INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Papillomaviruses are members of the papovavirus family and comprise both human and animal strains. In addition to causing common cutaneous warts, papillomaviruses can induce various skin and mucosal lesions (35, 36, 39). Some of these lesions may progress to malignant carcinomas, depending on the particular viral strain involved as well as environmental factors. Malignancies which are highly associated with papillomavirus infection include laryngeal, cervical, and other anogenital cancers, and it is noteworthy that human papillomavirus (HPV) types 16, 18, 31, 33, and 45 pose an especially high risk for females, as they are estimated to be present in at least 90% of cervical cancers (17, 27).

Bovine papillomavirus (BPV) has been used as a model for studying the biology of papillomavirus replication, particularly the DNA replication components. Upon entry of a host cell by wounding or abrasion, BPV replicates its small 8-kb genome to a multicopy episome which is maintained at steady-state levels of up to several hundred copies per cell (14). After a steady-state level of replication has been achieved, the viral DNA then replicates approximately once during each host cell cycle and is likely regulated by cell cycle controls (8, 18, 21). In order to initiate viral DNA replication, the E1 protein forms a multimeric complex with the E2 protein, the viral origin of replication, and several host cell factors (2, 6, 8, 13, 15, 26, 28, 33). The minimal BPV origin of replication is approximately 60 bp in length and contains, in order from 5' to 3', a 23-bp AT-rich element, an 18-bp imperfect palindrome which serves as an E1 binding site, and a 12-bp palindromic E2 binding site (20).

E1 functions as the initiator of papillomavirus DNA replication in vivo and plays several roles in this capacity. It is a multifunctional 68-kDa phosphoprotein which cooperates with the E2 protein to bind sequence-specifically to its cognate E1 binding element in the viral origin, facilitates origin DNA unwinding by acting as an ATP-dependent helicase, recruits the host cell DNA polymerase -primase complex, which then begins de novo DNA synthesis, and interacts with regulatory host cell factors such as cyclin E (8, 18, 23, 28, 32, 38). In addition, the phosphorylation state of E1 has been suggested as a regulatory device and link to the cell cycle control apparatus, while sumoylation is required for nuclear localization (8, 18, 19, 21, 24, 25, 39).

How initiation proteins such as E1 recognize and interact with specific origin sequences is a fundamental question in DNA replication. We previously demonstrated that an isolated E1 polypeptide, E1_121-311, retained specific origin recognition capacity similar to full-length E1 (16). The origin recognition activity of E1_121-311 indicated that the double-stranded DNA binding function of E1 resides in an independent domain that could be investigated in the absence of other E1 sequences. The precise demarcation of this functional domain has not yet been established, but the N-terminal boundary must lie near residue 160, as both E1_159-303 and E1_162-422 possess origin-binding activity in vitro while E1_162-308 does not (6, 10, 26).

Using alanine mutagenesis of E1_121-311 we identified two short hydrophilic regions, HR1 and HR3, that are critical for origin binding (12). These results were consistent with a previous mutational study by Thorner et al. that identified DNA-binding-negative mutants of E1 mapping to these same regions (34). By comparison with simian virus 40 (SV40) T antigen, we suggested that HR1 and HR3 were likely juxtaposed to form the DNA interaction surface of the E1 DNA binding domain (E1DBD). The subsequently determined crystal structure for this region confirmed this prediction and revealed that HR1 was located in a structured loop region positioned adjacent to -helix 4, which corresponds to HR3 (Fig. 1) (10). In addition, the crystal structure revealed an overall conformational similarity between the E1DBD and the T antigen DBD.

View larger version (34K): [in this window] [in a new window]   FIG. 1.   Summary of E1DBD mutations. (A) E1 amino acid sequence from residues 150 to 304. The locations of -helices and -sheets derived from the crystal structure (10) are diagramed above the sequence, along with the positions of the previously described hydrophilic regions HR1 and HR3 (12). Below the sequence are the origin-nonbinding ( row, squares) and origin-binding (+ row, triangles) mutants from this study (solid symbols) and our previous study (open symbols). (B) Three-dimensional ribbon diagram of E1 protein residues 159 to 303 from Enemark et al. (10). Residues whose mutation yields a nonbinding phenotype in the yeast one-hybrid assay are indicated in red on the left monomer, and binding-positive mutations are in blue on the right monomer. The boundaries of the HR1 loop region are indicated with red arrows, and the HR3 region boundaries in -helix 4 are indicated with blue arrows. Specific amino acid positions characterized in detail in the text are indicated (note that V246 is on the posterior side of -helix 4 and is not visible in this figure).

To further define critical residues for E1DBD function, we used a random mutagenesis procedure and an in vivo yeast one-hybrid screen to identify nonbinding mutants. The roles of individual residues in E1-DNA interaction are discussed, and sequence similarities between E1, T antigen, and EBNA1 are presented.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

E1DBD mutant library construction. The yeast expression construct pGAD424-E1DBD containing E1_121-311 has been described previously (12). Primers 5'GAD424 (5'-CCACTACAATGGATGATGTA-3') and 3'GAD424 (5'-TGCACAGTTGAAGTGAACTTG-3') were used to PCR amplify the E1DBD region in the presence of the nucleotide analog dITP. A 50-µl reaction contained 10 mM Tris-HCl (pH 9.0) at 25°C, 50 mM KCl, 0.1% Triton X-100, 5 mM MgCl₂, 200 µM dCTP, 200 µM dGTP, 200 µM dTTP, 20 µM dATP, 200 µM dITP (Sigma), 50 ng of template pGAD424-E1DBD, 25 pmol of primer 5'GAD424, 25 pmol of primer 3'GAD424, and 1.25 U of Taq polymerase (Promega). The reaction was run in an MJR PTC-200 thermocycler and used an initial denaturation at 94°C for 1 min followed by 30 cycles of 94°C for 15 s, 55°C for 30 s, and 72°C for 1 min. The single PCR product was purified using a Qiaquick gel extraction kit (Qiagen) and quantified by UV spectroscopy.

Evaluation of clones. (i) -Galactosidase colony-lift filter assay. The mutant library plasmid DNA was transformed into yeast strain YM4271[E1BST-LacZi] (hereafter referred to as YM4271[E1BST]) using a Frozen-EZ Yeast Transformation II kit (Zymo Research). Colonies were grown at 30°C on synthetic dropout medium plates lacking uracil and leucine and then lift-transferred onto nitrocellulose filters (Schleicher & Schuell). Filters were frozen in liquid nitrogen, thawed, and incubated on top of filter paper soaked in buffer containing 60 mM Na₂HPO₄ · 7H₂O, 40 mM NaH₂PO₄ · H₂O, 10 mM KCl, 1 mM MgSO₄ · 7H₂O, 0.2% -mercaptoethanol, and 265 µg of 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-gal) at 30°C for 3 h. The E1DBD open reading frame (ORF) was sequenced in reporter-negative colonies as described below. Selected reporter-negative clones were transformed into the control yeast reporter strain YM4271[p53BLUE] and assayed on nitrocellulose filters for reporter activity.

(ii) SSCP assay. Subsequent to filter assays, 60 reporter-positive yeast colonies were cultured and plasmid DNA was extracted with a Zymoprep yeast plasmid miniprep kit (Zymo Research). Two sets of primers were used in separate 50-µl PCRs to amplify and label the 5' and 3' halves of the E1DBD ORF from each clone. Oligonucleotide primers SSCP1 (5'-CCAAAAAAAGAGATC-3'), SSCP2 (5'-ATGAGATCTTTTTTGC-3'), SSCP3 (5'-GTTTCGAACTCCTAA-3'), and SSCP4 (5'-TTCATAGATCTCTGC-3') were purchased from the Texas A&M University Veterinary Pathobiology Core Facility. Reactions contained 10 mM Tris-HCl (pH 9.0) at 25°C, 50 mM KCl, 0.1% Triton X-100, 3 mM MgCl₂, 200 µM dATP, 200 µM dCTP, 200 µM dGTP, 200 µM dTTP (Sigma), 1.25 µl of template plasmid DNA extracted from yeast, 1.5 pmol of each primer from the SSCP1/SSCP2 primer pair or SSCP3/SSCP4 primer pair (Sigma-Genosys), and 0.25 U of Taq polymerase (Promega) in a 12.25-µl reaction volume. Ten microliters of each reaction product was added to 2 µl of single-strand conformation polymorphism (SSCP) sample buffer, which was made by combining 9.5 ml of deionized formamide, 0.4 ml of 0.5 M EDTA (pH 8.0), 0.05 ml of 10% bromophenol blue, and 0.05 ml of 10% xylene cyanol. Samples were boiled in a water bath for 5 min and immediately transferred to ice for 2 min. Samples were electrophoresed for 8 h at 350 V on an 8% polyacrylamide gel containing 5% glycerol and maintained at 19°C. The gel was dried and imaged using a Molecular Dynamics Phosphorimager. Samples which exhibited altered gel mobilities compared to a control were sequenced as described below.

(iii) Temperature-sensitive screens. The mutant libraries of pGAD-E1DBD were transformed into yeast strain YM4271[E1BST] and plated on synthetic dropout medium lacking uracil and leucine. Colonies were grown at 39°C for 1 week and then assayed for reporter activity by the filter assay described above. The same colonies were then grown for 3 additional days at 30°C and examined again for reporter activity by filter assay. Results at 30 and 39°C were compared, and colonies which were reporter positive at 30°C and reporter negative at 39°C were selected. Plasmid DNA was then extracted and sequenced as described below.

(iv) DNA sequencing. Yeast colonies were cultured and plasmid DNA was extracted using a Zymoprep yeast plasmid miniprep kit (Zymo Research). For reporter-negative clones, PCR was used prior to DNA extraction to confirm the presence of E1DBD ORF in the plasmid. After extraction, plasmid DNA was electroporated into E. coli (Epicurean Coli XL1-Blue; Stratagene). The E1DBD ORF was sequenced using an ABI Big Dye Terminator DNA sequencing kit (PE Biosystems) with primers against the pGAD424 plasmid, 5'GAD424, and 3'GAD424 (sequences listed above).

Purification of GST-E1DBD and E2 protein. The E1DBD coding region from pGAD424-E1DBD mutants F175S, N184Y/K288R, D185G, V193M, F237L, K241E, R243K, and V246D was subcloned into pGEX-5X-1 (Pharmacia) using the same EcoRI and SalI sites used to create the pGAD424-E1DBD mutant library. The subclones were electroporated into E. coli strain BL21 and verified by PCR and restriction digestion. Fusion proteins were purified as previously described (12).

In vitro DNA binding assays. The concentration of wild-type GST-E1DBD protein was determined by the Bradford assay (5). The concentrations of mutant GST-E1DBD proteins were estimated by Coomassie blue staining of acrylamide gels followed by densitometry and comparison to a known amount of wild-type protein. In vitro binding was assessed by a gel shift assay using a blunt-ended, double-stranded oligonucleotide comprising BPV-1 nucleotides 7926 to 29 as previously described (12, 37). Except where indicated otherwise, 10 ng of each wild-type or mutant GST-E1DBD protein was used in the binding assays; where E2 was also included, typically 20 ng of purified protein was used.

In vivo DNA replication assays. Mutations were constructed in heterologous expression vector pCGE1 for expression of full-length E1 mutants F175S, N184Y, D185G, V193M, F237L, K241E, R243K, and V246D. Heterologous expression vectors pCGE1 and pCGE2 at 1.0 and 0.1 µg, respectively, were mixed with 1.0 µg of the BPV origin-containing vector pBOR or 1.0 µg of pUC18 for positive and negative controls, respectively. Similar mixtures were made with each pCGE1 mutant, pCGE2, and pBOR. The combined DNAs were transected into Chinese hamster ovary (CHO) cells using the Lipofectamine transfection reagent (Gibco-BRL). At 48, 72, and 96 h posttransfection, plasmid DNA was harvested by Hirt extraction and digested with DpnI and HindIII (New England Biolabs). DNA samples were electrophoresed on an 8% agarose gel, transferred to a Nytran Supercharge nylon membrane (Schleicher & Schuell) by Southern transfer, and cross-linked to the membrane using a Stratalinker (Stratagene). A pBOR probe was radiolabeled using the Prime-a-Gene labeling system (Promega) and hybridized to the membrane using Rapid Hyb buffer (Amersham). The membrane was analyzed by Phosphorimager analysis.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Mutations in and near HR1 and HR3 decrease E1DBD-dependent reporter activity in yeast cells. We previously described the development of a yeast one-hybrid system for monitoring sequence-specific interactions of both full-length E1 and the E1DBD with chromosomal DNA containing three tandem E1 binding sites (E1BST) upstream of the lacZ gene (12) (Fig. 2A). Reporter gene expression in this system requires specific E1-E1BS interaction, as neither the pGAD424 parental in the YM4271[E1BST] strain nor the pGAD424-E1DBD construct in YM4271[p53BLUE] (which contains a p53 binding site instead of the E1 binding sites) is sufficient for -galactosidase production.

View larger version (41K): [in this window] [in a new window]   FIG. 2.   Yeast one-hybrid screening system and representative data. (A) Expression of yeast GAL4 transactivation domain (GAL4AD) fused to the E1 DNA binding domain (E1DBD) activates a lacZ reporter gene upon interaction of the E1DBD with three tandem copies of the E1 binding site (E1BST). (B) Nitrocellulose filter assay for reporter activity in yeast colonies harboring pGAD424-E1DBD constructs. YM4271[p53BLUE] is a control strain containing a p53 binding site in lieu of E1 binding sites, which confirms that interaction between the E1DBD and E1BST is sequence specific. Shown are both reporter-positive and reporter-negative mutants from our screens along with their genotypes. (C) SSCP assay was used to detect mutations in the E1DBD coding regions of reporter-positive yeast colonies. PCR was used to amplify and radiolabel the 5' and 3' halves of the 573-bp E1DBD open reading frame of wild-type pGAD424-E1DBD and candidate mutants. PCR products were then processed as described in Materials and Methods and electrophoresed on native polyacrylamide gels. Samples from the 5' and 3' amplimers are shown in the right and left panels, respectively. Lane designations are the clone names as listed in Table 2. Lanes marked C were amplified from the parental wild-type E1DBD clone.

Nonessential residues are rare in and near HR1 and HR3. We also screened the libraries for binding-functional E1DBD mutants using two different methods. First, we examined reporter-positive yeast clones for SSCP. Following plasmid DNA extraction from the yeast clones, we generated PCR products from the E1DBD ORF, which were then processed as described in Materials and Methods and examined on native polyacrylamide gels (Fig. 2C). Of 60 clones screened, 21 (35%) were found to display SSCP. Second, we used a temperature sensitivity screen by performing reporter assays at both 30 and 39°C. Using this method, we screened approximately 300 clones and found four mutant candidates (1.3%). Upon sequence analysis of clones which displayed conformation polymorphisms or temperature sensitivity, we identified a variety of mutations which were scattered throughout the E1DBD (Table 2 and Fig. 1). These binding-functional mutations were largely absent from the HR1 and HR3 regions, with a few exceptions which may represent functional papillomavirus polymorphisms or simply nonconserved residues (see Discussion). Furthermore, the paucity of nonsense and frameshift mutations in reporter-positive clones compared to reporter-negative clones is consistent with this tested E1 polypeptide being a minimized DNA binding domain.

HR1- and HR3-associated mutations affect sequence-specific DNA binding in vitro. Since we isolated the mutants by in vivo screening, it was important to test them in vitro for confirmation of their binding negativity as well as to control protein amounts, which may vary with expression in vivo. In a previous study we examined five substitution mutants of HR1 (K183A, D185A, K186A, T187A, and T188A) but only one mutant of HR3 (K241A). Our current yeast screens yielded two new mutations in HR1 (N184Y and D185G) and three new mutations in HR3 (K241E, R243K, and V246D), which gave us the opportunity to examine HR3 in more depth. In addition, the screen also identified several nonfunctional mutations which resided outside the HR1 and HR3 sequences. Mutants F175S, V193M, and F237L neighbor HR1 or HR3 and were chosen for additional study. Mutations at positions 210, 265, and 293 involved adding or removing a proline, which is likely to disrupt structure, and were not investigated further. Mutation S207C lies in the 3 helix and may be critical for dimerization of the E1DBD (10), and will be characterized as part of a later study.

View larger version (52K): [in this window] [in a new window]   FIG. 3.   In vitro analysis of GST-E1DBD interactions with a 50-bp double-stranded, sequence-specific DNA probe. (A) GST-E1DBD fusion proteins were purified by glutathione-Sepharose affinity chromatography. Shown are 10% polyacrylamide gels containing wild-type and mutant proteins and stained with Coomassie blue. Lanes M, size markers. (B) Electrophoretic mobility shift assay of wild-type (WT) and mutant GST-E1DBD fusion proteins. For each reaction, 10 ng of protein was incubated with the radiolabeled DNA except for mutant V246D, for which 3 ng was used in this experiment.

View larger version (51K): [in this window] [in a new window]   FIG. 4.   In vitro E2-mediated rescue of DNA binding by E1DBD mutants. Electrophoretic mobility shift assays were performed as in Fig. 3 using the amounts of proteins (in nanograms) indicated. The other single-substitution mutations shown in Fig. 3 were all negative for E2 rescue in this assay (not shown).

E1DBD mutants positive for E2 interaction support partial in vivo DNA replication. The ability of some binding-defective mutants to be rescued by E2 suggested that they may retain sufficient function to support DNA replication. Single point mutations F175S, N184Y, D185G, V193M, F237L, K241E, R243K, and V246D were constructed in the vector pCGE1, which expresses full-length E1 in mammalian cells for in vivo DNA replication assays. The N184Y mutation was separated from K288R in this experiment so that substitution at residue 184 was the only change present in E1. Heterologous expression vectors pCGE1 (wild-type or mutant) and pCGE2 were transfected along with the BPV-1 origin-containing plasmid pBOR into Chinese hamster ovary cells. Plasmid DNA was Hirt extracted at 48, 72, and 96 h posttransfection and examined by Southern blotting as described in Materials and Methods (Fig. 5A). DpnI-resistant replication products were quantified by Phosphorimager analysis, and a minimum of two independent transfections were performed for each mutant construct. From each experiment, replication efficiencies were calculated as the percentage of replication product made in the presence of wild-type E1, and the average replication efficiencies are plotted in Fig. 5B.

View larger version (37K): [in this window] [in a new window]   FIG. 5.   DNA replication in CHO cells. (A) Representative Southern blots of replication products recovered from CHO cells at 48, 72, and 96 h posttransfection. Cells were triple transfected at time zero with vectors for E1 and E2 expression and a plasmid containing the BPV origin, pBOR, or pUC18 as a non-origin-containing control. Wild-type (WT) and mutant E1 expression vectors were used as indicated. (B) Quantitation of DNA replication products. Band intensities on Southern blots were measured by Phosphorimager analysis. Replication efficiency of mutant E1 proteins was calculated by comparing the intensity of wild-type and mutant replication product bands at 96 h posttransfection. Average replication efficiencies for two or more independent experiments are shown. Data for V193M were generated in a separate study.

HR3 is unlikely to make the majority of base contacts with DNA. The distribution of functional versus nonfunctional mutations described above strongly implicates HR1 and HR3 as critical elements for sequence-specific E1-DNA interaction, but does not discriminate the individual roles of each element. We noted previously that the organization of these two E1 elements resembled that of the A and B2 regions of SV40 T antigen, which are similarly critical for sequence-specific DNA binding (12). The recent crystal data for the E1DBD confirmed an overall structural similarity between the T antigen and E1 DNA binding domains (10), and a manual alignment of the primary sequences shows a remarkable relatedness, particularly in the DNA interaction elements (Fig. 6A).

View larger version (24K): [in this window] [in a new window]   FIG. 6.   (A) Sequence alignment between DNA binding domains of SV40 T antigen and BPV-1 E1. Boxes indicate regions of sequence similarity. There is an extended -hairpin in E1 at residues 223 to 230 which is not found in T antigen. The locations of conserved regions HR1, HR3, A, and B2 are indicated. (B) HR1- and HR3-like elements are found in origin-binding proteins from three transforming DNA viruses: papillomavirus, SV40, and Epstein-Barr virus. E1, T antigen (T-AG), and EBNA1 contain regions of sequence similarity which are separated by approximately 50 amino acid residues and which are critical for sequence-specific DNA binding.

View larger version (32K): [in this window] [in a new window]   FIG. 7.   (A) Sequence alignment between E1 HR3 and SV40 T antigen (TAG) region B2. Also shown is the sequence in the HR3 region of the THR3 mutant. Multiple site-directed substitutions were made to create this chimeric mutant of the E1DBD whose HR3 element contains residues found in the T antigen B2 element. (B) Nitrocellulose filter assay showing sequence-specific interactions between THR3 and E1 binding sites in the yeast one-hybrid system. (C) Electrophoretic mobility gel shift assay as in Fig. 4, showing cooperativity of binding of THR3 with E2 on an origin-specific DNA probe. Amounts of proteins (in nanograms) are indicated.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We previously used site-directed mutagenesis to show that mutations in E1DBD elements HR1 and HR3 disrupt sequence-specific interaction with the DNA origin. In our current study, we used random PCR mutagenesis along with a yeast one-hybrid screen to identify additional mutants of the BPV-1 E1DBD which are defective for sequence-specific DNA interactions. As a complement, binding-functional mutants were also identified. Our goal in using this nondirected approach was to gain a more extensive understanding of the location and specific function of those amino acids most important for the origin-binding activity of E1.

The locations of the various binding-functional and -nonfunctional mutations are indicated on the E1 sequence and structure in Fig. 1, and the properties of selected nonbinding mutants are summarized in Table 3. While the mutations were mostly characterized in the context of the isolated E1DBD, two random and five site-directed mutations were subsequently analyzed in full-length E1 using the yeast one-hybrid system (data not shown). All seven mutations, four binding functional and three nonfunctional, had the same binding phenotypes in the E1DBD compared to full-length E1. Thus, the E1DBD seems to be a truly independent domain in E1, and results with the E1DBD should be generally reflective of the properties of full-length E1 for DNA interaction.

Residues necessary for sequence-specific DNA binding by E1 are concentrated in and near HR1 and HR3. Using the recently published crystal structure (10), previous studies of T antigen mutants (29, 30), and our new mutant data, we have arrived at several conclusions regarding possible functions of specific E1DBD residues. The first observation to emerge from sequence comparisons of the E1DBD ORF in reporter-negative yeast clones was the repeated occurrence of substitutions in and around the HR1 and HR3 regions (Fig. 1). Furthermore, a chi-square statistical test indicated that the clustering of critical mutations in HR1 and HR3 for the reporter-negative clones versus the reporter-positive clones had a less than 0.5% probability of occurring by chance. This correlation is further evidence that hydrophilic regions HR1 and HR3 are indeed necessary for sequence-specific DNA binding. Besides the HR1 and HR3 regions, the only other modest clustering of nonfunctional mutations occurred in -helix 3. This helix is involved in E1 dimerization, and the occurrence of several mutations in this region supports an important functional role for E1-E1 contacts in origin binding. As expected, functional mutations isolated from the reporter-positive clones were more evenly distributed throughout the E1DBD.

Role of HR1 residues in E1DBD-DNA interaction. Our previous study identified a cadre of hydrophilic residues in HR1 that were critical for origin binding (12). The recently published crystal structure of the E1DBD revealed that the HR1 region forms a structurally well-defined loop which potentially makes sequence-specific contacts with origin DNA (10). In the current study, two new mutations in HR1 were characterized, N184Y and D185G. N184Y was analyzed in the context of the double mutant E1DBD_N184Y/K288R; however, the K288R substitution is unlikely to be the critical change for disrupting DNA binding, since a nonconservative change at this position appeared in a binding-functional mutant (Table 2). Additionally, the single point mutant E1_N184Y exhibited a severe defect for in vivo DNA replication in CHO cells. Therefore, the N184Y substitution is most likely responsible for the defective DNA interaction both in vivo and in vitro as well as for E2 binding in vitro.

Role of HR3 residues in E1DBD-DNA interaction. We also isolated and characterized new mutations in the HR3 element. Mutant K241E was defective for in vivo and in vitro binding, E2 interaction, single-stranded DNA binding (unpublished data), and in vivo DNA replication. Residue K241 resides at the N-terminal side of -helix 4, contributes a positively charged amino group, and lies on the DNA binding surface of E1. Substitution of a negatively charged glutamic acid residue likely increases electrostatic repulsion between this position and the DNA backbone, though this does not exclude that the lysine might also normally make specific base contacts.

Role of critical residues outside the HR1 and HR3 regions. Other than substitutions at proline residues (P265 and P293), which likely severely disrupt overall structure, binding-critical residues outside HR1 and HR3 fell into two groups: hydrophobic amino acids (F175, V193, F237, and W277) and residues located in -helix 3 (F203, S207, and L210). From the crystal structure, it appears that these hydrophobic residues may contribute to a core that forms a scaffold beneath the HR1/HR3 structure and possibly provides an anchor function for the DNA interaction motif. A more detailed study of the role of this hydrophobic core will be presented elsewhere (unpublished data).

Residues nonessential for DNA binding by E1 occur throughout E1DBD and are rare in HR1 and HR3 regions. The SSCP and temperature sensitivity assays detected yeast clones expressing binding-functional mutations (Table 2). Most of these DNA binding-functional mutations were nonconservative, indicating that the side chains at these positions make little if any contribution to E1DBD structure. The lack of clustering of nonessential residues is also consistent with a simple role, such as providing appropriate spacing to promote proper orientation of essential residues in and near the critical HR1 and HR3 regions. It should be noted that these residues may be critical for other functions of the E1DBD, such as interaction with the E2 protein or host cell replication proteins, but this has not yet been determined. The only noncritical substitution which did occur in HR1 or HR3 was D185N in mutant D185N/E258G/F278S. However, this is a structurally conservative mutation and also can probably maintain the salt bridge with R243, as discussed above, so it is not surprising that this substitution was quite functional.