Sequences Adjacent to oriP Improve the Persistence of Epstein-Barr Virus-Based Episomes in B Cells

Abstract

Epstein-Barr virus (EBV) oriP and the EBV nuclear antigen 1 (EBNA-1) protein allow persistence of EBV-based episomes. A nuclear matrix attachment region (MAR) spans oriP and the adjacent region of the EBV genome containing the EBV-expressed RNAs. Here, we show that episomes with the MAR are retained significantly more efficiently in EBV-positive B cells than episomes containing oriP alone.

Epstein-Barr virus (EBV) is a lymphotrophic gammaherpesvirus that is capable of lifelong persistence as an episome in human B cells. In combination with the EBV nuclear antigen 1 (EBNA-1) gene, a 1.8-kb region of the EBV genome designated oriP is sufficient in cis for retention and replication of plasmids in eukaryotic cells (19, 20). However, oriP-based episomes are gradually lost from continuously dividing cultures in the absence of selective pressure for their retention. Here, we study the effect of sequences adjacent to oriP on episome retention.

oriP consists of two major functional regions (reviewed in reference 10). The family of repeats (FR) is composed of 20 imperfect copies of a 30-bp repeat, each of which binds an EBNA-1 dimer. The FR is separated by 1 kb from a region of dyad symmetry (DS) which contains a further four EBNA-1 binding sites. The FR and EBNA-1 protein are essential for episomal maintenance, while the DS acts as an EBNA-1-dependent replication origin (1). Plasmids carrying the FR alone fail to undergo DNA replication (9), unless they contain an alternative source of replication origins, such as genomic DNA (3, 9) or other EBV sequences (7, 12), which only uses the cellular replication machinery.

Analysis of the viral genome in Raji cells, an EBV-positive Burkitt's lymphoma (BL) cell line, has shown that EBV is attached to the nuclear matrix. The matrix attachment region (MAR) of EBV extends over both oriP and the adjacent EBV-expressed RNA (EBER) genes (6). The EBER genes do not encode protein and have not been ascribed a function although they are transcribed by RNA polymerase III throughout viral latency. They are primarily associated with the rough endoplasmic reticulum but associate with the chromosomes during metaphase (14). The EBER genes can be deleted from EBV without abrogating its ability to infect, transform, and either remain latently stable within B lymphocytes or enter the lytic cycle (15). However, a recent study suggests that the EBER genes may play a role in blocking cellular apoptosis, thereby promoting oncogenicity in BL (8).

We have previously shown that EBV episomes can carry large genomic DNA inserts for prolonged periods in cell culture (16) and that genomic DNA transgenes can be functionally expressed from these episomes in mammalian cell culture (17). Here, we investigate the functional significance of the MAR that spans oriP and includes the EBER genes, with a view to improved understanding of EBV episomal systems.

We constructed a series of mini-EBV replicons containing a 60-kb human genomic DNA insert and three variants of oriP (Fig. 1). The mini-EBV plasmid amplicons (pmEBV) contained oriP, defined as positions 7338 to 9520 in the B95-8 strain (19, 20). The amplicon pmEBV-EBER (Fig. 1) contains EBV DNA from position 4948 to 9520, which includes oriP and the entire region previously defined as attached to the nuclear matrix. pmEBV-ΔDS is the same as pmEBV with the DS removed by excision of the EcoRV/HpaI 140-bp fragment. All plasmids also contained the EBV terminal repeats, oriLyt and an EBNA-1 expression cassette. To allow the tracking of the episome in mammalian cells, expression cassettes for enhanced green fluorescent protein (GFP) and hygromycin phosphotransferase were included.

FIG. 1.

Structure of episomal plasmid pmEBV-EBER. Expression cassettes for GFP and hygromycin phosphotransferase (hph) are indicated. Hatched arrows, promoters; black arrows, genes. EBV elements (grey) included are oriP (FR and DS regions), EBER1 and -2, EBNA-1, the terminal repeats, the lytic origin of replication (oriLyt), and the adjacent internal repeat 2 site (IR2). Genomic DNA (the insert from PAC 220G18; Research Genetics) was cloned into the NotI site indicated. Bacterial elements (white arrows) are shown. pmEBV lacks the EBV region flanked by the NruI and EcoRI sites indicated. pmEBV-ΔDS also lacks the DS region, which was excised at the EcoRV and HpaI sites closely flanking the DS.

The vectors were transfected into HH514 cells, a BL cell line that contains a nontransforming strain of EBV (13), using a peptide-Lipofectin reagent described elsewhere (2). Transfected cells were selected with hygromycin B, and several clonal cell lines were isolated by dilution cloning: three carried pmEBV (HHmE-C1-3), three carried pmEBV-EBER (HHmE-EBER-C1-3), and one carried pmEBV-ΔDS (HHmEΔDS-C1). DNA was isolated from these lines and was transformed into bacteria to recover the episome, using protocols described elsewhere (16). All episomes rescued from all of the HHmE and HHmE-EBER cell lines were identical to the original plasmid. Five of six rescued episomes analyzed from HHmEΔDS-C1 were unaltered, but one appeared to have undergone a deletion of part of its genomic DNA insert (results not shown). Copy number analysis by quantitative Southern blotting of genomic DNA from the human cell lines (17) indicated that they carried between 35 and 64 episomes per cell (Table 1). The presence of the EBV genome in HH514 cells prevented the direct detection of EBER genes attributable to the introduced episomes in this cell line. However, EBER expression was detected from the pmEBV-EBER construct by Northern blot analysis, following transfection of the plasmid into EBV-negative 293 cells (not shown).

TABLE 1.

Characteristics of EBV episomes in clonal transformants

Cell line	Episome copy no.	% Initial green cells (a)	Episome loss rate (%/generation) (b)	% Loss rate (mean ± SD)a
HHmE-C1	41	90.2	2.8
HHmE-C2	64	93.0	2.6	2.8 ± 0.2
HHmE-C3	52	92.8	3.0
HHmEΔDS-C1	51	92.3	3.1
HHmEBER-C1	36	98.6	1.4
HHmEBER-C2	35	97.1	1.5	1.4 ± 0.1
HHmEBER-C3	39	98.3	1.3

^aMean loss rate values are for cell lines HHmE-C1 and HHmEBER cell lines, respectively. 

We then grew each cell line in both the presence and absence of hygromycin B selection. Every 5 days, flow cytometry was used to measure the proportion of cells expressing GFP and hence carrying the episome (Fig. 2). The loss of GFP expression allowed us to follow the loss of episome and thus to quantify episome retention efficiency. In these experiments, we assume that green fluorescence equates to episomal status because we did not observe a plateau phenomenon or reduction in fluorescence typical of cells in which an integration event has significantly outgrown the population (16). A cell was defined as green if it fluoresced more strongly than 99.5% of untransfected HH514 cells. The loss of green fluorescence of a cell only begins to occur once it has lost all episomes and when preexisting GFP protein degrades, so a change in their green status will lag somewhat behind the loss of the episome. To model this, the data was fitted with a line of the form y = ae^b(x−c) where y is the number of generations and x is the percentage of cells which remain green. The constant a is the percentage of green cells at the start of the relaxation, b is the rate of episome loss in terms of the proportion of cells losing the episome each generation, and c is the lag time in generations.

FIG. 2.

Assay of episome retention. The percentage of green cells (defined as those fluorescing more than 99.5% of untransfected cells) is shown for each cell line grown in the presence (◊) and the absence (□) of hygromycin. The best-fit line in the form y = ae^b(x−c) for the unselected cells is calculated from the data lacking the first two data points. The variable a is the percentage of cells that are green as measured at the zero time point for the cell line. The variable b calculates the first-order rate of episome loss (percentage of cells losing episome per cell division), and c models the lag time between episome loss and loss of GFP expression. All the data for the cell lines are summarized in Table 1.

The results shown in Fig. 2 and Table 1 indicate that episomes containing the MAR are retained significantly more efficiently than those lacking the EBER genes (P = 0.0007 by the Student t test), despite the slightly lower episome copy number of the pmEBV-EBER construct. Thus, the half-life of episome-carrying cells in the absence of selective pressure almost doubles, from approximately 25 to 49 cell divisions (31 to 61 days). The episome in which the DS was deleted from oriP was lost at a rate similar to those retaining a full oriP. These latter data are consistent with recent results for a recombinant EBV deleted for the DS, which was lost at the same rate as virus with wild-type oriP (12). We have not tested the effect of deletion of the DS in the presence of the MAR.

Unexpectedly, the HHmE-EBER cell lines consistently expressed a higher level of GFP (reflected by the value of a) (Table 1), although it does not influence the calculations of episome retention. This may be mediated by the transcription factor binding sequences in the regulatory elements of the EBER genes (4), either acting alone or in conjunction with FR. It is unlikely that the EBER transcripts are directly responsible for this effect, as the HH514 cells express EBER genes from their native virus in all of the cell lines studied. It may be interesting to assess whether a similar effect is observed for EBV promoters that are activated by FR, such as the EBNA genes' promoter, pC, and the latent membrane protein 1 promoter.

This study has not established whether the improved retention of oriP-MAR requires EBV in trans. Since the EBER genes are expressed by the EBV genome in HH514 cells in all the cell lines studied, we can conclude that the presence of the EBER genes (or nearby DNA elements) in cis significantly improves the retention of EBV-based episomes. These elements may or may not require the presence of the EBER RNAs themselves. Further study of these episomes in an EBV-negative cell line could address this question. Alternatively, EBER genes minimally mutated to abrogate expression (18) could be useful reagents to probe their role in episome retention.

A recent report has shown that the EBV region spanning positions 6400 to 8300 exhibits an extremely high sensitivity to micrococcal nuclease, indicating either an unusual nucleosome structure or a complete absence of nucleosomes (18). This region includes the FR and both EBER genes (but not the DS) and is entirely contained within both the previously defined MAR (6) and the region mediating improved episome retention described herein. Taken together, these data suggest that the region of the EBV genome covering the EBER genes and the FR adopts an unusual structure that promotes association with the nuclear matrix. This may be mechanistically related to metaphase chromosome association, which is critical in episome maintenance (5).

Recent results from Leight and Sugden (11) reinforce this hypothesis. The authors found that a stochastic, epigenetic event is required for efficient partitioning of oriP plasmids (11), for which they invoke a mechanism involving chromatin configuration. Given that sites of transcriptional regulation and origins of replication are often clustered around cellular and viral MARs (6), incorporation of the point mutations used by Wensing et al. (18) to modulate the EBER promoter in our vector would be an elegant test of this hypothesis. Further elucidation of the cis and trans elements that promote these effects promises to enhance our understanding of episomal maintenance, nuclear matrix attachment, and EBV nucleosome organization and also holds the promise of improved episomal gene transfer vectors.