Epstein-Barr Virus Exploits BSAP/Pax5 To Achieve the B-Cell Specificity of Its Growth-Transforming Program^▿

Electrophoretic mobility shift assays.

Electrophoretic mobility shift assays were carried out for BSAP binding as previously described (41).

Luciferase reporter assays.

All luciferase reporter constructs used in this work were generated from Wp440/GL2 (6), which contains Wp sequences from −440 to +173 (relative to the RNA start site) cloned between the BglII and HindIII sites of pGL2 Basic (Promega). WpΔ1 was created by deleting the sequences between −264 and −87 by digestion with AvrII and SnaBI, blunt-ending with Klenow DNA polymerase, and recircularizing the vector fragment. WpΔ2 was created by deleting the sequences between −440 and −264 by digestion with NheI (present in the multiple cloning site of pGL2 Basic upstream of the BglII site) and AvrII and recircularizing the vector in the presence of an adaptor oligonucleotide which restored both the original BglII and AvrII sites. WpΔ1+Δ2 was created by deleting the UAS1 sequence from WpΔ2 as described above. A further derivative, WpBSAPm, in which both the BSAP B and D binding sites are inactivated in the context of an otherwise wild-type Wp440 sequence, was generated by using a Quickchange site-directed mutagenesis kit (Stratagene). Briefly, this was done by replacing the sequence 5′ ATGGTT 3′ (−234 to −229) with 5′ AGTACT 3′ within the BSAP site D sequence (simultaneously introducing a ScaI restriction site) of Wp440Bm, a Wp derivative in which BSAP site B has already been mutated (41). Luciferase reporter assays were performed as described previously (6).

ChIP assay.

The chromatin immunoprecipitation (ChIP) assay was performed essentially as described previously (8). Briefly, cells were treated with formaldehyde to cross-link protein-DNA interactions and total chromatin was isolated. Immunoprecipitation reactions were then performed with anti-BSAP antibody (C-20X polyclonal; Santa Cruz Biotechnology), anti-EBNA1 antibody (Ch.EBNA1) (8), or an anti-mouse immunoglobulin gamma (IgG) control antibody (Jackson Immunoresearch). Immunocomplexes were collected with protein A-Sepharose beads, the protein removed, and the DNA eluted and purified. The amounts of cellular (CD19) and viral (Wp, oriP, and Qp) DNA sequences in each immunoprecipitate were then quantified by PCR. Quantitative PCR (Q-PCR) assays were performed in 25-μl volumes containing 5 μl of input DNA, 1× universal Taqman PCR mastermix (Applied Biosystems), and the primer/probe combinations and concentrations shown in Table 1. Samples were cycled for 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s.

Recombinant viruses.

Recombinant EBV genomes were created by using the EBV bacterial artificial chromosome (BAC) cloning system (11; data not shown).

B-cell infection.

B cells were positively selected from buffy coats by using CD19 Dynabeads (Invitrogen) followed by detachment with CD19 Detachabead (Invitrogen) according to the manufacturer's protocol. The purified B cells were incubated overnight at 37°C with a recombinant virus preparation, with or without EBNA2KO virus, at known multiplicities of infection (MOI). The delivery of recombinant EBV genomes was assessed by fluorescence in situ hybridization (FISH) using a fluorescein-labeled EBV DNA cosmid probe (36). The level of EBV antigen expression was determined by immunofluorescence staining using JF186 (EBNA-LP), PE2 (EBNA2), and CS1-4 (LMP1) antibodies followed by a Cy3-conjugated anti-mouse IgG secondary antibody (Stratech). EBV transcripts were quantified by quantitative reverse transcriptase PCR (Q-RT-PCR) using previously described primer/probe combinations (7) on RNA isolated from 1 × 10⁶ infected cells per time point. Cell proliferation was examined by FACS analysis of 5- (and 6-) carboxyfluorescein diacetate (CFSE)-labeled B cells infected with each recombinant virus, with or without EBNA2KO virus coinfection, and 8-week transformation assays as described previously (36).

Epithelial cell infection.

AdAH cells were infected by using the transfer infection method (37) and examined by immunofluorescence cell staining with R4 (EBNA1), CS1-4 (LMP1) and BZ1 (BZLF1) antibodies followed by a suitable Cy3-conjugated secondary IgG antibody.

BSAP binding sites within Wp.

Based on reporter gene assays using progressive 5′ deletions of the 440-bp sequence upstream of the transcription start site, Wp contains a lineage-independent activating region, UAS2, with two YY1 binding sites, and a B-cell-specific region, UAS1, with RFX-, CREB-, and two BSAP-binding sites (Fig. 1A). We created a new set of reporter constructs in which these two regulatory regions had been deleted, either individually (Δ1 and Δ2) or together (Δ1+Δ2), or in which both BSAP sites had been mutated (BSAPm) using nucleotide substitutions previously shown to abrogate BSAP binding to site B (41) and shown here to abrogate BSAP binding to site D (Fig. 1B); note that the BSAP site D mutation introduced a ScaI restriction site (underlined in Fig. 1A) that could serve as a marker during recombinant virus construction. Transient transfection assays confirmed that the wild-type Wp440 construct was more active in B-cell lines (BJAB and DG75) than in non-B-cell lines (K562 and Jurkat) and that the Δ1 deletion markedly reduced reporter activity in B cells, while the Δ2 deletion had the greater effect in a non-B-cell environment (Fig. 1C and data not shown). Importantly, in all cases the effects of the Δ1 deletion could be reproduced by mutation of the BSAP sites within an otherwise wild-type construct.

We then used ChIP assays to ask whether an interaction could be detected between BSAP and Wp sequences in the viral genome of EBV-infected cells. To this end, a BSAP-specific antibody was used to precipitate protein-DNA complexes from chromatin preparations of two different LCLs; in parallel, we used an EBNA1-specific antibody, previously shown to detect EBNA1 bound to its two sites on the EBV genome, oriP and Qp (8), and an irrelevant control antibody. In each case, the immunoprecipitated DNA was amplified alongside the original input DNA in a series of Q-PCR assays using one of four primer pairs specific for Wp UAS1, for the BSAP-regulated CD19 gene promoter as a positive control for BSAP binding (24), and for oriP and Qp as positive controls for EBNA1 binding (8). The anti-BSAP immunoprecipitates showed a clear enrichment of both CD19 DNA and Wp UAS1 sequences compared to the levels in the control antibody and no enrichment of oriP and Qp sequences. Conversely, anti-EBNA1 immunoprecipitates were preferentially enriched in oriP and Qp sequences but not in Wp or CD19 DNA (Fig. 1D). These results were reproducible and provide the first evidence that BSAP does indeed bind to Wp sequences on the resident EBV genome in a latently infected B cell.

Recombinant viruses carrying Wp mutations.

We then engineered a series of recombinant EBVs based on BAC clone 2089, containing the B95.8 EBV genome (11) and carrying the different Wp sequences shown in Fig. 1A. Naturally occurring EBV isolates contain different numbers of copies of the BamHI W fragment (3), up to the 11 copies present in BAC clone 2089 (here called 11W). The manipulation of Wp sequences in the context of the viral genome required us to reduce the number of BamHI W copies to a minimum; note that this inevitably reduced the achievable size of one of the transforming proteins, EBNA-LP, since the W1 and W2 exons within BamHI W encode the N-terminal repeat domain of the EBNA-LP protein (35, 38). However, earlier work had shown that a virus with only two copies retains the transforming function (21) and so, using the 11W virus as a template, we generated a set of recombinants containing just two BamHI W repeats (Fig. 2A), in which both Wp sequences were either wild type (2W), had UAS1 (Δ1), UAS2 (Δ2), or UAS1 and UAS2 (Δ1+Δ2) deleted, or were mutated in both their BSAP sites (BSAPm) (data not shown).

To compare the efficiencies with which these different viruses could enter B cells, we used standard methods of FISH with an EBV cosmid probe capable of detecting individual EBV genomes in cell nuclei (10). As illustrated by the control results shown in Fig. 2B, uninfected B cells gave no detectable signals, whereas the Namalwa-BL reference line had the expected two genome copies per cell, seen as fluorescent green spots by microscopy, and the Raji-BL line had around 50 copies per cell (Fig. 2B). In a typical experiment, when resting B cells were exposed to a standard dose (50 MOI) of each of the recombinant viruses and analyzed by FISH 24 h later, nuclear viral genomes (typically 1 to 2 per infected cell) were detectable in each case (Fig. 2B) and roughly equal percentages of cells (25 to 35%) had become genome positive (Fig. 2C).

Effects of Wp mutations on B-cell-transforming infection.

We then used a series of Q-RT-PCR assays with transcript-specific EBV primer-probe combinations (7) to track the initial burst of Wp transcription that occurs within the first 1 to 2 days postinfection (p.i.) and drives the expression of EBNA2 and EBNA-LP mRNAs and is then followed by the activation of transcription from Cp. Cells exposed to the 2W wild-type virus showed easily detectable Wp-initiated transcripts and appropriately spliced EBNA2 mRNAs at day 1 p.i. and Cp-initiated transcripts by day 2 (Fig. 3A). Cells exposed to the Δ2 virus gave similar values, albeit with somewhat slower activation of EBNA2 transcription in this particular experiment. However, we never observed detectable Wp or Cp activity or EBNA2 transcription at any time up to day 10 p.i. with either the Δ1 deletion, the Δ1+Δ2 double deletion or, most significantly, the BSAPm virus (Fig. 3A and data not shown).

The Q-RT-PCR results indicated that the overall levels of viral transcripts seen from the 2W wild-type virus within the first 2 days p.i. were substantially lower than those seen with an 11W virus (Fig. 3A). While this might partly reflect an influence of Wp copy number per se on transcript levels immediately p.i., another possible factor was the inability of the 2W virus to make a sufficiently large EBNA-LP. To control for this, we carried out coinfections with a recombinant EBNA2 gene knockout (EBNA2KO) helper virus (4) which, though nontransforming, retains 11 wild-type BamHI W copies and can provide a full-length EBNA-LP in trans. Fig. 3B shows the transcript levels observed in cells infected with the 2W wild-type or BSAPm viruses used at an MOI of 50, either alone as above or with an equivalent MOI of the helper virus. Note that the EBNA2KO virus alone produces high levels of Wp transcription (expressed relative to that seen with 2W wild-type virus alone) but does not progress to Cp activation. Coinfection of the 2W wild-type virus with the EBNA2KO virus leads to a further increase in total Wp transcription and to levels of Cp and EBNA2 transcripts that are markedly higher than those seen with 2W virus alone; we infer that these Cp transcripts, like those of EBNA2, are expressed from the 2W virus. By contrast, coinfection with the EBNA2KO virus did not rescue any detectable expression from the BSAPm virus genome, whether monitored by an enhancement of overall Wp activity or by the appearance of either Cp-initiated or EBNA2 transcripts (Fig. 3B). When tested in parallel, the Δ2 virus behaved like the 2W wild type, whereas the Δ1 and Δ1+Δ2 viruses were as inactive as the BSAPm virus (data not shown).

In further experiments, cells exposed to 2W wild-type or mutant viruses (with or without the EBNA2KO helper) were harvested at day 4 p.i. and the progress of infection was monitored by immunofluorescence staining for EBNA-LP, EBNA2, and LMP1. Representative EBV antigen stains (red) and 4′,6′-diamidino-2-phenylindole (DAPI) counterstains (blue) of the same fields are shown in Fig. 4A, and the percentages of antigen-positive cells seen in one such experiment are given in Fig. 4B. Infection with 2W wild-type or Δ2 virus alone produced small numbers of EBNA-LP- and EBNA2-positive cells on day 4; LMP1 staining was first detected either on day 4, as seen here in the 2W Δ2 virus infection, or slightly later (data not shown). However, no EBNA-LP, EBNA2, or LMP1 staining was observed in cultures infected with Δ1, Δ1+Δ2, and BSAPm virus, even with the analysis extended up to 10 days p.i. In line with the Q-RT-PCR data, the EBNA2KO helper virus alone induced detectable EBNA-LP expression in the majority of cells, but never EBNA2 or LMP1. This helper virus caused a small but significant increase in the number of cells expressing EBNA2 and LMP1 in response to 2W and Δ2 virus infection, seen consistently in several experiments. However, even in the presence of the helper virus, we never observed EBNA2- or LMP1-positive cells following Δ1, Δ1+Δ2, or BSAPm virus infection.

The ultimate test of a virus's B-cell-transforming activity requires the analysis of cultures for morphological change and for cell proliferation leading to successful LCL outgrowth. Cultures infected with the 2W wild-type and Δ2 recombinants did not contain visible foci at day 5 p.i., but foci clearly did develop later and were easily detectable by day 28 (Fig. 5A); these same viruses did induce foci by day 5, however, in coinfections with the EBNA2KO helper. Note that the original 11W wild-type virus, included here for comparison, formed foci by day 5 with and without the helper. By contrast, no foci developed in cultures infected with Δ1, Δ1+Δ2, or BSAPm viruses, whether alone or with the helper. To quantitate the early proliferative events, B cells were stained with CFSE prior to infection with each of the above viruses plus the EBNA2KO helper. FACS profiles of CFSE levels on days 4 and 8 p.i. (Fig. 5B) confirmed proliferation induced by the 2W wild-type and Δ2 viruses, albeit less extensive than in 11W-infected cultures, whereas no proliferation was detectable using the other mutants; note that the helper virus (control, Fig. 5B) was also incapable of driving cells into cycle. Finally, we titrated the 2W wild-type and BSAPm viruses in long-term transformation assays, infecting B cells with these agents at doses of 0.5 to 100 MOI (with and without 50 MOI of helper virus), plated the cells at 10⁴ cells/well in 96-well plates, and after 8 weeks, scored all wells with active growth leading, where subcultured, to LCL establishment. The highest dose of 2W wild-type virus induced transformation in every well if given with the helper virus and in >50% of wells if given alone (Fig. 5C); LCLs were obtained down to 2W virus doses of 2 and 30 MOI in the presence and absence of the helper, respectively. The Δ2 virus (data not shown) gave equivalent yields. However, the BSAPm virus (and the Δ1 and Δ1+Δ2 viruses, data not shown) never yielded LCLs either in this type of assay or in other high-MOI infections seeding cells at higher densities to optimize the chances of outgrowth.

Effects of Wp mutations on nontransforming infection of epithelial cells.

Finally, we compared the same viruses for their ability to initiate epithelial cell infection by using the recently developed protocol of transfer from the surface of virus-loaded resting B cells (37). Note that epithelial cell infections lead to the activation of Qp rather than Wp/Cp and to the expression of EBNA1 in the absence of the other EBNAs; in addition, some infected cells express LMP1 and/or the immediate-early lytic cycle antigen BZLF1 (23, 37). Cells of the AdAh epithelial cell line were cocultivated for 10 min with B cells that had been recently exposed to 25 MOI of 2W wild-type virus or the 2W mutants and then cultured for 4 days before being stained for EBNA1, LMP1, and BZLF1. Representative examples of staining are shown in Fig. 6A, and the percentages of antigen-positive cells seen in one such experiment are shown in Fig. 6B. All viruses showed active epithelial cell infection in this and other experiments, with no consistent differences between the recombinants in the levels of EBNA1, LMP1, or BZLF1 staining observed. Significantly, there was no detectable Wp or Cp activity in these cells and no detectable staining for EBNA2 (data not shown). This strongly suggests that the Wp mutant viruses that fail to transform B cells are perfectly competent to initiate infections in a cell environment that does not naturally involve Wp activation.