Cellular Immediate-Early Gene Expression Occurs Kinetically Upstream of Epstein-Barr Virus bzlf1 and brlf1 following Cross-Linking of the B Cell Antigen Receptor in the Akata Burkitt Lymphoma Cell Line ^▿

Cell lines.

Akata cells, with and without the EBV genome, were gifts from Kenzo Takada (37, 39, 40). The cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) (catalog no. SH30070.03; HyClone), penicillin, 50 U/ml, streptomycin, 50 U/ml, and amphotericin B, 1 μg/ml. 2089 cells harboring an EBV bacmid (17), gifts from H.-J. Delecluse, were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (Invitrogen), 50 U/ml penicillin and streptomycin, 1 μg/ml amphotericin B, and 100 μg/ml hygromycin B (Calbiochem).

Reagents.

Anti-IgG (A042301-2; Dako) was dialyzed in calcium and magnesium-free phosphate-buffered saline (PBS). Akata cells were resuspended in fresh medium, and anti-IgG was added at 7.5 μg/ml to induce the lytic cycle. Cycloheximide (239764; Calbiochem) at 33 μg/ml and anisomycin (A9789; Sigma) at 100 μM were used to inhibit protein synthesis. These reagents were added 15 min prior to or at intervals after anti-IgG. Bisindolylmaleimide X (B3931; Sigma) at 5 μM was used to inhibit protein kinase C.

Microarray analysis.

Total RNA was isolated using an RNeasy minikit (74104; Qiagen) from 1.5 × 10⁷ EBV⁺ Akata cells untreated or treated with anti-IgG for 2 h. RNA was quantified by UV spectrometry. RNA samples from three independent experiments were pooled and used for microarray analysis to examine the mRNA expression profile in collaboration with the W. M. Keck Foundation Biotechnology Resource Laboratory at Yale University (http://keck.med.yale.edu). In this study, Affymetrix gene chip microarrays for the human genome (U133; Plus 2.0) were used. Cellular mRNA expression level in anti-IgG-treated Akata cells was compared with that in uninduced Akata cells. A total of 672 gene entries were changed 2-fold or more in anti-IgG-induced samples. These included known genes, expressed sequence tags, hypothetical genes, and repeated entries. Among the changes, 420 were increases and 252 were decreases. A total of 74 gene entries were changed 8-fold or more in anti-IgG-induced samples. Among the changes, 56 were increases and 18 were decreases. This report analyzes the expression patterns of 6 genes that were induced 9- to 128-fold on the microarray (see Table 1).

RNA isolation and quantitative real-time reverse transcription (RT)-PCR.

Total RNA was isolated from 5 × 10⁶ EBV-positive or EBV-negative Akata cells using the RNeasy minikit (Qiagen). Each RNA sample was treated with 30 units DNase I using the RNase-Free DNase set (Qiagen). RNA concentration was measured by UV spectrometry. Total RNA (200 ng) was used as a template for real-time RT-PCRs using the iScript one-step RT-PCR kit with SYBR (Bio-Rad). The primer pair for EBV brlf1 spanning the only intron has been reported previously (44). The primer pair for EBV bzlf1 spans the second intron. The primers for cellular genes egr1, egr2, egr3, nab2, nr4a1, and nr4a3 were also designed to span one or multiple introns when possible. Sequences of the primers will be provided upon request. PCRs were run on a MyiQ2 two-color real-time PCR detection system (Bio-Rad). Serial dilutions of total RNA from anti-IgG-induced EBV-positive Akata cells were used as standards for mRNA quantification.

Northern blot analysis.

Preparation of total cellular RNA, radioactive probes, gel conditions, and transfer have been described (44).

Immunoblots.

Preparation of extracts of cellular proteins, gels, electrophoresis, and transfer to nitrocellulose have been described (14). Antibodies to the following proteins were used: EBV ZEBRA (rabbit polyclonal) (41), EBV Rta (rabbit polyclonal) (34), EBV EA-D (mouse monoclonal R3.1 (33), EGR1 (rabbit polyclonal, Santa Cruz 20689), EGR2 (rabbit polyclonal, Santa Cruz 20690), EGR3 (rabbit polyclonal, Santa Cruz no. 191), NAB2 (mouse monoclonal, Santa Cruz 28323; rabbit polyclonal, Santa Cruz 22815), β-actin (mouse monoclonal; Sigma 5316). Antigen-antibody reactions were visualized by ¹²⁵I-labeled protein A or enhanced chemiluminescence (ECL) (Amersham RPN 2106).

Plasmids.

pCMV-XL5-EGR1, which contains human egr1 cDNA under the control of a cytomegalovirus (CMV) promoter/enhancer, was purchased from OriGene. pcDNA-EGR2, pcDNA-NAB2, pcDNA-NR4A1, and pcDNA-NR4A3, which express EGR2, NAB2, NR4A1, and NR4A3 proteins, respectively, were constructed by ligating cDNA fragments of these genes amplified from Akata cells into a pcDNA3.1 vector (Invitrogen). Specifically, 200 ng of total RNA from anti-IgG (2.5 h)-induced Akata cells was used as a template for RT-PCRs to amplify full-length cDNA fragments of EGR2, NAB2, NR4A1, and NR4A3. To facilitate subcloning, a BamHI restriction site was added to the 5′ primers, and XbaI (for NR4A3) or XhoI (for EGR2, NAB2, and NR4A1) sites were added to the 3′ primers. cDNA fragments were digested with BamHI and XbaI (for NR4A3) or with BamHI and XhoI and ligated into the pcDNA3.1 vector linearized with the same enzymes. The sequences of the PCR primers are as follows. EGR2, 5′-TGGGATCCCACCATGATGACCGCCAAGGCCGTAG (forward) and 5′-GTCTCGAGTCAAGGTGTCCGGGTCCGAGAG (reverse); NAB2, 5′-CTGGATCCCACCATGCACAGAGCGCCTTCCC (forward) and 5′-TCCTCGAGTCACTGCCGGCTGGCCTCAGC (reverse); NR4A1, 5′-AGGGATCCCACCATGCCCTGTATCCAAGCCCAATA (forward) and 5′-GGCTCGAGTCAGAAGGGCAGCGTGTCCATGA (reverse); NR4A3, 5′-TAGGATCCCACCATGCCCTGCGTCCAAGCCCAAT (forward) and 5′-GCTCTAGATTAGAAAGGTAGGGTGTCCAGG (reverse). CAAX-GFP vector, which expresses a membrane-bound form of the jellyfish (Aequorea victoria) green fluorescence protein (GFP), was provided by Tony Hunter of the Salk Institute.

Transfection of Akata cells and 2089 cells.

Akata cells were transfected with the Amaxa Nucleofector II device according to the manufacturer's recommended protocol. Briefly, 4 × 10⁶ Akata cells were transfected with 0.5 μg CAAX-GFP plasmid and 1.5 μg control plasmid (pcDNA3.1 empty vector) or with 0.5 μg CAAX-GFP plasmid and 0.375 μg each of pCMV-XL5-EGR1, pcDNA-EGR2, pcDNA-NAR4A1, and pcDNA-NR4A3 plasmids. At 18 to 20 h after transfection, cells were left untreated or treated with anti-IgG or CHX (see Fig. 9). Cells were centrifuged and resuspended in ice-cold fluorescence-activated cell sorter (FACS) buffer (1× PBS supplemented with 2% FBS) and sorted with a FACS Vantage SE system. Two subpopulations of cells, namely, GFP-negative and GFP-positive cells, were collected and analyzed for bzlf1 mRNA expression. 2089 cells were transfected with DMRIE-C reagent (Invitrogen) as follows: 4 μg of total DNA and 12 μl of DMRIE-C reagent (2 mg/ml) were diluted in 0.5 ml of OPTI-MEM1 reduced serum medium (Invitrogen) and mixed, and the mixture was incubated at room temperature for 30 min. The mixture was added to 2089 cells grown on six-well plates. Cells were incubated at 37°C in a CO₂ incubator for 6 h before the addition of serum to a final concentration of 10%. Cells were cultured for 20 h and then harvested for a qRT-PCR assay for bzlf1 and brlf1 mRNA.

Kinetics of expression of brlf1 and bzlf1 mRNAs in Akata cells treated with anti-IgG.

To explore the hypothesis that initial phases of synthesis of bzlf1 and brlf1 mRNAs might be immediate-early in nature, we first determined, using sensitive qRT-PCR technology, how soon an increase of brlf1 and bzlf1 mRNA could be detected in anti-IgG-treated Akata cells relative to that in untreated cells (Fig. 1). Neither viral mRNA was induced by 1 h; both viral mRNAs were stimulated above background level by 1.5 h after anti-IgG (8.5-fold for brlf1 mRNA and 43-fold for bzlf1 mRNA in this experiment). Maximal stimulation of brlf1 and bzlf1 mRNAs occurred 3 h after exposure to anti-IgG. At 1.5 h after treatment with anti-IgG, expression of bzlf1 mRNA was induced 5-fold more than brlf1 mRNA; at 2 and 3 h, bzlf1 mRNA was 2-fold more than brlf1. Thereafter, the relative difference between the stimulated levels of the two mRNAs was no longer observed. At each time after 3 h, the stimulation level of the two mRNAs progressively decreased, suggesting that both these mRNAs were unstable. By 8 h after addition of anti-IgG, the brlf1 mRNA was 26% and the bzlf1 mRNA 12.4% of the maximal level at 3 h.

Activation of the EBV brlf1 and bzlf1 genes by anti-IgG is blocked by inhibitors of protein synthesis.

In order to determine whether at any time the bzlf1 and brlf1 mRNAs behaved with immediate-early gene kinetics, activation of bzlf1 and brlf1 was compared in anti-IgG-treated cells with or without the addition of CHX, an inhibitor of protein synthesis. Whenever CHX was present, expression of bzlf1 and brlf1 was inhibited. At 1.5 h, CHX blocked the appearance of brlf1 mRNA by 87% and bzlf1 mRNA by 93%. CHX inhibited the maximal 3-h induction of brlf1 by 99% and that of bzlf1 by 98% (Fig. 1).

To determine the time interval when de novo protein synthesis was required for expression of brlf1 and bzlf1, Akata cells were treated with anti-IgG and inhibitors of protein synthesis, either CHX (Fig. 2 A, B) or anisomycin (ANIS) (Fig. 2B). The inhibitors were added either simultaneously with anti-IgG or at intervals after anti-IgG. brlf1 and bzlf1 mRNAs were detected by Northern blotting. These experiments demonstrated that new protein synthesis was required for induction of brlf1 and bzlf1 mRNAs some time during the first 2 h after addition of anti-IgG. When the experiment was analyzed at 8 h (Fig. 2A), CHX, when added at time zero, inhibited bzlf1 expression by 71% and brlf1 expression by 82%. When added 2 h or 4 h after anti-IgG, CHX did not inhibit but stabilized the brlf1 and bzlf1 mRNAs by 2- to 3-fold compared to that of anti-IgG-treated cells that did not receive CHX (Fig. 2A, compare lane 3 with lanes 5 and 6). Figure 2B shows that both CHX and ANIS blocked induction of brlf1 and bzlf1 expression when the inhibitors were added simultaneously or 1 h after addition of anti-IgG. When added at 2 h after anti-IgG, the protein synthesis inhibitors did not appreciably alter the expression of the viral lytic activator genes. In Akata cells, brlf1 and bzlf1 mRNAs were relatively unstable during the period 2 to 8 h after anti-IgG but could be stabilized by CHX in the period between 2 and 6 h (Fig. 2A, lanes 5 and 6). This stabilization effect of CHX or ANIS was not evident when the experiment was analyzed after 3 h (Fig. 2B).

Kinetics of inhibition of anti-IgG induction of brlf1 and bzlf1 mRNA by CHX.

Figure 2B shows that induction of brlf1 and bzlf1 mRNA became resistant to CHX or ANIS between 1 and 2 h after addition of anti-IgG. To determine more precisely when, within this interval, new protein synthesis was required, CHX was added together with anti-IgG or at 15-min intervals thereafter. The abundance of brlf1 and bzlf1 mRNAs was measured by qRT-PCR (Fig. 3). Up to 30 min after administration of anti-IgG, expression of the mRNAs of the two lytic activator genes was >97% sensitive to CHX. Between 45 min and 1 h after addition of anti-IgG, expression of bzlf1 and brlf1 mRNA displayed some resistance to CHX but was still inhibited by 60% to 70%. However, brlf1 and bzlf1 expression became completely resistant to CHX by 1.25 h after anti-IgG. At 1.75 h after anti-IgG, the levels of bzlf1 and brlf1 mRNA were 1.35- and 1.64-fold higher in cells treated with CHX than in cells which were not treated with the inhibitor. These results suggest that one set of newly synthesized proteins were required to activate the viral genes within the first hour after anti-IgG. Other newly synthesized proteins may play a role in decreasing bzlf1 and brlf1 mRNA expression beginning 1.5 to 2 h after anti-IgG. Taken together, the data of Fig. 1 to 3 establish that brlf1 and bzlf1 are not immediate-early genes, even under the conditions of very rapid induction of the EBV lytic cycle by anti-IgG in Akata cells.

Identification of cellular genes whose expression was activated within 2 h after addition of anti-IgG to Akata cells.

The results illustrated in Fig. 1 to Fig. 3 provoked the hypothesis that newly synthesized proteins, presumably cellular in origin, synthesized in the first 2 h after anti-IgG treatment might be required for activation of the EB viral lytic cycle in Akata cells. We used gene expression microarrays, comparing mRNAs from untreated Akata cells and those treated with anti-IgG for 2 h, as a preliminary screen to identify some candidate cellular genes for this function. We identified 21 unique cellular genes whose expression was equal to or >8-fold higher in Akata cells treated with anti-IgG for 2 h than in untreated cells. In Table 1, in which a subset of these genes is listed in decreasing order of the extent of activation, it can be seen that early growth response (egr1, egr2, egr3, nab2) and nuclear orphan receptors (nr4a1, nr4a3) were strongly activated at 2 h after anti-IgG treatment of Akata cells.

egr1, egr2, egr3, nr4a1, and nr4a3 are cellular immediate-early genes that are induced by anti-IgG treatment of Akata cells with and without the EBV genome.

We selected the six abundantly expressed cellular genes listed in Table 1 for further study. These were four early growth response genes, egr1, egr2, egr3, and EGR1 binding protein 2 (nab2), and two nuclear orphan receptors, nr4a1 and nr4a3. We used qRT-PCR to confirm the results of gene expression microarrays (Fig. 4). Expression of each of the six genes was induced in both EBV-positive and EBV-negative Akata cells by treatment with anti-IgG. With the exception of egr3, which was weakly induced (2-fold) in EBV-negative Akata cells, the remaining five genes were induced robustly in Akata cells with and without the EBV genome. Five of the six genes could be classified as immediate early (IE). The expression of these five genes was greatly enhanced when both EBV-positive and EBV-negative Akata cells were treated with anti-IgG in the presence of CHX. The enhancement of expression by CHX ranged from 6-fold to 29-fold in EBV-positive Akata cells and 16-fold to 65-fold in EBV-negative Akata cells. CHX increased expression of the five cellular IE genes to a 2- to 3-fold-greater extent in EBV-negative Akata cells than in EBV-positive Akata cells.

The expression of nab2 was inhibited by CHX in both EBV-positive and EBV-negative Akata cells (Fig. 4D). This was not an unexpected result, as EGR1 is a major activator of nab2 expression (25). In the experiment illustrated in Fig. 4, EBV bzlf1 and brlf1 were studied in parallel with the cellular genes. The two viral mRNAs were detected only in anti-IgG-treated Akata cells that were EBV positive. Expression of the two viral genes was completely inhibited by CHX, as predicted by the results of Fig. 1 to 3.

Kinetics of expression of cellular immediate-early genes relative to expression of EBV brlf1 and bzlf1 in Akata cells treated with anti-IgG.

One criterion that might imply a functional role of a cellular IE gene product in the regulation of EBV brlf1 and bzlf1 is that the cellular IE gene should be expressed before the viral genes. To investigate this hypothesis, Akata cells were treated with anti-IgG or untreated, and cellular and viral gene expression were monitored at frequent intervals (Fig. 5; Table 2). Expression of brlf1 and bzlf1 in anti-IgG-treated Akata cells was first detectably above that in untreated cells at 90 min and rose progressively thereafter to 3 h when the experiment was terminated (Fig. 5G and H). Expression of the six cellular genes preceded the expression of the viral genes and also reached a peak earlier than 3 h. Among the cellular genes there appeared to be three kinetic classes based on the earliest time of detection. egr1 mRNA was expressed above background levels by 15 min, egr2, egr3, and nr4a1 by 30 min, and nr4a3 and nab2 by 60 min in anti-IgG-treated cultures. The time of maximum expression of these six cellular genes also differed: egr1 peaked at 1 h, egr2, nr4a1, and nr4a3 by 1.5 h, and egr3 and nab2 by 2 h. Among the cellular genes there were also notable changes in mRNA abundance with time. The levels of egr1 and nr4a1 decreased more rapidly than those of egr2, nr4a3, and nab2.

Expression of EGR1 protein precedes that of EBV Rta and ZEBRA.

The hypothesis derived from the data of Fig. 1 to 3 predicts that cellular IE gene protein synthesis should be temporally upstream of bzlf1 and brlf1 expression in order to contribute to regulation, either activation or repression, of the viral genes. Since it has been reported that ZEBRA activates the expression of egr1 in epithelial cells (9), we examined the effects of anti-IgG on the synthesis of EGR1 protein in EBV-positive and EBV-negative Akata cells (Fig. 6 A, B). These experiments showed that induction of EGR1 protein by anti-IgG was independent of the presence of the EBV genome. In EBV-negative Akata cells, EGR1 protein was detected by 1 h after anti-IgG treatment, reached and maintained a near-maximal level from 2 to 4 h after anti-IgG treatment, and decayed thereafter (Fig. 6A). In EBV-positive Akata cells, EGR1 protein was also detected by 1 h after anti-IgG treatment and reached a maximal level by 2 h, but EGR1 rapidly decreased in abundance beginning 3 h after anti-IgG treatment (Fig. 6B). Thus, the duration of persistence of EGR1 differed in Akata cells with and without the EBV genome.

EBV Rta and ZEBRA were first detected by immunoblotting 4 h after anti-IgG and continued to increase at 6 to 7 h. At 23 h after anti-IgG treatment EBV Rta was present at near maximal level, while the ZEBRA protein was 56% of its maximal level. In contrast, at 23 h the level of EGR1 protein had decreased to 8% of the maximal level. This experiment demonstrates a remarkable difference in the kinetics of expression of the EGR1 cellular IE gene protein and EBV ZEBRA and Rta.

Relative kinetics of expression of four cellular IE proteins.

In addition to EGR1 (Fig. 6A and B), EGR2, EGR3 (Fig. 6C and D), and NR4A1 (not shown) were detectable with commercially available antibodies. However, the antibodies to EGR2, EGR3, and NR4A1 were of low titer and required the use of chemiluminescence for detection on immunoblots. Therefore, the relative magnitude of expression levels cannot be compared among the four proteins. However, the relative protein levels can be compared at different times for each individual protein. EGR2 and EGR3 proteins were detectable at 1 h after anti-Ig. ZEBRA protein was at background levels at 1 h when EGR2 and EGR3 proteins were stimulated. EGR2 continued to increase in abundance at 2 h and was maximal by 3 h, but the level of EGR3 decreased at these times. NR4A1 protein first appeared above background level at 3 h, and its expression was transient. Based on these studies, EGR1, EGR2, and EGR3 proteins are kinetically upstream of the EBV lytic cycle proteins, ZEBRA and Rta.

Kinetics of expression of NAB2.

The kinetics of expression of NAB2 protein differed from those of EGR1, EGR2, and EGR3 in several respects. NAB2 appeared later. It was not significantly induced above background level until 3 h in both EBV-negative (Fig. 7A) and EBV-positive (Fig. 7B) Akata cells. Unlike that of EGR1, the expression level of NAB2 was relatively stable over time. NAB2 abundance was maintained at maximal level for at least 23 h in EBV-negative Akata cells (Fig. 7A) and at 50% maximal level for 23 h in EBV-positive Akata cells (Fig. 7B). Although a low level of NAB2 was detectable at 2 h, maximal levels of NAB2 coincided with induced expression of the EBV lytic proteins, ZEBRA and EA-D, at 3 and 4 h after anti-IgG treatment of EBV-positive Akata cells (Fig. 7C). The relative temporal patterns of expression of EGR1 and NAB2 proteins corresponded to the relative kinetics of expression of the egr1 and nab2 transcripts (Table 2).

Expression of EGR1, NAB2, and the EBV early lytic proteins induced by anti-IgG is blocked by an inhibitor of protein kinase C (PKC).

Cross-linking of the B cell antigen receptor by anti-IgG initiates a complex signal transduction cascade (8, 26, 42). One component of this cascade is PKC (26, 42). Therefore, we tested whether expression of EGR1, NAB2, ZEBRA and Rta, following treatment of Akata cells with anti-IgG, was affected by bisindolylmaleimide X (Bis X), a potent and specific inhibitor of PKC (7). EGR1 protein appeared at 1 h (Fig. 8A), while NAB2 (Fig. 8B and C), ZEBRA, and Rta (Fig. 8A) were significantly above background level at 3 h, in agreement with the data in Fig. 6 and 7. Bis X completely blocked anti-IgG induction of all four proteins. As expected, Bis X also inhibited expression of bzlf1 and brlf1 mRNA (not shown). This experiment is most consistent with a model in which anti-IgG induces signal transduction that leads to the expression of cellular IE genes whose protein products are temporally and presumably functionally upstream of EBV brlf1 and bzlf1 expression (see Discussion).

EGR1 activates expression of EBV BZLF1 and BRLF1 mRNAs.

Another criterion for the importance of a cellular immediate-early gene in activating the EBV lytic cascade is that overexpression of the cellular protein might activate the expression of viral lytic cycle genes. Figure 9 shows that transfection of a mixture of plasmids encoding four cellular immediate-early products activates BZLF1 expression from the endogenous EBV genome in Akata cells. Since standard electroporation of Akata cells is inefficient, these cells were “nucleofected” with GFP plus the cellular genes or GFP plus control vector. EBV lytic gene expression was assessed in the GFP-positive and GFP-negative populations. The conditions varied in four iterations of this experiment: either no treatment (panel A), CHX treatment to stabilize the BZLF1 mRNA (panel B), a suboptimal time of harvest (panel C), or a suboptimal amount of anti-IgG (panel D). Transfection of a mixture of four plasmids expressing cellular IE genes has been repeated six times. The mean ratio of BZLF1 expression in GFP-positive cells transfected with the four cellular immediate-early genes versus cells transfected with control vector was 2.85 (range, 2.34 to 3.94). In the GFP-negative population this mean ratio was 1.2 (range, 1.06 to 1.33). In preliminary experiments we were unable to see any effect of the individual candidate cellular genes in Akata cells, possibly because of the low transfection efficiency or requirement for cooperation among the cellular IE genes. However, in 2089 cells (293 cells with an EBV bacmid) we found that EGR1 induced expression of BRLF1 and BZLF1 mRNA 5- to 6.6-fold above the background level when assayed 24 and 48 h after transfection (Fig. 10).

bzlf1 and brlf1 are not immediate-early genes in Akata cells.

Since engagement of the B cell antigen receptor activates a complex and diverse cell signaling cascade (26, 42), it was possible, even likely, that in Akata cells the bzlf1 and brlf1 genes were directly activated by events downstream of signal transduction and behaved with true IE kinetics. Our data (Fig. 1 to 3) showed that bzlf1 and brlf1 expression was initially exquisitely sensitive to CHX and anisomycin, did not begin to escape the effects of CHX until 45 min after anti-IgG, and was completely resistant to the protein synthesis inhibitors only 1.25 h after anti-IgG. The model (19) which proposes that the dampening effects of protein synthesis inhibitors are solely the result of blocking autostimulation by low levels of ZEBRA and Rta protein synthesized from true viral IE genes seems unlikely from our experiments for at least two reasons. First, by 2 h after anti-IgG treatment, CHX inhibited the expression of brlf1 and bzlf1 by 92% to 94% (Fig. 1). Thus, the overwhelming majority of these transcripts did not display IE kinetics. Second, ZEBRA and Rta protein were not detected above background levels until 3 or 4 h after anti-IgG (Fig. 6 to 8) and hence are not likely to explain the significant levels of activation (36-fold for brlf1 and 79-fold for bzlf1) observed by 2 h after anti-IgG.

Our experiments in which CHX was added at different times after anti-IgG may provide possible clues to the inconsistencies in the literature about the effects of inhibitors of protein synthesis on the levels of bzlf1 and brlf1 mRNA in Akata cells (19, 32, 40). qRT-PCR analysis indicates that these mRNAs are unstable during the period 3 to 8 h after anti-IgG (Fig. 1). When CHX is added 2 h after anti-IgG, these mRNAs are stabilized 2- to 3-fold (Fig. 2B). Therefore, whether or not protein synthesis inhibitors seem to have an inhibitory effect is markedly influenced by the time when they are added.

Cellular IE genes are activated by anti-IgG treatment of Akata cells.

Of the cellular genes whose expression was 9-fold higher in anti-IgG-treated cells than in untreated cells on gene expression microarrays, we selected the early growth response and nuclear orphan receptor genes for our initial study because these were among the most actively stimulated genes, because they are transcription factors, and because antibodies to the EGR proteins were available commercially (Table 1). The stabilizing effects of CHX on expression levels of the early growth response genes and nuclear orphan receptor genes should be contrasted with the inhibitory effect of CHX on bzlf1 and brlf1 expression (Fig. 4). The expression level of these cellular genes was significantly enhanced when Akata cells were treated with a combination of anti-IgG and CHX. Only nab2, an EGR1 binding protein gene that requires EGR1 protein for its expression, was inhibited by CHX. A significant observation was that the early growth response and nuclear orphan receptor genes were activated to a high level and their expression was stabilized by CHX in Akata cells lacking the EBV genome (Fig. 4). These findings demonstrate that none of these genes is dependent on an EBV-encoded product for their synthesis. In fact, the transcripts of some cellular IE genes such as EGR1 and EGR2 were expressed at a higher level in EBV-negative Akata cells (Fig. 4) and EGR1 protein was more stable in EBV-negative Akata cells (Fig. 6). This phenomenon may be attributable to a shut-off cellular gene expression during the lytic cycle in EBV-positive Akata cells (36).

Others have reported that in EBV-infected 293 cells and in nasopharyngeal cancer cell lines bzlf1 expression is followed by expression of EGR1 (9). In these cell systems, small interfering RNA (siRNA) to bzlf1 decreases EGR1 expression. Therefore, EGR1 is considered to be a downstream target of ZEBRA. Our data do not demonstrate this pathway in Akata cells treated with anti-IgG. EGR1 protein is expressed in the absence of EBV (Fig. 6A). Expression of EGR1 protein precedes expression of ZEBRA in EBV-positive Akata cells (Fig. 6B, Fig. 8). Furthermore, EGR1 protein expression decays while ZEBRA protein expression increases. While it is known that EGR1 is expressed downstream of activation of the B cell antigen receptor (22), a novel aspect of our study was to demonstrate that expression of several cellular IE genes, including egr1, was kinetically upstream of bzlf1 and brlf1 (Fig. 5; Table 2).

Does EGR1 play an essential role in activation of the EBV lytic cascade in Akata cells?

Here we present new types of indirect evidence that favor the hypothesis that egr1 may play a role in activating the EBV lytic cascade. egr1 is kinetically upstream of bzlf1 and brlf1 expression after anti-IgG treatment of Akata cells (Fig. 5; Table 2). Inhibition of protein kinase C blocks expression of EGR1 protein and ZEBRA and Rta proteins (Fig. 8). In preliminary experiments transfection of an EGR1 expression plasmid activated bzlf1 and brlf1 in 293 cells carrying an EBV bacmid (Fig. 10).

Several earlier studies provided other types of indirect evidence that correlates egr1 expression and activation of the EBV lytic cycle. Zalani and coworkers showed that a vector expressing EGR1 protein activated reporters containing Rp, the promoter of the EBV BRLF1 gene, in EBV-positive D98-HE-R1 and EBV-negative NIH 3T3 cells (47). Moreover, EGR1 protein that was prepared in Escherichia coli bound to Rp in electrophoretic mobility shift (EMSA) and DNase I footprinting assays. B95-8 marmoset lymphoblastoid cells treated with TPA expressed EGR1 protein that bound to Rp in EMSA. Rp contains two EGR1 binding sites; deletion of these sites abolished the response of Rp reporters to exogenously expressed EGR1 and to TPA-induced activation. Thus, Rp is a plausible target for EGR1.

Feng et al. showed that gemcitabine and doxorubicin activated BZLF1 and BRLF1 expression in human lymphoblastoid cells (18). Both drugs activated Zp and Rp reporters in EBV-negative DG75 cells. Elimination of the two EGR1 binding sites at −41 and −123 in Rp abolished activation of the reporters by the two chemotherapeutic agents. Some 293 human embryonic kidney cell lines infected with EBV spontaneously undergo lytic cycle activation; they express ZEBRA and Rta (9). siRNA to egr1 inhibits the spontaneous lytic activation in these cells.

Evidence against an activating role of EGR1.

Arguments can also be marshaled against an essential positive role of egr1 in activating the EBV lytic cycle. From 1.5 to 3 h after anti-IgG treatment of Akata cells, the 0.9-kb BZLF1 mRNA expressed from Zp is preferentially activated (11) (Fig. 1 and 2B). bzlf1 mRNA is 2- to 3-fold more abundant than the 3.0-kb bicistronic mRNA initiated from Rp. Zp is not known to be directly activated by EGR1 and does not contain canonical EGR1 binding sites for 1 kb upstream of the start of transcription. This result suggests that more than one cellular IE gene must be expressed to activate the EBV lytic cycle. Many cellular IE genes, in addition to egr1, are expressed after cross-linking of the B cell antigen receptor. It is not yet possible to discriminate among these IE genes for their role in EBV lytic activation. Some of these cellular IE gene products might specifically target Zp and others Rp.

Although EGR1 efficiently activates Rp reporters, such experiments cannot be extrapolated to effects on the intact EBV genome. For example, the ZEBRA mutant Z(S186A) can activate Rp reporters but cannot activate BRLF1 from the latent EBV genome (20). This may be related to epigenetic modifications of Rp in the latent genome (5, 14). The proximal EGR1 site in Rp overlaps Sp1 sites (35). Sp1 can activate Rp reporters in insect cells that do not contain endogenous Sp1 (45). Therefore, any effects of lytic cycle-inducing agents on Rp activation must discriminate between an effect of EGR1 and an effect of Sp1.

Expression of cellular IE genes, including EGR1, may not be sufficient for lytic cycle activation or may be inhibitory and promote a cellular state that is refractory to lytic activation. For example, phorbol esters strongly activate EGR1 expression in the Jijoye Burkitt lymphoma cell line but do not activate the lytic cascade (35). When the EBV-containing Burkitt lymphoma HH514-16 cell line is treated with HDAC inhibitors, Na butyrate and TSA, as lytic cycle-inducing agents, expression of cellular IE genes, such as stat3 and c-fos, is markedly induced. However, expression of these cellular IE genes is associated with a refractory state to lytic cycle induction (15). Therefore, it must be proved, for any cellular IE gene that is induced by an agent that activates the EBV lytic cycle, that the cellular IE gene is expressed in the same cells that are activated into the lytic cycle. The association of STAT3 expression with a state refractory to lytic induction is based on a method to separate HH514-16 cells that do or do not respond (4). A system for separating Akata cells that express ZEBRA could be useful to determine which cellular IE genes are expressed in the same cells that express ZEBRA protein.

A revised model for EBV lytic activation in Akata cells.

Based on the available evidence, anti-IgG activates a complex signaling pathway that leads to cellular IE gene expression. The signaling cascade may also lead to posttranslational modification of cellular IE proteins. These cellular IE gene products, individually or in combination, are likely to promote the formation of two distinct populations of Akata cells. One subpopulation responds to cellular IE proteins by expression of the EBV-encoded lytic activator genes. Another subpopulation responds to cellular IE products by establishing a refractory state. A goal for future research on the upstream problem will be to identify newly synthesized cellular proteins that specifically promote the refractory or lytic state.