The amino terminus of Epstein-Barr Virus (EBV) nuclear antigen 1 contains AT hooks that facilitate the replication and partitioning of latent EBV genomes by tethering them to cellular chromosomes

Abstract

During latency, Epstein-Barr virus (EBV) is stably maintained as a circular plasmid that is replicated once per cell cycle and partitioned at mitosis. Both these processes require a single viral protein, EBV nuclear antigen 1 (EBNA1), which binds two clusters of cognate binding sites within the latent viral origin, oriP. EBNA1 is known to associate with cellular metaphase chromosomes through chromosome-binding domains within its amino terminus, an association that we have determined to be required not only for the partitioning of oriP plasmids but also for their replication. One of the chromosome-binding domains of EBNA1 associates with a cellular nucleolar protein, EBP2, and it has been proposed that this interaction underlies that ability of EBNA1 to bind metaphase chromosomes. Here we demonstrate that EBNA1's chromosome-binding domains are AT hooks, a DNA-binding motif found in a family of proteins that bind the scaffold-associated regions on metaphase chromosomes. Further, we demonstrate that the ability of EBNA1 to stably replicate and partition oriP plasmids correlates with its AT hook activity and not its association with EBP2. Finally, we examine the contributions of EBP2 toward the ability of EBNA1 to associate with metaphase chromosomes in human cells, as well as support the replication and partitioning of oriP plasmids in human cells. Our results indicate that it is unlikely that EBP2 directly mediates these activities of EBNA1 in human cells.

FIG. 1.

(A) Schematic diagram of EBNA1. Domain A spans aa 33 to 89, and domain B spans aa 328 to 378. The presumptive AT hooks within domains A and B are compared to the known AT hooks of HMGA1a (GenBank accession number NM_145899) and MBD2a (GenBank accession number NP_003918). Amino acid mismatches are italicized and underlined. (B) Bacterially expressed domains A and B bind AT-rich DNA in vitro. One hundred twenty-five nanograms of labeled poly(dA-dT) probe was incubated in the presence of 1 μg of poly(dA-dC)(dG-dT), and increasing amounts of bacterially expressed HMGA1a, domain A or domain B, after which reactions were stopped and filtered through nitrocellulose. Nitrocellulose filters were washed and scintillation counted. The binding curve obtained with each protein is indicated by the legend within the graph. (C) The association between AT-rich DNA and domains A and B is specific. We incubated 0.1 μM HMGA1a, 0.2 μM domain A, 0.2 μM domain B, or 0.1 μM bEBNA1 with 125 ng of labeled poly(dA-dT) probe, 1 μg of poly(dA-dC)(dG-dT), and increasing amounts of cold poly(dA-dT), poly(dG-dC), or a HindIII digest of phage lambda as competitor DNA prior to assay by filter binding. For bEBNA1, a fragment of AGP74 that contains oriP was also used as a cold competitor. The binding curve obtained with each competitor is indicated by the legend within each graph. For all of the proteins tested, only cold poly(dA-dT) functioned as an effective competitor. (D) Distamycin A inhibits the association of HMGA1a, domain A, and domain B with poly(dA-dT). Increasing amounts of distamycin A were preincubated with the labeled poly(dA-dT) probe (125 ng) for 10 min. At this time, 0.2 μM HMGA1a, domain A, or domain B was added, along with 1 μg of poly(dA-dC)(dG-dT). Reactions were incubated for a further 20 min and then assayed by filter binding, followed by scintillation counting.

FIG. 1.

(A) Schematic diagram of EBNA1. Domain A spans aa 33 to 89, and domain B spans aa 328 to 378. The presumptive AT hooks within domains A and B are compared to the known AT hooks of HMGA1a (GenBank accession number NM_145899) and MBD2a (GenBank accession number NP_003918). Amino acid mismatches are italicized and underlined. (B) Bacterially expressed domains A and B bind AT-rich DNA in vitro. One hundred twenty-five nanograms of labeled poly(dA-dT) probe was incubated in the presence of 1 μg of poly(dA-dC)(dG-dT), and increasing amounts of bacterially expressed HMGA1a, domain A or domain B, after which reactions were stopped and filtered through nitrocellulose. Nitrocellulose filters were washed and scintillation counted. The binding curve obtained with each protein is indicated by the legend within the graph. (C) The association between AT-rich DNA and domains A and B is specific. We incubated 0.1 μM HMGA1a, 0.2 μM domain A, 0.2 μM domain B, or 0.1 μM bEBNA1 with 125 ng of labeled poly(dA-dT) probe, 1 μg of poly(dA-dC)(dG-dT), and increasing amounts of cold poly(dA-dT), poly(dG-dC), or a HindIII digest of phage lambda as competitor DNA prior to assay by filter binding. For bEBNA1, a fragment of AGP74 that contains oriP was also used as a cold competitor. The binding curve obtained with each competitor is indicated by the legend within each graph. For all of the proteins tested, only cold poly(dA-dT) functioned as an effective competitor. (D) Distamycin A inhibits the association of HMGA1a, domain A, and domain B with poly(dA-dT). Increasing amounts of distamycin A were preincubated with the labeled poly(dA-dT) probe (125 ng) for 10 min. At this time, 0.2 μM HMGA1a, domain A, or domain B was added, along with 1 μg of poly(dA-dC)(dG-dT). Reactions were incubated for a further 20 min and then assayed by filter binding, followed by scintillation counting.

FIG. 2.

(A) Schematic representation of EBNA1 and derivatives lacking domain B but containing one or more copies of domain A. The three A domain derivatives contained one, two, or three copies of domain A (aa 33 to 89) fused to the EBNA1 NLS (aa 379 to 386), and the DBD (aa 451 to 641). In derivatives that contain two or more A domains, adjacent A domains are separated by a short peptide linker with the sequence QSWS. The version of EBNA1 used here contains only five copies of the Gly-Gly-Ala repeat between domains A and B but functions like wild-type EBNA1 (2, 48). Retroviral vectors expressing EBNA1, DBD, 1A-DBD, 2A-DBD, and 3A-DBD were used to transduce 293 cells and establish cell lines that stably express each of these proteins. (B) An immunoblot analysis of clones of 293 cells that express the proteins described above. Cell extracts were prepared from 293 cell clones and immunoblotted with a rabbit polyclonal antibody that recognizes the DBD of EBNA1. The cell line used is indicated above each lane. All of the derivatives migrated at their expected mobilities relative to wild-type EBNA1.

FIG. 3.

Domain B is required for EBNA1 to associate with the cellular protein EBP2. 293 cells were transiently transfected with a FLAG-EBP2 expression plasmid and expression plasmids for the indicated derivative of EBNA1 (indicated above each lane) or with just the FLAG-EBP2 expression plasmid (lane No EBNA1). Extracts were prepared as described in the text and immunoprecipitated with a rabbit polyclonal antibody to the DBD of EBNA1. Immunoprecipitates were analyzed by SDS-PAGE and then immunoblotted with a rat monoclonal antibody to EBNA1 (A) or a mouse monoclonal antibody to FLAG-EBP2 (B). Only EBNA1 and EBNA1Δ386-450, which both contain domain B, were observed to immunoprecipitate FLAG-EBP2. (C) An immunoblot against FLAG-EBP2 with a 5% load of whole-cell extract from each of 293 cell cotransfections, indicating that the expression of FLAG-EBP2 was approximately equal in all transfections. The antibodies used for immunoprecipitation (IP) or immunoblotting (IB) are indicated adjacent to each blot. The positions of molecular mass markers on the SDS-PAGE gels are also indicated in kilodaltons.

FIG. 4.

1A-DBD, 2A-DBD, and 3A-DBD bind metaphase chromosomes. Metaphase chromosomes were isolated from 293, 293/DBD, 293/EBNA1, 293/1A-DBD, 293/2A-DBD, and 293/3A-DBD cells stalled in mitosis by Colcemid treatment. Indirect immunofluorescence was performed by using the K67.3 rabbit polyclonal antibody against the DBD of EBNA1. Images of individual layers (40 z-sections of 100 nm each) were captured using a ×100 objective, with ×1.5 optical enhancement, and deconvolved by using softWoRx. The cell line used for metaphase chromosome isolation is indicated above each panel. All three proteins were observed to localize to metaphase chromosomes in discrete punctate spots that resemble those observed previously with wild-type EBNA1 or HMGA1a-DBD (48).

FIG. 5.

EBNA1 derivatives 1A-DBD, 2A-DBD, and 3A-DBD support the episomal replication of oriP plasmids. (A) The oriP replication reporter plasmid, AGP74 (20FR), was transfected into 293/EBNA1, 293/1A-DBD, 293/2A-DBD, or 293/3A-DBD cells, along with a GFP expression plasmid. Two days after transfection, live transfected cells were recovered and subjected to puromycin selection for an additional 5 days (total = 7 days) or 19 days (total = 21 days). At this time, episomal DNAs were recovered, digested with DpnI, linearized, and quantified by Southern blotting. The cell line transfected is indicated above each lane, as is the time point at which episomal plasmids were recovered. The migration of standards is indicated by an arrowhead and “A,” and the linearized plasmid DNA is indicated by an arrowhead and “B.” Copy numbers of replicated plasmids per cell at each time point from three replicates of this experiment are indicated in Table 1. 1A-DBD only supported the replication of AGP74 (20FR) up to 7 days posttransfection but not at the later time point. 2A-DBD and 3A-DBD supported replication of this plasmid at both time points. (B) The replication defect in 1A-DBD is compensated by a plasmid, AGP264, that contains 30 EBNA1-binding sites in FR. AGP74 (20FR) or AGP264 (30FR) were transfected into 293/1A-DBD cells, which were processed as in panel A. Replicated episomal DNAs were detected for both plasmids at 7 days posttransfection, but only for AGP264 (30FR) at 21 days posttransfection, indicating that increasing the number of EBNA1 binding sites in FR compensated for the defect in 1A-DBD. Copy numbers for replicated plasmids per cell at each time point from two replicates of this experiment are indicated in Table 1. As a control for DpnI digestion, AGP74 was transfected into 293 (C) or 293/EBNA1 cells (D). Hirt DNAs were recovered at 4, 5, or 6 days posttransfection, digested with DpnI, linearized, and quantified by Southern blot. The DpnI fragments are indicated by “C,” whereas DpnI-resistant DNAs and markers are indicated as previously.

FIG. 6.

(A) Schematic representation of the EBP2-DBD fusion protein in which the EBNA1 amino terminus (aa 1 to 450) is replaced by human EBP2. EBNA1's NLS (aa 379 to 386) was added to the fusion junction. A retrovirus vector expressing EBP2-DBD was used to transduce 293 cells and create a 293 derivative cell line that stably expresses this protein (293/EBP2-DBD). (B) Endogenous EBP2 localizes to nucleoli in interphase human cells. The stain or antibody used is indicated above each panel. HeLa cells were transfected with an expression plasmid for the nucleolar protein MPP10 fused to GFP to locate nucleoli (57). MPP10-GFP largely localizes to DAPI-negative regions (compare panel DAPI to panel DAPI+MPP10). EBP2 largely colocalizes with MPP10 (compare panel DAPI+MPP10 to panels DAPI+EBP2 and MERGE). This colocalization indicates that the majority of EBP2 is localized to nucleoli as observed previously (10). (C) EBP2-DBD colocalizes with EBP2 in 293 cells. 293/EBP2-DBD cells were transfected with an expression plasmid for FLAG-EBP2, and interphase cells were visualized by using antibodies against the DBD (FITC) or FLAG-EBP2 (rhodamine). EBP2-DBD and FLAG-EBP are largely localized to regions of the nucleus that stain poorly with DAPI (compare panel DAPI to panels DAPI+EBP-DBD and DAPI+FLAG-EBP2). Also, EBP2-DBD localizes similarly to EBP2 within cells (compare panel DAPI+EBP2-DBD to panels DAPI+FLAG-EBP2 and MERGE).

FIG. 7.

EBP2-DBD does not support the replication of oriP plasmids in human cells. (A) An oriP replication reporter plasmid (i.e., plasmid 1033) was transfected into 293/EBNA1 and 293/EBP2-DBD cells, after which cells were propagated in the absence of selection and harvested 4 or 21 days posttransfection. The transfected cell line is indicated above each lane, and the time point is given above each set of lanes. DNAs were extracted and digested with DpnI, linearized with XbaI, and quantified by Southern analysis with a probe generated from plasmid 1033 digested with HincII. The amounts of standards loaded are indicated above each lane, and their electrophoretic mobilities are indicated by “A.” “B” indicates the electrophoretic mobility of the linearized replicated plasmid 1033. Quantified results from two repetitions of this experiment are given in Table 3.

FIG. 8.

EBP2-DBD and EBP2 do not associate with condensed mitotic chromatin within cells or with isolated metaphase chromosomes. (A) 293/EBP2-DBD cells were treated with Colcemid and subjected to whole-cell immunofluorescence with the K67.3 antibody to the DBD of EBNA1. In these images, the cells are mitotic, the nuclear envelope has broken down, but the cell membrane is intact. EBP2-DBD is visualized with a FITC-conjugated secondary antibody, whereas condensed mitotic chromatin is counterstained with DAPI. The upper row shows deconvolved images from three cells visualized with a ×100 objective lens, indicating that EBP2-DBD stains wherever mitotic chromatin is absent. The lower row shows a single cell also taken with a ×100 objective lens, but with 1.5× optical enhancement from an independent experiment, also indicating an exclusion of EBP2-DBD from condensed mitotic chromatin within intact cells, although the protein can be detected readily within such cells. (B) Endogenous EBP2 is not bound to metaphase chromosomes isolated from 293/EBNA1 cells. Metaphase spreads were recovered from 293/EBNA1 cells and subjected to coimmunofluorescence with K67.3 (rhodamine-conjugated secondary antibody) and a monoclonal antibody against endogenous human EBP2 (FITC-conjugated secondary antibody). As previously observed (48), EBNA1 (red) coats metaphase chromosomes in a distinct, punctate manner; EBP2 (green) is not found on the isolated metaphase chromosomes. (C) Overexpressed FLAG-EBP2 is not bound to isolated metaphase chromosomes. Metaphase spreads were recovered from 293/EBNA1 cells transiently transfected with a FLAG-EBP2 expression plasmid and subjected to coimmunofluorescence as in panel B. EBNA1 was detected and visualized as described above, whereas FLAG-EBP2 was detected by using a monoclonal anti-FLAG antibody (Chemicon) and visualized by using an FITC-conjugated secondary antibody. Even when overexpressed, EBP2 was not detected on metaphase chromosomes. (D) Localization of EBP2 and EBNA1 within a cell in early metaphase isolated by mitotic shake-off. 293/EBNA1 cells in mitosis were isolated by shake-off, fixed on slides, and stained with a monoclonal antibody against endogenous EBP2, followed by an FITC-conjugated secondary antibody, and the K67.3 rabbit polyclonal antibody against EBNA1 (rhodamine-conjugated secondary). (E) Localization of EBP2 and EBNA1 within a cell in late anaphase. 293/EBNA1 cells were isolated and processed as in panel D.

FIG. 8.

EBP2-DBD and EBP2 do not associate with condensed mitotic chromatin within cells or with isolated metaphase chromosomes. (A) 293/EBP2-DBD cells were treated with Colcemid and subjected to whole-cell immunofluorescence with the K67.3 antibody to the DBD of EBNA1. In these images, the cells are mitotic, the nuclear envelope has broken down, but the cell membrane is intact. EBP2-DBD is visualized with a FITC-conjugated secondary antibody, whereas condensed mitotic chromatin is counterstained with DAPI. The upper row shows deconvolved images from three cells visualized with a ×100 objective lens, indicating that EBP2-DBD stains wherever mitotic chromatin is absent. The lower row shows a single cell also taken with a ×100 objective lens, but with 1.5× optical enhancement from an independent experiment, also indicating an exclusion of EBP2-DBD from condensed mitotic chromatin within intact cells, although the protein can be detected readily within such cells. (B) Endogenous EBP2 is not bound to metaphase chromosomes isolated from 293/EBNA1 cells. Metaphase spreads were recovered from 293/EBNA1 cells and subjected to coimmunofluorescence with K67.3 (rhodamine-conjugated secondary antibody) and a monoclonal antibody against endogenous human EBP2 (FITC-conjugated secondary antibody). As previously observed (48), EBNA1 (red) coats metaphase chromosomes in a distinct, punctate manner; EBP2 (green) is not found on the isolated metaphase chromosomes. (C) Overexpressed FLAG-EBP2 is not bound to isolated metaphase chromosomes. Metaphase spreads were recovered from 293/EBNA1 cells transiently transfected with a FLAG-EBP2 expression plasmid and subjected to coimmunofluorescence as in panel B. EBNA1 was detected and visualized as described above, whereas FLAG-EBP2 was detected by using a monoclonal anti-FLAG antibody (Chemicon) and visualized by using an FITC-conjugated secondary antibody. Even when overexpressed, EBP2 was not detected on metaphase chromosomes. (D) Localization of EBP2 and EBNA1 within a cell in early metaphase isolated by mitotic shake-off. 293/EBNA1 cells in mitosis were isolated by shake-off, fixed on slides, and stained with a monoclonal antibody against endogenous EBP2, followed by an FITC-conjugated secondary antibody, and the K67.3 rabbit polyclonal antibody against EBNA1 (rhodamine-conjugated secondary). (E) Localization of EBP2 and EBNA1 within a cell in late anaphase. 293/EBNA1 cells were isolated and processed as in panel D.

FIG. 9.

Model for the partitioning of oriP plasmids on a per-replicon basis. (A) High-resolution indirect immunofluorescence comparing the localization of wild-type EBNA1 and HMGA1a-DBD on metaphase chromosomes by using an antibody to the DBD of EBNA1. The localizations of both proteins were similar; many pairs of sister chromatids contained an equal number of dots for both proteins. In several instances, the dots were symmetrically positioned on both sister chromatids. (B) Model of a portion of a chromosome based on the models constructed by Laemmli and coworkers (20, 45, 46). For convenience we have depicted a metaphase chromosome with Q and R bands containing AT-rich SARs. Laemmli and coworkers have demonstrated that sequences present as SARs on interphase chromosomes are present in R bands in metaphase chromosomes (45, 46). We propose that EBNA1 or HMGA1a-DBD tethers oriP plasmids to SARs that are present relatively infrequently compared to other sequences on chromosomes. Upon S phase, when there is replication of chromosomes, we hypothesize that the replicated oriP plasmids are partitioned between the sister SARs on sister chromatids, by a distribution of EBNA1 or HMGA1a-DBD to each sister SAR. Key to this model is that the actual partitioning event is concomitant with replication and occurs during S phase. The plasmids remain tethered to sister chromatids and piggyback on the sisters during mitosis (9).

FIG. 9.

Model for the partitioning of oriP plasmids on a per-replicon basis. (A) High-resolution indirect immunofluorescence comparing the localization of wild-type EBNA1 and HMGA1a-DBD on metaphase chromosomes by using an antibody to the DBD of EBNA1. The localizations of both proteins were similar; many pairs of sister chromatids contained an equal number of dots for both proteins. In several instances, the dots were symmetrically positioned on both sister chromatids. (B) Model of a portion of a chromosome based on the models constructed by Laemmli and coworkers (20, 45, 46). For convenience we have depicted a metaphase chromosome with Q and R bands containing AT-rich SARs. Laemmli and coworkers have demonstrated that sequences present as SARs on interphase chromosomes are present in R bands in metaphase chromosomes (45, 46). We propose that EBNA1 or HMGA1a-DBD tethers oriP plasmids to SARs that are present relatively infrequently compared to other sequences on chromosomes. Upon S phase, when there is replication of chromosomes, we hypothesize that the replicated oriP plasmids are partitioned between the sister SARs on sister chromatids, by a distribution of EBNA1 or HMGA1a-DBD to each sister SAR. Key to this model is that the actual partitioning event is concomitant with replication and occurs during S phase. The plasmids remain tethered to sister chromatids and piggyback on the sisters during mitosis (9).