INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The Epstein-Barr virus (EBV) genome is maintained as an autonomously replicating, multicopy plasmid in most latently infected cells (9, 20, 27). Each plasmid is replicated once per cellular S phase, and the progeny plasmids are equally partitioned to daughter cells during mitosis (1, 38). Replication and maintenance of the genome during latency require three viral components: an origin of DNA replication, a DNA maintenance element (the family of repeats [FR]), and the EBV nuclear antigen 1 (EBNA-1) (reviewed in reference 45). FR is composed of 21 imperfect copies of a 30-bp repeat, 20 of which contain an EBNA-1 binding site (31). In addition to a dimerization and DNA binding domain, EBNA-1 contains multiple elements that mediate binding to mitotic host cell chromosomes (15, 16, 24, 43). It is believed that EBNA-1 performs an essential function in viral genome maintenance through simultaneously binding the reiterated sites in FR and host cell chromosomes, thereby tethering the EBV genome to the host cell chromosomes during cell division (15, 16, 24, 43). The analysis of replicative intermediates by electron microscopy and two-dimensional agarose gel electrophoresis revealed that latent cycle DNA replication can initiate at multiple sites on the EBV chromosome (8, 9, 21, 28). One of these origins is located close to FR, and, together, these two elements are referred to as oriP (Fig. 1) (8, 44). Recombinant plasmids bearing a 2.2-kb fragment of EBV DNA encompassing oriP are replicated in a cell cycle-regulated manner and maintained efficiently in cells expressing EBNA-1 (47, 48). Thus, small oriP plasmids provide an experimentally tractable system for studying EBV latent-cycle DNA replication and genome maintenance. Because all of the proteins that are required for the replication of oriP plasmids, with the exception of EBNA-1, are provided by the host cell, elucidation of the mechanism by which replication initiates at oriP may provide insight into the nature of mammalian origins of DNA replication.

View larger version (7K): [in this window] [in a new window]   FIG. 1.   Organization of EBV oriP. The EBV DNA (nt 7333 to 9516) present in pHEBo-1 and pHEBo-1.1 and the locations of FR and DS are shown. Replication initiates within, or close to, DS (8). The arrangement of EBNA-1 binding sites (ovals) and repeat elements a, b, and c (arrows) within nt 8996 to 9137 are shown at the bottom.

The replicator of oriP is located ca. 1 kb from FR and is referred to as DS due to the presence of a 65-bp dyad symmetry element (10, 28, 32, 34, 46). DS contains four EBNA-1 binding sites arranged as two pairs with 21-bp center-to-center spacing and contains, or is close to, the site at which DNA replication initiates (Fig. 1) (8, 31). Plasmids bearing DS are replicated in an EBNA-1-dependent manner but are not stably maintained in the absence of FR (10, 34, 46). The observation that EBNA-1 differs from eukaryotic viral replication initiator proteins in that it lacks the ability to unwind DNA has led to the hypothesis that EBNA-1 participates in the recruitment of host cell factors required for the initiation of DNA replication from within oriP (7, 11). EBNA-1 interacts with a human single-stranded DNA binding protein (RPA) and a 32-kDa acidic host cell protein (p32/TAP) that may function in latent-cycle DNA replication, but interactions with a DNA helicase or other proteins required for DNA replication have not been reported (41, 42, 49). Data from our previous investigation of the EBNA-1 binding sites within DS showed that the EBNA-1 binding sites function as pairs and that replicator activity was not abolished by the inactivation of one pair of these binding sites (10). This raised the question of why the replicator of the EBV isolate used in our studies (B95-8) contains two pairs of EBNA-1 binding sites. To address this question, we utilized a quantitative short-term replication assay to assess the impact of inactivating individual and paired EBNA-1 binding sites within the oriP replicator. We found that all four binding sites are required for full replicator activity in an EBV-positive Burkitt's lymphoma cell line. These results are in agreement with the recently published findings of Yates et al. (46), and our studies further show that the paired binding sites are not functionally equivalent. We also measured the effects of deletions within oriP on replication efficiency, and the results suggest that interactions between EBNA-1 and a host cell protein bound to repeats flanking the EBNA-1 binding sites within DS are required for minimal replicator activity. Finally, we reinvestigated the role of sequences to the right of EBNA-1 binding site 1 and determined that one or more elements within this region are required for full oriP replicator function.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. pHEBo-1 and pHEBo-1.1, plasmids bearing EBV (B95-8 strain) nucleotides (nt) 7333 to 9516 and a hygromycin resistance gene (hph) expression cassette, were described previously (11, 37). pHEBo-1.19138-9516 was derived from pHEBo-1.1 by deleting EBV nt 9138 to 9516 and 56 nonessential nucleotides from the 3' end of the hph gene by partial digestion with BsaI followed by complete digestion with HpaI. The hph gene was restored upon ligation of a 471-bp PCR product that contained the appropriate hph sequences followed by BamHI and HpaI sites. The resulting plasmid contains a BamHI site immediately following EBV nt 9137. pHEBo-1.19138-9459 was constructed using the same basic strategy; however the PCR product ligated to the digested pHEBo-1.1 fragment contained EBV nt 9459 to 9516. pHEBo-1.19138-9459+M13 was created by introducing a 683-bp BamHI-BglII fragment from M13mp18 at the BamHI site of pHEBo-1.19138-9459. pHEBo-2.2 was derived by deleting sequences (EBV nt 8996 to 9135) between the EcoRV and HpaI sites of pHEBo-1.19138-9516. pHEBo-2.2.1 was constructed by substituting the EcoRV-HpaI fragment containing DS sequences from pHEBo-1.1(dpm1+2) (10) for the small EcoRV-HpaI fragment of pHEBo-1.19138-9516. pHEBo-2.2.5, -2.2.6, -2.2.7, and -2.2.8 were created by annealing oligonucleotides with 3' complementary ends, synthesizing the complementary strands with the Klenow fragment of Escherichia coli DNA polymerase I, digesting the double-stranded products with BamHI and/or BglII, and inserting these fragments at the BamHI site of pHEBo-2.2. pHEBo-2.2.20 and -2.2.24 were generated by PCR amplification of DS sequences using pHEBo-1.1 as the template, digesting the products with BamHI and BglII (pHEBo-2.2.20) or EcoRV and BglII (pHEBo-2.2.24) and inserting the products at the BamHI site of pHEBo-2.2 (pHEBo-2.2.20) or between the EcoRV and BamHI sites of pHEBo-1.19138-9516 (pHEBo-2.2.24). pHEBo-1.1(dpm1), pHEBo-1.1(dpm2), pHEBo-1.1(dpm3), pHEBo-1.1(dpm4), pHEBo-1.1(dpm1+2), pHEBo-1.1(dpm3+4), pHEBo-1.1(in1/2[5]), pHEBo-1.1(in2/3[5]), and pHEBo-1.1(in3/4[5]) were described previously (10). pHEBo-1.1(dpm3+4;2=HAS) and pHEBo-1.1(dpm1+2;3=HAS) contain mutations that are predicted to increase the affinity of EBNA-1 for sites 2 and 3, respectively (2). Base pair transitions were introduced at the 3, +3, and +8 positions in site 2 in pHEBo-1.1(dpm3+4) or the 4 and +8 positions in site 3 in pHEBo-1.1(dpm1+2) by oligonucleotide-directed mutagenesis as previously described (11). pBS/KS[BclI] was made by inserting a BclI linker into the unique XhoI site of pBluescript KS (Stratagene). A SalI-EcoRV fragment from pHEBo-1.1 containing EBV nt 7333 to 8994 was introduced between the same sites in the polylinker region of pBS/KS[BclI] to create pBS/FR. Similarly, a SalI-BamHI fragment from pHEBo-1.19138-9516 containing EBV nt 7333 to 9137 was inserted between the same sites in the polylinker of pBS/KS[Bcl] to create pBS/FR.DS. The EBV nucleotides present in each of the plasmids created for this study are given in Table 1. Plasmid DNA was isolated from dam⁺ E. coli DH-1 or DH-5 cells and purified by isopycnic centrifugation on CsCl-ethidium bromide gradients. All plasmids were sequenced after purification to confirm their identity and to confirm that the intended mutations were the only mutations present within DS.

Sequence analysis of oriP from multiple EBV strains. Whole-cell DNA was prepared from Raji (30), Namalwa (29), FF41 (6), Jijoye (12), and Akata (40) cells as previously described (23). Jijoye and Akata cells were induced to enter the EBV lytic cycle by adding either goat anti-human immunoglobulin M (IgM; Jijoye) or anti-human IgG (Akata) antibodies (HyClone Laboratories, Inc.) to the growth medium at a final concentration of 1%. Whole-cell DNA was isolated after 24 h of incubation in the presence of inducing antibodies. A 782-bp fragment of the EBV genome containing oriP DS and flanking sequences (B95-8 nt 8587 to 9363) was amplified by PCR using primers containing EcoRI and HindIII sites. The sequence between nt 8976 and 9142 was determined by direct sequencing of the PCR products or by sequencing pBluescript KS derivatives containing the amplified sequences.

Short-term replication assays. Raji cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (HyClone Laboratories) and 100 U each of penicillin and streptomycin per ml at 37°C. Sixty-one micrograms each of the reference (pHEBo-1 or pKS/oriP) and test plasmid DNAs purified from dam⁺ E. coli were combined, precipitated with ethanol, and dissolved in 61 µl of 10 mM Tris-HCl (pH 8.0), 1 mM EDTA (TE). One microliter was reserved for determination of the ratio of test to reference plasmid DNA introduced into cells by quantitative Southern blot analysis. Logarithmic-phase Raji cells were washed with cold growth medium and suspended at 4 × 10⁷ cells per ml in cold growth medium. Electroporations were performed in triplicate by mixing 10⁷ cells with 20 µl of the combined test and reference plasmid DNA. This mixture was transferred to a 0.4-cm-wide cuvette and electroporated with a Bio-Rad Gene Pulser set at 250 V and 960 µF. Cells were returned to culture in 20% conditioned-80% fresh medium and maintained in an actively dividing state for 72 h (three population doublings), at which time low-molecular-weight DNA was isolated from 2 × 10⁷ to 5 × 10⁷ cells by the method of Hirt (13). Briefly, cells were washed twice with phosphate-buffered saline, suspended in 2 ml of 10 mM Tris-HCl (pH 8.0)-10 mM EDTA, mixed gently with an equal volume of 10 mM Tris-HCl (pH 8.0)-10 mM EDTA-1.2% sodium dodecyl sulfate and incubated for 20 min at room temperature. High-molecular-weight DNA was precipitated upon the addition of 800 µl of 5 M NaCl and incubation at 4°C for 12 to 16 h and pelleted by centrifugation at 27,000 × g for 30 to 45 min at 4°C. The supernatant was treated with 40 µg of RNase A per ml for 1 h at 37°C and, subsequently, 10 µg of proteinase K per ml at 37°C for 1 h. The samples were extracted once with phenol-chloroform-isoamyl alcohol (25:24:1) and once with chloroform-isoamyl alcohol (24:1), and the DNA was precipitated with ethanol. The DNA was pelleted by centrifugation at 25,000 × g for 45 min at 0°C, dissolved in 0.3 M sodium acetate (pH 7.5), and reprecipitated with ethanol. The precipitates were dissolved at 2 × 10⁵ cell equivalents per µl in TE containing 20 µg of RNase A per ml.

Determination of replication efficiencies. Southern blot quantitation was performed with a Molecular Dynamics Storm 860 phosphorimager. The replication efficiencies of test plasmids were calculated as the ratios of replicated test plasmids to the replicated reference plasmid (pHEBo-1 or pKS/oriP). These ratios were adjusted for the relative amounts of the two plasmids present at the time of electroporation and normalized to the amount of pHEBo-1.1, which was set at 100%. The efficiency with which plasmids replicated per cell generation was calculated using the formula N_t/N_r = [(1 + )ⁱ  (1  )ⁱ]/2ⁱ, where N_t/N_r is the normalized ratio of replicated test plasmid molecules to replicated reference plasmid molecules after i cell generations and  is the replication efficiency of the test plasmid per cell generation. This calculation assumes that the test plasmid is maintained in dividing cells with the same efficiency as the reference plasmid and that plasmids that are not replicated in one cellular S phase may be replicated in subsequent S phases. It also takes into account the elimination of plasmid DNA that is never replicated by digestion with DpnI. Statistical analysis of the short-term replication assay data was performed by one-way analysis of variance. Tukey's "honestly significant difference" test was used to test pairwise comparisons between the means, and P values of <0.05 were considered significant.

Long-term plasmid replication and maintenance assays. The efficiency of transformation of Raji cells to hygromycin resistance by pHEBo-1.1 and its derivatives was determined by electroporating 10⁷ cells with 2 µg of test plasmids as described above. Forty-eight hours after electroporation, cells were plated in 24-well dishes in growth medium supplemented with 300 µg of hygromycin B (Boehringer Mannheim) per ml at 10⁴, 10³, and 10² cells per well as previously described (46). Fresh medium containing 300 µg of hygromycin B per ml was added to the wells every 6 to 7 days, and the number of wells containing clones was determined 21 days after electroporation. Transformation frequencies were derived by applying the Poisson distribution (36). Two clones which were derived by plating the smallest number of electroporated cells per well were expanded under selection until 4.6 × 10⁶ to 1.0 × 10⁷ cells were available for analysis. Low-molecular-weight DNA was isolated as described above and was analyzed as either uncut or digested molecules by Southern blotting. Blots were probed with ³²P-labeled pHEBo-1.1 (see Fig. 5) or the large SacII-EcoRV fragment of pHEBo-1.1 (see Fig. 3).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Conservation of sequences in oriP DS. Our previous analysis of EBNA-1 binding sites within oriP DS demonstrated that replicator function was not eliminated by inactivation of two sites provided the remaining sites were separated by 21 bp (10). However, four EBNA-1 binding sites, arranged as pairs with 21-bp center-to-center spacing, are present not only within the B95-8 strain oriP replicator used in our studies but also within the Raji and C15 EBV genomes (3, 14, 26, 31, 35). To determine if EBV isolates with fewer than four oriP DS EBNA-1 binding sites exist and to extend the comparisons further into flanking sequences, EBV (B95-8 isolate) nt 8587 to 9363 from four EBV isolates were amplified by PCR and the sequence of nt 8977 to 9142 was determined. These viral sequences were obtained from three Burkitt's lymphoma cell lines (Raji, Jijoye, and Akata) (12, 30, 40) and a marmoset cell line established by in vitro immortalization of B cells with virus from an infectious mononucleosis patient (FF41) (6). Two EBV types (EBV-1 and -2) have been distinguished (33), and both type 1 (B95-8, Raji, FF41, and Akata) and type 2 (Jijoye) viruses were included in this comparison.

View larger version (52K): [in this window] [in a new window]   FIG. 2.   Sequence of oriP DS and flanking nucleotides from type 1 and type 2 EBV isolates. EBV nucleotides 8977 to 9142 from the B95-8 strain and the same sequence from the Raji, Jijoye, Akata, FF41, and C15 EBV strains are shown. The four EBNA-1 binding sites (boxes) and nonamer repeats a, b, and c (arrows) are indicated. Differences between the B95-8 sequence and the other viral sequences are underlined. The sequences from the B95-8 and C15 viral genomes were previously reported (3, 35).

Contribution of individual EBNA-1 binding sites in DS to replicator activity. The results of this survey suggested that all of the EBNA-1 binding sites within DS, although individually not essential for replicator function, contribute to its activity. Two assays were performed to determine if derivatives of an oriP-bearing plasmid (pHEBo-1.1) with disabling mutations within individual or paired EBNA-1 binding sites in DS are replicated less efficiently. First, the frequencies with which pHEBo-1.1 and mutant derivatives transformed Raji cells to hygromycin resistance were determined (37, 46). The results of two independent experiments are shown in Table 2. No hygromycin-resistant clones were obtained with a pHEBo-1.1 derivative that lacks oriP DS (pHEBo-2.2) at 10⁴ cells/well, yielding a transformation frequency of <3 × 10⁶, while pHEBo-1.1 transformed Raji cells to hygromycin resistance with a frequency of 3 × 10² (experiment 1) to 2 × 10³ (experiment 2). This high transformation efficiency is due to the ability of pHEBo-1.1 to be maintained as an autonomously replicating plasmid in cells expressing EBNA-1 and is similar to previously reported frequencies (37, 46). Each of the plasmids containing inactivating mutations within one of the four EBNA-1 binding sites transformed Raji cells at reduced frequency compared to pHEBo-1.1, suggesting that all four sites are required for full replicator activity. Because the binding of EBNA-1 to sites 1 and 4 enables binding to sites 2 and 3, respectively (10, 39), we expected pHEBo-1.1 derivatives with disabling mutations within site 1 or site 4 to resemble plasmids with mutations within both sites 1 and 2 or sites 3 and 4 in their ability to transform Raji cells. However, pHEBo-1.1(dpm1+2) and pHEBo-1.1 (dpm3+4) transformed Raji cells at even lower frequencies than pHEBo-1.1(dpm1) and pHEBo-1.1(dpm4) (30- and 300-fold reduction in transformation efficiency, respectively; Table 2).

View larger version (66K): [in this window] [in a new window]   FIG. 3.   Analysis of plasmid DNA in Raji transformants established with pHEBo-1.1 and its derivatives bearing mutations within single and paired EBNA-1 binding sites. Low-molecular-weight DNA isolated from Raji cells (lane 1) or hygromycin-resistant Raji clones transformed by pHEBo-1.1 (lanes 2 and 3), pHEBo-1.1(dpm1) (lanes 4 and 5), pHEBo-1.1(dpm2) (lanes 6 and 7), pHEBo-1.1(dpm3) (lane 8), pHEBo-1.1(dpm4) (lanes 9 and 10), pHEBo-1.1(dpm1+2) (lane 11), or pHEBo-1.1(dpm3+4) (lanes 12 and 13) was analyzed by Southern blotting with radiolabeled pHEBo-1.1 DNA. The samples were either digested with SacII (A) or analyzed without prior restriction enzyme digestion (B). The positions of covalently closed circular (fI), nicked circular (fII), and linear (fIII) pHEBo-1.1 DNA are indicated on the right. Also indicated are the positions of Raji EBV DNA containing oriP that was generated by SacII digestion (A) or mechanical shearing during isolation of low-molecular-weight DNA (B).

View larger version (62K): [in this window] [in a new window]   FIG. 4.   Short-term replication assay of pHEBo-1.1 derivatives containing inactivating mutations within EBNA-1 binding sites. Low-molecular-weight DNA isolated from Raji cells 72 h after electroporation with pHEBo-1 and a test plasmid [pHEBo-2.2 (lane 3), pHEBo-1.1 (lane 4), pHEBo-1.1(dpm1) (lane 5), pHEBo-1.1(dpm2) (lane 6), pHEBo-1.1(dpm3) (lane 7), pHEBo-1.1(dpm4) (lane 8), pHEBo-1.1(dpm1+2) (lane 9), or pHEBo-1.1(dpm3+4) (lane 10)] was digested with DpnI, EcoRV, and PstI (lanes 4 to 10) or DpnI, EcoRV, PstI, and SacI (lane 3) and analyzed by Southern blotting with a probe containing oriP FR. Low-molecular-weight DNA from Raji cells that had not been subjected to electroporation (lane 1) and that had been supplemented with 400 pg of pHEBo-1 DNA from dam+ E. coli (lane 2) was digested with DpnI, EcoRV, and PstI and included in the analysis. The positions of the diagnostic PstI-EcoRV (2,873 bp) and EcoRV-EcoRV (2,276 bp) fragments derived from pHEBo-1.1 and pHEBo-1, respectively, are indicated on the right. Also indicated are two fragments derived from pHEBo-1 and pHEBo-1.1 replicative intermediates containing DpnI-sensitive sites adjacent to oriP FR. EBV (-SacI), 4.4-kb PstI-EcoRV fragment from Raji EBV DNA containing oriP FR. Digestion of Raji EBV DNA with SacI, in addition to EcoRV and PstI, results in a 2.7-kb fragment bearing oriP FR which migrates slightly more rapidly than the pHEBo-1.1 diagnostic fragment (lane 3). This digestion allowed for the quantitation of DpnI-resistant pHEBo-2.2 DNA, which migrates slightly more rapidly than the Raji EBV DNA fragment generated by digestion with EcoRV and PstI (asterisk to the left of lane 3, fragment's expected position).

The sequence context and spatial organization of EBNA-1 binding sites influence their contribution to replicator function. EBNA-1 binds cooperatively to the paired sites within DS in vitro, and double point mutations that eliminate binding to site 4 also eliminate binding to site 3 (10, 39). We therefore expected the replication efficiencies of pHEBo-1.1(dpm4) and pHEBo-1.1(dpm3+4) to be the same. However, as suggested by the results of the Raji transformation assay (Table 2), pHEBo-1.1(dpm4) replicated more efficiently than pHEBo-1.1(dpm3+4) (P < 0.01) and the replication efficiencies of pHEBo-1.1(dpm3) and pHEBo-1.1(dpm4) were the same (P = 1; Table 3). These results suggest that, in vivo, either EBNA-1 can bind sites 3 and 4 independently or the mutations that eliminate EBNA-1 binding in vitro do not completely eliminate binding in vivo. Both scenarios could be explained by interactions between EBNA-1 and a host cell protein(s) bound to neighboring sequences in vivo, and interactions between EBNA-1 dimers could also facilitate the occupation of a crippled site. We expected the relative effects of mutations within EBNA-1 binding sites 1 and 2 to be the same as the effects of mutations in sites 3 and 4 but, instead, observed that the mutations within site 2 had a greater impact on replicator function than mutations within site 1 (P < 0.01). One notable difference between sites 1 and 2 is the presence of nonamer c adjacent to site 1 and the absence of a similarly positioned nonamer repeat next to site 2 (Fig. 1 and 2). Taken together, these results suggest that interactions between EBNA-1 and a host cell protein(s) bound to the nonamer repeats and between EBNA-1 dimers increase the ability of EBNA-1 to bind sites containing inactivating mutations.

View larger version (62K): [in this window] [in a new window]   FIG. 5.   Analysis of plasmid DNA in Raji transformants established with pHEBo-1.1 and its derivatives bearing insertion and point mutations in oriP DS. Low-molecular-weight DNA isolated from Raji cells (lane 1) or hygromycin-resistant Raji clones transformed by pHEBo-1.1 (lanes 2 and 3), pHEBo-1.1(in1/2[5]) (lanes 4 and 5), pHEBo-1.1(in2/3[5]) (lanes 6 and 7), pHEBo-1.1(in3/4[5]) (lanes 8 and 9), pHEBo-1.1(dpm3+4;2=HAS) (lanes 10 and 11), or pHEBo-1.1(dpm1+2;3=HAS) (lanes 12 and 13) were analyzed by Southern blotting with radiolabeled pHEBo-1.1 DNA. The samples were either digested with SacII (A) or analyzed without prior restriction enzyme digestion (B). The positions of covalently closed circular (fI) and linear (fIII) pHEBo-1.1 DNA and Raji EBV DNA containing oriP that was generated by SacII digestion or mechanical shearing during isolation are indicated on the right. Nicked circular plasmid DNA is present in panel B, lanes 2, 3, 6, 7, 12, and 13, between the wells and the largest viral DNA fragments complementary to the probe.

The low affinity of EBNA-1 binding sites 2 and 3 relative to that of sites 1 and 4 is not required for the paired sites to function in the initiation of oriP plasmid DNA replication. We considered the possibility that the arrangement of higher-affinity (sites 1 and 4) and lower-affinity (sites 2 and 3) EBNA-1 binding sites within oriP DS is important for replicator function. To test this hypothesis, we changed site 2 within pHEBo-1.1(dpm3+4) and site 3 within pHEBo-1.1(dpm1+2) to high-affinity EBNA-1 binding sites. Base pair transitions were introduced at positions 4 and +8 within site 3 and at positions 3, +3, and +8 within site 2 (2). No differences in the ability of EBNA-1 to bind the altered sites or the adjacent intact sites in each pair were detected by DNase I footprinting performed with full-length EBNA-1 (data not shown). Both pHEBo-1.1(dpm3+4;2=HAS) and pHEBo-1.1(dpm1+2;3=HAS) were maintained as monomeric circles (Fig. 5) and yielded replication efficiencies in the short-term replication assay (Table 3) similar to those for pHEBo-1.1(dpm3+4) and pHEBo-1.1(dpm1+2) (P = 1). These results indicate that the pairing of higher- and lower-affinity EBNA-1 binding sites is not required for replicator function.

An element(s) located to the right of nonamer c contributes to oriP replicator activity. Previous investigations of the boundaries of oriP relied on the ability of mutated oriP plasmids to replicate and be maintained over many cell generations (4, 32, 44). The reduced copy numbers observed with plasmids lacking the sequence to the right of EBV nt 9134 in several of these studies suggested that sequence elements in addition to EBNA-1 binding sites, though not essential for oriP function, may contribute to replicator activity (see Fig. 1 and 2 for the location of nonamer c and nucleotide coordinates) (32, 44). The oriP deletion derivatives examined in these previous studies lacked part or all of nonamer c in addition to missing all other right-hand EBV sequences. To determine if the sequence to the right of the EBNA-1 binding sites within DS is required for full replicator activity and to distinguish between a requirement for nonamer c and nucleotides to its right, a pHEBo-1.1 derivative lacking all EBV nucleotides to the right of nonamer c was tested for its ability to be replicated and maintained in Raji cells. This plasmid, pHEBo-1.19138-9516, transformed Raji cells to drug resistance with sixfold-reduced frequency compared to pHEBo-1.1, and its copy number in two Raji cell clones was also reduced in comparison to that of pHEBo-1.1 (average of 2.5 copies per cell for pHEBo-1.19138-9516 compared to 7.1 copies per cell for pHEBo-1.1; Table 2). In agreement with these results, the replication efficiency of pHEBo-1.19138-9516 was only 70% per cell generation (Table 4). The replication defect of pHEBo-1.19138-9516 may be due to the loss of an auxiliary element(s) that facilitates the initiation of DNA replication or, alternatively, a negative effect of placing plasmid sequences adjacent to DS (see, e.g., reference 4). To address the latter possibility, the right-hand flanking sequence was changed in two ways. A 683-bp fragment from phagemid M13mp18 was substituted for EBV nt 9138 to 9459 in pHEBo-1.1 and the ability of this plasmid and its parental plasmid lacking M13 DNA (pHEBo-1.19138-9459) to be replicated was determined. The EBV sequences present in pHEBo-1.19138-9516 (nt 7333 to 9137) were also introduced into pBluescript KS, and the replication efficiency of this plasmid (pBS/FR.DS) was compared to that of pBluescript derivatives which contain EBV nt 7333 to 9516 (pBS/oriP) or EBV nt 7333 to 8994 (pBS/FR). pBS/FR replicated very inefficiently, as expected for a plasmid lacking DS. pHEBo-1.19138-9516, pHEBo-1.19138-9459, pHEBo-1.19138-9459+M13, and pBS/FR.DS replicated with the same reduced efficiency (approximately 70 to 80% per cell generation), indicating that the reduced replication efficiency of pHEBo-1.19138-9516 was not likely due to changes in flanking sequences (Table 4). The effect of deleting EBV nt 9138 to 9516 from a plasmid with inactivating mutations within EBNA-1 binding sites 1 and 2 was also examined. If the right-hand sequence contributes to replicator activity independently of EBNA-1 binding sites 1 and 2, the replication efficiency of this plasmid (pHEBo-2.2.1; Fig. 6) would be expected to be lower than the replication efficiency of pHEBo-1.1(dpm1+2). The replication efficiencies of these two plasmids, however, were the same (P = 0.98). Taken together, these results suggest that the deletion of EBV nt 9138 to 9459 either disrupts or eliminates an element(s) required for full activity of the oriP replicator and this putative element functions in conjunction with EBNA-1 binding sites 1 and 2. 

View larger version (17K): [in this window] [in a new window]   FIG. 6.   Replication efficiencies of pHEBo-1.1 and -2.2 derivatives with sequences in oriP DS deleted. The replication efficiencies (Repl. Eff.) of pHEBo-1.19138-9516 and pHEBo-2.2 derivatives lacking sequences in oriP DS were determined in triplicate and are given as percent per cell generation with standard deviations relative to that of pHEBo-1.1. These plasmids all contain oriP FR and the sequences normally present between FR and DS (EBV nt 7333 to 8995; see Fig. 1). The DS-derived sequences present in each plasmid are represented schematically at the right, and the nucleotide coordinates are at the top. Ovals, EBNA-1 binding sites; a, b, and c, nonamer repeats; EBNA-1 binding sites labeled X, sites containing inactivating mutations at the +5 and 5 positions (10). The replication efficiency of pHEBo-1.1 was 102 ± 15%, and that of pHEBo-1.19138-9516 was determined in the experiment presented in Table 4.

Minimal sequence requirements for the oriP replicator. The results of our mutational analysis of EBNA-1 binding sites (10) showed that the paired sites are essential components of the oriP replicator, and data presented here suggest that at least one additional sequence element, located between EBV nt 9138 and 9516, is required to ensure that replication of oriP plasmids will initiate once per cell cycle. To identify the minimal fragment from oriP DS that provides replicator activity and to determine if there are additional, nonessential components that contribute to replicator function, we constructed derivatives of pHEBo-1.19138-9516 with mutations between EBV nt 8995 and 9137 and analyzed their ability to be replicated in the short-term assay. The approach taken was to consider the 14-bp repeats flanking EBNA-1 binding sites 4 and 3 (containing nonamers a and b) and nonamer c potential functional elements and to test the EBNA-1 binding sites in various combinations with the nonamer repeats and other flanking sequences. We chose to focus on sites 3 and 4 because they contribute more to replicator activity than sites 1 and 2 (Tables 2 and 3). The oriP DS sequences present in these plasmids are summarized in Table 1 and are shown schematically in Fig. 6. A plasmid containing EBNA-1 binding sites 3 and 4 but lacking all other sequences from DS and right-hand flanking sequences (pHEBo-2.2.6) resembled the parental plasmid lacking DS (pHEBo-2.2) in its inability to be replicated in Raji cells (P = 1), suggesting that one or more elements, in addition to these EBNA-1 binding sites, are required to constitute minimal replicator activity. Inclusion of either of the 14-bp repeats flanking sites 3 and 4 did not give rise to replicator activity (compare pHEBo-2.2 with pHEBo-2.2.7 and -2.2.8; P = 1), but a plasmid containing EBNA-1 binding sites 3 and 4 and both 14-bp repeats replicated, albeit inefficiently (compare pHEBo-2.2 with pHEBo-2.2.5; P < 0.01). Inefficient replication was also observed when all four EBNA-1 binding sites and the 17-bp between the paired sites containing nonamer b were present (compare pHEBo-2.2 with pHEBo-2.2.20; P < 0.01). Replication efficiency increased significantly when the 24 bp between nt 8994 and 9019 were restored (compare pHEBo-2.2.5 with pHEBo-2.2.24; P < 0.01). The latter result indicates that the inclusion of this sequence either provides an element that enhances the activity of EBNA-1 binding sites 3 and 4 and flanking repeats or places this minimal replicator at an appropriate distance from an auxiliary element located between oriP FR and DS.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Assessment of oriP replicator function. Previous genetic analyses of the EBV oriP replicator carried out by this and other laboratories relied predominantly on quantitation of the ability of mutated oriP plasmids to transform EBNA-1-expressing cells to drug resistance and the copy number of the plasmids in stably transformed cells cultured over many cell generations (4, 10, 11, 22, 32, 44, 46). The transformation assays used in these previous experiments suffered from the absence of an internal control for transfection efficiency. Furthermore, the copy number of an oriP plasmid is determined not only by the replication efficiency of the plasmid but also by the amount of plasmid DNA introduced into a cell at the time of transfection (47). For these reasons, these experimental approaches have provided only a semiquantitative measure of replicator function and have been unable to allow accurate comparisons between plasmids with mutations that reduce but do not eliminate replicator activity (see, e.g., references 10 and 44). These limitations are highlighted by the results presented in this study. Although the transformation frequencies of plasmids reported here generally reflected relative replication efficiencies, there were several notable exceptions that suggested the influence of experimental variation [e.g., pHEBo-1.1(dpm3+4;2=HAS), pHEBo-1.1(dpm1+2), and pHEBo-1.1(in1/2 [5]); Tables 2 and 3]. Note, too, that the plasmid copy numbers for Raji clones transformed by pHEBo-1.1(dpm4) and pHEBo-1.1(dpm3+4) were similar and did not reflect the 20-fold difference in transformation efficiency (Table 2) and different replication efficiencies (Table 3) exhibited by these plasmids. In several previous studies, short-term replication assays were also used to characterize mutated oriP plasmids. These assays either lacked an internal control for transfection efficiency and experimental error in the isolation of plasmid DNA from transfected-cell populations (10) or compared the amount of replicated plasmid DNA to the total amount of plasmid DNA recovered from the cells (34). The latter comparison assumes that all of the plasmid DNA isolated from the cells is nuclear and available for replication and does not take into account possible contamination with extracellular plasmid DNA or plasmid DNA that is in an intracellular compartment separate from the nucleus. To solve these problems and enable the identification of genetic elements that are essential for replicator function or augment the activity of essential elements, we adopted a quantitative, internally controlled, short-term replication assay which compares the replication efficiency of a test plasmid to that of a previously characterized reference plasmid that is replicated once per S phase (19). Using this assay, we characterized the minimal sequences that are required for initiation of DNA replication from within oriP and identified additional elements that are required to ensure that replication will initiate once per cellular S phase.

Multiple functional elements of the oriP replicator. Our analysis of pHEBo-1.1 derivatives bearing mutations in EBNA-1 sites that inhibit binding in vitro showed that all four EBNA-1 binding sites within oriP DS are required for full replicator activity in Raji cells. These findings provide an explanation for the conservation of all four EBNA-1 binding sites in EBV type 1 and type 2 isolates. Although we have not conducted quantitative analyses of the replication efficiencies of these plasmids in other EBNA-1-expressing cells, the reduced copy number of these plasmids in the D98/Raji hybrid cell line suggests that the requirement for all four sites is not restricted to B lymphocytes (10). Yates et al. reported that oriP plasmids bearing six-base substitutions near the center of each of the EBNA-1 binding sites in DS transformed Raji cells to drug resistance less efficiently than a plasmid containing four intact sites. In addition, the copy numbers of these mutated plasmids were reduced in an EBNA-1-expressing human osteosarcoma cell line as well as in Raji cells (46). Because the mutations created for their experiments may have altered genetic elements in addition to the EBNA-1 binding sites, it was not certain that the effects of the mutations were due to the disruption of the EBNA-1 binding sites. The data presented here substantiate previous results and support two conclusions: all four EBNA-1 binding sites are required for full replicator activity, and this requirement is likely universal and not cell type specific.

The minimal replicator. EBNA-1 binding sites 3 and 4 did not support the initiation of DNA replication by themselves, but minimal replicator activity was observed when both of the 14-bp repeats that flank these EBNA-1 binding sites were also present. It is unlikely that this effect is due to an alteration in the position of the EBNA-1 binding sites relative to flanking sequences because the presence of only repeat a or repeat b adjacent to EBNA-1 binding sites 3 and 4 did not provide for detectable activity (Fig. 6) and because substitution mutations within the 14-bp repeats reduced plasmid copy number when EBNA-1 binding site 1 or sites 1 and 2 were deleted (46). Because the EBNA-1 binding sites are essential components of the oriP replicator (10) and because sites 3 and 4 contribute more to replicator activity than sites 1 and 2, we conclude that the minimal replicator is contained within the 65-bp fragment encompassing sites 3 and 4 and the flanking 14-bp repeats. Yates et al. showed that substitution mutations that substantially altered all three of the nonamer repeats had no apparent effect on plasmid maintenance. This result led them to conclude that the nonamer repeats are not part of the minimal replicator. These same mutations had a profound effect, however, when they were introduced into oriP plasmids that contained only "one-half" of the DS. Based on these and other results, they proposed that the minimal replicator of oriP is composed of two EBNA-1 binding sites with 21-bp center-to-center spacing (46). Although it is not clear why the mutation of all three nonamer repeats was found to have no apparent effect when assayed in the context of an otherwise wild-type replicator, results presented here demonstrate that the EBNA-1 binding sites do not provide for significant replicator activity in isolation.