Nucleoprotein Structure of Immediate-Early Promoters Zp and Rp and of oriLyt of Latent Epstein-Barr Virus Genomes

Abstract

Genomic footprints across Rp, Zp, and oriLyt of Epstein-Barr virus (EBV) have been conducted in a panel of latently infected B-cell lines. Close protein-base contacts were found about 360 nucleotides upstream of the Zp initiation site. Gel shifts and transient transfection assays indicated that an Sp1-NF1 locus may serve as a repressive transcriptional element against Zp induction from latent EBV genomes.

Although the lytic cycle of latent Epstein-Barr virus (EBV) is strongly repressed in B cells, activation can be achieved through overexpression of the viral immediate-early proteins Rta (also called BRLF1 or R) (34, 49) and Zta (also called BZLF1, Z, EB1, or ZEBRA) (35, 43, 45), which is also the viral origin binding protein for oriLyt (12, 38). Currently, a cooperative model of mutual activation between Rta and Zta is favored (1, 5, 32). The molecular architecture of the immediate-early locus of EBV resembles that of the major immediate-early locus of human cytomegalovirus, with the exception that there is no strong enhancer in EBV. Therefore, Zta can be expressed from both the BRLF1 and BZLF1 promoters (Rp and Zp) (24). Zta and Rta coactivate oriLyt, a key cis regulatory element for the progress of the lytic cycle that is essential for virus production (7, 9, 12, 15, 16, 38, 39). Many transcriptional elements important for the induction of the three EBV regulators (3, 8, 13, 14, 21-23, 33, 39, 41, 42, 47, 48, 50) and also some elements important for the repression of Zp in latency (19, 22, 25, 26, 40) have been characterized so far by in vitro binding studies and reporter gene experiments (for review, see reference 43). However, there is a fundamental difference between naked DNA as used for in vitro analyses and chromatin as it occurs in the living cell. Therefore, understanding in vivo protein-DNA interactions at Zta, Rta, and oriLyt on latent genomes may shed some light on the mechanisms through which EBV is kept latent in B cells. Since they have not been examined on the chromatin of latent EBV genomes at nucleotide resolution, we undertook a survey of in vivo protein binding at these regulatory elements in latent EBV genomes in a panel of B-cell lines.

Little overall in vivo protein binding to lytic cycle regulatory elements.

The cell lines Raji, LCL 721, Mutu BLI-Cl216, Mutu BLIII-Cl99, and Rael are in a strictly latent state, as documented by terminal repeat analysis (44) and by the absence of early antigens or their coding mRNAs (J. Minarovits, unpublished data). All cell lines belong to latency type I (Rael and Mutu I) or III (LCL721, Mutu III, and Raji), as described before (37). Ligation-mediated PCR (LM-PCR) (27) was applied under the same conditions as described earlier (29, 30, 37) in order to obtain a comprehensive set of genomic footprints of those regulatory elements from the five cell lines. For each footprint reaction, 10⁷ exponentially growing cells were treated with 5 μl of dimethyl sulfate for 1 min at room temperature. Reactions were stopped, footprinted DNAs were isolated and subjected to piperidine treatment, and footprints were visualized from 2 μg of sequenced or footprinted DNA by LM-PCR, under the same reaction conditions as described earlier (29, 30, 37). LM-PCR primers are listed in Fig. 1. One-fifth of each sample was separated on a standard 5% sequencing gel and autoradiographed. Footprints were mostly identical in all cell types. Subtle signals for protein-DNA interactions were found at many locations on both strands of Zp and Rp, and for oriLyt, footprint lanes were mostly similar to the G-tracks (data not shown). These subtle protections and hypersensitivities were generally not correlated with previously described in vitro binding sites at the promoters. The nature of the proteins causing faint signs of in vivo protein-DNA interactions at immediate-early promoters Zp and Rp and at oriLyt is unclear; they await further identification. Possibly, they might be caused by general chromatin packaging factors that do not contact the nucleotide bases, but the phosphate backbone. The absence of strong genomic footprints was not due to methodical problems, since the same sets of footprinted DNAs were used for the present analyses that had been used previously (37), and all experiments were performed several times with independent preparations of genomic DNAs with identical outcomes. Furthermore, strongly positive control footprints on the DS element of oriP comparable to the ones in Raji cells (30) have been obtained from the same set of genomic DNAs (H. H. Niller, unpublished data). These data most likely document the absence of activating transcription factors from lytic regulators in EBV latency. Activating factors for Zp, Rp, and oriLyt characterized through previous in vitro binding and reporter gene assays may play a crucial role at the second phase of the lytic cycle, after latency is interrupted, and viral genomes have been moved to the ND10 domains for lytic replication (4).

FIG. 1.

Genomic footprint analyses of oriLyt (A and B) and Zp (C and D). Panels A and C show the upper strands, and panels B and D show the lower strands. Lanes: 1, G-track from LCL 721 DNA; 2 to 6, footprint lanes; 2, LCL 721 cells; 3, Mutu III cells; 4, Rael cells; 5, Raji cells; 6, Mutu I cells; 7, G-track from Mutu I DNA. To the left of each panel are indicated the locations of protected (lollipops) and of hypersensitive (arrows) guanines. To the right of each panel are given nucleotide numbers according to the EBV sequence of Baer et al. (2). Solid symbols indicate strong footprints, and open symbols indicate subtle signals. The LM-PCR primer coordinates are given in reference to the nucleotide numbers of the viral sequence (2) as follows: primer set A, primer 1, 53176 to 53200; primer 2, 53202 to 53226; and primer 3, 53224 to 53248; primer set B, primer 1, 53680 to 53657; primer 2, 53668 to 53644; and primer 3, 53629 to 53605; primer set C, primer 1, 103698 to 103674; primer 2, 103690 to 103666; and primer 3, 103666 to 103641; and primer set D, primer 1, 103403 to 103427; primer 2, 103453 to 103478; and primer 3, 103468 to 103492. In the case of a single nucleotide polymorphism at nucleotide 103654, a wobble base was included in primer 3 of primer set C.

Protein binding at oriLyt.

In contrast to the overall absence of close protein-base contacts, there were some closer contacts at the upper strand of the downstream element of oriLyt (Fig. 1A); however, again more subtle footprints were found on the lower strand (Fig. 1B). Like in Zp and Rp, footprints on oriLyt were mostly identical for all cell types. Footprints generally did not correlate with previously characterized elements. There were no signs of protein binding visible at the Sp1 boxes (3, 14) on the parts of oriLyt examined. However, the protein contact around nucleotide 53539 was located at RRE1 (36), a previously characterized transcriptional and replicative element of oriLyt. Since oriLyt and both the BHLF1 and BHRF1 genes are silent in EBV latency, protein binding in the absence of Zta hints at cellular repressors being bound to oriLyt in latency.

Specific protein binding at Zp.

A strong genomic footprint was found at a GC box, 360 nucleotides upstream of Zp transcription initiation (Fig. 1C and D). The GC box was located in the vicinity of YY1 site D (25, 26), HI elements γ and β (40), and an NF1-like site (13) and has not been described previously (see Fig. 3). However, it might correspond to a footprint previously seen by in vitro DNase I analyses on Zp with HeLa cell nuclear protein (26). For the GC box locus of Zp, we characterized candidate binding proteins through electrophoretic mobility shift experiments. Mutu I nuclear extracts were made according to the standard of Dignam et al. (10), with modifications as described previously (37). For gel shifts, 1 ng of T4 kinase-radiolabeled double-stranded oligonucleotides was incubated with 0.5 μg of poly(dI-dC)-poly(dI-dC), 5 μg of nuclear protein, and a 50-fold excess of unlabeled double-stranded competitor oligonucleotide, as indicated, in 25 μl of bandshift buffer (10 mM Tris-HCl [pH 7.5], 5 mM MgCl₂, 80 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, 12.5% glycerol) for 20 min. Protein complexes were resolved by electrophoresis on native 5% polyacrylamide gels (29 + 1) in 0.25× Tris-borate-EDTA (TBE) buffer at 20 mA for several hours. The shift experiments demonstrated that the four binding sites, YY1, Sp1, NF1, and HI, contributed to specific in vitro protein binding at this locus (data not shown). Sequence-specific binding of NF1 and Sp1 proteins from Mutu I cells to their respective binding sites is shown in Fig. 2. Specific binding was found not only with nuclear protein from Mutu I cells, but also with protein from the other cell lines used in this study (data not shown). The in vivo protection together with the in vitro binding of this locus are a hint that NF1 and Sp1 proteins really play a role in the nucleoprotein structure of silent Zp. To elucidate further a possible functional role of the major footprinted area of Zp, transient transfection assays were done in DG75 cells, an EBV-negative Burkitt's lymphoma cell line, with luciferase reporter constructs. ZpLuc contained promoter sequences between EBV coordinates 103194 and 103730 in front of the luciferase gene of plasmid pGL2 (Promega). This corresponds to Zp sequences from nucleotide positions −536 to +1 relative to the transcriptional start site. Construct ZpLucΔ1, which contains a deletion of both the Sp1 and NF1 sites of Zp from nucleotides 103542 to 103570, was made with the QuikChange mutagenesis kit (Stratagene) from ZpLuc by using the mutant oligonucleotide GGTCAGTTCGTCCAAATGGCTGTCCACATATGGCTGCTTC and the corresponding opposite-strand oligonucleotide. The mutant construct (Fig. 3) was confirmed by sequencing both strands of the deleted promoter area. Twenty micrograms of each double-cesium chloride-purified plasmid was transfected into DG75 cells by the DEAE-dextran method as described previously (28). Four hours after transfection, cells were induced with tetradecanoyl phorbol acetate (TPA) at 40 ng/ml for 20 h, and relative light units from cellular extracts were measured with the Promega dual luciferase reporter assay system on a Berthold luminometer (Lumat LB 9501). Since Renilla luciferase control constructs were themselves strongly induced by TPA, standardization was not done with Renilla luciferase, but by using equal amounts of protein extract for luciferase assays and repeating the experiment several times. Wild-type reporter construct ZpLuc and deletion construct ZpLucΔ1 yielded the same low baseline activities. Upon induction through TPA, luciferase activity increased strongly for both constructs. However, inducibility of the deletion construct through TPA was on average 1.6-fold higher than that of the wild-type construct (Fig. 3B). The GC locus protection in the promoter-distal part of Zp together with the absence of activator binding in lytic regulators might contribute to keeping EBV tightly latent. NF1 and Sp1 may, together with YY1 and the HI binding factor, form an inhibitory protein complex on Zp. Previously, endogenous Zp was found to be so efficiently silenced that it is not active even under conditions that activate exogenously transfected Zta under the control of Zp (18, 20). Therefore, the Sp1-NF1 locus might contribute to a stronger inhibition of Zp on endogenous genomes compared to the transient effect from transfection experiments in DG75 cells. Rp, Zp, and oriLyt of latent EBV genomes are efficiently silenced in latency through the absence of activator binding, through the presence of repressive factors, and through a chromatin structure that favors inactivity. The latent state is stabilized through DNA methylation and histone deacetylation (6, 11, 17, 31). It will be interesting to examine the function of the newly found protein-DNA interactions on the Sp1-NF1 locus of Zp in an experimental system closer to latent EBV genomes, possibly through the construction of a mutant virus.

FIG. 3.

(A) Sequence around the −360 upstream area of Zp. Previously described promoter elements YY1D (25, 26), HIγ and HIβ (40), and the newly found Sp1 and NF1 sites are indicated by boxes. The deleted sequence of mutant ZpLucΔ1 is indicated by a dashed line. The location of protected (lollipops) and hypersensitive (arrows) guanines from Fig. 1C and D is indicated. (B) Activities of ZpLuc and ZpLucΔ1 transcriptional elements in DG75 cells treated with TPA. The data from a representative transfection experiment are given in relative light units on the ordinate, and the plasmids and treatment of cells are indicated on the abscissa.

FIG. 2.

Electrophoretic mobility shift assay demonstrating specific protein binding activities in Mutu I cell nuclear protein to the Sp1 (lanes 1 to 5) and NF1 (lanes 6 to 10) sites from Zp. Double-stranded (only one strand given) labeled oligonucleotides for shifts and unlabeled oligonucleotides for competition of complexes were either taken from Zp sequences or contained consensus binding sites for the respective transcription factors (46). Lanes: 1, ZpSp1 (GGATCCACGAGGGGGCGGGTGCCATG; Zp nucleotides 103542 to 103557) probe only, no protein added; 2 to 5, nuclear extract added; 3, competition with unlabeled Sp1 consensus oligonucleotide (GGATCCACGAGGGGCGGGGTG CCATG); 4, competition with NF1 consensus oligonucleotide (GATCGGGTGGCACTGTGCCAAGGATC); 5, competition with ZpHI oligonucleotide (CATGGCGGTCCATCTGTCCACGATCC; Zp nucleotides 103563 to 103577); 6, ZpNF1 (GATCGGGTGGCTCAGGTCCATCCATG; Zp nucleotides 103553 to 103570) probe only, no protein added; lanes 7 to 10, nuclear extract added; lane 8, competition with unlabeled NF1 consensus oligonucleotide; 9, competition with Sp1 consensus oligonucleotide; 10, competition with ZpHI oligonucleotide. Lowercase letters with arrows represent unshifted probes (a and d), specific Sp1-related complexes (b and c), specific NF1-related complexes (e and f), and nonspecific complexes (g).

Nucleotide sequence accession numbers.

Rp promoter sequences may be found under GenBank accession no. AJ422215, AJ422216, AJ422217, and AJ422218, and Zp promoter sequences may be found under GenBank accession no. AJ422219, AJ422220, AJ422221, and AJ422222 for the cell lines Rael, Raji, Mutu BLI clone 216, and Mutu BLIII clone 99, respectively.