins (38) are indicated by arrows. Indicated below E1 are truncated constructs of the protein that we have found in previous studies to be sufficient for catalyzing DNA replication in a cell-free system, for stable binding to the origin, for interaction with the transactivation domain of E2 (19) or the p70 subunit of polymerase ␣ primase (25), or for oligomerization of the protein (20). The position of a minimal E1-E1 interaction domain mapped between amino acids 353 and 416 using the yeast two-hybrid system (20) is indicated by a hatched box.
E1 (GenBank TM accession number L41216) except for eight silent mutations (nt 1297 C to A, nt 1495 T to A, nt 1540 C to T, nt 2170 C to T, nt 2363 T to C, nt 2396 T to C, nt 2530 G to A, and nt 2548 T to C) and two mutations encoding conservative changes (nt 1670 G to T, changing Val 280 to Leu, and nt 1741 C to A, changing Asp 303 to Glu). 2 The amino acid sequence for HPV6a E1 is 88% identical to that for HPV11 E1.
E1 Expression-SF21 cells were maintained in SFM-900 medium (Life Technologies) at cell densities between 1 and 2 ϫ 10 6 /ml. Infections were carried out for 48 -72 h using a multiplicity of infection of 5. Cells were harvested and frozen rapidly in liquid nitrogen before being stored at Ϫ80°C. All subsequent steps were performed at 0 -4°C. For nuclear extraction, frozen cell pellets were thawed and resuspended in one volume of cell lysis buffer (20 mM Tris, pH 8, 5 mM ␤-mercaptoethanol, 5 mM KCl, 1 mM MgCl 2 , 1 mM Pefabloc (Pentapharm Ltd.), 1 g/ml pepstatin, 1 g/ml leupeptin, and 1 g/ml antipain) for 15 min and then broken with a Dounce homogenizer (20 strokes for 40 ml, pestle B) followed by centrifugation at 2500 ϫ g for 20 min. Supernatant (cytosol) was discarded, and nuclei were resuspended to 1.4 volumes with extraction buffer A (20 mM Tris, pH 8, 5 mM ␤-mercaptoethanol, 2 mM Pefabloc, 2 g/ml pepstatin, 2 g/ml leupeptin, and 2 g/ml antipain), followed by the addition of an equal volume of extraction buffer B (20 mM Tris, pH 8, 5 mM ␤-mercaptoethanol, and 0.02% Triton X-100 for full-length E1 or the same buffer supplemented with NaCl to a final concentration of 450 mM for truncated proteins). The nuclei were incubated with rocking for 30 min before ultracentrifugation at 148,000 ϫ g for 45 min. Glycerol was added to the supernatant to a final concentration of 10%, and the extract was frozen rapidly on dry ice and stored at Ϫ80°C.
His-E1 Purification-All steps were performed at 4°C. A 5-ml Hi-Trap chelating column was charged with NiSO 4 according to the manufacturer's instructions (Amersham Pharmacia Biotech) and then washed with equilibration buffer (20 mM Tris, pH 8, 5 mM ␤-mercaptoethanol, 500 mM NaCl, 10 mM imidazole, and 10% glycerol). Nuclear extracts from 5 liters of culture were thawed rapidly, the NaCl concentration was adjusted to 500 mM for full-length E1 that was initially extracted in low salt, and the material was loaded onto the Hi-Trap column. The column was washed with 50 ml of equilibration buffer followed by 10 volumes of washing buffer (equilibration buffer with 50 mM imidazole). His-tagged E1 was eluted in 1-ml fractions with equilibration buffer containing 180 mM imidazole and dialyzed in 20 mM MES, pH 7.0, 500 mM NaCl, 1 mM DTT, 0.05 mM EDTA, and 10% glycerol before being frozen on dry ice and stored at Ϫ80°C. E1 truncations were eluted with 250 mM (E1-(72-649)) or 350 mM (E1-(353-649)) imidazole and were not dialyzed prior to freezing, since dilution of the protein into assay buffers was found to give insignificant residual imidazole concentrations. E1 preparations were tested for the presence of DNA using the fluorescent dye PicoGreen (Molecular Probes, Inc., Eugene, OR), according to published procedures (51). The concentrations of purified proteins were determined using the Coomassie Plus reagent (Pierce), with bovine serum albumin serving as a standard.
Analytical Ultracentrifugation-A velocity run for HPV11 E1-(72-649) was carried out using the XL-A analytical ultracentrifuge (Beckman Coulter), in an An-60 Ti rotor at 40,000 rpm and 11°C. E1-(72-649) was exchanged into a buffer composed of 20 mM Tris, 500 mM NaCl, 10% glycerol, 1 mM DTT, and 0.1 mM EDTA, pH 8.0, using a NAP-5 column (Amersham Pharmacia Biotech). The protein was diluted to ϳ40 g/ml (600 nM) and analyzed using a 1.2-cm double sector charcoalfilled Epon centerpiece and quartz windows pretreated with Sigmacoat silconizing agent (Sigma). Data were acquired at 230 nm and at 0.005-cm intervals, with five replicate readings taken at each point. Scans were spaced as closely as the absorbance optics would allow, approximately every 2 min. E1-(72-649) was found to be stable over the time required to acquire velocity data, although it began to lose activity after several hours at 4 -11°C. Buffer density was determined to be 1.05659 Ϯ 0.00007 g/ml at 22°C using a pyknometer and corrected to a value of 1.05851 g/ml at 11°C (53). The partial specific volume for E1-(72-649) (0.7285 cm 3 /g) and the solvent viscosity (1.810 centipoise) at 11°C were calculated using the program Sednterp (53). Data were analyzed with the programs DCDTϩ 1.13 (54,55) and SVEDBERG 6.38 (56), which gave similar results.
E2-dependent E1-DNA Binding-E2-dependent binding of E1 to the HPV11 origin was measured using a modification of the procedure previously used for in vitro translated proteins (19). Binding reactions (80 l) contained ϳ0.4 ng of probe DNA, 2 g of salmon sperm DNA, and the indicated amounts of E1 and E2. Radiolabeled probe DNA was obtained by PCR amplification of the HPV11 origin from plasmid pN9. HPV11 E2 was expressed in SF21 insect cells infected with baculovirus Ac11E2 (obtained from R. Rose, University of Rochester) and purified by DNA affinity chromatography in a procedure similar to that of Seo et al. (57). After 1-h binding reactions, detection was performed by a modification of the previously described procedure in which protein A scintillation proximity assay beads (SPA beads) were used in place of protein A-Sepharose. Anti-E1-coated SPA beads were added to binding reactions, and after an additional 1 h, plates were spun briefly to pellet beads and counted using a TopCount NXT microplate scintillation counter (Packard Instruments). Blanks were subtracted before calculating relative activities. Synthesis of Helicase Substrates-M13mp18 ssDNA was obtained by infection of E. coli according to standard procedures (58). Substrate oligonucleotides were synthesized (trityl-off) on an Applied Biosystems 394 DNA/RNA synthesizer and purified by gel electrophoresis (58). To synthesize the substrate used for most of the work in this report, 125 pmol of the purified oligonucleotide 5Ј-TTC CCA GTC ACG TTG T-3Ј was mixed with an equimolar amount of M13mp18 in 500 l of TEN buffer (10 mM Tris, 80 mM NaCl, 1 mM EDTA), heated in a water bath to 95°C, and then allowed to cool slowly overnight. dCTP was then added (to give 230 M), together with [␣-33 P]dATP (125 pmol or 250 Ci) and unlabeled dATP (400 pmol), to give a final volume of 700 l, and the oligonucleotide sequence was extended with the Klenow fragment at 30°C to give the final sequence 5Ј-TTC CCA GTC ACG ACG TTG TAA AAC-3Ј (radiolabeled residues in boldface). The labeled partial duplex was purified using a freshly packed 50-ml column of Sepharose 4B equilibrated in TEN buffer. Pooled partial duplex fractions (ϳ10 ml) were concentrated to ϳ2 ml using Centriprep 50 filters (Millipore Corp.). The typical yield was 80 pmol of purified product at a specific activity of 10 7 cpm/pmol.
To synthesize substrates for testing the effect of 3Ј-or 5Ј-unannealed tails, 125 pmol of each oligonucleotide were 5Ј-labeled using 30 units of polynucleotide kinase and 125 pmol of [␥-33 P]ATP. The reaction was stopped by the addition of EDTA to 70 mM, and labeled oligonucleotide was separated from remaining [␥-33 P]ATP using a Chromaspin 10 column (CLONTECH). Radiolabeled oligonucleotides were then annealed to M13mp18 and purified by Sepharose 4B chromatography as above. Oligonucleotides used were as follows: GTA AAA CGA CGG CCA GTG CCA AGC (nontailed), GTA AAA CGA CGG CCA GTG CCA AGC AAT GTA AGA TAG CAT CTC CGT (3Ј-tailed substrate), and AAG AAG AAG AAG AAG AAG GTA AAA CGA CGG CCA GTG CCA AGC (5Јtailed substrate). The sequence complementary to M13 is underlined.
Helicase Assays-Unless stated otherwise, reactions were run for 90 min in an incubator maintained at 37°C and 70% humidity in Microfluor 2 96-well plates (Dynex) and in a buffer consisting of 20 mM HEPES, pH 7.5, 1 mM DTT, 0.05 mM EDTA, and 0.01% IGEPAL-CA630 (Sigma; equivalent to Nonidet-P40). M13 partial duplex substrates were used at 0.05 nM, with 0.3 mM ATP and 1 mM MgOAc. Reaction products were analyzed using either gel electrophoresis or streptavidincoated SPA beads; the two methods gave equivalent results. For detection by gel electrophoresis, an aliquot of the reaction was mixed with one-half volume gel-loading buffer (60 mM EDTA, pH 8, 0.6% SDS, 15% Ficol 400, and 0.09% of the nonintercalating dye Orange G (Sigma)). Samples (10 l) were loaded onto 20% acrylamide gels and electrophoresed at 100 V until the orange band had migrated two-thirds of the way down the gel. Gels were then dried onto DE-81 paper (Whatman) and exposed to a storage phosphor screen. Unwinding was quantified using a STORM 860 imaging system (Molecular Dynamics, Inc., Sunnyvale, CA). Results were calculated as the percentage of substrate unwound, where this value was measured relative to the intensity of the oligonucleotide band from a reaction blank sample boiled for 15 min prior to electrophoresis. Blank reactions, containing no enzyme, were run in parallel with all helicase reactions, and the resulting blank value, typically 2-3% of the boiled sample, was subtracted from enzyme reactions. For detection by SPA beads (Amersham Pharmacia Biotech), 45 l of helicase reaction was mixed with 20 l of stop solution (100 mM HEPES, pH 7.5, 300 mM NaCl, 20 mM EDTA, 1% SDS) containing a biotinylated oligonucleotide complementary to the substrate oligonucleotide at 20 nM. After 30 min, 20 l of streptavidin SPA beads suspended at 3.3 mg/ml in 20 mM HEPES with 0.05% NaN 3 were added. After brief mixing and a further 30-min incubation, plates were centrifuged at 1300 ϫ g for 5 min and counted using a TopCount NXT (Packard Instruments). SPA-detected reactions were quantified in the 2 L. Bourgon and L. Doyon, personal communication. same way as above, using a boiled reaction mixture as a 100% reference and subtracting the value for reaction blanks.
ATPase Assays with Purified Proteins-Reactions were performed using the same buffer as used for helicase reactions in either Dynex Microfluor 2 plates or Packard Optiplates. Unless stated otherwise, reactions were typically performed using 0.3 mM ATP and 1.0 mM MgOAc or 1 M ATP and 0.5 mM MgOAc. Reactions at higher ATP concentrations were carried out at 37°C, and phosphate product was quantified using an ammonium molybdate/malachite green colorimetric assay (59). The same procedure was used for assays with the seven other nucleotide triphosphates and with manganese and magnesium chloride. We found the practical limit of sensitivity for this method to be ϳ25 M phosphate product. Reactions at low ATP concentrations were supplemented with [␥-33 P]ATP (Amersham Pharmacia Biotech or PerkinElmer Life Sciences) to at least 1 nCi/45-l reaction, and unless stated otherwise they were carried out at room temperature. For reactions run to determine the value of K m (ATP), ATP and radiolabeled ATP were maintained at a constant ratio, and the magnesium concentration was adjusted to give a constant excess, usually 500 M, since the ATP-magnesium complex is the true substrate for the reaction, and the K d for ATP chelation of magnesium is ϳ50 M (60). Radiolabeled phosphate product was detected either by thin layer chromatography or by binding a phosphate-molybdate complex to streptavidin SPA beads. For TLC detection, reactions were stopped by the addition of one-half volume of cold EDTA (500 mM), and 1-l aliquots were spotted onto polyethyleneimine-cellulose TLC plates (Sigma). Spots were allowed to dry and samples were eluted in a running buffer of 1 M formic acid plus 1 M LiCl. Plates were then dried and exposed on a storage phosphor screen, and the resulting signal was quantified using a STORM 860 imaging system (Molecular Dynamics). Blank reactions containing no E1 were run in parallel, and this background signal, typically 3-5%, was subtracted. The same TLC procedure was used for reactions using [␣-33 P]GTP, UTP, or dATP, since the nucleoside diphosphates were found to have R f values similar to that for orthophosphate. The SPAbased detection procedure will be described in detail elsewhere, 3 but briefly this involves quenching reactions with a mixture of polyvinyl toluene SPA beads and ammonium molybdate in HCl. The resulting phosphomolybdate complex binds to SPA beads by hydrophobic forces, and binding is enhanced by the subsequent addition of cesium chloride and citric acid. Signals were detected by scintillation counting using a TopCount NXT (Packard Instruments), and values for no-enzyme blank reactions run in parallel, typically 2-4%, were subtracted. To correlate the detected signal with the absolute amount of phosphate product produced at each time point, in each assay, some reactions were detected with both the TLC and SPA procedures. Plots of cpm (SPA) versus nM phosphate (TLC) were linear up to at least 100 M phosphate, and the slope of this line was used as a conversion factor for all reactions in the assay.
Data Analysis-IC 50 inhibition curves were fit to a logistic using SAS, and kinetic data were analyzed using the appropriate equations in the program GraFit 3.0 (Erithicus Software Ltd.). For kinetic experiments, reaction aliquots were taken (in triplicate) at multiple time points between 20 and 120 min. Observed rates were calculated using data at 20% conversion or lower, to maintain initial rate conditions. ATPase Assay Using in Vitro Translated E1-Truncated E1 proteins were synthesized and labeled in vitro with [ 35 S]methionine by coupled transcription/translation using the TNT reticulocyte lysate system (Promega), as described previously (19). 150-l transcription/translation reactions were programmed with deletion constructs of the E1 open reading frame cloned into plasmid pTM1, which have been described previously (19,20). In vitro synthesized E1 proteins were then immunoprecipitated using either the K72 polyclonal antibody or, when indicated, an anti-FLAG M2 monoclonal antibody (Sigma) coupled to protein A-or protein G-Sepharose beads (Amersham Pharmacia Biotech), respectively. Immunoprecipitations were carried out for 90 min in binding buffer (20 mM Tris-HCl, pH 7.6, 100 mM NaCl, 3 mM MgCl 2 , 1 mM EDTA, 1 mM DTT). The immunoprecipitated proteins bound to beads were washed three times with 1 ml of ATPase wash buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 1 mM DTT) followed by two times with 1 ml of ATPase buffer (50 mM HEPES, pH 7.5, 1 mM MgOAc, 0.05 mM EDTA, 1 mM DTT). ATPase reactions were performed by resuspending the bead-bound E1 immunocomplexes in 30 l of ATPase buffer supplemented with 1 mM ATP followed by incubation at 37°C for 2 h. Inorganic phosphate generated during the reaction was quantified using an ammonium molybdate/malachite green colorimetric microassay (59). ATPase activities above background levels were normalized to the amount of 35 S-labeled E1 protein present in each immunoprecipitate, as determined by SDS-PAGE analysis and autoradiography. Quantification of 35 S-labeled E1 was performed with a STORM 860 imaging system (Molecular Dynamics) with a correction for the number of methionine residues present in each E1 truncation.
DNA-binding Assay-75-l binding reactions were performed in triplicate in 96-well plates (Dynex), with the indicated concentrations of purified HPV11 E1 protein, either E1-(353-649) or E1-(72-649), in binding buffer (20 mM Tris pH 7.6, 100 mM NaCl, 5 mM ATP, 3 mM MgCl 2 , 1 mM EDTA, 1 mM DTT, 0.02% Nonidet P-40). Binding reactions were started by adding a mixture of probe and competitor DNA. The probe consisted either of 1.25 ng of a 63-mer oligonucleotide, endlabeled by phosphorylation with [␥-33 P]ATP (ssDNA probe), or of a 370-base pair 33 P-labeled PCR fragment (dsDNA probe) identical to the probe used in the E2-dependent E1-DNA-binding assay described above. Binding reactions were performed at 23°C for 1 h. E1-DNA complexes were immunoprecipitated using the anti-E1 K72 polyclonal antibody coupled to protein A-Sepharose beads, washed three times with 200 l of binding buffer, and then filtered on a 96-well 0.45-m Durapore membrane filter plate (Millipore). The amount of radiolabeled DNA probe present in the retained E1-DNA complexes was quantified with a TopCount NXT (Packard Instruments).

Expression and Purification of HPV11 and HPV6 E1-We
initially attempted to purify E1 after overexpression in E. coli. However, we consistently found HPV E1 expressed in bacteria to be largely insoluble and the soluble protein to be partially degraded. Preparations of E1 proteins with amino acid substitutions of the essential Lys 484 in the Walker A phosphatebinding motif retained helicase and ATPase activities, demonstrating that bacterially produced E1 copurified with contaminating helicases and ATPases (data not shown). As a result of these experiences, we decided to express HPV E1 in insect cells using a baculovirus expression system. Recombinant baculoviruses were constructed to express full-length HPV11 and HPV6 E1 as well as a slightly shorter version of HPV11 E1, encompassing amino acids 72-649. A similar truncation of HPV33 E1 was reported to be expressed at higher levels than the full-length protein in E. coli (61). Furthermore, we have shown previously that this truncated E1, when made by in vitro translation, retains the ability to support cell-free DNA replication (25). These three E1 proteins were expressed as fusions with an N-terminal 6-histidine tag to facilitate their rapid purification by metal affinity chromatography and hence minimize losses of activity over the course of the procedure. Finally, in order to rule out the possibility that the enzymatic activities associated with our E1 preparations were due to contaminating proteins, we also expressed several catalytically inactive HPV11 E1 proteins containing single amino acid substitutions at lysine 484. We showed previously, using in vitro translated proteins, that HPV11 E1 in which Lys 484 is substituted by alanine, histidine, isoleucine, or arginine retains the ability to form a specific complex with E2 at the HPV origin (19). These substitutions were made in the context of the fulllength HPV11 E1 protein, and the alanine and arginine changes were also made in E1- .
In agreement with previous work, we found that during expression in baculovirus-infected SF21 cells E1 accumulated in the nuclei, with only a small proportion in the cytosolic fraction. E1 was extracted from nuclei with a buffer containing 500 mM NaCl, but in further experiments, we found that fulllength HPV6 and HPV11 E1 could be extracted from nuclei using a buffer with no added salt. This unusual low salt extraction was as efficient as a more standard high salt procedure at extracting soluble wild type E1, with the advantage that significantly fewer contaminating proteins were present in the low salt extracts, so that the final product obtained after chromatography was more pure (data not shown). Due to the higher expression level obtained for E1-(72-649), very pure E1 could be isolated in relatively good yield after high salt extraction. Following extraction, all three E1 proteins were purified in a single nickel affinity chromatography step. The E1 isolated by this procedure was Ͼ80% pure as estimated by SDS-PAGE (Fig. 2, A-D). We routinely obtained ϳ200 g of full-length E1 or 800 g of E1-(72-649) per liter of insect cell culture.
The purified HPV11 E1 mutant proteins were as active as WT E1 in binding cooperatively with E2 at the origin, indicating that they were folded properly. However, they had no detectable helicase or ATPase activity (Fig. 2E). Collectively, these results demonstrate that E1 is responsible for the activity detected in WT E1 preparations. Similar results were obtained for mutant HPV11 E1-(72-649) proteins (data not shown). HPV6 E1 preparations were not validated in the same way; but given the complete lack of activity of HPV11 mutant proteins purified by the same procedure and the similarity of HPV6 E1 to HPV11 E1 in terms of sequence, behavior during purification, and enzymatic properties as reported below, it is very unlikely that contaminating proteins are responsible for the activity of HPV6 preparations.
It has been reported that HPV11 E1 preparations obtained from nuclear extracts contain significant concentrations of contaminating DNA (42). We used a sensitive assay with the fluorescent intercalating dye PicoGreen (51) to determine that our purified E1 contains only trace amounts of DNA, ϳ1 ng/g of protein (data not shown).
Quaternary Structure of Purified HPV11 E1-(72-649)-HPV11 E1 expressed in insect cells with a Glu-Glu epitope tag or in native form and purified by immunoaffinity has been shown previously to consist of a mixture of oligomers, with monomers or hexamers predominating, depending on the E1 concentration (14). Similar results have been reported for BPV E1 expressed in insect cells (15), although in other studies BPV E1 has also been purified in essentially monomeric form from insect cells (44) or when expressed in bacteria (11,15). Preliminary investigations of the oligomeric state of our preparations of polyhistidine-tagged HPV11 E1 by analytical size exclusion chromatography suggested that the full-length protein was a mixture of hexamers and a much larger species, whereas E1-(72-649) existed in solution predominately as hexamers. Further biophysical characterization was therefore focused on the more homogenous E1-(72-649).
We have previously shown by chemical cross-linking experiments that HPV11 E1 produced by in vitro translation exists in solution predominately as monomers that form hexamers upon the addition of single-stranded DNA (20). Similar crosslinking experiments performed with purified HPV11 E1-(72-649) revealed that it exists predominately or entirely as a hexamer in solution over a concentration range of 0.02-2.5 M, even in the absence of ssDNA (Fig. 3A). These experiments were carried out in a buffer containing 500 mM NaCl, since high salt concentrations were found to reduce the tendency of E1 to aggregate at high concentration.
Sedimentation velocity analytical ultracentrifugation was also performed to assess the oligomeric state of purified HPV E1-(72-649) (Fig. 3B). Data were well fit by a model assuming a single thermodynamic species with an estimated molecular mass of 355,000. This is close to the value expected for a hexamer (399,000 kDa), but we cannot rule out the possibility that a low concentration of E1 monomers or other lower order species was also present in equilibrium with the hexamer. Hence, the results from both cross-linking and sedimentation experiments demonstrate that purified E1-(72-649) is predominantly hexameric.
Helicase and ATPase Activities of Purified E1 Proteins-Characterization of the helicase activity of the purified E1 proteins was performed with DNA substrates made from short, 33 P-radiolabeled oligonucleotides annealed to bacteriophage M13 single-stranded DNA. Similar to reaction conditions in other reports, we found activity to be optimal at pH 7.5. Low concentrations of bovine serum albumin were found to stimulate activity (data not shown) but were not used in the experiments reported here. As illustrated for HPV11 E1, 200 M ATP-Mg was sufficient for maximal unwinding. The addition of excess magnesium did not have a significant effect on activity, except possibly at low levels of ATP (Fig. 4A). At these low ATP concentrations, we also found that a significant proportion of the ATP was consumed during the 90-min incubation (data not shown). The linearity of unwinding and ATPase activities over time (to ϳ90 min for helicase and 120 min for ATPase) is proteins. E, purified wild type and Lys 484 mutant E1 proteins (200 ng) were able to form a complex with E2 (10 ng) at the origin, as shown by a binding assay in which the complex of E1, E2, and 33 P-labeled origin fragment was detected by binding to an anti-E1 antibody bound to protein A SPA beads (black). Results of the same proteins in helicase (dark gray)/ATPase (light gray) assays show that the mutant preparations lack enzymatic activity. Enzymatic assays were run at 37°C using each E1 at 270 nM in a buffer containing 25 mM MES, 0.1 mM EDTA, and 10% glycerol at pH 7.0, with 1 mM ATP, 2 mM MgCl 2 , and 0.1 nM helicase substrate (nontailed substrate in Fig. 5A). After 2 h, separate samples were used for analysis of helicase and ATPase activities, using the SPA and colorimetric procedures described under "Experimental Procedures." In the reaction with WT E1, 36% of the substrate was unwound, and 20% of the ATP was hydrolyzed. illustrated in Fig. 4, B and C. At the concentrations used in Fig.  4, HPV6 and 11 E1 had similar helicase activity, while E1-(72-649), tested at a lower concentration, had approximately 5-fold greater specific activity (see also Fig. 8B). In contrast, the HPV11 proteins had similar ATPase activities, while HPV6 E1 was less active.
We found that M13 DNA could inhibit HPV11 helicase activity with an IC 50 of 2 g/ml or 7 M nucleotides (Fig. 4D). Interestingly, however, very little inhibition was observed if M13 DNA was added shortly after mixing E1 with the substrate (data not shown), demonstrating that binding of hexam-eric E1 to circular single-stranded DNA is rapid and dissociation is slow. Furthermore, M13 was found to be a much more potent inhibitor of E1-(72-649) than of wild type E1 (IC 50 ϳ0.2 g/ml), suggesting that this truncated form of the enzyme may bind more tightly to ssDNA.
We also examined the DNA substrate specificity of purified E1. As shown previously for BPV E1 (44) and some other helicases (62,63), unwinding is often more efficiently initiated on tailed substrates that possess a DNA fork, formed at the junction of dsDNA with noncomplementary single strands. To test if this was the case for HPV E1, unwinding reactions were Although we expect that an ATP-Mg complex is the natural substrate for E1, nucleotide hydrolysis and helicase activities could be observed for WT HPV11 E1 using eight natural nucleoside triphosphates (Table I). CTP and dCTP were the poorest substrates tested but were still hydrolyzed at 20 -30% the rate of ATP. Manganese could be substituted for magnesium, but with approximately a 50% loss of activity.
As seen in Fig. 4B, helicase activity was only slightly affected by decreasing the concentration of ATP from 1 to 0.1 mM. This suggested that the value of K m (ATP) might be significantly lower than 100 M, in contrast to the much higher values reported previously (41,42,45,49). The colorimetric assay we used to initially characterize the ATPase activity of HPV6 and 11 E1 was not sensitive enough to measure ATP hydrolysis at low micromolar concentrations of ATP, so we turned to more sensitive assays using radiolabeled ATP. Two different formats of this assay were used. In one, the released radiolabeled phosphate was detected by thin layer chromatography, while in the other it was complexed with ammonium molybdate and selectively bound to scintillation proximity beads (see "Experimental Procedures"). These two assays gave equivalent results, but we found the bead procedure more suitable for kinetic time courses, which require a large number of data points. We determined values for K m (ATP) of 6 and 12 M for HPV11 and HPV6 E1, respectively (Table II). HPV6 E1 had consistently higher values for K m and lower values for k cat . Similar values were obtained for HPV11 E1-(72-649). We obtained similar values for the alternate nucleoside triphosphate substrates GTP, UTP, and dATP and also found that higher levels of free magnesium did not affect these kinetic parameters (data not shown). Plots of observed rate versus substrate concentration were hyperbolic, suggesting that individual active sites of the E1 hexamer hydrolyze ATP independently (see Fig. 6B).
The ATP analog ATP␥S inhibited the ATPase activity of HPV11 E1 with an IC 50 of 6 M at 1 M ATP, consistent with the value for K m (ATP) shown in Table II. As anticipated, ATP␥S was a much weaker inhibitor of ATPase or helicase activity at 300 M ATP (IC 50 ϳ120 M, data not shown), consistent with it binding in competition with ATP. In other experiments using 1 M ATP, we found that ADP inhibited HPV-11 E1 ATPase activity with an IC 50 of 40 M, whereas no inhibition was observed with 600 M AMP (data not shown).
For many helicases, ATPase activity is stimulated severalfold by ssDNA (62,63), and for BPV E1 it has been shown that this stimulation correlates with the ability of ssDNA to stimulate E1 oligomerization (46). The experiments we described above were done in the absence of ssDNA. The addition of M13 ssDNA to HPV11 E1 or E1-(72-649) ATPase reactions containing 0.3 mM ATP had only a modest effect on activity (Fig. 4D).
Effect of the E2 Transactivation Domain on ATPase Activity-The finding by others that ATP hydrolysis somehow promotes dissociation of BPV E1 and E2 at the origin (17) prompted us to examine whether complex formation between HPV E1 and E2 would in turn affect E1 ATPase activity. To do so, a GST fusion of the HPV11 E2 transactivation domain (GST-TAD, amino acids 1-209), the portion of the protein that binds to E1 (19, 64 -66), was purified after expression in E. coli. Titration of GST-TAD into ATPase assays using HPV11 E1 and 1 M ATP resulted in inhibition at nanomolar concentrations (Fig. 6A). Inhibition was much weaker in reactions using 5 or 20 M ATP, and GST-TAD had no effect on helicase assays performed at 300 M ATP (data not shown). Several control experiments were performed to verify the specificity of this inhibition. GST had no effect on ATPase activity. Most importantly, a GST-TAD protein containing the amino acid substitution E39A, known to weaken E1-E2 binding but not transactivation (67, 68), did not affect ATP hydrolysis. As expected, the full-length E2 protein also inhibited ATPase activity. In no   FIG. 5. Unwinding by HPV6 or 11 E1 is not stimulated by using a forked substrate. E1 helicase reactions run using three substrates with different radiolabeled oligonucleotides annealed to the same region of M13, but containing either no tail, a 21-nucleotide 3Ј-tail, or an 18-nucleotide 5Ј-tail. For each composite gel, lanes are shown for boiled substrate and reaction blank, followed by reactions with HPV6 E1, HPV11 E1, and then HPV11 E1- . Unwinding percentages shown are values obtained relative to the boiled substrate, after subtraction of the blank value. Reaction conditions were as described for Fig. 4A except that the ATP concentration was 1 mM. case, however, did we observe complete inhibition of ATPase activity. This lack of complete inhibition and the ATP concentration dependence that we observed on TAD inhibition suggested that the TAD-E1 complex might still possess ATPase activity, but with altered kinetic parameters. In fact, we determined an apparent value for K m (ATP) in the presence of saturating (1 M) HPV11 GST-TAD, which was 7-fold higher than the value obtained in the absence of GST-TAD, while the apparent value of k cat was not altered (Fig. 6B and Table II).
Since ATP binds more weakly to the E1-E2 TAD complex, it was expected that ATP-Mg would in turn weaken the interaction of E1 with E2 and affect their cooperative binding to the origin in vitro. We in fact found this to be true (Fig. 6C) for both HPV11 E1 and HPV11 E1- . Inhibition by ATP-Mg was detected only at low concentrations of E1 and E2; at higher, saturating, concentrations, little effect was observed (data not shown). Interestingly, ATP-Mg had no effect on the interaction of two catalytically inactive mutants of E1-(72-649) (Fig. 6C), confirming that the observed effects were due to the productive binding of substrate. We observed very similar results using ATP␥S (data not shown), which also affected the binding of full-length E1 or E1-(72-649) to a much greater degree than that of the two inactive mutants.
Identification and Mutational Analysis of a Minimal ATPase Domain-As part of our characterization of the ATPase activity of E1, we wished to identify the smallest domain of the protein that retains ATPase activity. To achieve this, we made use of the fact that the ATPase activity of in vitro translated E1 protein can be assayed qualitatively following immunoprecipitation of the protein from a rabbit reticulocyte lysate (see "Experimental Procedures"). By performing such an assay on a series of truncated HPV11 E1 proteins, we determined that the C-terminal half of E1, comprising amino acids 353-649, is necessary and sufficient for ATPase activity (Fig. 7).
As mentioned in the Introduction, this portion of the protein encompasses four highly conserved regions termed A, B, C, and D that are shared between E1 and the large T antigens of SV40 and other polyomaviruses (Fig. 1). To confirm the importance of these regions for ATPase activity, mutant HPV11 E1-(72-649) proteins containing amino acid substitutions at highly conserved residues of region A and D and of the three conserved NTP-binding motifs were synthesized in vitro and tested for ATPase activity (Table III). Amino acid substitutions in the NTP-binding domain, of lysine 484 in motif a (P-loop) or of the two aspartate residues at positions 522 and 523 in motif b, dramatically affected ATPase activity, as anticipated. Another substitution in the P-loop, P479S, also reduced but did not completely abolish ATPase activity, in contrast to what was observed for E1 of BPV1 (41) and HPV6b (49), but in agreement with a previous report for full-length HPV11 E1 (42). This result is also consistent with our earlier observation that the P479S substitution does not completely abolish the ability of HPV11 E1 to support transient HPV DNA replication in vivo (19). Substitutions T566A and N568A in motif c also reduced ATPase activity. Although the exact function of motif c is not yet clear, these results support its importance in ATPase activity. Altogether, these results confirm the importance of motifs a, b, and c for ATPase activity.
The ATPase activity of E1 also proved to be affected by some but not all amino acid substitutions in conserved region A, in the minimal E1 oligomerization region. Three substitutions, Y380A, N389A, and F393A, severely hampered activity while two others, A390G and Q399A, reduced activity slightly. Only the F378A mutant protein had activity comparable with that of FIG. 6. Binding of E2 inhibits E1 ATPase activity, and productive binding of ATP perturbs the interaction of E1 with E2. A, ATPase reactions with 1 nM HPV11 E1 in the presence of other proteins. Each protein was used at 1 nM (black), 10 nM (dark gray), 100 nM (light gray), or 1000 nM (white). All values are shown relative to the activity in the absence of added protein (23 Ϯ 2% hydrolysis). Reactions were run for 120 min at 23°C using 1 M ATP and 500 M magnesium acetate. B, plots of observed rate versus ATP concentration for HPV11 E1 in the absence (ࡗ) or presence (OE) of 1 M GST-TAD. Reactions were carried out using 2 nM HPV11 E1 and the indicated concentration of ATP with sufficient magnesium acetate to give a constant excess of 500 M. Observed rates were determined from a time course as described under "Data Analysis" under "Experimental Procedures." Lines represent the best fits to the Michaelis-Menten equation for this data set (K m (ATP) ϭ 5.9 Ϯ 0.9 M, or 49 Ϯ 9 M in the presence of GST-TAD); parameters given in Table II are  wild type E1. These substitutions and those in the NTP-binding motifs were shown to have at most a slight effect on the interaction of E1 with E2 (19,20), and they also do not affect interaction with the p70 subunit of polymerase ␣ primase, 4 indicating that their deleterious effect on ATPase activity is not related to a dramatic change in the structure of E1. Hence, these results support the notion that region A is needed for ATPase activity. Similarly, the fact that one of the two region D substitutions, N609A, has a negative effect on ATPase activity is consistent with a role for this region in ATP-binding and/or hydrolysis, either directly or indirectly.
Enzymatic Activities of the Purified E1 Minimal ATPase Domain-To further characterize the enzymatic activities of the minimal E1 ATPase domain, we expressed it in insect cells as a polyhistidine fusion protein using a baculovirus expression system. For validation purposes, two catalytically inactive mutant derivatives (K484A and K484R) were also expressed in parallel. All three proteins were purified to near homogeneity (Fig. 8A) by nickel affinity chromatography (see "Experimental Procedures"). The purified wild type E1-(353-649) was found to be active in ATPase assays, with activity comparable with full-length E1 (Table II). As expected, both mutant derivatives were inactive, confirming that the ATPase activity observed for the wild type E1 subdomain was not due to contaminating polypeptides.
Given that the C-terminal domain of E1 has ATPase activity and that we previously showed that this domain is able to assemble into hexamers (20), we wished to determine if it was sufficient for helicase activity. The purified C-terminal domain was indeed able to unwind M13-based substrates with efficiency similar to full-length HPV11 E1 or one-half that of E1-(72-649) (Fig. 8B). Improved activity was observed for E1-(353-649) relative to full-length E1 at higher enzyme concentrations, suggesting that preparations of this protein contain lower levels of residual nuclease activity. As anticipated, the two ATPase-defective mutant proteins were also inactive (data not shown). These results indicate that the two known enzymatic activities of E1, helicase and ATPase, are encoded within 4 D. Fink, personal communication.

TABLE III
Effect of amino acid substitutions on HPV11 E1-(72-649) ATPase activity ATPase activity was measured as described under "Experimental Procedures" and in the legend of Fig. 7. Activity of E1-(72-649) was set arbitrarily at 100%. The amino acid sequence of each region or motif is indicated, with the substituted amino acids in boldface type.  35 S-labeled by coupled in vitro transcription/translation. Proteins were immunoprecipitated with a polyclonal antibody directed against the C-terminal 14 amino acids of E1. An autoradiogram of SDS-polyacrylamide gels, used to quantify the amount of each immunoprecipitated E1 protein, is shown on the right. For each protein, the amount of ATPase activity was normalized to the amount of radiolabeled protein contained in the E1 immunoprecipitate (see "Experimental Procedures"). The relative ATPase activity (in percent) of each truncated protein is indicated relative to that of E1-(1-649), which was set arbitrarily at 100%. B, same as A but using C-terminally truncated E1 proteins containing a FLAG epitope (indicated by a hatched box). Proteins were immunoprecipitated with the anti-FLAG monoclonal antibody. ATPase activity was normalized relative to the amount of immunoprecipitated protein as for A and is indicated relative to that of FLAG-tagged E1-(72-649), which was set arbitrarily at 100%. the C-terminal 297 amino acids.
Binding of Purified E1 to Single-stranded and Doublestranded DNA-The finding that E1-(353-649) is able to unwind partial duplex DNA substrates suggested that this domain of the protein is able to bind to single-stranded DNA. This is of significance, since the ssDNA-binding domain of E1 has not been mapped previously. To examine more directly the binding of E1-(353-649) to DNA, we measured its binding to radiolabeled ssDNA or dsDNA probes (Fig. 8, C and D). In parallel, we also examined the DNA binding ability of E1-(72-649) and of the two mutant forms of E1-(353-649). In these assays, E1-DNA complexes were immunoprecipitated using anti-E1 antibodies coupled to beads, and the amount of co-precipitated DNA was detected by scintillation counting (see "Experimental Procedures"). All four purified proteins displayed similar ssDNA binding activity over the concentration range tested (Fig. 8C). These results suggest that the ssDNA binding activity of E1 is indeed associated with the C-terminal domain of the protein and is not affected by the two amino acid substitutions in the P-loop. The C-terminal domain of E1 was also able to bind to dsDNA (Fig. 8D), but in contrast to what was observed with ssDNA, it showed a 75% reduction in activity compared with that measured for E1- . Hence, sequences located between amino acids 72 and 353 contribute to dsDNA binding, as anticipated from the fact that they are also needed for origin binding (see Introduction). Collectively, these results suggest that the ssDNA-binding domain of E1 is distinct from the origin-binding domain.
Since the C-terminal helicase domain of E1 is able to bind to both ss and dsDNA, we performed competition experiments to determine if their binding to E1 is mutually exclusive. In these assays, ϳ4 ng (8 nM) of unlabeled 20-mer oligonucleotide or 30 ng (20 nM) of a longer 60-mer oligonucleotide were required to inhibit the binding of E1 to the ssDNA probe by 50% (data not shown). In contrast, 1300 ng of double-stranded linearized plasmid DNA were necessary to achieve a similar percentage of inhibition, suggesting that E1-(353-649) binds preferentially to ssDNA over dsDNA and that it binds to both forms of DNA in a mutually exclusive manner. Finally, competition by plasmid DNA was the same regardless of whether the plasmid had been linearized or cut into 23 different fragments, indicating that the weak binding of E1-(353-649) to dsDNAs does not occur through the ends of the dsDNA molecules (data not shown). DISCUSSION E1 plays a key role in the replication of human papillomavirus; thus, understanding the activities of this protein is an important step in the search for improved antiviral therapies. The expression and purification conditions outlined in this report have allowed us to produce sufficient quantities of fulllength and truncated HPV E1 proteins to study ATP hydrolysis and nonspecific helicase activities in more detail. Unlike the in vitro translated HPV E1 proteins we have used in previous studies, the protein expressed in baculovirus-infected insect cells is isolated as preformed hexamers rather than monomers. These preformed HPV E1 hexamers could not be easily converted into monomers in vitro, similar to what was observed by Rocque et al. (42) with their preparations of HPV11 E1 isolated from the cytosolic fraction of baculovirus-infected insect cells. As a result, we have been unable to compare the enzymatic activities of monomeric and hexameric E1. However, in the case of SV40 T antigen, direct comparison of the monomeric protein and the preformed hexamers indicated that these have equal ATPase activity and similar helicase activity on partial duplex DNA substrates.
In agreement with earlier studies on BPV E1 (46) and SV40 T antigen (69), we observed that preformed HPV11 E1 hexam- ers have nonspecific helicase activity but are inactive in catalyzing cell-free viral DNA replication. These results suggest that HPV11 E1, like BPV E1 and SV40 T antigen, must be isolated in monomeric form to be active in cell-free DNA replication assays. In support of this suggestion, we observed previously that HPV11 E1 expressed as monomers by in vitro translation (25) is active in a cell-free DNA replication assay. The oligomeric nature of our purified E1 is also most likely the reason why the ATPase and nonspecific helicase activities could not be stimulated by ssDNA or by a forked DNA substrate, respectively. Indeed, previous studies of BPV E1 indicated that ssDNA enhances ATPase and helicase activity by promoting oligomerization of the monomeric protein rather than by affecting the activity of preformed oligomers (46).
The unwinding activity of our HPV11 E1 preparations appears to be similar to that reported previously for BPV E1 and SV40 T antigen and more recently for HPV11 E1. BPV E1 was reported to unwind 0.7 pM substrate/h/nM E1 monomer (5). By comparison, SV40 T antigen was found in one study to unwind 3 pM substrate/h/nM enzyme (70), while in another a somewhat lower value of 0.3 pM/h/nM was measured (71). Dean et al. (69) reported values of either 1 or 0.6 pM/h/nM depending on whether assays were initiated with monomeric or hexameric protein, respectively. From Fig. 4B, we calculated a comparable value of 0.3 pM/h/nM for purified HPV11 E1. This rate of unwinding appears similar to that observed by Rocque et al. (42) for their preparations of HPV11 E1, which they reported had approximately 4-fold less unwinding activity than T antigen. The helicase activity we measured for E1-(72-649), ϳ1.6 pM/ h/nM enzyme, is approximately 5-fold greater than that of wild type E1, despite the fact that both enzymes have similar ATPase activity (as shown in Fig. 4C or by values of k cat and K m (ATP)). The first 71 amino acids of E1 contain a high proportion of acidic residues, and removal of this region may account for the improved helicase activity as well as improved affinity of this truncated E1 for ssDNA (Fig. 4D). We found this truncated E1 to be less prone to aggregation, which may account in part for its improved activities.
Our characterization of the NTPase activity of purified E1 revealed that eight different nucleoside and deoxynucleoside triphosphates could serve as substrates in NTP hydrolysis and helicase assays. These results extend those reported by Rocque et al. (42) and are consistent with our previous findings that the same NTPs are also able to stimulate E1-DNA binding and oligomerization (20). We also found that ATP␥S and ADP (but not AMP), shown previously to stimulate E1-DNA binding and oligomerization, inhibit ATPase activity. Collectively, these findings reinforce our earlier suggestion that the E1 catalytic site can accommodate a wide variety of substrates. However, the reproducibly lower activity we obtained with CTP, dCTP, and TTP suggests that there are nevertheless some constraints on substrate binding. Surprisingly, using both TLC and scintillation proximity procedures, we obtained much lower values for K m (ATP) or other nucleoside triphosphates than those reported previously for HPV E1 (230 -380 M) (42,49) or BPV E1 (750 -1100 M) (41,45). A similar value of 600 M was also reported for SV40 T antigen (72), but in a later report a much lower value of 4 M was given (73). This latter value is consistent with another finding that a fluorescent analog of ATP has a K d of 0.4 M (74) as well as the finding that, in the absence of DNA, T antigen monomers oligomerize at half-maximal efficiency upon the addition of 10 M ATP (69). Our values for E1 are more consistent with the value of 4 M for T antigen. Although it is difficult to explain these differences, it is possible that some of the previously reported values were not obtained under true initial velocity conditions. Furthermore, in one re-port on BPV E1, Santucci et al. (45) observed biphasic rate versus ATP curves, which they suggest could be due to a contaminating ATPase in their preparations. It is also noteworthy that the value of 230 M reported for HPV6 E1 was obtained from a preparation purified as an E1-E2 complex (49); the presence of E2 may have affected the reported value (see below). Rocque et al. (42) did report a turnover value for the HPV11 E1 ATPase reaction of 8 min Ϫ1 with saturating ATP, quite similar to our value for k cat of 16 min Ϫ1 .
We observed that the transactivation domain of E2, either in the context of full-length E2 or as a fusion with GST, had a significant effect on E1 ATPase activity at low ATP concentrations. Even high concentrations of TAD had no effect at higher ATP concentrations despite the fact that E2 can still bind to E1 under these conditions (19). In apparent contrast with our results, a previous study found that the presence of BPV E2 had no significant effect on the ATPase activity of BPV E1 over a wide range of ATP concentrations (45). In that study, however, the activity of partially purified E1 was compared with that of a partially purified E1-E2 complex obtained by coinfection of insect cells with baculoviruses expressing both E1 and E2. E2 was not titrated into the ATPase reaction, and it is not possible to estimate the percentage of E1 that remained associated with E2 under the assay conditions. The inhibition we observed was specific, since neither GST alone nor GST-TAD with a change of Glu 39 to Ala, which is known to weaken E1 binding (67), was inhibitory.
GST-TAD appeared to modulate ATPase activity such that saturating concentrations of E2 increased the apparent value of K m (ATP) by 7-fold, while the value of k cat was unchanged. A simple thermodynamic consequence is that saturating concentrations of ATP should weaken the binding of E2 to E1 to a similar degree (75), which we have provided evidence for using an E2-dependent E1-DNA binding assay, which requires interaction of E1 with E2. In this assay, we observed inhibition only at low protein concentrations, consistent with the modest 7-fold decrease in affinity expected from ATPase experiments. Binding of E2 to catalytically inactive E1 proteins was not significantly affected by ATP-Mg, suggesting that only productive substrate binding weakens the E1-E2 protein-protein interaction. Since E2 affects ATP binding (K m ) rather than hydrolysis (k cat ), we would expect that ATP binding suffices to weaken the interaction of E1 with E2 (75), probably by inducing conformational changes in E1. In virus-infected cells, where E2 is present at low concentrations but ATP is in large excess, this perturbation of E2 binding could promote efficient viral DNA replication, since previous work by us (25) and others (23,24) has suggested that E2 must dissociate from E1 prior to association with the polymerase ␣ primase. Hence, saturation of E1 with ATP subsequent to E1 oligomerization may help release E1 from E2, and allow for its interaction with pol ␣ primase. It has been proposed that ATP hydrolysis is required for displacement of E2 from the origin concomitantly with the conversion of the initial E1-E2 origin complex into an enzymatically active multimeric E1-origin complex (17). While our results suggest that ATP binding plays a direct role in weakening the E1-E2 interaction, displacement of E2 from DNA may in addition require ATP hydrolysis, for example to drive distortion and opening of the origin by E1 hexamers. Additional work, ideally with purified monomeric E1, will be necessary to fully understand the sequence of binding and enzymatic events involved in the initiation of HPV DNA replication.
We determined by deletion analysis that the C-terminal domain of E1 comprising amino acids 353-649 is necessary and sufficient for ATPase activity. Shorter E1 fragments, which still retain the three ATPase motifs characteristic of superfam-ily 3 helicases (38), were inactive, indicating that sequences encompassing conserved regions A and D (Fig. 1) are also necessary for activity. The requirement for these two conserved regions, and for the three NTP-binding motifs, was demonstrated by site-directed mutagenesis of highly conserved residues. Region A is part of the minimal E1-E1 interaction domain, and the effect of region A substitutions on E1 ATPase activity remarkably paralleled their effect on oligomerization of the protein, which we reported previously (20). The in vitro synthesized E1-(72-649) that was used in these ATPase assays (either WT or mutant sequences) was probably monomeric, based on previous chemical cross-linking studies (20), although we cannot rule out that some of the protein may have oligomerized under the assay conditions and/or during immunocapture of E1 on beads. Further studies will be required to address this issue. Nevertheless, our results point to a functional relationship between the minimal E1-E1 interaction domain and ATPase activity, perhaps underlying the fact that oligomerization enhances ATPase activity.
Contrary to our expectation, we found that the minimal E1 ATPase domain (amino acids 353-649) is also sufficient for nonspecific helicase activity and for binding to ssDNA, despite the fact that it lacks the minimal origin-binding domain. A recent report indicated that the analogous C-terminal domain of SV40 T antigen, also lacking the origin-binding domain, is sufficient for binding to ssDNA, probably as a hexamer (76). It has not been determined if this domain of T antigen has enzymatic activity, but on the basis of our findings with E1, this seems a likely possibility. Despite encoding all of the sequences needed for nonspecific helicase activity, the C-terminal domain of E1 made by in vitro translation is unable to support cell-free DNA replication (25), presumably because it lacks upstream sequences between amino acids 166 and 353 required for binding to double-stranded DNA.
It has been suggested previously that during the initiation of viral DNA, replication E1 must first function as a sequencespecific DNA-binding protein, albeit of low sequence specificity, prior to assembling into enzymatically active hexamers (9,16,77,78). Our results suggest that these two activities of E1, sequence-specific DNA binding and helicase activity, require different domains of the protein, with the N-terminal half of E1 being required primarily to function as a dsDNA-binding protein and the C-terminal half being sufficient for unwinding. One could speculate that these two main activities of E1 may have arisen from the fusion of two independent functional domains during the course of evolution. A similar domain organization is also likely to apply to other origin-binding replicative helicases and in particular to the related large T antigens of SV40 and polyomaviruses. Recent electron microscopic images of T antigen double hexamers bound at the SV40 origin have indeed revealed that this protein is composed of two structurally distinct domains corresponding to the origin-binding and C-terminal domains, respectively (79).
More generally, we note that the 34-kDa C-terminal helicase domain of E1 is similar in size to the smallest naturally occurring hexameric helicase identified to date, the RepA protein (30 kDa) from the autonomously replicating RSF1010 plasmid (80). Furthermore, enzymatically active C-terminal domains of similar size have been identified for both E. coli DnaB (81-83) and phage T7 gene 4 protein (84), which are hexameric helicases in the same F4 family as RepA (85). Interestingly, these catalytic C-terminal domains possess the same organization as E1-(353-649), with an oligomerization region in the N-terminal portion followed by additional motifs required for ATPase and helicase activities. This striking similarity between E1 and T antigen and these more distantly related helicases suggests that many hexameric replicative helicases may have a similar architecture, with a C-terminal domain of 30 -40 kDa that is sufficient to support oligomerization around single-stranded DNA as well as ATP hydrolysis and DNA unwinding.