Epstein-Barr Virus BHRF1 Micro- and Stable RNAs during Latency III and after Induction of Replication^▿

Cell culture and antibodies.

B95-8 (46, 59), IB4 (65), recently derived LCLs, NPC C666-1 (18), and EBV-infected or uninfected BJAB (25), BL41 (14), and Akata (63, 64) cells were maintained in RPMI 1640 medium (Gibco-BRL) supplemented with 10% or 20% Fetal Plex animal serum complex (Gemini). C15 and C17 nasopharyngeal tumors were passaged in nude mice as xenografts (9). MAbs PE2 (71), A10 (45), and 3E8 (42) are specific to EBNA2, EBNA3C, and BHRF1, respectively.

Western blot analysis.

Cells were harvested at indicated times after virus replication induction and lysed in 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer. Cell extracts were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Bio-Rad). Following incubation in blocking buffer (phosphate-buffered saline containing 0.1% Tween 20 and 5% nonfat milk), membranes were incubated with diluted primary MAbs overnight at 4°C with gentle shaking. Following washes and secondary antibody incubation, specific binding was visualized by ECL enhanced chemiluminescence (Amersham Biosciences).

RNA preparation and Northern analysis.

Total RNA was prepared using TRIzol reagent (Invitrogen). Cytoplasmic and nuclear RNAs were fractionated from Akata cells at 48 h after virus replication induction by immunoglobulin G (IgG) cross-linking. Briefly, cells were resuspended in 4 volumes buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 1.5 mM MgCl₂) for 30 min on ice, Dounce homogenized, and spun at 2,000 rpm at 4°C for 5 min. The supernatant was extracted with TRIzol reagent to isolate cytoplasmic RNA. The buffer A-washed pellet was resuspended in 3 volumes of buffer B (20 mM HEPES [pH 7.9], 20% glycerol, 420 mM NaCl, 1,5 mM MgCl₂, 0.2 mM EDTA), incubated for 30 min on ice, and spun at 12,000 rpm for 20 min at 4°C. The separated supernatant was extracted with TRIzol reagent to isolate nuclear RNA. For miRNA or mRNA detection, 30 μg total RNA per lane was separated in 15% polyacrylamide gel or in denaturing 1% agarose-2.2 M formaldehyde gel, transferred onto GeneScreen Plus hybridization transfer membranes (Perkin-Elmer), immobilized by baking at 80°C for 2 h, and hybridized with ³²P-labeled RNA probes at 37°C overnight or ³²P-labeled DNA probes at 42°C overnight. ³²P-labeled RNA probes specific for each miRNA were prepared by in vitro transcription using the mirVana miRNA probe construction kit (Ambion) and primers (Table 1). ³²P-labeled 5′, intron, ORF, and 3′ EBV BHRF1 DNA probes were prepared by random primer DNA labeling (Invitrogen) using BHRF1 5′-untranslated, intron, ORF, and 3′-untranslated sequences as templates. Membranes were hybridized in Ultrahyb-Oligo hybridization buffer (Ambion). The signals in Northern blots were detected and quantified using PhosphorImager software (Amersham Biosciences).

Induction of EBV replication.

Virus replication was induced in latency I Akata cells or latency III B95-8 cells. About 2 × 10⁶ to 5 × 10⁶ Akata cells were treated with 50 μg per ml mouse anti-human IgG (Dako Cyto, Japan) at 37°C for 4 h. Fresh medium was added to keep the cells at 5 × 10⁵ cells per ml until harvest. EBV replication in B95-8 cells was induced by adding 4-hydroxytamoxifen to latency III-infected B95-8 LCL cells that stably express an RNA encoding a 4-hydroxytamoxifen-dependent fusion of the EBV immediately-early gene coding for BZLF1 (Zta) with a mutant estrogen receptor (34). Cell surface EBV gp350 late gene expression was assayed with MAb 72A1 as an indicator of late virus replication (53, 66).

5′-RACE.

Rapid amplification of cDNA 5′ ends (5′-RACE) was done as instructed by the manufacturer (Invitrogen). Briefly, the cDNAs were initially synthesized from 2 μg of total RNAs by SuperScript reverse transcriptase using a BHRF1 ORF-specific Racer1 primer (Table 1 and Fig. 1A). After RNase H treatment, polydeoxycytosines were added to the 3′ ends of the cDNA by terminal deoxynucleotide transferase. PCR was done with nested Racer2 primer (Table 1 and Fig. 1A) and universal primer (Gibco-BRL) in a total volume of 50 μl with Taq DNA polymerase (Invitrogen) at 95°C for 2 min with 35 cycles at 94°C for 50 s, 50°C for 50 s, and 72°C for 1.5 min, followed by extension at 72°C for 10 min. The amplicons were analyzed by 1.5% agarose gel electrophoresis, gel purified, cloned into pCR4-TOPO (Invitrogen), and sequenced. The same sequence was obtained from at least four individual clones for each EBV specific PCR product.

Nucleotide sequence accession number.

The DNA sequence of the EBV Akata strain BHRF1 gene has been deposited in GenBank under accession no. EF192979.

MiR-BHRF1-1, -2, and -3 are expressed at high level in latency III B cells, whereas miR-BART1 and -2 are expressed at high level in NPC cells.

Northern hybridization was done with ³²P-labeled miRNA probes to validate our detection of EBV miRNAs. MiR-BHRF1-1, -2, and -3 were clearly detected in latency III EBV-infected BJAB, BL41, B95-8, IB4, and recently derived LCLs, but not in latency II NPC C666-1 (18), C15 cells (9), or latency I-infected Akata BL cells (Fig. 1B) (data not shown). Cell miR-16 blots are included as loading controls (38). In contrast, miR-BART1 and -2 were expressed at lower levels in latency III-infected B cells and latency I-infected Akata BL cells than in latency II-infected NPC C666-1 and C15 cells (Fig. 1B), in accord with previous results (11, 16, 27, 36, 50, 56, 61). EBV miRNAs were not detected in NPC C17 cells, presumably related to the low EBV genome presence and expression (9). Thus, high-level miR-BHRF1-1, -2, and -3 are tightly associated with latency III, as opposed to latency I and II (11) (Table 2).

Latency III is associated with expression of a stable, unspliced, 1.3-kb BHRF1 ORF-containing RNA.

To further analyze BHRF1 latency III-associated RNAs, Northern hybridization was done with ³²P-labeled 5′, intron, BHRF1 ORF, and 3′ EBV DNA probes (Fig. 1A and Table 1). Northern analysis of total RNA detected an unspliced, 1.3-kb BHRF1 RNA in latency III-infected BJAB, BL41, and IB4 cells with BHRF1 ORF, 5′, and intron probes, but not with the 3′ probe (Fig. 1A and C and Table 2) (data not shown). This RNA was not detected in uninfected BJAB and BL41 BL cells, EBV latency I-infected Akata cells, or latency II-infected NPC C666-1, C15, or C17 cells (Fig. 1A and C) (data not shown), indicating that the 1.3-kb BHRF1 RNA is associated with miR-BHRF1-1, -2, and -3 expression and EBV latency III but not latency I or II infection.

To further characterize the BHRF1 1.3-kb RNA detected in Northern blots, the 5′ sequence of the BHRF1 RNAs from IB4 and B95-8 cells was obtained by 5′-RACE analysis using latency I Akata cell RNA as a negative control. A 0.75-kbp PCR product was amplified from BHRF1 1.3-kb RNA from IB4 and B95-8 cells (Fig. 1D). Four 0.75-kbp PCR product clones were sequenced. The 1.3-kb BHRF1 RNA began with 5′- AAGAAGGG-3′ (GenBank accession no. EF192979) (Fig. 1A and Table 2), which is the predicted 3′ Drosha cutting site for miR-BHRF1-1 (50). Since a Drosha cleavage 5′ of miR-BHRF1-2 would result in a 1,312-nt RNA versus 1,430 nt for cleavage 5′ of miR-BHRF1-3, the size of the RNA, detection with BHRF1 intron probe, and a specific inability to detect the 1.3-kb RNA with the BHRF1 3′ probe are evidence that the 1.3-kb BHRF1 RNA extends from 3′ of miR-BHRF1-1 to 5′ of miR-BHRF1-2. No EBV-specific PCR product was detected with RNAs from latency I-infected Akata cells (Fig. 1D) (sequence data not shown). These data indicate that latency III EBV infection is associated with the accumulation of miR-BHRF1-1, -2, and -3 and of a stable unspliced BHRF1 1.3-kb RNA and support a model that these RNAs are Drosha cleavage products of EBNA transcripts or introns.

BHRF1 RNAs after induction of EBV replication in latency I-infected Akata cells.

To test whether EBV BHRF1 miRNAs or 1.3-kb RNAs are also expressed as a consequence of early replication-associated BHRF1 mRNA expression or latency III EBNA expression late in EBV replication in latency I-infected Akata cells (72), EBV replication was induced in latency I-infected Akata cells (63, 64). Cell surface EBV gp350 late gene expression was assayed with MAb 72A1 as an indicator of late virus replication (53, 66). EBV gp350 expression was detected as early as 12 h, increased at 24, and peaked at 48 h, when 40% of Akata cells were gp350 positive by microscopy (data not shown).

Early EBV replication-associated BHRF1 protein (Fig. 2B) and spliced and polyadenylated BHRF1 1.4-kb mRNA (Fig. 3 and Table 2) were detected by 12 h after induction of EBV replication. BHRF1 1.4-kb mRNA [including 100-nt poly(A) tail] increased somewhat at 24 to 72 h and persisted at high levels (Fig. 3, ORF and 3′ probe). BHRF1 protein increased at 24 and 48 h and somewhat thereafter (Fig. 2B). As expected, BHRF1 mRNA hybridized to the ORF and 3′ probes but not the intron probe (Fig. 1A and 3 and Table 2). An EBV-specific 0.32-kbp PCR product with the beginning sequence 5′-TTTCATC-3′ (Fig. 1A and Table 2) was amplified by 5′-RACE analysis of RNA from induced Akata and B95-8 cells at 48 h (Fig. 1D, lanes 5 and 6). Similar products were not detected from latency I Akata or latency III B95-8 cell RNAs without induction of virus replication (Fig. 1D, lanes 3 and 4). Four clones of the 0.32-kbp 5′-RACE product were sequenced. The sequence start site was at nt 41501 in the EBV genome (GenBank accession no. AJ507799), 9 nt 5′ to nt 41510, the previous start site determined by S1 analysis (19), and 25 nt 3′ to the end of the BHRF1 TATTA sequence (TATA box).

At 12 h to 48 h, a small amount of 0.9-kb BHRF1 3′-cleaved mRNA was detected by BHRF1 ORF and 5′ probes but not by intron or 3′ probes (Fig. 3B and Table 2) (data not shown). Intron probe detected the 1.3-kb latency III RNA and a nonspecific 1.0-kb RNA (Fig. 3A). The 0.9-kb BHRF1 RNA significantly increased in abundance at 48 to 72 h (Fig. 3B). As previously described (72), late-replication-onset latency III-associated EBNA2, EBNA3C, and EBNA1 expression were detected at 24 to 48 h and were abundant at 72 h (Fig. 2B). In parallel with latency III EBNA expression, pre-miR-BHRF1-1, -2, and -3 were readily detected at 48 h and were abundant by 72 h (Fig. 2A). By 72 h, miR-BHRF1-1, -2, and -3 (Fig. 2A) and 1.3-kb Drosha-cleaved EBNA intron RNA were abundant (Fig. 3). The 1.3-kb RNA hybridized to 5′, intron, and ORF probes but not to the 3′ probe, consistent with cleavage 5′ to miR-BHRF1-2 or -3 (Fig. 3) (data not shown). The 1.3-kb RNA was identical in size to the latency III-associated EBNA intron RNA, which is Drosha cleaved 3′ to miR-BHRF1-1 and 5′ to miR-BHRF1-2, versus the 1.43 kb expected from Drosha cleavage 5′ to miR-BHRF1-3. Thus, activation of latency III EBNA transcription is the likely basis for the high-level miR-BHRF1-1, -2, and -3 and 1.3-kb residual EBNA intron RNA late in Akata cells (Fig. 3 and Table 2).

As miR-BHRF1-1, -2, and -3 became abundant at 48 to 72 h (Fig. 2A), the 0.9-kb ORF-positive and 3′-truncated RNA also became more abundant (Fig. 3B and Table 2). Thus, the 0.9-kb RNA has the size and sequence content consistent with cleavage of BHRF1 mRNA 3′ to miR-BHRF1-2 by Drosha in the nucleus or by miR-BHRF1-2 in the cytoplasm.

To further clarify the genesis of 0.9-kb BHRF1 3′-truncated RNA, cytoplasmic and nuclear RNAs were isolated from EBV-infected Akata cells at 48 h after virus replication induction, and RNAs were analyzed by Northern hybridization with ³²P-labeled BHRF1 ORF probe. As a control, the isolated RNAs were separated in 5% polyacrylamide gel and probed with ³²P-labeled DNA oligonucleotide probes specific for EBV-encoded EBER1 and EBER2 small nuclear RNAs (Table 1) (5, 24, 32, 41). As shown in Fig. 3D, EBERs were in the nuclear RNA fraction from Akata cells before and 48 h after induction of virus replication. Furthermore, the latency III EBNA-associated 1.3-kb unspliced and putatively Drosha-cleaved BHRF1 RNA was threefold more abundant in the 48-h nuclear versus cytoplasmic RNA (Fig. 3D and Table 2). Moreover, BHRF1 1.4-kb mRNA was fourfold more abundant in 48-h cytoplasmic versus nuclear RNA. Importantly, the 0.9-kb spliced and 3′-truncated BHRF1 RNA was 1.6-fold more abundant in the cytoplasm than in the nucleus, consistent with this RNA being mostly a product of miR-BHRF1-2 cleavage of BHRF1 mRNA in the cytoplasm, as postulated (38).

miR-BART1 and -2 expression did not change following induction of EBV replication in latency I-infected Akata cells (Fig. 2B).

EBV replication in latency III-infected B95-8 cells does not result in significantly increased miR-BHRF1-1, -2, or -3 or 1.3-kb RNA levels.

To evaluate the effects of preexistent latency III EBV infection on EBV replication-associated RNA levels, EBV replication was induced in B95-8 cells (34). Surface gp350 expression increased at 12 to 24 h and peaked at 48 h, when 50% of B95-8 cells were gp350 positive by microscopic evaluation (data not shown). In contrast to latency I-infected Akata cells, EBNA proteins (Fig. 4B), pre-miR- and miR-BHRF1-1, -2, and -3, and 1.3-kb residual Drosha-cleaved RNA barely changed at any time following induction of replication in latency III-infected B95-8 cells (Fig. 4A) (data not shown). BHRF1 mRNA and protein increased significantly by 12 h and were at high levels thereafter (Fig. 4B and C). BHRF1 0.9-kb RNA was also detected at a constant level by BHRF1 ORF and 5′ probes but not by intron and 3′ probes from 12 h onward (Fig. 4C and Table 1) (data not shown). These data indicate that EBV replication in preexisting latency III-infected cells is associated with little change in EBNA, miR-BHRF1-1, -2, or -3, or 1.3-kb RNA expression, early turn on of BHRF1 1.4-kb mRNA and 0.9-kb cleavage product, and a constant ratio of BHRF1 mRNA and cleavage product throughout replication.