Requirements for Epstein-Barr Nuclear Antigen 1 (EBNA1)-induced Permanganate Sensitivity of the Epstein-Barr Virus Latent Origin of DNA Replication*

Epstein-Barr nuclear antigen 1 (EBNA1) activates DNA replication from the Epstein-Barr virus latent origin of DNA replication, oriP. EBNA1 binds cooperatively to four recognition sites in the dyad symmetry (DS) element oforiP, causing alterations in the origin DNA structure, which can be detected by the increased sensitivity of one Thy residue in two of the binding sites to permanganate oxidation. To better understand the significance of this EBNA1-induced origin distortion, we have investigated the DNA sequence and EBNA1 amino acid requirements for permanganate sensitivity. We have shown that the EBNA1 DNA binding and dimerization domains are sufficient to induce permanganate sensitivity and that amino acids 463–467, which form an extended chain that travels along the minor groove of the EBNA1 recognition site, play an important role in generating the DNA distortion. The EBNA1-induced permanganate sensitivity is independent of cooperative interactions between EBNA1 molecules on the origin and requires a specific sequence within the EBNA1 binding site. Using synthetic EBNA1 binding sites, we found that the inversion of a single AT base pair in the EBNA1 recognition sequence is sufficient to confer EBNA1-induced permanganate sensitivity. These studies indicate that permanganate oxidation can detect very minor alterations in DNA structure.

Epstein-Barr nuclear antigen 1 (EBNA1) activates DNA replication from the Epstein-Barr virus latent origin of DNA replication, oriP. EBNA1 binds cooperatively to four recognition sites in the dyad symmetry (DS) element of oriP, causing alterations in the origin DNA structure, which can be detected by the increased sensitivity of one Thy residue in two of the binding sites to permanganate oxidation. To better understand the significance of this EBNA1-induced origin distortion, we have investigated the DNA sequence and EBNA1 amino acid requirements for permanganate sensitivity. We have shown that the EBNA1 DNA binding and dimerization domains are sufficient to induce permanganate sensitivity and that amino acids 463-467, which form an extended chain that travels along the minor groove of the EBNA1 recognition site, play an important role in generating the DNA distortion. The EBNA1-induced permanganate sensitivity is independent of cooperative interactions between EBNA1 molecules on the origin and requires a specific sequence within the EBNA1 binding site. Using synthetic EBNA1 binding sites, we found that the inversion of a single AT base pair in the EBNA1 recognition sequence is sufficient to confer EBNA1-induced permanganate sensitivity. These studies indicate that permanganate oxidation can detect very minor alterations in DNA structure.
The first step in the initiation of DNA synthesis in all replication systems involves the melting of the origin DNA. In many systems the disruption of the hydrogen bonds between the bases is a function of the origin binding protein, but in others the interaction of the origin binding protein with the origin is not sufficient to melt the DNA and additional factors are required (for review, see Refs. [1][2][3]. Origin binding proteins that do not melt DNA might, however, contribute to the melting process by destablizing the structure of the DNA helix. To better understand how localized DNA distortions caused by origin-binding proteins contribute to origin melting, we have examined the structural changes in origin DNA induced by Epstein-Barr virus nuclear antigen 1 (EBNA1). 1 DNA replication from the Epstein-Barr virus latent origin of DNA replication, oriP, occurs once every cellular S-phase and is activated by EBNA1 (4 -6). EBNA1 binds as a dimer to an 18-bp recognition site present in multiple copies in the two sub-elements of oriP, the dyad symmetry (DS) element and the family of repeats (FR) (7)(8)(9)(10). The DS element contains four EBNA1 binding sites and likely the initiation site for DNA synthesis (7,(11)(12)(13)(14). The FR element contains 20 EBNA1 binding sites and enhances replication from the DS element (7,13).
The interaction of EBNA1 with the DS element has been studied by several laboratories. In vivo studies have shown that EBNA1 remains bound to the DS throughout most or all of the cell cycle (12,15) and that the minimum requirement for origin activation is the binding of EBNA1 to the two recognition sites in either half of the DS element (sites 1 and 2 or sites 3 and 4) (14). In vitro, EBNA1 dimers have been shown to assemble cooperatively on the four sites of the DS element (14,16), generating a single EBNA1 complex that sharply bends the DS DNA (17). EBNA1 on its own does not appear to melt oriP DNA sequences; however, two lines of evidence indicate that EBNA1 does cause localized structural distortion of the DS element. First, EBNA1 binding to the DS element induces the permanganate sensitivity of one Thy residue in each of the two outer recognition sites (sites 1 and 4). This permanganate sensitivity has been demonstrated both in vivo and in vitro (15,18,19). Second, modeling studies with the EBNA1-DNA co-crystal structure suggest that a change in the DNA structure is necessary to accomodate binding of EBNA1 to two adjacent sites in the DS element (20).
The nature of the structural change in the DS DNA that leads to the EBNA1-induced permanganate sensitivity is not known nor is it clear if the DNA distortions predicted from the modeling and permanganate studies are the same or different. Localized distortion of the DS DNA by EBNA1 is of interest because it could be the first step in the melting of the origin for the initiation of DNA synthesis. To further understand the nature and importance of the EBNA1-induced distortion of the DS DNA, we have examined the requirements for the EBNA1induced permanganate reactivity of sites 1 and 4.

EXPERIMENTAL PROCEDURES
DNA-The construction of pGEMdyad, which contains the DS element of oriP, and pGEMs1, which contains only EBNA1 binding site 1 from the DS element, have been previously described (17,21). Plasmids containing the palindromic EBNA1 consensus binding site (CS) or the consensus binding site with an inversion of the AT base pairs at position 6 and Ϫ6 (CSA/T) were constructed from the oligonucleotides 5Ј-GGG-TAGCATATGCTACCC-3Ј and 5Ј-GGGAAGCATATGCTTCCC-3Ј, re-spectively. Each of these oligonucleotides was annealed to itself and then cloned into the SmaI site of pBluescriptKS to generate pCS and pCSA/T. The construct for the expression of EBNA WF was generated by amplifying the EBNA1 gene in two fragments, which encode amino acids 452 to 463 and amino acids 464 to 641, by PCR. The 452-463 fragment was amplified using the primers 5Ј-CGTCGACATATGGGT-CAGGGTGAT-3Ј (452 primer) and 5Ј-CCCTCCTTTTTTGCGCCT-3Ј. The 464 -641 fragment was amplified using the primers 5Ј-GCGGCGG-GAAAGCATCGTGGTCAA-3Ј and 5Ј-CCTCCAGGATCCTCACTCCT-GCCCTTCCTCACC-3Ј (641 primer), which converts the Trp and Phe residues at positions 464 and 465 to alanines. The two fragments of the EBNA1 gene were phosphorylated using T4 polynucleotide kinase and ligated together. The ligation products were amplified by PCR using the 452 and 641 primers listed above, which generates an NdeI site at the N terminus and a BamHI site at the C terminus. PCR products were digested with NdeI and BamHI and inserted between the NdeI and BamHI sites of expression vector pET15b (Novagen), downstream of a 6-histidine tag. The EBNA1 fragment in this construct was sequenced and confirmed to be correct except that amino acid 465 was converted to a serine instead of an alanine.
Protein Expression and Purification-bEBNA1 was produced in SF-9 insect cells and purified to homogeneity as described previously (9). EBNA 459 -619 , EBNA 463-607 , and EBNA 468 -607 were produced in E. coli and purified to homogeneity as described previously (16,22). For expression of EBNA WF , the pET15b expression construct was used to transform the BL21(DE3) pLysS strain of E. coli (23), and transformants were grown at 37°C in 2 liters of Super Broth supplemented with 100 g/ml ampicillin and 34 g/ml chloramphenicol until the A 600 was 0.6 -0.8. Protein expression was induced with 1 mM isopropyl-␤-Dthiogalactopyranoside, and cells were harvested 3 h postinduction. Cell pellets were rinsed in phosphate-buffered saline and then stored at Ϫ70°C. For protein purification, cells were thawed in 10 ml of lysis buffer (50 mM Tris-HCl, pH 7.5, 10% sucrose, 350 mM NaCl, 1 mM EDTA, 1 mM benzamidine and 1 mM PMSF) and sonicated to reduce the viscosity. Lysates were clarified by centrifugation at 100,000 ϫ g for 30 min, and the supernatant was applied to a DE52 column containing 4 g of resin equilibrated in lysis buffer. The DE52 flow-through was diluted with 50 mM HEPES, pH 7.5, to a final NaCl concentration of 200 mM and then loaded onto a 20-ml heparin-agarose column equilibrated with buffer A (50 mM HEPES, pH 7.5, 200 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM dithiothreitol, 1 mM PMSF, 1 mM benzamidine). The column was washed with buffer A and then developed with buffer A containing 750 mM NaCl. The eluted protein was dialyzed against buffer B (50 mM HEPES, pH 7.5, 750 mM NaCl, 10% glycerol, 1 mM PMSF, 1 mM benzamidine) and loaded onto a 1-ml metal chelating column charged with nickel and equilibrated in buffer B plus 5 mM imidazole. The column was washed with buffer B plus 5 mM imidazole and then with buffer B plus 50 mM imidazole. The 6-histidine-tagged EBNA WF was eluted with buffer B plus 300 mM imidazole and was dialyzed against 50 mM HEPES, pH 7.5, 500 mM NaCl, 10% glycerol. CaCl 2 was added to 2.5 mM, and the 6-histidine tag was removed from EBNA WF by thrombin digestion (10 units/mg protein) overnight at 4°C followed by dialysis against 50 mM HEPES, pH 7.5, 500 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM dithiothreitol.
Permanganate Assays-Permanganate assays were performed as described previously (18). Briefly, EBNA1 proteins were incubated with 0.5 g of supercoiled pGEMdyad, pGEMs1, pCS, or pCSA/T for 10 min at 37°C in 20 l of 50 mM HEPES, pH 7.5, 250 mM NaCl, 5 mM MgCl 2 . KMnO 4 was then added to a final concentration of 20 mM, and, after a 4-min incubation at 37°C, the reaction was quenched, the protein was removed, and the plasmid DNA was heat-denatured. 32 P-End-labeled oligonucleotide primers were then hybridized to the denatured plasmids and extended with the Klenow fragment of DNA polymerase I to detect the position of any oxidized thymine residues, which act as a block to the polymerase. The extension products were separated on 8% polyacrylamide/50% urea sequencing gels and visualized by autoradiography. The positions of the oxidized thymines were determined by comparison with dideoxy sequencing ladders generated from the same primers used for the permanganate assays.
Methylation Protection Footprints-EBNA1 proteins were incubated with 0.5 g of supercoiled pGEMdyad, pGEMs1, pCS, or pCSA/T in 20 l of 20 mM Tris-HCl, pH 7.5, 250 mM NaCl, 5 mM MgCl 2 for 10 min at 37°C prior to the addition of 10 mM dimethyl sulfate (DMS). After a 5-min incubation at 37°C, the reactions were stopped, phenol was extracted, and the DNA was cleaved with piperidine as described previously (18). The DNA fragments generated were denatured and hybridized to the same primers used in the permanganate assays, and the primers were extended with the DNA polymerase I Klenow fragment as for the permanganate assays. Extension products were separated on an 8% polyacrylamide/50% urea sequencing gel and visualized by autoradiography.

EBNA1 Amino Acids Required for DS DNA Distortion-
Structural changes in the DS element of oriP, which can be detected by permanganate oxidation, are induced by wild type and internally deleted versions of EBNA1 (15,18,19). We wished to determine if the DNA binding and dimerization domains of EBNA1 were sufficient to induce this DNA distortion or if other EBNA1 domains were required. To this end, we compared the permanganate reactivity patterns of the DS element when bound by EBNA 459 -619 , which contains only the DNA binding and dimerization domains, with that induced by a biologically active version of EBNA1 (bEBNA1; Fig. 1). As shown in Fig. 2, bEBNA1 and EBNA 459 -619 generated identical permanganate reactivity patterns when titrated onto the DS element. Both proteins caused one Thy residue within EBNA1 binding site 1 and one Thy residue within EBNA1 binding site 4, on the opposite DNA strand, to become oxidized by KMnO 4 . For simplicity, results are shown for one DNA strand only (detecting the oxidized T in site 1); however, in all cases where permanganate sensitivity was detected in site 1, it was also detected in site 4 (data not shown).
We have previously shown that amino acids 461 to 469 from the EBNA1 flanking DNA binding domain forms an extended amino acid chain that travels along the minor groove of the EBNA1 recognition site (20) (see Fig. 3). An EBNA1 truncation mutant lacking this extended chain (EBNA 468 -607 ) retains the ability to bind to high affinity EBNA1 recognition sites, such as sites 1 and 4, but is severely impaired in its ability to bind to lower affinity sites, such as sites 2 and 3 of the DS element (16).
To determine if this interaction with the minor groove of the DNA causes the distortion that results in permanganate sensitivity, we repeated the permanganate assays using EBNA 468 -607 and EBNA 463-607 , which lack all or part of the extended chain, respectively. The results in Fig. 2A show that EBNA 463-607 , but not EBNA 468 -607 , causes the permanganate reactivity pattern seen with bEBNA1. The same results were obtained for site 4 on the opposite DNA strand (data not shown). These results suggest that the minor groove-extended chain of EBNA1 is responsible for the permanganate sensitivity.
To verify that the failure of EBNA 468 -607 to induce permanganate reactivity was not due to failure to bind the DS element, we performed methylation protection footprints with EBNA 468 -607 on the DNA templates used for the permanganate experiments. The results in Fig. 2B show that all of the EBNA1 proteins, including EBNA 468 -607 , bound to the DS element at the protein concentrations used in the permanganate experiments. For bEBNA1, EBNA 459 -619 , and EBNA 463-607 , simultaneous filling of the four recognition sites of the DS was observed, at 10 to 30 pmol of protein for both bEBNA1 and EBNA 459 -619 and at 90 pmol for EBNA 463-607 . This is the same pattern of filling of the DS that has been reported previously for bEBNA1 (18). In this assay, binding to each of the four binding sites is indicated by the lightening of the bands indicated by the arrows in Fig. 2B. Binding to site 4 is also detected by the appearance of the band indicated by the asterisk, which represents a DMS-hypersensitive adenine. Unlike the larger EBNA1 proteins, EBNA 468 -607 was observed to bind only to sites 1 and 4 of the DS (at 90 -270 pmol of protein); interactions with sites 2 and 3 were not detected at any of the protein concentrations tested.
DNA Requirements for EBNA1-induced DS Distortion-The failure of EBNA 468 -607 to induce permanganate reactivity within sites 1 and 4 of the DS element could be interpreted in either of two ways. First, it might indicate that EBNA1 sequences between amino acid 463 and 468 are required to distort sites 1 and 4. Second, it might indicate that binding to sites 2 and/or 3 is required for the induction of permanganate reactivity within sites 1 and 4 since EBNA 468 -607 was the only EBNA1 protein tested that did not bind sites 2 and 3. To distinguish between these possibilities, we repeated the permanganate assays on DNA templates that contained only EBNA1 binding site 1 (pGEMs1). The results in Fig. 4 show that the binding of EBNA 459 -619 and EBNA 463-607 to site 1 alone is sufficient to cause permanganate oxidation of the same Thy residue that is reactive in the complete DS sequence. As was the case with the DS DNA templates, induction of perman- ganate reactivity within site 1 with the pGEMs1 template was not induced by EBNA 468 -607 . These results indicate that EBNA1 sequences between amino acids 463 and 468 are required to distort site 1 and that binding to additional sites in the DS is not required for this distortion. This DNA distortion is independent of the supercoiling of the DNA template since the same permanganate-sensitive T was induced in site 1 when experiments were performed on linear, as opposed to supercoiled, DNA templates (Fig. 4A).
The methylation protection footprints shown in Fig. 4B verify that all of the EBNA1 proteins tested, including EBNA 468 -607, bound to site 1 at the concentrations used in the permanganate reactions. The degree of protection of the diagnostic band in site 1 (arrow) by EBNA 468 -607 was somewhat less than that seen for the larger EBNA1 proteins; however, the intensity of this band was still reduced compared with that seen in the absence of protein (in the absence of protein the diagnostic band is equal in intensity to the band just above it; with EBNA 468 -607 it is less intense than the band above it). These results suggest that EBNA 468 -607 may bind less stably to site 1 than the other EBNA1 proteins.
The Role of Trp-464 and Phe-465 in Inducing the Permanganate Reactive T in Site 1-The crystal structure of the EBNA1 DNA binding and dimerization domains bound to DNA revealed a peculiar arrangement of the Trp-464 and Phe-465 side chains within the minor groove of the DNA (20) (see Fig. 3). The aromatic rings of these residues are oriented parallel with the minor groove and appear to be pushing on the two sugar phosphate backbones. The minor groove of the DNA is widened by 2-3 Å at this point (20). To determine whether Trp-464 and Phe-465 are responsible for generating the DNA disortion that leads to permanganate reactivity within site 1, we repeated the permanganate assays using a version of EBNA1 that contains the complete DNA binding and dimerization domains in which Trp-464 and Phe-465 have been changed to alanine and serine residues, respectively (EBNA WF ; see Fig. 1). EBNA WF was shown by electrophoretic mobility shift assays to bind site 1 with a 10 -20-fold reduced affinity relative to the same EBNA1 fragment containing the wild-type sequence. 2 When EBNA WF was titrated onto pGEMs1, it was observed to induce permanganate oxidation of the same Thy residue as that seen with wild-type EBNA1 (Fig. 5). The amount of protein required for this reactivity (30 -90 pmol) corresponded to the amount of protein required to bind site 1, as determined by methylation protection footprints perfomed on the same DNA template (Fig.  5). Therefore, the aromatic side chains of Trp-464 and Phe-465 are not responsible for the DNA distortion that leads to permanganate reactivity within site 1.
EBNA1 Recognition Site Sequence Requirements for EBNA1induced Permanganate Sensitivity-The results described above show that EBNA1 binding to some individual recognition sites causes a change in DNA structure leading to permanganate sensitivity. Ultimately, we would like to determine the nature of the structural change in the DNA that results in permanganate reactivity. Previously we solved the structure of the EBNA1 DNA binding domain bound to a single EBNA1 binding site (20). This binding site is a palindromic consensus sequence (Fig. 6) to which EBNA1 binds with high affinity but which does not actually exist in oriP. In the EBNA1-consensus DNA structure, there was no disruption of the hydrogen bonds between the bases of the two DNA strands nor was there any other obvious distortion of the DNA structure that might be expected to cause permanganate sensitivity. However, since little is known about the DNA structural requirements for permanganate oxidation, we tested the EBNA1-induced permanganate sensitivity of the consensus EBNA1 binding site. For these experiments, the consensus 18-bp binding site shown in Fig. 6 was cloned into pBluescriptKS to generate pCS. When bEBNA1 and EBNA 459 -619 were titrated onto pCS, binding to the consensus site was detected by DMS footprint analysis at as little as 10 pmol of protein (Fig. 7B), but no EBNA1-induced permanganate reactivity was detected at any amount of protein tested (10 -90 pmol) (Fig. 7A). Some degree of permanganate reactivity was observed for Thy residues at positions 2 and Ϫ1 within the consensus sequence, but this sensitivity was independent of EBNA1.
A comparison of the sequences of the EBNA1 binding sites in the origin and the consensus EBNA1 binding site is shown in Fig. 6. For both of the EBNA1 binding sites that have been 2 A. Fleming, and L. Frappier, unpublished data. shown to be permanganate sensitive (sites 1 and 4 of the DS element), the permanganate-sensitive T residue is located at position 6, relative to the axis of symmetry of the 18-bp palindromic sites (see Fig. 6). Of the 24 EBNA1 bindings sites in oriP (4 in the DS and 20 in the FR element), only sites 1 and 4 of the DS have a Thy in this position. For the other EBNA1 binding sites (and for the other half of the imperfect palindrome in sites 1 and 4), there is an Ade at position 6 (and a Thy at position Ϫ6) i.e. an inversion of the AT base pair relative to the permanganate-sensitive Thy in sites 1 and 4. The lack of permanganate reactivity of the Thy residues at position Ϫ6 suggests that inversion of the AT base pair at position 6/Ϫ6 affects the structure of the EBNA1-bound DNA. To determine if the inversion of this AT base pair is sufficient to elicit the EBNA1induced structural change, we constructed a perfect palindromic EBNA1 binding site that is identical to the consensus site except that the AT base pair at position 6/Ϫ6 in both halves of the palindrome has been inverted (Fig. 6, AT inversion). When bEBNA1 and EBNA 459 -619 were titrated with a plasmid containing this sequence, both proteins were observed to induce permanganate oxidation of the Thy at position 6 ( Fig. 7). Therefore, the inversion of the AT base pair at position 6/Ϫ6 of the EBNA1 binding site enables the structural change that accompanies EBNA1 binding. DISCUSSION Permanganate oxidation is commonly used to detect alterations in the structure of DNA helices, but the structural requirements for permanganate sensitivity are not well understood. While double-strand DNA is largely insensitive to permanganate oxidation, pyrimidines (particularly Ts) in sin-gle-strand DNA are oxidixed at the 5,6-double bond by this reagent to form pyrimidine glycols (24,25). Therefore permanganate is useful for the detection of melted regions of doublestrand DNA (26). In some cases, however, permanganate-sensitive residues have been detected in regions of double-strand DNA that, by other criteria, are not melted. For example, the lac repressor (27), SV40 large T antigen (28), repressor activator protein 1 (RAP1) (29), and EBNA1 (18,19) have all been reported to induce localized structural changes in the DNA to which they bind, resulting in the increased sensitivity of one or more T residues to permanganate oxidation. The nature of the structural changes that cause these permanganate sensitivities are not clear.
Although EBNA1 does not appear to melt oriP DNA, several observations suggest that the interaction of this protein with its recognition sites in the DS element of oriP structurally alters the DNA helix. First, electron microscopic studies of EBNA1 on the DS element show pronounced bending of the DS DNA at the position where the four EBNA1 dimers are assembled (17). Second, modeling studies using the EBNA1-DNA co-crystal structure suggest that the cooperative assembly of EBNA1 dimers on adjacent binding sites in the DS must be accompanied by a change in structure of the DNA, most likely unwinding and/or unbending of the DNA (20). Third, EBNA1 binding to the DS element, both in vivo and in vitro, has been shown to induce permanganate sensitivity in one Thy residue in each of the two outer EBNA1 binding sites (15,18,19).
The studies presented here were conducted to better understand the nature of and the requirements for the EBNA1induced distortion of the DS element that results in permanganate sensitivity. We have shown that this permanganate sensitivity occurs on single EBNA1 binding sites, and therefore the DNA distortion detected by permanangate is distinct from that predicted by the modeling studies to accompany cooperative assembly of EBNA1 on two adjacent sites of the DS. This conclusion is consistent with the results of Harrison et al. (14), who showed that point mutations in sites 2 or 3 in the DS element did not abrogate the EBNA1-induced permanganate sensitivity of sites 1 and 4 and that this sensitivity was independent of the spacing between the EBNA1 binding sites. In their experiments, Harrison et al. (14) used the full-length EBNA1 protein, which contains a domain that mediates interactions at a distance between DNA-bound EBNA1 molecules (21, 30 -32). Therefore, although the cooperative filling of adjacent sites was disrupted by virtue of their DNA templates, it is likely that interactions occurred between EBNA1 molecules on distant sites (i.e. between sites 1 and 4, 2 and 4, or 1 and 3, and between different DNA molecules). Since EBNA1 interac- tions at a distance might have substituted for the interaction between EBNA1 molecules on adjacent sites in inducing permanganate sensitivity, we felt that it was important to test permanganate sensitivity under conditions in which neither type of EBNA1 interaction was possible (i.e. on individual binding sites using EBNA1 proteins that lack the domain that mediates interactions at a distance).
We have explored the EBNA1 amino acids required to induce permanganate reactivity and found that amino acids 463-607, containing the DNA binding and dimerization domains of EBNA1, are sufficient to elicit this response. Deletion of part of the flanking DNA binding domain that forms an extended chain positioned along the base of the minor groove of the EBNA1 binding site (amino acids 463-467) abrogated the ability of EBNA1 to induce permanganate reactivity. We have previously shown, using electrophoretic mobility shift assays, that this deletion had little effect on the ability of EBNA1 to bind to site 1 from the DS element (K d for EBNA 463-607 and EBNA 468 -607 was 110 and 125 nM, respectively) (16). The methylation protection footprints presented here also confirm that EBNA 468 -607 binds site 1, both in the context of the DS element and in isolation. However, the weaker footprint generated by this protein suggests that EBNA 468 -607 binds less tightly to site 1 than the larger versions of EBNA1 tested. Therefore, the requirement for amino acids 463-467 for permanganate reactivity could either reflect a specific requirement for the extended chain in the minor groove of the DNA or a requirement for tight binding. We favor the former possibility because other mutations in the EBNA1-flanking DNA binding domain that reduce binding to site 1 still induce permanganate sensitivity. An example of this is the EBNA WF mutant. This mutant gives a weaker footprint on site 1 (Fig. 5) and binds site 1 with an affinity that is approximately 15-fold lower than the same fragment of EBNA1 with the wild-type sequence (as determined by electrophoretic mobility shift assays), 2 yet EBNA WF still induces permanganate reactivity in site 1.
Investigation of the DNA sequence requirements for EBNA1induced permanganate reactivity revealed that the differences in permanganate sensitivity of different EBNA1 binding sites is soley due to the inversion of a single AT base pair at position 6/Ϫ6 relative to the axis of dyad symmetry of the palindromic binding site. Permanganate sensitivity is only seen when the Thy residue is at position 6, generating the sequence CCCT-TCGTA. This suggests that EBNA1 binding to this sequence causes a distortion in the DNA structure that does not occur when EBNA1 binds to recognition sites containing the sequence CCCATCGTA. This distortion might be facilitated by the 6-bp polypurine/polypyrimidine tract that is generated by the AT inversion since polypurine/polypyrimidine tracts have a tendency to adopt alternative DNA structures (33). Alternatively, the structures of the EBNA1-bound CCCTTCGTA and CCCATCGTA binding sites might be the same, and the Thy residue at position Ϫ6 might simply be inaccessible to the permanganate reagent due to postioning of EBNA1 amino acids. These two alternative interpretations can only be resolved by solving the high resolution structure of the EBNA1 DNA binding and dimerization domains on binding sites that are and are not permanganate-sensitive.
We have already solved the structure of the EBNA1 DNA binding and dimerization domains on the consensus binding site that is not permanganate sensitive (20). In keeping with the results of the permanganate assays on the consensus site, distortion of the DNA helix at position 6/Ϫ6 was not apparent from this structure. We have also solved the structure of the EBNA1 DNA binding domain on the permanganate-sensitive In panel B, bands from the two DNA templates that become protected by EBNA1 are indicated by the arrow. site 1 at 2.8 Å resolution. 3 The structure of the site 1 DNA was indistinguishable from that of the consensus site, however the particular packing arrangement of EBNA1 in the crystal form used to solve the structure would permit the EBNA1-DNA complex to crystallize in either of two orientations. As a result, the final structure of the site 1 DNA might represent an average of the two halves, and hence even moderate structural perturbations on one side of the pseudopalindromic DNA may not be revealed. To conclusively determine the structure of EBNA1-bound permanganate-sensitive DNA, we require a palindromic permanganate-sensitive EBNA1 binding site in which the structural distortion would be manifested on both halves of the palindrome. We have shown that we can generate such a binding site by inverting the AT base pair at position 6/Ϫ6 in both halves of the consensus palindrome. This finding should facilitate the structural determination of permanangate-sensitive DNA bound by EBNA1, which in turn will help us to understand the structural requirements for permanganate oxidation.