Role of the Epstein-Barr Virus Rta Protein in Activation of Distinct Classes of Viral Lytic Cycle Genes

Abstract

Initiation of the Epstein-Barr virus (EBV) lytic cycle is controlled by two immediate-early genes, BZLF1 and BRLF1. In certain epithelial and B-cell lines, their protein products, ZEBRA and Rta, stimulate their own expression, reciprocally stimulate each other’s expression, and activate downstream viral targets. It has been difficult to examine the individual roles of these two transactivators in EBV-infected lymphocytes, as they are expressed simultaneously upon induction of the lytic cycle. Here we show that the Burkitt lymphoma cell line Raji represents an experimental system that allows the study of Rta’s role in the lytic cycle of EBV in the absence and presence of ZEBRA. When expressed in Raji cells, exogenous Rta does not activate endogenous BZLF1 expression, yet Rta remains competent to transactivate certain downstream viral targets. Some genes, such as BaRF1, BMLF1, and a late gene, BLRF2, are maximally activated by Rta itself in the absence of detectable ZEBRA. The use of the Z(S186A) mutant form of ZEBRA, whose transactivation function is manifest only by coexpression of Rta, allows identification of a second class of lytic cycle genes, such as BMRF1 and BHRF1, that are activated in synergy by Rta and ZEBRA. It has already been documented that of the two activators, only ZEBRA stimulates the BRLF1 gene in Raji cells. Thus, there is a third class of viral genes activated by ZEBRA but not Rta. Moreover, ZEBRA exhibits an inhibitory effect on Rta’s capacity to stimulate the late gene, BLRF2. Consequently ZEBRA may function to repress Rta’s potential to activate some late genes. Raji cells thus allow delineation of the combinatorial roles of Rta and ZEBRA in control of several distinct classes of lytic cycle genes.

The reactivation of Epstein-Barr virus (EBV) from latency is associated with the expression of two immediate-early genes, BZLF1 and BRLF1, whose products are transcriptional activators that drive the lytic cascade of the virus (12, 36, 44, 51, 54). These two genes lie in overlapping transcriptional units and are expressed simultaneously during induction of the lytic cycle (35, 50). Until recently ZEBRA, the product of BZLF1, had been thought to be the only viral protein capable of initiating the lytic cycle (12, 22, 45, 51). Recently Rta, the product of BRLF1, was also shown to be able to disrupt latency in epithelial cells and in certain B-cell lines (44, 54). In those cells Rta leads to activation of Zp, the promoter of BZLF1, expression of ZEBRA, and thereby stimulation of early lytic genes, DNA replication, and late gene expression.

In the past there had been suggestions that cell line and cell type differences played a role in determining the extent to which Rta is able to induce the lytic cycle. Both early and recent studies failed to find any evidence that Rta mediated disruption of latency in Raji cells, a Burkitt lymphoma (BL) cell line (1, 13). Bogedain et al. suggested that Rta might be more active than ZEBRA in activating early lytic cycle genes in lymphoblastoid cell lines, while Rta had little effect in BL cell lines (6). Zalani et al. reported that Rta was able to induce the lytic cycle in an epithelial cell-specific manner but not in B cells (54). We have shown that Rta does have the capability to activate lytic cycle genes in B-cell lines such as the HH514-16 derivative of the P3JHR-1 (HR1) BL cell line (44).

These varied and sometimes contradictory results have caused us to examine more closely the effects of viral and cellular background on the activity of Rta. Here we describe the activity of Rta in the BL cell line Raji and compare its behavior to that in the HH514-16 cell line. In HH514-16 cells Rta activates the promoter of BZLF1 (Zp) and that of its own gene, BRLF1 (Rp) (44). However, we show here that in Raji Rta activates neither of these two promoters and fails to stimulate detectable ZEBRA expression. Nonetheless, Rta is a competent transactivator of other downstream lytic cycle genes in Raji cells. These genes include one class which responds maximally to Rta and another group which requires the concomitant activities of Rta and ZEBRA. Moreover, in the absence of ZEBRA, Rta can bypass a requirement for DNA synthesis in stimulating expression of the late gene BLRF2 in Raji cells. ZEBRA exerts an inhibitory effect on Rta’s action on late genes.

MATERIALS AND METHODS

Cell lines.

Cells were maintained in 5% CO₂ at 37°C in RPMI 1640 supplemented with 8% fetal calf serum. HH514-16 is a clonal derivative of the P3J-HR-1 B-cell line derived from an EBV-positive BL that is permissive for viral replication (43); Raji is a human B-cell line derived from a BL containing an EBV strain that is defective for DNA replication and late gene expression (42); B95-8 is a marmoset B-cell line transformed with EBV (37).

Antisera.

Anti-Rta is a polyclonal rabbit antiserum raised to an N-terminal fragment of Rta (44). Anti-ZEBRA and anti-BLRF2 are polyclonal rabbit antisera raised to TrpE-BZLF1 and TrpE-BLRF2 fusion proteins (26, 48). R3 is a mouse monoclonal antibody to BMRF1 (39), and SJ is a human serum with reactivity to EBNA1 and BFRF3 (48).

Chemical induction and PAA treatment.

Cells were subcultured 3 days prior to drug treatment or transfection. Drug treatment consisted of the addition of 10 ng of tetradecanoylphorbol-13-acetate (TPA) per ml and 3 mM sodium butyrate to the culture medium. In assays for late gene expression, the viral polymerase inhibitor phosphonoacetic acid (PAA) was added to the culture medium to 0.4 mM at the same time as the drug treatment or immediately following transfection (52).

Expression vectors and reporter plasmids.

The Rta and ZEBRA expression vectors pRTS/Rta, pBXG1-genomic Z (pBXG1/genZ), and CMV (cytomegalovirus)-genomic Z(S186A) [pCMV/genZ(S186A)] and their parent vectors pRTS, pBXG1, and pHD1013 have been described previously (14, 18, 44). The RpCAT constructs were generated by cloning the PCR-amplified sequence spanning nucleotides 106123 to 107143 of the EBV genome (relative to the B95-8 prototype [2]) into the XbaI and SphI sites of the chloramphenicol acetyltransferase (CAT) expression plasmid pCAT basic (Promega). PCRs were performed with total cellular DNA from Raji, HH514-16, and B95-8 cells. The ZpCAT constructs were made in a similar manner by cloning the PCR-amplified BZLF1 promoter fragment (nucleotides 103182 to 103742 of the EBV genome) into pCAT basic. The cloning sites were incorporated into the PCR primers. E4CAT, containing a minimal adenovirus E4 promoter, Z3E4CAT, containing three consecutive ZIIIB sites upstream of this minimal promoter, and the luciferase control vector pGL2 basic+HMP have been described elsewhere (9, 49).

Transfections.

Transfections were carried out by electroporation (49). For Western and Northern analyses, 10 μg of plasmid DNA (5 μg of each expression vector or 5 μg of expression vector plus 5 μg of empty vector, either pBXG1 or pRTS) was used. For reporter assays, 10 μg of CAT constructs plus 5 μg of expression vector or pBXG1 or pRTS and 1 μg of pGL2 basic+HMP were used.

Reporter assays.

CAT and luciferase assays were performed as described elsewhere (49). CAT results represent averages of at least two separate transfections and are standardized for transfection efficiency with luciferase activity.

Protein extracts and Western blots.

Cells were collected by centrifugation, washed once in phosphate-buffered saline, and resuspended in sodium dodecyl sulfate sample buffer at 10⁶ cells/10 μl. Prior to separation on sodium dodecyl sulfate–12% polyacrylamide gels, samples (20 μl) were heated to 100°C for 5 min. Following electrophoresis, the proteins were transferred to nitrocellulose membranes by electroblotting and blocked in 5% nonfat dry milk overnight at 4°C. The blots were incubated with antisera, diluted in 5% nonfat dry milk, at 25°C for 2 h, then washed three times for 10 min each in TS (10 mM Tris [pH 7.5], 200 mM NaCl, 5% Tween 20), incubated with ¹²⁵I-protein A for 1 h, and washed again. The membranes were exposed overnight with intensifying screens to Kodak XAR-5 film at −70°C.

RNA isolation and Northern blots.

Samples of 2.5 × 10⁶ cells were harvested 30 h following transfection. Cellular RNA prepared by using an RNeasy Mini kit (Qiagen) was electrophoresed through a 1% agarose–6% formaldehyde gel in 20 mM MOPS (morpholinepropanesulfonic acid [pH 7.0]); the RNA was transferred to Nytran and probed with ³²P-radiolabeled oligonucleotides. The BZLF1 oligonucleotide used spanned nucleotides 93 to 128 in exon I of the BZLF1 gene. A 531-bp excised EagI fragment from EBV BamHI-M was used to detect the BMRF1 and BaRF1 messages, while a 1.3-kb EcoRI/BamHI fragment from BamHI-M was used to detect BMLF1 mRNA. The BHRF1 probe consisted of a 2.477-kb SacI/BamHI fragment from EBV BamHI-H. A radiolabeled 370-bp NcoI-PstI fragment of H1 component of RNase P was used as a probe to control for RNA loading (3).

Sequencing.

Genomic Rp and Zp sequences were PCR amplified by using the high-fidelity Pfu DNA polymerase (Stratagene) and primers I (5-CTC GTT AAC TGA GAG C), II (5-GCT CTA GAC TGC CTG ACT GCG CTG A), III (5-GGA TAG CAG CGG TCC ACC), and IV (5-CAC TGG GAA CAG CTG AGG). Primers I and II were used for Rp; primers III and IV were used for Zp. Resulting PCR products were gel purified and then sequenced by the HHMI Biopolymer Laboratory & W. M. Keck Foundation Biotechnology Resource Laboratory at Yale University with internal Rp and Zp primers. Each promoter segment was sequenced at least two to three times from independent PCRs.

RESULTS

Cell background affects autostimulation of BRLF1 and BZLF1 in HH514-16 and Raji cells.

In permissive cell lines, such as HH514-16, transfection of a Rta expression vector leads to stimulation of its own gene, BRLF1, activation of the other immediate-early gene, BZLF1, and induction of downstream EBV early lytic cycle genes (44). As shown in Fig. 1, transfection of HH514-16 cells with either Rta or ZEBRA activated transcription from the Rp and Zp promoters, yielding the characteristic 3.0-kb bicistronic and 1.0-kb monocistronic messages, which were also induced by treating the cells with the chemicals TPA and sodium butyrate. Although Rta induced expression of the BZLF1 and BRLF1 genes less strongly than chemicals or ZEBRA, its effect in HH514-16 cells was highly reproducible. In the BL cell line Raji, by contrast, Rta was incapable of activating either gene. ZEBRA induced transcription solely from Rp (Fig. 1B), as has been previously documented (29, 31). The 1.3-kb mRNA originates from the ZEBRA expression vector (29). Only treatment with inducing chemicals led to the stimulation of both the Rp and Zp promoters in Raji cells.

FIG. 1.

Rta and ZEBRA stimulate the BRLF1 and BZLF1 genes to different extents in HH514-16 and Raji cells. Northern blot analysis of total cellular RNA isolated 30 h after chemical treatment or transfection is shown. Cells were either untreated (lanes 1 and 7), induced with TPA and sodium butyrate (lanes 2 and 8), or transfected with empty vector (lanes 3, 5, 9, and 11) or Rta and ZEBRA expression vectors (lanes 4, 6, 10, and 12). Northern blots were probed with a 30-base oligonucleotide from within exon I of BZLF1, which detects the 1.0-kb mRNA originating at Zp, the 3.0-kb bicistronic message from Rp, and a 1.3-kb vector transcript from BXG1/genZ. A probe for the H1 component of RNase P was used to control for RNA loading. (A) HH514-16 cells; (B) Raji cells.

Rta’s inability to activate either BRLF1 or BZLF1 was not due to poor Rta expression. In Raji cells, Rta was expressed to similar levels whether the cells had been transfected with RTS/Rta or BXG1/genZ or treated with inducing chemicals (Fig. 2). The transfected Rta protein was functional, as it induced the expression of the early antigen complex EA-D. Furthermore, Rta strongly activated the putative late gene product of BLRF2, a component of the viral capsid (5, 48). However, in cells transfected with ZEBRA or induced by TPA and sodium butyrate, BLRF2 was minimally induced. ZEBRA efficiently induced the expression of endogenous Rta at the protein level as well as at the RNA level, while Rta failed to induce detectable ZEBRA.

FIG. 2.

Immunoblots comparing activation of lytic cycle genes following the transfection of Rta and ZEBRA into Raji cells. Cells were either untreated (lane 1), chemically induced with TPA and sodium butyrate (T/B; lane 2), or transfected with 10 μg of plasmid DNA (lanes 3 to 7). In lanes 4 and 6, cells received 5 μg of activator (ZEBRA [Z] or Rta [R]) and 5 μg of empty vector. In lanes 3 and 5, cells received only vector pBXG1 (V1; lane 3) or pRTS (V2; lane 5). In lane 7, vectors expressing both ZEBRA and Rta were transfected. Immunoblots prepared 72 h following transfection were probed sequentially with the indicated antisera for EBNA1, Rta, EA-D, ZEBRA, BLRF2 (LR2), and BFRF3 (FR3).

Rta does not detectably activate ZEBRA in Raji cells.

Although Northern and Western analyses did not reveal any BZLF1 mRNA or ZEBRA protein in Raji cells following transfection of Rta, we used a highly sensitive reporter assay intended to detect even trace amounts of biologically active ZEBRA. The Z3E4CAT reporter contains three copies of the ZIIIB ZEBRA response element (ZRE) immediately upstream of the adenovirus minimal E4 promoter fragment linked to the CAT gene in pCAT (9). ZEBRA synergistically activates this reporter due to the presence of oligomerized ZREs, which, as we will show, render the reporter extremely responsive to even minute amounts of ZEBRA protein. The responsiveness of Z3E4CAT to ZEBRA is shown in Fig. 3, illustrating a titration in which low amounts of ZEBRA expression vector produced a dramatic stimulatory effect on the CAT activity. For example, 1 ng of expression vector yielded 8.7-fold stimulation of CAT activity over the level in cells that had been transfected with E4CAT, the corresponding negative control vector containing the minimal E4 promoter without ZREs. With increasing amounts of BXG1/genZ, the stimulation index (the ratio of Z3E4CAT activity over E4CAT activity) quickly rose to well above 100. By contrast, transfection of Raji cells with 5 μg of RTS/Rta resulted in only a twofold increase in CAT activity over the negative control. If Rta had induced biologically significant amounts of ZEBRA, we would have expected a dramatic increase in CAT activity.

FIG. 3.

Rta does not activate ZEBRA expression in Raji cells. (A) Reporter assay. Raji cells were transfected with 16 μg of total plasmid DNA consisting of the indicated amount of Rta or ZEBRA expression vector (made up to 5 μg with pBXG1; U, only pBXG1), 10 μg of E4CAT or Z3E4CAT reporter vector, and 1 μg of luciferase control vector. CAT and luciferase assays were performed 72 h following transfection. CAT activity was standardized for transfection efficiency on the basis of the luciferase data. Fold activation is the ratio of CAT activity generated by the reporter E4CAT or Z3E4CAT in the presence of Rta or ZEBRA divided by the activity in the presence of empty expression vector. (B) Western blot. Protein extracts made from the same experiment were probed on an immunoblot for EA-D and ZEBRA.

To quantitate the sensitivity of this assay, we examined protein extracts of the transfected Raji cells for ZEBRA and EA-D expression in the same experiment (Fig. 3B). Expression of EA-D would indicate the presence of sufficient quantities of either Rta or ZEBRA to be biologically active on downstream targets in the viral genome. We could not detect any ZEBRA in cells transfected with 5 μg of Rta expression vector, even after a prolonged (15 days) exposure of the blot to autoradiography film. ZEBRA was detectable in cells that had been transfected with 50 and 100 ng of ZEBRA expression vector. At these two BXG1/genZ concentrations EA-D was also induced to appreciable levels. The level of EA-D induced by 5 μg of transfected Rta expression vector was approximately equivalent to the level induced by 10 to 50 ng of ZEBRA expression vector. Were this stimulation due indirectly to the ability of Rta to activate endogenous ZEBRA expression, rather than a direct effect of Rta, we would expect a stimulation index between 55- and 111-fold in the CAT assay (Fig. 3A). However, only a twofold stimulation by Rta was observed. In summary, three lines of evidence favor the conclusion that Rta does not activate BZLF1 in Raji cells: (i) no BZLF1 mRNA is detectable by Northern blotting, (ii) Rta does not activate detectable expression of ZEBRA, and (iii) Rta does not activate a synthetic reporter that is exquisitely sensitive to ZEBRA.

Rta maximally activates some downstream targets in the absence of ZEBRA in Raji cells.

Having established that Rta is unable to induce the expression of ZEBRA in Raji cells, we next explored the capacity of Rta to activate viral targets by itself. The level of lytic cycle activation was compared in Raji cells transfected with Rta or ZEBRA expression vectors or treated with inducing chemicals. After a 30-h incubation at 37°C, total RNA isolated from the cells was analyzed by Northern blotting. The same blot was probed sequentially for messages from a representative group of viral lytic cycle genes. These included BRLF1, the gene for Rta (23), BaRF1, the ribonucleotide reductase small subunit gene (19, 20), BMRF1, the viral polymerase processivity factor (28, 32), BMLF1, the early post transcriptional activator (8, 27), and BHRF1, the bcl2 homologue (38). For a loading control, the blot was probed for the RNA H1 component of the cellular RNase P (3). As shown in Fig. 4, all of the genes were strongly activated by treating the cells with TPA and sodium butyrate (lane 2) or by transfecting of BXG1/genZ (lane 6). Transfection of Raji cells with RTS/Rta, on the other hand, did not lead to uniform activation of the genes selected for study (lane 4). Rta did not activate BRLF1, and it only weakly activated BMRF1 and BHRF1. However, the remaining two genes, BaRF1 and BMLF1, were activated by Rta at least as strongly as by inducing chemicals or by ZEBRA (Fig. 4; compare lanes 2, 4, and 10). These results suggested that BaRF1 and BMLF1 were maximally stimulated by Rta in the absence of ZEBRA in Raji cells. Their activation by transfected ZEBRA can be explained by the capacity of ZEBRA to activate Rta from the endogenous virus.

FIG. 4.

EBV lytic cycle genes activated by Rta alone or together with Z(S186A). Cells were either untreated (lane 1), chemically induced with TPA and sodium butyrate (lane 2), or transfected with 10 μg of plasmid DNA (lanes 3 to 10). In lanes 4, 6, and 8, cells received 5 μg of activator and 5 μg of empty vector. In lanes 3, 5, and 7, cells received only vector pRTS (lane 3), pBXG1 (lane 5), or pCMV (lane 7). In lanes 9 and 10, Rta was transfected with ZEBRA and the mutant Z(S186A), respectively. Total RNA prepared 30 h following transfection was analyzed by Northern blotting using probes for the indicated genes (see Materials and Methods). The blot was stripped between probes. Classification of the genes according to primary activator(s) is indicated to the right (see Discussion). The extra band above the expected size of the BRLF1 mRNA in lane 10 is most likely the result of an Rta-activated transcript from the Z(S186A) expression vector.

The ZEBRA mutant Z(S186A) allows the classification of genes that respond in synergy to Rta and ZEBRA.

Genes such as BMRF1 and BHRF1, that were only weakly activated by Rta, were maximally activated by ZEBRA. This result could be explained if the genes were exclusively controlled by ZEBRA or if they were controlled by a combination of ZEBRA and Rta. To distinguish between these two possibilities, we utilized the Z(S186A) mutant of ZEBRA. This mutant no longer initiates the lytic cascade of EBV due to its inability to activate Rp (1, 17). Adding exogenous Rta rescues Z(S186A), and the lytic cycle is successfully triggered. Consequently, the mutant can still act in synergy with Rta. Supplying Raji cells with either Rta, Z(S186A), or both permitted us to determine whether target genes were activated in synergy by Rta and ZEBRA. Raji cells were transfected with RTS/Rta, CMV/genZ(S186A), and RTS/Rta plus CMV/genZ(S186A). Total RNA isolated 30 h following transfection was examined on the same Northern blot. The ZEBRA mutant Z(S186A) was unable to activate any of the examined genes (Fig. 4, lane 8). When Rta was supplied along with Z(S186A), however, both BHRF1 and BMRF1 were activated to levels far above those achieved by Rta alone. Moreover, the effects of Z(S186A) together with Rta were indistinguishable from the effects of Rta together with wild-type ZEBRA or ZEBRA by itself. BHRF1 and BMRF1 therefore represent genes that are activated in synergy by Rta and ZEBRA. The signals for BaRF1 and BMLF1, on the other hand, were no stronger when cells had been transfected with Rta and the Z(S186A) mutant than with Rta alone, which indicates that these two genes are exclusive targets of Rta and do not require ZEBRA. BRLF1 represents yet another class of genes since it was not activated by Rta, by Z(S186A), or by the two together. Rta did not compensate for the Z(S186A) mutation in the activation of BRLF1. Thus, in Raji cells Rp is regulated only by ZEBRA and not by Rta, and not by a combination of the ZEBRA and Rta proteins.

Rp and Zp do not differ significantly in HR1 and Raji cells.

The inability of Rta to activate either BRLF1 or BZLF1 in Raji cells may be due to cell line differences or to inherent differences in the viral genomes. The promoter sequences of BRLF1 and BZLF1 could contain pertinent point mutations or deletions which might interfere with or result in the loss of regulatory elements within the promoters. To examine these possibilities, the nucleotide sequences of the Rp and Zp regions from both Raji and HH514-16 cells were compared to the sequences in the B95-8 prototype (2). The results of the analysis are shown in Fig. 5A (Rp) and B (Zp). Rp from Raji cells differs from the B95-8 sequence in four locations within the region examined (−965 to +140 relative to the transcription start site). In three locations (−840, −708, and +74), a G has been changed to an A; in one (−1), a C has been converted to an A. The HH514-16 sequence deviates only in one location from B95-8, with the same C-to-A transversion at position −1. The −1 position is the only location which falls into a known regulatory element of the promoter, a YY1 binding site that overlaps the transcriptional start site (53). This mutation within the YY1 site is shared by the Raji and HH514-16 cell lines. The two point mutations at positions −708 and −840 of Raji Rp are likely to be too far upstream to have any effect; the mutation at position +74 is significantly downstream of the transcriptional start site and beyond the 5′ splice site of the Rp message.

FIG. 5.

Comparison of Rp (A) and Zp (B) promoter sequences among Raji, HH514-16, and B95-8 genomes. Promoters were amplified from total Raji and HH514-16 cell DNA by PCR, and sequences were compared to that of the B95-8 prototype. Deviations from B95-8 and their locations relative to the transcriptional start site are indicated in bold; known regulatory sites affected by the mutations are indicated on the right. The diagrams represent the promoters with their known proximal regulatory elements. ZRE, ZIIIA, and ZIIIB, ZEBRA response elements; Z1 and ZIA through -D, AT-rich sequences, reported to bind Sp1, Sp3, and MEF2D (7, 33, 34); SRE, serum response element; OCT, AP-1-like octamer; TRE, TPA response element.

Within Zp, the HH514-16 sequence deviates at seven locations from the B95-8 prototype. All are point mutations matching those previously reported by Jenson et al. (25) and reside at least 99 bases upstream of the transcriptional start. Five of the changes are transitions, while the other two are transversions. The Raji genome shares two of these mutations, at positions −364 and −459. The former falls within a ZIIIA ZEBRA response element and likely destroys it (30). A mutation within the HH514-16 sequence that is not shared with Raji lies within another regulatory element, the ZIC site covering position −140. ZI sites have previously been described as AT-rich sequences involved in the TPA-mediated induction of Zp. They have been shown to bind Sp1, Sp3, and members of the MEF2 family of transcription factors (7, 16, 33, 34). However, the −140 mutation in ZIC lies outside the Sp1/3 binding site (33, 34). Although this mutation does fall within the MEF2D recognition sequence, ZIC is the only member of the ZI sites that does not bind MEF2D (34). Consequently, the overall function of the variant ZI site is unlikely to be affected by the transition. The remaining four mutations in HH514-16 Zp do not fall within any known regulatory sequences.

Rta activates RpCAT from B95-8, HR1, and Raji genomes equally well in Raji cells.

We next determined whether the minor sequence differences observed in Rp and Zp between the HR1 and Raji genomes altered the response of the two promoters to activation by Rta in a transient reporter assay in Raji cells. We have shown previously that Rta not only activates Rp from the endogenous virus in HH514-16 cells but also efficiently activates an RpCAT reporter in these cells (44). The promoter sequences from the B95-8, Raji, and HR1 genomes were cloned into pCAT basic and transfected with or without Rta into Raji cells. As shown in Fig. 6, CAT reporters bearing Rp sequences from the three different genomes were activated equally well by Rta in Raji cells. These results suggest that the differences in response of the endogenous Rp to Rta in Raji and HH514-16 cells is not likely to be directly attributable to sequence variations in the Rp promoter. The same comparison was performed using cloned ZpCAT constructs cotransfected with Rta into Raji cells (data not shown). Again there was no difference in the response of the three ZpCAT constructs to Rta, a result consistent with the conclusion that the minor sequence differences do not significantly account for the major differences in regulation observed in the two cell backgrounds.

FIG. 6.

Activation of RpCAT containing sequences from B95-8, HH514-16, and Raji genomes by Rta in Raji cells. See the legend to Fig. 3 for experimental details. Fold activation is the ratio of CAT activity generated by a reporter in the presence of Rta divided by the CAT activity in the presence of empty vector. pCAT, reporter lacking promoter or enhancer sequences; RpCAT, Rp from the three respective genomes cloned into pCAT.

Rta bypasses the requirement for DNA replication in the activation of a late gene, BLRF2, in Raji cells.

Since Raji cells are deficient for expression of late genes (4, 55), the ability of Rta to strongly activate expression of the specific mRNAs of BLRF2 (5) in Raji cells was unexpected (Fig. 2). BLRF2 has previously been classified as a late gene whose expression is dependent on viral DNA replication (5, 48). Identification of late genes is achieved by treatment of cells with viral polymerase inhibitors, such as PAA, along with the inducing stimulus. Analysis of viral RNA will fail to detect late gene expression in PAA-treated cells but will reveal the full lytic repertoire in cells not treated with PAA.

Using this definition, we first sought to reconfirm that BLRF2 is a true late gene in HH514-16 cells. Expression of viral genes was examined both at the protein and RNA levels. Western blots were probed for the presence of Rta itself, BLRF2, and BFRF3, another late gene product (52). EBNA1 was included as a loading control (Fig. 7A). At the RNA level, BLRF2 expression was analyzed with an oligonucleotide probe on a Northern blot; the H1 component of RNase P served as a loading control (Fig. 7B). Induction by TPA and sodium butyrate or by transfection of ZEBRA or Rta led to expression of BLRF2 (lanes 3, 7, and 11), yet under each induction condition, BLRF2 expression was inhibited by PAA (lanes 4, 8, and 12). By contrast, the expression of Rta itself and EBNA1 were unaffected by PAA treatment. The results showed clearly that in HH514-16 cells BLRF2 behaved as a late gene, similar to BFRF3.

FIG. 7.

Rta activates BLRF2, a late gene. (A and B) HH514-16 cells; (C and D) Raji cells. (A and C) Western analysis of protein extracts from cells that were uninduced (lanes 1 and 2) or had been chemically induced (lanes 3 and 4) or transfected with 5 μg of empty vector (lanes 5, 6, 9, and 10) or 5 μg of expression vector (lanes 7, 8, 11, and 12). For each condition, the cells had also been either untreated (−) or treated with the viral DNA polymerase inhibitor PAA (+). Immunoblots were prepared 30 h following transfection and probed sequentially with antisera to the indicated proteins (LR2, BLRF2; FR3, BFRF3). (B and D) Northern analysis of RNA prepared from cells of the same transfections as above. The blots were probed with a ³²P-labeled oligonucleotide from within BLRF2 and a fragment of the H1 component of RNase P as loading control.

The same analysis was performed in Raji cells with a different outcome (Fig. 7C and D). The Raji genome has a deletion within the BamHI A fragment, containing among others the BALF2 gene, which codes for a DNA binding protein essential for DNA replication. Accordingly, treatment of Raji cells with TPA and sodium butyrate induced the lytic cycle and the expression of Rta but not the late gene BLRF2 (Fig. 7C and D, lanes 3). The addition of PAA had no effect on the expression of either protein. Transfection of Raji cells with Rta, however, strongly induced BLRF2 at the protein and RNA levels (Fig. 7C and D, lanes 7). The addition of PAA did not abrogate this effect. In Raji cells, therefore, exogenously expressed Rta can bypass the requirement for DNA replication and lead to the activation of BLRF2. Expression of Rta did not significantly induce BFRF3, however (data not shown). BLRF2 may thus be a specific target of Rta. The data implies that Rta may function in activation of late lytic genes.

ZEBRA, on the other hand, had an inhibitory effect on the capacity of endogenous Rta to activate the expression of BLRF2. Following transfection of ZEBRA, which activated the endogenous Rta, only minute levels of BLRF2 protein were synthesized (Fig. 7C and D, lanes 7 and 11) and the BLRF2 message was barely detectable. This inhibition of Rta’s ability to activate BLRF2 was also seen when Rta and ZEBRA were cotransfected (Fig. 2, lane 7). Similarly, following chemical treatment, which induced both Rta and ZEBRA (Fig. 2), only trace amounts of BLRF2 protein were expressed. Thus, in Raji cells ZEBRA appeared to counteract the potential stimulatory effect of both endogenous and exogenous Rta on late gene expression.

DISCUSSION

In this report, we present three novel findings regarding the role of Rta in activation of lytic cycle gene expression in the BL cell line Raji. First, when transfected into Raji cells, Rta does not detectably activate either Zp or Rp, and therefore no ZEBRA synthesis ensues. Nonetheless, Rta activates lytic gene products. Second, among the genes stimulated by Rta some are activated maximally by Rta alone, and others are activated as the result of synergy between ZEBRA and Rta. The use of the Z(S186A) mutant, which by itself lacks the capacity to activate lytic cycle genes, permits the identification of genes which respond synergistically to Rta and ZEBRA. Third, Rta can activate a subset of late genes in Raji cells, bypassing the requirement for viral DNA synthesis. ZEBRA suppresses this activity.

Extensive data presented here show that in the Raji cell line, Rta fails to activate the Rp and Zp promoters. ZEBRA expression was not detected at either the RNA or protein level. Moreover, a sensitive reporter assay, performed in parallel with assays for endogenous viral gene expression, strongly suggests that ZEBRA is not induced following transfection of Rta. Yet in this cell background, Rta remains a competent transactivator capable of activating several lytic genes such as BaRF1 and BMFL1 to a maximal level and other genes, such as BMRF1 and BHRF1, to a lesser extent. Raji cells transfected with Rta represent a de facto ZEBRA knockout system in which the role of Rta in the activation of lytic genes can be explored. This system will ultimately allow identification of all early genes that are responsive to Rta, directly or indirectly, in the absence of ZEBRA and viral DNA replication.

Classes of EBV lytic cycle genes in Raji cells.

Using Raji cells, we describe a method with which one can ascertain precisely whether a gene is predominantly activated by Rta or by the combination of Rta and ZEBRA. The ZEBRA mutant Z(S186A) is impaired in its ability to activate all viral genes, including Rta (1, 17, 18). Coexpression of Rta and Z(S186A), however, rescues the deficient mutant phenotype and the usual program of gene activation ensues. The change of a serine to an alanine in the DNA binding domain of ZEBRA may reduce the affinity of the protein for its response elements within Rp, or it may impair the ability of ZEBRA to be phosphorylated, a possible prerequisite for interaction with other proteins involved in stimulation of Rp (1, 17, 18). Nonetheless the mutation does not significantly reduce the ability of ZEBRA to synergize with Rta on downstream viral targets. Using this information, we classify early lytic viral genes in Raji cells into three distinct groups. Class Z genes, represented by BRLF1, respond only to Z(S186A). Rta has no effect on BRLF1 by itself or in combination with ZEBRA. Class R genes, e.g., BaRF1 and BMLF1, are dominantly activated by Rta in Raji cells. The level of stimulation of these genes by the combination of Z(S186A) and Rta does not exceed the level observed with Rta alone. ZEBRA also stimulates these genes maximally as a result of its ability to induce endogenous Rta protein. Class RZ genes, including BMRF1 and BHRF1, respond in synergy to Rta and ZEBRA. Rta by itself activates these genes to low levels, but addition of the Z(S186A) mutant achieves maximal stimulation. There are likely to be additional classes of lytic cycle genes on which Rta by itself has no activity. One class is represented by BZLF1, which in Raji cells appears under control of factors other than Rta and ZEBRA. Another class would be genes other than BRLF1 that are exclusively controlled by ZEBRA. Definition of this class will require mutants that lack Rta expression even in the presence of ZEBRA. Finally, there may be genes that are only stimulated by the combination of Rta and ZEBRA.

Functional differences of Rta in Raji and HH514-16 cells.

Rta is able to stimulate Rp and Zp in HH514-16 cells but not in Raji cells. This difference in function can be attributed to cell background or viral genome differences or both. Only a few point mutations distinguish Rp and Zp in the EBV strains carried by Raji and HH514-16 cells. In general, these mutations do not fall within known regulatory elements. Two mutations that may potentially affect such elements that are present in the prototype B95-8 strain are shared by Raji and HH514-16. Moreover, Rta is competent to activate RpCAT and ZpCAT from all EBV strains when presented as plasmids in Raji cells. Although we cannot discount the possibility that Rta activates the reporter constructs from initiation sites different from those used in the endogenous virus, we consider this possibility improbable. The low background activation of the pCAT vector by Rta suggests an absence of cryptic initiation sites within the vector. Since the RpCAT constructs require the promoter TATA box for activation by Rta, transcription is likely to initiate properly (44). For all of these reasons, it seems unlikely that primary sequence differences in Rp and Zp account for the dramatic differences in behavior. However, our experiments have not explored the possible contributions of viral sequence alterations elsewhere in the HR1 and Raji genomes. For example, the HH514-16 cell line carries the HR1 strain, which lacks EBNA2, and the Raji genome lacks EBNA3C, BARF1, BALF1, and BALF2 (15, 24, 40). The absence of these genes may directly alter viral gene regulation or result in the creation of different cellular environments through their failures to activate or repress specific cellular genes. Alternatively, the different response to Rta may be a consequence of cellular background effects on chromatinization, methylation, or the presence or absence of cellular coactivators or repressors. These cofactors may also be specific for their viral targets only in the context of the endogenous chromatinized viral genome, which might explain why Rta could still activate the reporter constructs in Raji cells.

Potential role of Rta in late gene activation.

Rta plays a role in the activation of expression of a late gene, BLRF2, in Raji cells. This was an unexpected finding because a defining defect of Raji cells is the absence of DNA replication and late gene expression. Using the standard assay to classify late viral genes, namely, inhibition by viral polymerase inhibitors (52), we found that BLRF2 displays unambiguous late character in HR1 cells. It is expressed only if viral DNA replication is allowed to proceed. In Raji cells, Rta can circumvent this requirement. ZEBRA, on the other hand, appears to downregulate Rta mediated activation of BLRF2. The late gene is only stimulated by Rta in the absence of ZEBRA. In chemically treated or ZEBRA-transfected cells which express both transactivators, BLRF2 is expressed poorly or not at all. Two explanations could account for the repressive effect of ZEBRA on late gene expression. In the early lytic cycle, ZEBRA may bind to the late BLRF2 promoter and act as a repressor. The BLRF2 promoter does contain several putative ZREs. Upon DNA replication, ZEBRA is displaced or modified so that it loses its affinity for the promoter, allowing access to Rta and other putative late activators. In the absence of viral DNA replication, such as occurs in Raji cells, ZEBRA remains bound to the BLRF2 promoter, thus inhibiting its activation. The second scenario posits ZEBRA’s ability to sequester a factor that together with Rta is essential for late gene expression. The capacity of ZEBRA to bind this essential factor is lost after DNA replication. In Raji cells, in the absence of ZEBRA, the factor is therefore available to work together with Rta in the activation of BLRF2. While the expression of Rta unaccompanied by ZEBRA expression is not likely to occur during the normal EBV life cycle, this pattern of gene expression uniquely encountered in Raji cells seems to provide attractive new insights into the function of the two activators in regulation of late genes.

Implications of the findings.

Unlike regulation of prokaryotic gene expression, the regulation of eukaryotic gene expression is rarely controlled by a single activator or repressor (41). Instead, the combinatorial effects of multiple activators and repressors controls expression of eukaryotic genes. The switch from latency to the lytic cycle of EBV is similarly under the control of an intricate regulatory mechanism seeming to involve activators and repressors of both cellular and viral origin. For a long while, ZEBRA was thought to represent the single essential viral activator. However, it is now clear that the cellular mechanisms that govern the lytic cycle switch simultaneously promote expression of two viral proteins, ZEBRA and Rta, that are involved in a complex interplay that is beginning to be defined in this and other recent reports (1, 17, 44, 54). As we show here, interactions between ZEBRA and Rta occur at the level of autostimulation and reciprocal stimulation of Rp and Zp. Furthermore, Rta and ZEBRA demonstrate complex interactions in their control of different sets of downstream lytic cycle viral genes.

ZEBRA and Rta are both likely to possess multiple functions. ZEBRA’s capacity to activate transcription, to participate in replication (46, 47), and to cause cell cycle arrest (10, 11) have all been well described. In earlier studies, ZEBRA has been implicated in interference with Rta’s ability to activate synthetic Rta responsive promoters (21). Here we demonstrate that ZEBRA possesses a potential role as a repressor of Rta-activated late gene expression under some conditions in the context of an intact viral genome (Fig. 7C and D). The ability of ZEBRA to prevent late gene expression activated by Rta may ensure the proper stage-specific progression into the lytic cycle. The repressive effect of ZEBRA on late gene expression may be relieved by DNA replication. In all likelihood, Rta will also eventually be found to carry out different functions in the viral life cycle. Here we define its capacity to dominate expression of some genes and to facilitate the activity of ZEBRA on other genes (Fig. 4). Other functions of Rta may be revealed as the panel of genes known to be controlled by ZEBRA, Rta, and cell factors is expanded.

In summary, detailed analysis of the roles of Rta and ZEBRA in the EBV lytic cycle switch is beginning to reveal all the beauty and complexity of control of eukaryotic gene expression and should continue to serve as a model for this central process in biology.