Epstein-Barr Virus Nuclear Antigen 1 Replication and Segregation Functions in Nasopharyngeal Carcinoma Cell Lines

Abstract

Nasopharyngeal carcinomas (NPC) are usually Epstein-Barr virus (EBV) positive, but, with the exception of C666-1 cells, these cells lose the EBV genomes when grown in culture. Maintenance of EBV requires the viral EBV nuclear antigen 1 (EBNA1) protein, which ensures the replication and mitotic segregation of the genomes through interactions with OriP. Here we compare the abilities of C666-1 and NPC cells that have lost EBV genomes to replicate and segregate OriP plasmids. We found that either cell line can replicate and maintain OriP plasmids for extended periods under conditions where low levels of EBNA1 are expressed but that high EBNA1 levels selectively interfered with mitotic segregation.

TEXT

Epstein-Barr virus (EBV) is causatively associated with the development of nasopharyngeal carcinoma (NPC), as almost all undifferentiated NPCs are comprised of monoclonal expansions of cells latently infected with EBV. These tumor cells express the viral Epstein-Barr virus nuclear antigen 1 (EBNA1) protein and the secreted BARF1 protein and sporadically express latent membrane protein 1 (LMP1) (11, 17). EBNA1 is essential for the persistence of the EBV episomal genomes due to its important roles in both the replication and the mitotic segregation of the EBV episomes (reviewed in references 5, 6, and 10). EBV persistence also requires the OriP sequence that contains both an origin of DNA replication (DS element) and a segregation element (FR). DNA replication requires EBNA1 binding to the DS element, while segregation requires EBNA1 binding to the FR as well as EBNA1-mediated tethering of the episomes to the host mitotic chromosomes. While virtually all undifferentiated NPCs are EBV positive, studies of the persistence of EBV in these cells have been hampered by the fact that these cells lose the EBV genomes when grown in culture. In fact, while EBV episomes are stably maintained in many B-cell lines, they are generally not stably maintained in epithelial cell lines (4, 13). A notable exception to this rule is the C666-1 cell line, derived from a xenograft of an undifferentiated NPC, which stably maintains the EBV episomes (in nonintegrated form) (2, 3). Here we examine whether the ability of C666-1 cells to maintain EBV episomes better than other NPC cell lines reflects differences in OriP-mediated functions in replication or segregation.

We began by comparing the abilities of various cell lines to maintain a plasmid containing OriP (shown in Fig. 1 A). Two EBV-negative NPC cell lines, CNE2Z (16) (Fig. 1B and C) and HK-1 (8) (Fig. 1D), were transfected with an OriP plasmid that expresses EBNA1 from a cytomegalovirus (CMV) promoter (OriPE) and then propagated without selection for the plasmid for the numbers of cell doublings indicated in Fig. 1 (one cell doubling is 1 day for these cell lines). Plasmids were then recovered from the cells, linearized, and treated with DpnI in order to digest any of the input plasmid that had not undergone DNA replication (the majority of the plasmid in the 3-day samples). As expected, neither of these cell lines was able to replicate or maintain the negative-control OriP plasmid that lacked EBNA1 expression (Fig. 1C, lanes 1 to 5; Fig. 1D, lanes 1 to 3). In contrast, OriPE plasmids replicated, resulting in DpnI-resistant bands in all of the cell lines after three cell doublings (Fig. 1B, lane 13; Fig. 1C, lane 10; Fig. 1D, lane 8). Comparison of the amounts of DpnI-resistant OriPE at various time points showed that these plasmids were not maintained past 14 doublings and that there was a noticeable loss of plasmid between the 7- and 14-doubling time points for CNE2Z. This plasmid loss rate was consistent with that observed in HeLa cells (Fig. 1B).

Fig. 1.

Comparison of maintenance of OriP plasmids in multiple cell lines. (A) Schematic representation of the OriP and OriPE plasmids used in this study. The DS element and FR of OriP are indicated by the dark boxes. OripE,GA is the same as OriPE except that the EBNA1 cDNA is 670 bp larger. hCMV, human CMV. (B) The abilities of CNE2 and HeLa cells to replicate and maintain the OriPE plasmid were compared. Cells grown in a 10-cm dish were transfected with 8 μg of pc3oriP plasmid expressing EBNA1 (OriPE) (14) using Lipofectamine 2000 (Invitrogen) and, 24 h later, were moved to a 15-cm dish, where they were allowed to grow for 48 h. At this point, 5 × 10⁶ cells were harvested (3-day or 3-cell-doubling time point) to monitor replication of the plasmid. The remaining cells were replated and harvested after 7, 14, 21, and 28 cell doublings to monitor plasmid loss. In all cases, cells were lysed by the Hirt method, and low-molecular-weight DNA was isolated as previously described (1, 14). Recovered plasmids were linearized with XhoI and, where indicated, were also incubated with DpnI to digest unreplicated plasmids. For samples without DpnI digestion, 1/10 of the amount of sample was loaded relative to DpnI-digested samples. DNA was separated by agarose gel electrophoresis, Southern blotted, and probed with ³²P-labeled pc3oriP. The position of the DpnI-resistant plasmid is indicated by the asterisk. A similar experiment performed with an OriP plasmid lacking EBNA1 after 3 cell doublings is shown for HeLa cells in lanes 1 and 2 as a negative control. Markers of the positions of linear OriP and OriPE plasmids are shown in lanes 3 and 4. (C) The abilities of CNE2 cells to replicate and maintain plasmids OriP, OriPE, and OriPE,GA (expressing a version of EBNA1 with a long Gly-Ala repeat [described in reference 14]) were monitored as described for panel A. Markers of the positions of linear OriP and OriPE are shown in lanes 7 and 8. (D) Replication and maintenance of OriP, OriPE, and OriPE,GA plasmids in HK-1 cells were followed as described for panel B. (E) Replication and maintenance of pc3oriP and the parental pCAN plasmid without OriP (negative control) were assayed in C666-1 (EBV-positive) cells as described above. Plasmid maintenance was monitored for 3 to 42 cell doublings as indicated (corresponding to 6 to 84 days).

We also followed the maintenance of an OriP plasmid in the EBV-positive C666-1 cell line (Fig. 1E). Since this cell line already expressed EBNA1, EBNA1 expression was not supplied on the OriP plasmid. The OriP plasmid was seen to persist at stable levels for at least 28 doublings and was still detectable after 42 doublings (equivalent to 84 days, since the C666-1 cell doubling time is approximately 48 h). The pCAN plasmid lacking OriP sequences was used as a negative control and, as expected, showed no replication or maintenance (Fig. 1E, lanes 1 to 4). These results show that C666-1 cells can maintain an OriP-based plasmid considerably better than any of the other epithelial cell lines.

EBNA1 contains a Gly-Ala repeat region that is variable in length in different isolates. C666-1 cells express a version of EBNA1 with a long Gly-Ala repeat region, whereas the EBNA1 expressed on the OriPE plasmid has very little of this Gly-Ala repeat. In order to determine if the length of the Gly-Ala repeat affected plasmid maintenance, we repeated the plasmid loss experiments in CNE2Z and HK-1 cells using a version of OriPE that expressed EBNA1 with the long Gly-Ala repeat (from the B95-8 strain of EBV) (Fig. 1C and D, OriPE,GA lanes). However, OriP plasmid maintenance was not improved with the larger EBNA1.

The stable maintenance of OriP plasmids requires a combination of two EBNA1-mediated functions, DNA replication and mitotic segregation. The replication of the OriP plasmids can be measured directly by determining the level of DpnI-resistant plasmid after 2 to 3 cell doublings (because at such an early time point plasmids are maintained in the cells regardless of whether they have a segregation mechanism) (14). To determine whether the increased plasmid maintenance observed in C666-1 cells was due to increased efficiency of OriP plasmid replication, the levels of DpnI-resistant plasmids were compared in C666-1, CNE2Z, and HeLa cells after 3 cell doublings and quantified relative to the amount of total OriP plasmid recovered at the same time point (input) to account for any differences in initial transfection efficiencies (Fig. 2 A and B). The results show that C666-1 cells do not replicate OriP plasmids more efficiently than the other cell lines, suggesting that the increased ability to maintain plasmids is due to more-efficient mitotic segregation.

Fig. 2.

Comparison of replication of OriPE plasmids in HeLa, CNE2, and C666-1 cells. (A) Cells were transfected with OriPE plasmids and harvested after 3 cell doublings (3 days for HeLa and CNE2 cells and 6 days for C666-1 cells) as described in the legend for Fig. 1B. Plasmid DNA was isolated and linearized as described for Fig. 1B, and 1/10 of the sample was analyzed directly as a recovery control (Input). The remaining sample was incubated with DpnI to digest any plasmid that was not replicated in the human cells (+DpnI). Samples were then analyzed by Southern blotting as described for Fig. 1B, except that pCAN (the backbone of the OriP plasmid) was used as the probe to avoid hybridization with the EBV genomes in C666-1 cells. Each experiment is shown in triplicate (lanes 1, 2, and 3). Only the DpnI-resistant plasmid is shown in the top panel. A marker of the position of linearized OriPE is shown in the first lane. (B) Quantification of the plasmid replication results shown in panel A. Linearized plasmid bands were quantified by phosphorimager analysis using ImageQuant software (Molecular Dynamics), and the intensities of the DpnI-resistant bands were quantified relative to the input band for the same sample. Average values are shown relative to values from CNE2 cells. Error bars indicate standard deviations.

To more directly compare the abilities of CNE2Z and C666-1 cells to maintain OriP plasmids, we integrated an expression cassette for EBNA1 in CNE2Z cells to generate CNE2E cells. These cells express EBNA1 at levels similar to C666-1 cells (15), enabling us to follow the maintenance of the same OriP plasmid used for assays with C666-1 cells. The OriP plasmid was maintained in CNE2E cells considerably longer than was OriPE in CNE2 cells, as it was still detected after 42 days (Fig. 3 A, lanes 10 to 18). These results are similar to those seen with C666-1 cells (Fig. 1D), indicating that CNE2Z cells are not inherently unable to maintain OriP plasmids.

Fig. 3.

Plasmid maintenance and replication assays in CNE2E cells. (A) CNE2E cells were transfected with pCAN (negative control), OriP, or OriPE plasmid as indicated, harvested after 3 to 56 cell doublings (3 to 56 days), and processed as described in the legend for Fig. 1B. The amount of DpnI-resistant plasmid at each time point (marked by the asterisk) was quantified by phosphorimager analysis from three experiments and plotted relative to the 3-cell-doubling amount, which was set to one (histogram). Average values with standard deviations are shown. (B) Transient-replication assays comparing the bands from OriP and OriPE plasmids after 3 cell doublings before (Input) and after (+DpnI) DpnI digestion. The positions of DpnI-resistant bands are indicated by the asterisk. The histogram shows quantification of DpnI-resistant bands relative to input bands from three experiments. Results are plotted relative to the signal from the OriP plasmid (set to one). Error bars indicate standard deviations. (C) Western blot confirming the overexpression of EBNA1 in OriPE samples compared to endogenous EBNA1 in CNE2E. Total cell lysates (30 μg) are shown from CNE2E 3 days posttransfection with pCAN, OriP, or OriPE plasmid. Blots were probed with antibodies against EBNA1 (a mix of R4 rabbit serum [7] and OT1X monoclonal antibody) and actin (CP01; Calbiochem). Values at left are molecular size markers in kilodaltons. (D) CNE2E cells were transfected with OriP or OriPE plasmid as for replication assays. Cells (1 × 10⁵) were then plated in four 6-cm dishes and harvested and counted 1, 3, 5, or 7 days later. Average cell numbers from three independent experiments are plotted, with standard deviations.

A major difference in the experiments comparing maintenance of OriP plasmids and OriPE plasmids is that the EBNA1 levels are considerably higher when expressed from OriPE than the levels expressed in CNE2E or C666-1 cells. Therefore, we asked whether the higher levels of EBNA1 expressed from OriPE were detrimental for plasmid maintenance. This was tested by comparing maintenance of OriP and OriPE plasmids in CNE2E cells (Fig. 3A) as well as in C666-1 cells (Fig. 4 A), both of which already express low levels of EBNA1. In both cell lines, we found that the OriP plasmid was maintained for longer periods of time than the OriPE plasmid. However, comparison of replication of OriP and OriPE showed that the higher EBNA1 levels did not decrease replication efficiency in either cell line (Fig. 3B and 4B). In fact, relative to the total plasmid taken up by the cell (input), the OriPE plasmid replicated somewhat more efficiently than the OriP plasmid in C666-1 cells (histogram in Fig. 4B). The increased replication efficiency shows that OriPE is not cytotoxic to the cells. EBNA1 overexpression from OriPE is confirmed in the Western blots shown in Fig. 3C (CNE2E cells) and Fig. 4C (C666-1 cells). In addition, the proliferation rates of CNE2E (Fig. 3D) and C666-1 (Fig. 4D) cells were not significantly affected by the presence of OriPE compared to OriP, despite the fact that the majority of the cells contained the plasmids (60% and 80% transfection efficiencies for C666-1 and CNE2 cells, respectively, as determined by immunofluorescence microscopy for EBNA1 in OriPE). Taken together, the results indicate that higher EBNA1 levels are detrimental to plasmid segregation, resulting in decreased OriP plasmid maintenance.

Fig. 4.

Plasmid maintenance and replication assays in C666-1 cells. (A) C666-1 cells were transfected with pCAN (negative control), OriP, or OriPE plasmid as indicated, harvested after 3 to 42 cell doublings (6 to 84 days), and processed as described in the legend for Fig. 1B, except that pCAN (the backbone of the OriP plasmid) was used as the probe to avoid hybridization with the EBV genomes in C666-1 cells. DpnI-resistant bands were quantified as described in the legend for Fig. 3A, and results are shown in the histogram. (B) Transient-replication assays comparing the bands from OriP and OriPE plasmids after 3 cell doublings (6 days) before (Input) and after (+DpnI) DpnI digestion. The positions of DpnI-resistant bands are indicated by the asterisk. The double bands seen in lanes 5 and 8 are due to incomplete linearization of OriPE, resulting in both nicked (upper band) and linearized (lower band) forms of the plasmid. The histogram shows quantification of DpnI-resistant bands relative to input bands from three experiments. Results are plotted relative to the signal from the OriP plasmid (set to one). Error bars indicate standard deviations. (C) Western blot confirming the overexpression of EBNA1 in OriPE samples (EBNA1) compared to endogenous EBNA1 in C666-1 cells (GA), which is larger due to having a long Gly-Ala repeat region. Total cell lysates (30 μg) from C666-1 cells 6 days posttransfection with pCAN, OriP, or OriPE plasmid were Western blotted as described for Fig. 3C. Values at left are molecular size markers in kilodaltons. (D) C666-1 cells were transfected with OriP or OriPE plasmid as for replication assays. Cells (3 × 10⁵) were then plated in four 6-cm dishes and harvested and counted 4, 8, 12, and 14 days later. Average cell numbers from three independent experiments are plotted, with standard deviations. (E) C666-1 cells were transfected with OriP or OriPE plasmid, and cells were propagated for the indicated numbers of cell doublings before they were lysed in radioimmunoprecipitation assay (RIPA) buffer and total DNA was isolated. Quantitative RT-PCR on the EBV genomes was performed using a primer set for the BZLF1 gene, and results were normalized to the host GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene. The histogram shows average levels of the EBV genomes and standard deviations from two independent experiments, where the signal from the 3-doubling time point is set to one.

We then asked whether introducing OripE into C666-1 cells would lead to loss of the EBV episomes from C666-1 cells due to interference with episome segregation. However, quantification of the EBV episomes by quantitative real-time PCR (RT-PCR) (using primers to the BZLF1 region) showed no obvious decrease at any time point up to 24 cell doublings after transfection with OriPE (Fig. 4E). This suggests that, while higher EBNA1 levels may disrupt segregation of plasmids containing only OriP, additional factors affect whether or not full EBV episomes are maintained. This likely includes selective pressure to maintain the EBV genomes due to expression of viral proteins and RNA molecules on which the C666-1 cells may have become dependent for growth. This is in contrast to EBV-negative NPC cell lines, which are clearly not dependent on EBV, likely due to acquired genetic alterations that enable continued proliferation without EBV.

In summary, our results show that NPC cells that have lost EBV genomes through growth in culture are able to support EBNA1-mediated replication and segregation from OriP and therefore are not fundamentally different from C666-1 cells in that regard. However, EBNA1-mediated segregation was found to be sensitive to high levels of EBNA1, suggesting that the level of EBNA1 expression in the initial EBV-positive cells may be a factor (although not the only factor) in whether or not the episomes are stably maintained in epithelial cells when cells are placed in culture. The requirement of a precise EBNA1 level for segregation may reflect the need for the formation of a ternary complex (for example, interactions between FR-bound EBNA1 and EBP2 on mitotic chromosomes [9]), which would be disrupted by excess EBNA1, and may be one reason that EBNA1 expression from Qp (the EBNA1 promoter used in epithelial cell infection) is subject to autoregulation (12, 18).