How Epstein-Barr Virus Induces the Reorganization of Cellular Chromatin

ABSTRACT We have discovered how Epstein-Barr virus (EBV) induces the reorganization of cellular chromatin (ROCC), in which host chromatin is compacted and marginated within the nucleus, with viral DNA replication occurring in the chromatin-free regions. Five families of DNA viruses induce ROCC: herpesviruses, adenoviruses, parvoviruses, baculoviruses, and geminiviruses. These families infect a variety of hosts, including vertebrates, insects, and plants. They also share several characteristics: they replicate and encapsidate their genomes in the host nucleus and package their genomes unbound by histones. We have identified the viral genes and processes required for EBV’s ROCC. Each of EBV’s seven core DNA synthesis genes and its origin of lytic replication (oriLyt), in trans, are required, while its protein kinase, BGLF4, and its true late genes are not. Following these findings, we tested the role of EBV lytic DNA amplification in driving ROCC. Surprisingly, the inhibition of EBV’s lytic DNA synthesis still supports chromatin compaction but blocks its margination. We propose a two-step model for ROCC. First, the initiation of viral lytic DNA synthesis induces a cellular response that results in global chromatin compaction. Second, the histone-free, productive viral DNA synthesis leads to the margination of compacted chromatin to the nuclear periphery. We have tested this model by asking if the histone-associated simian virus 40 (SV40) DNA synthesis could substitute for oriLyt-mediated synthesis and found that EBV’s ROCC is incompatible with SV40 DNA replication. Elucidating EBV’s induction of ROCC both illuminates how other viruses can do so and indicates how this spatial control of cellular chromatin benefits them.

IMPORTANCE Five families of viruses support the reorganization of cellular chromatin (ROCC), the compaction and margination of host chromatin, upon their productive infection. That they all share this phenotype implies the importance of ROCC in viral life cycles. With Epstein-Barr virus (EBV), a herpesvirus, we show that the viral replication complex and origin of lytic replication (oriLyt) are essential for ROCC. In contrast, its protein kinase and true late genes are not. We show that, unexpectedly, the viral lytic amplification is not required for chromatin compaction but is required for its margination. We propose a two-step model for ROCC: first, global chromatin compaction occurs as a cellular response to the initiation of viral DNA synthesis; then, the accumulation of newly synthesized, histone-free viral DNA leads to cellular chromatin margination. Taken together, our findings provide insights into a process contributing to the productive phase of five families of viruses.
KEYWORDS Epstein-Barr virus, chromatin reorganization, productive infection, virus-host interactions V iruses as obligate cellular parasites evolve to manipulate cellular functions for their own benefit. An example of one such manipulation is illustrated by Epstein-Barr virus (EBV) and found in members of four additional families of viruses which expropriate control of nuclear structure during their productive phases. These viruses foster the ganciclovir (GCV) and phosphonoacetic acid (PAA), we unexpectedly found that detectable lytic DNA synthesis is not required for chromatin compaction but is required for its margination. Thus, we propose a model in which ROCC requires two steps: first, initiation of viral DNA synthesis induces a cellular response that results in global chromatin compaction; then, extensive accumulation of histone-free, newly synthesized viral DNA leads to the margination of the compacted cellular chromatin to the periphery of the nucleus.
In order to understand ROCC further, we have explored it comparatively. We examined EBV's ROCC in cells cotransfected with simian virus 40 (SV40), a DNA virus that packages chromatinized DNA and does not support ROCC (22,23). We found that cells in which the first step of ROCC has been triggered do not support SV40 DNA amplification. Thus, EBV's ROCC is incompatible with SV40's histone-associated DNA synthesis.

RESULTS
EBV's oriLyt and DNA synthesis genes are required for ROCC. EBV's ROCC occurs in parallel with lytic DNA synthesis; therefore, we investigated any requirement of EBV's DNA synthesis genes for ROCC. The origin of lytic replication (oriLyt) and seven early genes (BALF5, BALF2, BBLF2/3, BBLF4, BSLF1, BMLF1, and BMRF1) were found to be necessary and, in combination with BZLF1 and BRLF1, sufficient to support this synthesis (24,25) (Table 1 contains a list of genes and their functions). Cells from either of two libraries of EBV (26,27), in which mutants null for a single EBV gene were maintained in 293 cells, were used for transcomplementation assays (Fig. 1A). Each cell line was transfected with an expression vector for enhanced green fluorescent protein (eGFP)-H2B or mCherry-H2B to visualize cellular chromatin and with an empty expression vector (uninduced [ in the supplemental material). The proportion of cells in the population that supported ROCC was normalized to that of the transcomplemented (I1t) condition (Table 1; see  also Table S1). In all eight sets of assays, uninduced cells did not support ROCC.  Table S1), from a combination of three transfection replicates. *, P value , 0.01; **, P value , 0.05, Fisher's exact test of I versus I1t group pairs.
Samples in the induced group had zero or near-zero cells supporting ROCC, and 2 to 14% of cells in the transcomplemented group supported ROCC. These various levels reflect clonal differences in the rates of entry into the lytic phase. ROCC did not occur in the absence of any of these eight viral genes. Thus, oriLyt and each of the seven core lytic DNA synthesis genes are required for ROCC. EBV's protein kinase, BGLF4, is not required for ROCC. EBV's serine/threonine (S/T) kinase, also known as BGLF4 or EBV-PK, has previously been reported to support a phenotype of premature chromosome condensation in transfected cells (28,29). Given that ROCC involves the spatial condensation of cellular chromatin, we asked if BGLF4 contributes to ROCC, using the same transcomplementation assay as described in Fig. 1. In contrast to EBV BACmids (shuttle vectors for amplifying EBV plasmids in E. coli) deleted for oriLyt or for any of the seven core lytic DNA synthesis genes, EBV was capable of supporting ROCC in the absence of BGLF4 (Table 1, induced versus induced 1 transcomplemented proportions of ROCC 1 cells are not significantly different). These cells were null for BGLF4, as confirmed by BMRF1, a substrate of BGLF4, not being phosphorylated (Fig. S2). This result indicates that, in 293/EBV cells, BGLF4 is not required for ROCC.
EBV's true late genes are not required for ROCC. An additional insight can be gleaned from the DoriLyt transcomplementation assay. The expression of EBV's true late gene requires an oriLyt sequence in cis (on the same DNA), as well as oriLyt-mediated DNA replication of the EBV genome containing these genes (27). A true late gene is defined as one that absolutely requires lytic DNA synthesis for its expression, whereas a leaky late gene is expressed at low levels prior to lytic DNA synthesis, with an increase in its expression upon lytic DNA synthesis (30). In the induced 1 transcomplemented group, the 293/EBV DoriLyt cell line was transfected with a plasmid encoding oriLyt. Under this condition, oriLyt is present in trans to the late genes, and Djavadian et al. have shown that this configuration does not support the expression of true late genes (27). Despite the lack of late gene expression, DoriLyt induced 1 transcomplemented cells support ROCC (Table 1; Table S1). Thus, EBV's true late genes are not required for ROCC.
The role of true late genes for ROCC was tested independently using cells defective for the expression of true late genes. DBVLF1 and DBcRF1 cell lines have been previously characterized to be incapable of expressing true late genes (27). BVLF1 and BcRF1 are two of six genes unique to beta-and gammaherpesviruses that form a viral preinitiation complex (vPIC) responsible for true late gene transcription (27,31). EBV with defective vPICs lacks expression of true late genes (27,30). Transcomplementation assays of DBVLF1 and DBcRF1 showed that these cells can support ROCC in the absence of BVLF1 and BcRF1 (Table 1; Table S1). Higher proportions of cells supporting ROCC were found in the induced (I) than in the induced 1 transcomplementation (I1t) group of these cells. One likely explanation for this observation is that the I1t group was capable of expressing true late genes, following which virus particles were assembled and released, resulting in cell death. At 48 h postinduction (hpi), some cells would have completed their lytic phase and died (13,32). On the other hand, lytic cells in the induced group were incapable of expressing true late genes, as they were missing a vPIC component, and therefore could survive longer in the lytic phase, leading to their being observed upon cell counting at 48 hpi and adding to the apparent proportion of ROCC-positive cells. Collectively, our findings indicate that EBV's true late genes are not required for ROCC.
The enzymatic function of EBV's DNA polymerase is required for ROCC. Given that EBV's true late genes as well as one of the early genes, BGLF4, are dispensable for ROCC, we further scrutinized the role of EBV's lytic DNA synthesis for ROCC using mutants of EBV's DNA polymerase, BALF5. All herpesvirus DNA polymerases belong to the alpha-like polymerase family, which has seven highly conserved regions (33)(34)(35). The most conserved region, region I, is centered around the amino acid residue sequence YGDTDS, which is invariant in herpesvirus polymerases, those of many other viruses, and the human DNA polymerase alpha ( Fig. 2A) (34,36,37). Region I is located within the palm domain of the DNA polymerase (Fig. S3) and is involved in the coordination of metal ions at the catalytic site (38,39). Three BALF5 mutants were generated, with substitutions in the YGDTDS residues: D755N/D757N, Y753F, and S758T. The two aspartic acid residues (D755 and D757 in EBV's BALF5) appear to be particularly sensitive to mutations, as their substitutions typically lead to functionally null polymerases, as has been shown in both human DNA polymerase alpha and herpes simplex virus 1 (HSV-1) DNA polymerase (Pol) (37,40).
We constructed FLAG-tagged expression vectors for the D755N/D757N, Y753F, and S758T BALF5 mutants and used them in transcomplementation assays in the 293/EBV DBALF5 cell line. Control transcomplementation experiments using FLAG-tagged wildtype BALF5 showed that the tagged protein was functional in supporting lytic DNA synthesis and ROCC. Western blotting assays and quantitative PCR (qPCR) measurements verified the expression of these mutant constructs and the effects of these mutations on BALF5's activity ( Fig. 2B and C; see also Table S2). All three mutants were null for BALF5 activity. These BALF5 mutants failed to support ROCC (Fig. 2C). This finding indicates that the complete absence of lytic EBV DNA synthesis leads to a complete abrogation of ROCC.
Detectable EBV lytic DNA synthesis is not required for some features of ROCC. The observed requirement for ROCC for all EBV genes essential for viral DNA amplification made it likely that blocking DNA synthesis would block ROCC, too. We tested this hypothesis and uncovered a surprising result. We treated the lytic-phase-inducible, EBV-positive cell line iD98/HR1 (13) with ganciclovir (GCV; 40 mg/mL) or phosphonoacetic acid (PAA; 100 mg/mL) to inhibit viral DNA synthesis. GCV is a guanosine analogue that specifically inhibits the elongation of viral DNA (41,42). PAA is a pyrophosphate analogue that binds directly to the viral DNA polymerase, thus specifically inhibiting the viral DNA synthesis (43,44). The iD98/HR1 cells were induced for the lytic phase 24 h following treatment with GCV or PAA and pulsed with 5-ethynyl-29-deoxyuridine (EdU) for 30 min at 24 and 48 h following induction (Fig. 3A). At 48 hpi, the cells were either harvested for qPCR or fixed for immunofluorescence assays. The viability of cells at 48 hpi was also assessed, and cells were found to be viable (Fig. S4). qPCR measurement for EBV genomes confirmed that GCV or PAA treatment inhibited lytic DNA synthesis to levels comparable to that of uninduced, nonlytic cells ( Fig. 3B and C). For immunofluorescence, fixed cells were permeabilized, following which click chemistry was performed to detect EdU signals, and the cells were subjected to immunofluorescence to detect the early lytic protein, BMRF1, and 49,6-diamidino-2-phenylindole (DAPI) staining to detect DNA. The distribution of BMRF1 differs in cells treated with GCV or PAA and covers the periphery of the nucleus, as has been observed in similarly treated human cytomegalovirus (HCMV)-infected cells (12). Cells that did not express BMRF1 were considered nonlytic, did not accumulate EdU in viral amplification factories, and did not display ROCC ( Fig. 4A and B). Only cells found to express BMRF1 were WT, wild type. (C) 293/EBV DBALF5 cell transcomplementation assay with BALF5 variants. Bar plot displaying relative EBV genome measurement by qPCR, normalized to that under the uninduced condition. The corresponding fraction of ROCC 1 cells is noted below, normalized to induced 1 transcomplementation with WT BALF5. All three BALF5 mutants are null for its DNA polymerase activity and are incapable of supporting ROCC. Error bars show standard deviations. More than 600 eGFP-positive cells were examined for ROCC per condition, from a combination of three transfection replicates. qPCR measurements were performed on three technical replicates per condition per transfection replicate, from a total of three transfection replicates. *, P value of ,0.05, Welch's t test of each condition compared to the induced (without transcomplementation) condition. considered to be in the lytic phase. As expected, these cells did have EdU signals in viral amplification factories and did display ROCC. The GCV-or PAA-treated lytic cells, also characterized by BMRF1 expression, largely had no detectable EdU signals (10% or 0% EdU 1 cells in GCV-or PAA-treated cells, respectively), confirming the inhibition of viral DNA synthesis in them. Two types of ROCC were identified: ROCC type I, in which the cellular chromatin was spatially condensed but not marginated, and ROCC type II, in which the cellular chromatin was spatially condensed and marginated, as seen in wild-type viral productive phase ( Fig. 4A and B; see also Fig. S5). These findings were also observed in a different EBV-positive cell line, the gastric carcinoma cell line AGS-Akata (Fig. S6). Quantification of the ROCC types in iD98/HR1 showed that, in those cells treated with GCV or PAA that entered the lytic phase as indicated by their expression of BMRF1, there was a significant increase in ROCC type I ( Fig. 4C and D). This finding indicates that detectable EBV lytic DNA synthesis is not required for cellular chromatin condensation but contributes to its margination. Along with findings from the transcomplementation assays with BALF5 mutants, this observation gives us a mechanistic insight into ROCC: the initiation of EBV DNA synthesis is required for ROCC type I.
It is yet unknown what other factors are required to support ROCC type I or whether the initiation of EBV's lytic DNA synthesis is sufficient. Our results indicate that, following the global chromatin compaction characterizing ROCC type I, extensive amplification of EBV genomes gives rise to chromatin margination (ROCC type II).
EBV's ROCC is incompatible with the DNA replication of the non-ROCC SV40. We have tested our model of EBV's lytic DNA amplification mediating the margination of chromatin, thus driving ROCC type I into ROCC type II, by asking if a different kind of viral DNA amplification, which lacks EBV's oriLyt, could substitute for EBV's oriLyt-mediated DNA amplification in promoting ROCC type II. For this substitution, we chose SV40, a polyomavirus which, like EBV, amplifies and encapsidates its genomes within the nucleus. However, unlike EBV, it encapsidates histone-bound genomes (22) and Samples were fixed and subjected to click chemistry for detection of EdU, immunofluorescence for the early protein BMRF1, and DAPI staining for DNA. Imaging was performed using scanning confocal microscopy. (B and C) qPCR measurements of EBV genome from GCV and PAA experiments, normalized to the uninduced, untreated condition. qPCR measurements were performed on three technical replicates per condition per transfection replicate, from a total of three transfection replicates. Error bars show standard deviation.

EBV Reorganizes Cellular Chromatin mBio
does not induce ROCC (23). iD98/HR1 cells were plated and treated with 100 mg/mL PAA 24 h prior to mock or SV40 transfection, and then cells were induced to enter EBV's lytic phase (Fig. 5A). At 48 h posttransfection, cells were incubated for 1 h with EdU. Cells were scored for their expression of EBV's BMRF1, indicating entry into EBV's lytic phase, and SV40's large T antigen, indicating expression from SV40's viral DNA.
The proportions of cells that express BMRF1 alone (BMRF1 1 only), T antigen alone (TAg 1 only), or both (BMRF1 1 /TAg 1 ) were measured (Fig. 5B). Control experiments showed that, in the presence of PAA, all cells that entered the lytic phase, as noted by their expression of BMRF1, failed to synthesize EBV DNA and were EdU negative (Fig. S7). Moreover, control samples that were pretreated with PAA, mock transfected, and induced to enter EBV's lytic phase did not incorporate EdU (0% EdU 1 cells, n > 100 cells counted). We therefore used EdU incorporation to identify cells supporting SV40 replication, in those cells that were transfected with SV40 viral DNA. From this experiment, we found that we could not answer our initial question: whether SV40's DNA amplification could substitute for EBV's oriLyt-mediated DNA amplification in driving ROCC type I into ROCC type II, as it appears that EBV's ROCC and SV40's DNA amplification are incompatible (Fig. 5C). Of cells that had entered EBV's lytic phase and were transfected with SV40 DNA (BMRF1 1 /TAg 1 ), only those that had no ROCC were found to support SV40 replication (EdU 1 ). In addition, the proportion of total EdU 1 cells was reduced in the BMRF1 1 /TAg 1 population compared to cells FIG 4 Detectable lytic DNA synthesis is dispensable for chromatin condensation but required for its margination. iD98/HR1 cells were treated with GCV or PAA and induced to enter the lytic phase (Fig. 3A). (A and B) Representative cells from GCV (A) and PAA (B) experiments are shown. Uninduced (nonlytic) cells do not express BMRF1 and do not display ROCC. Induced, lytic cells robustly express BMRF1, have EdU signals localizing to viral replication compartment(s), and support ROCC type II, in which cellular chromatin is both condensed and marginated. Lytic cells treated with GCV or PAA robustly express BMRF1, and $90% lack EdU signals, confirming the inhibition of viral DNA synthesis in them. The majority of these cells display ROCC type I, in which cellular chromatin is condensed but not marginated. All images have the same scale; scale bar is 10 mm. (C and D) ROCC types were classified for each imaged cell (all cells in uninduced samples and BMRF1 1 cells in induced samples) and quantified for each treatment group. Induced cells treated with GCV or PAA support increased levels of ROCC type I compared to induced, untreated cells, which support mostly ROCC type II. n > 30 cells per group. *, P value of ,0.01, 2 Â 3 Fisher's exact test.
In both the BMRF1 1 -only and BMRF1 1 /TAg 1 populations of this experiment, there is a higher proportion of cells supporting ROCC type I than those supporting ROCC type II (Fig. 5E), an apparent discrepancy with the population distribution of ROCC types in previous experiments (Fig. 4). This difference can be explained by one condition of the experiments involving transfections of SV40 viral DNA. When EBV-positive cells are induced for the lytic phase, they enter the lytic phase at the beginning of their next S phase (13). Following transfection, cells tend to have a delayed growth/cycle, likely resulting in a later entry into the lytic phase, thus affecting the proportion of ROCC 1 cells at 48 hpi. Taking this delay into account, the proportions of cells supporting ROCC type I and type II in the experiment involving SV40 transfections are consistent with those previously seen in iD98/HR1 cells induced to enter the lytic phase in the presence of PAA (Fig. 4D).
Our findings from the cotransfection of SV40 viral DNA into EBV lytic cells led us to an additional mechanistic insight: EBV inhibits cellular DNA synthesis during its lytic phase, and we now found that its ROCC is incompatible with SV40 DNA synthesis. Both cellular and SV40 DNA synthesis require cellular DNA polymerases and also are histone associated. Polyomaviruses, such as SV40, as well as papillomaviruses, encapsidate histone-bound genomes and do not induce ROCC (22,45). In contrast, all five families of DNA viruses that induce ROCC lack cellular histones in their virions (7-10, 46, 47). In EBV, cellular DNA synthesis does not occur once viral DNA synthesis begins. Thus, it is likely that the mechanism that underlies the failure of cellular DNA synthesis extends to that of SV40 through a shared mechanism involving the inhibition of histone-associated DNA synthesis.

DISCUSSION
Five families of DNA viruses elicit the reorganization of cellular chromatin (ROCC) during their productive infections. However, little was known about the mechanisms by which they mediate ROCC, nor how this compaction of cellular DNA benefits their virus production. We have examined the ROCC driven by EBV to illuminate these unknowns. Our studies with EBV differ from those with other viruses that induce ROCC because EBV usually enters its productive phase having first established its latent phase in newly infected cells.
Given that EBV's ROCC can first be detected shortly before or coincident with the amplification of its DNA (13), we tested genetically EBV's DNA synthesis genes for their potential roles in ROCC. Both EBV's oriLyt and its genes encoding the core lytic DNA synthesis complex are required for ROCC ( Fig. 1 and Table 1). However, neither BGLF4, which encodes the viral protein kinase, nor EBV's true late genes contribute to ROCC. These findings yielded the first mechanistic insight into EBV's eliciting ROCC: the viral productive DNA replication is integral to forming ROCC.
This insight led us to ask if EBV DNA synthesis per se is essential for ROCC. We used inhibitors of viral DNA synthesis, ganciclovir (GCV) and phosphonoacetic acid (PAA), and found that an early form of ROCC (type I), in which cellular chromatin became compacted but was not moved to the periphery of the nucleus, took place (Fig. 4). Complete ROCC (type II) required extensive EBV DNA amplification. These surprising findings were extended on examining mutants of EBV's DNA polymerase encoded by BALF5. Catalytically dead derivatives of EBV's DNA polymerase supported neither type I nor type II of ROCC (Fig. 2). These observations led to a second insight into EBV's mechanism of mediating ROCC: EBV induces a compaction of cellular chromatin prior to observable amplification of its DNA. It is likely from our findings with the BALF5 mutants and GCV/PAA experiments that this initial compaction requires the formation of a viral replication complex and the initiation of, but not extensive, viral DNA synthesis. Taking these findings together with our results from the transcomplementation assays (Fig. 1), it is apparent that EBV's oriLyt and the core components of the lytic DNA synthesis complex are required for both type I and type II of ROCC, as neither form was observed in the absence of oriLyt or each of the core lytic DNA synthesis genes.
Not all families of DNA viruses which replicate their genomes in the nucleus induce ROCC. We asked whether a different kind of viral DNA amplification, which lacks EBV's oriLyt, could substitute for EBV's oriLyt-mediated DNA amplification in promoting ROCC type II. In using SV40, a virus which replicates its DNA in the nucleus but does not induce ROCC (23), we found that EBV's ROCC and SV40's DNA replication are incompatible. SV40 can replicate in cells that carry EBV but not ones that have induced ROCC (Fig. 5). It has been shown previously that EBV's lytic phase inhibits cellular DNA synthesis, and we now found that EBV's ROCC is incompatible with SV40 DNA replication. A third insight into the mechanism of EBV's ROCC comes from the observations that both cellular and SV40 DNA synthesis are blocked during EBV's lytic DNA synthesis. Thus, ROCC occurs when DNA synthesis requiring histone deposition does not occur while EBV's histone-free DNA synthesis does.
Our three insights into the mechanisms of ROCC have led us to propose a two-step model (Fig. 6). First, cellular chromatin is triggered to coalesce by the formation of a viral replication complex initiating synthesis in the absence of the deposition of cellular histones. This first step, which we call ROCC type I, is reminiscent of the "honeycombed" nuclear structure previously observed in early lytic EBV cells (13) and the fiberlike structure of chromatin in cells infected with geminiviruses (19). Second, the synthesis of histone-free viral DNA drives the cellular chromatin to the periphery of the nucleus, resulting in its characteristic margination.
EBV's induction of ROCC may seem counterintuitive to the success of its own lytic DNA replication. While the cellular chromatin is being compacted and marginated, EBV continues to synthesize its DNA. This apparent escape from the global chromatin compaction may be mediated by the viral transactivator protein, Zta. The chromatinized circular EBV genome has many Zta binding sites (48). Zta has been shown to be a pioneer factor, capable of binding "closed" chromatin and keeping it in an "open" conformation (49,50). It is possible that the binding of multiple Zta proteins along the circular EBV genome contributes to their serving as the templates for lytic DNA synthesis. It has also been demonstrated that newly synthesized, histone-free EBV DNA is capable of serving as the template for DNA synthesis (13). Both events could allow EBV's lytic DNA synthesis to continue in parallel with ROCC.
Much evidence supports the involvement of cellular factors in the global chromatin EBV Reorganizes Cellular Chromatin mBio compaction that characterizes ROCC type I. The five families of viruses that induce ROCC vary in their genetic content, with geminiviruses and parvoviruses containing as few as two to four genes. Their restricted genetic content means they are unlikely to encode the machinery required directly to condense the cellular chromosomal DNA. For example, some geminiviruses multimerize their capsids to accommodate longer DNAs rather than condensing them (51). Thus, it is likely that viruses promote the chromatin compaction in ROCC by inducing their host to use its own condensation and chromatin remodeling machineries. Cellular factors known to contribute to the remodeling of chromatin structures could contribute to ROCC type I; these include the condensin and cohesin complexes, histone modifications associated with repressed gene expression and heterochromatin, such as H3K9me3, H3K27me3, or macroH2A, and insulator proteins such as CTCF (CCCTC-binding factor).
We have considered two mechanisms to explain how EBV mediates the margination of cellular chromatin. The first depends on EBV's replication compartments moving the cellular chromatin as they expand. The compartments are thought not to be bound by membranes and do expand and fuse as the productive cycle progresses (32). Although the viral DNA in them can accumulate to 30% of the level of cellular DNA, the accompanying increase in volume of the nucleus means that the concentration of the amplified viral DNA is less than or similar to that of the cellular DNA prior to EBV's productive phase (32). In this proposed mechanism, the expanding compartment with viral DNA unbound by histones would drive the cellular chromatin by charge repulsion. DNA carries a high negative charge which is about twice that of nucleosome-bound DNA (52). However, the charge dispersion by ions has been found to screen the electrostatic potentials of the DNA helices from electrostatic repulsion beyond a center-to-center distance of #30 Å (53), so that electrostatic repulsion alone is unlikely to drive chromatin margination. To illustrate, 10 4 EBV DNA molecules each separated by 30 Å would occupy a volume of 6 mm 3 , a small fraction of a nuclear volume. A second plausible mechanism is that of liquidliquid phase separation (LLPS) which has been found likely to contribute to the formation of early replication compartments during HCMV's productive infection (54). This work found that two early genes of HCMV essential for replication are required for the detected LLPS. The requirement for EBV's core replication proteins for it to mediate ROCC makes LLPS an attractive mechanism to consider further.
No current model explains all cases of ROCC. We found that 1/10 to 1/4 of the cells in which viral DNA synthesis was inhibited displayed ROCC type II (Fig. 4). These cells had less than 2% of the viral DNA produced in the uninhibited cell population as assayed by qPCR ( Fig. 3B and C; compare induced with and without GCV or PAA treatment) and no detectable DNA synthesis as assayed by EdU incorporation (Fig. 4A and  B). It is apparent that, in a minority of cases, the condensation of cellular chromatin triggered by formation of EBV's replication complex can lead to the margination of ROCC type II without substantive viral DNA replication.
How do the many viruses that induce ROCC benefit from this striking reorganization of their host's chromatin? All these viruses express essential viral genes from chromatin-bound templates to carry out their productive infections. They all package viral DNAs free of cellular histones. ROCC provides them a common pathway to segregate the progeny DNA they will encapsidate from the DNA templates used for transcription. Condensed and inaccessible chromatin is also typically associated with a repression of gene expression. In addition to being condensed, chromatin in ROCC is marginated to the periphery of the nucleus. The nuclear periphery is typically a repressive environment, in which case ROCC may resemble the repressed chromatin of lamina-associated domains (LADs) (55,56). ROCC could contribute to the suppression of host gene expression to benefit the replication and production of progeny viruses.
ROCC is mediated by multiple families of viruses infecting diverse hosts; it is likely that it has an important role in viral life cycles. Viruses that support ROCC share similarities to our model system, EBV. Other herpesviruses, as well as adenoviruses, parvoviruses, baculoviruses, and geminiviruses, all replicate in the host nucleus, are chromatinized within the host, and package histone-free genomes in their virions. As with EBV, ROCC may confer a benefit on these viruses in segregating histone-rich host chromatin from histone-free regions of viral DNA replication. Compacted and marginated chromatin is also associated with suppressed gene expression, in which case ROCC may play a role in the manipulation of host gene expression at the genomic level. A recent proposed megataxonomy of viruses indicates that the five families of ROCC-inducing viruses do not share a common ancestry (57). Thus, it appears that ROCC has evolved independently in five families of viruses as a shared phenotype that supports productive lytic infection. ROCC also uniquely illustrates viruses' capacity to control the nuclear structures of their hosts.

MATERIALS AND METHODS
Cell lines and culture. All cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) (Gibco) supplemented with 10% fetal bovine serum (FBS) (HyClone Laboratories) and appropriate antibiotics for selection (described below). All cell culture media were supplemented with 200 U/mL penicillin (MP Biomedicals) and 200 mg/mL streptomycin (MP Biomedicals). All cells were grown at 37°C in a 5% CO 2 humidified atmosphere. iD98/HR1 is a fusion between D98 cells and the Burkitt lymphoma cell line P3HR1. Both iD98/HR1 and iD98/HR1 eGFP-H2B cells have been previously described (13); inducible Z-ER (EBV's immediate early protein Zta fused to the estrogen receptor ligand-binding domain) was selected and maintained in these cells with 1 mg/mL puromycin. The following 293 cells infected with mutant EBV BACmids were kind gifts from Ya-Fang Chiu (Chang-Gung University, Taoyuan, Taiwan) and have been previously described (26) Generation of mutant EBV-positive 293 cell lines. 293/EBV DBVLF1 and 293/EBV DBcRF1 were generated using BM2710 Escherichia coli carrying the invasin gene from Yersinia pseudotuberculosis and the hly gene encoding listeriolysin O from Listeria monocytogenes, which allow gene transfer of intact BACmids to some mammalian cells in vitro. BACmids EBV DBVLF1 (MI-383) and EBV DBcRF1 (MI-27) have been previously described as a part of a comprehensive EBV mutant library (26). BM2710 E. coli stably transformed with EBV DBVLF1 or EBV DBcRF1 was obtained from Eric Johannsen (University of Wisconsin-Madison, USA). These EBV-positive BM2710 E. coli cells were grown overnight at 32°C in brain heart infusion media (37 g/L [wt/vol]) supplemented with 0.5 mM 2,6-diaminopimelic acid (DAP) (Alfa Aesar) and 25 mg/mL spectinomycin (Alfa Aesar) and selected with 50 mg/mL kanamycin. One milliliter of BM2710 E. coli overnight culture was added to 293 cells at 80 to 90% confluence in a 60-mm cell culture plate, in 1Â DMEM supplemented with 0.5 mM DAP and 25 mg/mL spectinomycin, and incubated for 2 h. Following incubation, the medium was removed, the cells were washed with 1Â DMEM to remove as many bacteria as possible, and fresh 1Â DMEM plus 10% FBS medium was added, supplemented with 50 mg/mL gentamicin (Gibco). The following day, the medium was replaced with 1Â DMEM plus 10% FBS. EBV-positive 293 cells were selected and maintained with 200 mg/mL hygromycin B.
Generation of BALF5 mutant expression vectors. Expression vectors encoding the wild type (Addgene no. 192454) and D755N/D757N (Addgene no. 192455), Y753F (Addgene no. 192456), and S758T (Addgene no. 192457) mutants of EBV's BALF5 were generated using Gibson assembly. The pCMV5.1/FLAG-BALF5 plasmid, encoding wild-type BALF5 sequence, was digested with NheI-HF (New England Biolabs [NEB], R3131) and FastDigest KflI (isoschizomer of SanDI; Thermo Fisher Scientific; catalog no. FD2164), and the digested DNA was purified using a QIAquick PCR purification kit (Qiagen; catalog no. 28104). Point mutation D755N/D757N, Y753F, or S758T was introduced by Gibson assembly using NEBuilder HiFi DNA assembly master mix (NEB; catalog no. E2621) following the manufacturer's protocol, with gBlocks gene fragments containing mutated YGDTDS sequences (ordered from Integrated DNA Technologies [IDT]) as inserts at a vector/insert molar ratio of 1:2. Due to GC-level constraints, the gBlocks fragments were designed with several silent nucleotide changes. A gBlocks fragment containing these nucleotide changes but that was wild type in amino acid sequence was used to construct a "codonshuffled" wild-type BALF5 vector and was used as a control. Resulting Gibson mixes were transformed into NEB 10-beta electrocompetent E. coli (NEB; catalog no. C3020K) and selected by plating onto agar plates containing 100 mg/mL carbenicillin. The integrity of the resulting plasmid clones was verified by Sanger sequencing of the BALF5 regions and Oxford Nanopore sequencing of whole plasmids through Plasmidsaurus (https://www.plasmidsaurus.com/), with results from the latter sequencing deposited onto Addgene as referred to above.
ROCC transcomplementation assay. 293/EBV single-gene null-mutant cells were plated on 30-mm MatTek plates (MatTek Corporations; catalog no. P35G-1.5-14-C) and transfected under the following conditions: uninduced (eGFP-H2B 1 carrier), induced (eGFP-H2B 1 Zta 1 Rta 1 carrier), and induced 1 transcomplemented (eGFP-H2B 1 Zta 1 Rta 1 expression vector for null gene). pCDNA3 was used as a carrier. A total of 5 mg DNA and 5 mL of Lipofectamine 2000 (Invitrogen) were used per plate, and transfection was carried out following the manufacturer's protocol, with the following modifications: the cells were washed with DMEM, 1 mL DMEM was readded to each plate (half of the typical growth medium volume), and the transfection reagent mixture was gently added to the medium. The transfection mix was incubated with the cells for 4 h in the 37°C cell incubator (humidified, with 5% CO 2 ), following which cells were washed with 2 mL of DMEM with 10% FBS (D10F) before another 2 mL of D10F was added as growth medium. Cells were grown for 48 h, and then live-cell imaging was performed using a Zeiss Axiovert 200M fluorescence microscope. ROCC-positive cells were counted and normalized against the number of eGFP-positive cells, and >400 eGFP-positive cells were counted per condition (see Table S1 in the supplemental material) from a total of three transfection replicates. Fisher's exact test was used for statistical analysis of I versus I1t group pairs of each knockout (KO) cell line. Representative images of cells from transcomplementation assays were taken using the Nikon A1RS HD confocal microscope, to show clearly ROCC from a single, static z-slice.
Assay for the inhibition of lytic DNA synthesis. iD98/HR1 eGFP-H2B cells (13) were plated on 22by 22-mm coverslips in six-well plates and grown overnight. These cells were then treated with either the viral DNA synthesis inhibitor ganciclovir (GCV; 40 mg/mL; Calbiochem) or phosphonoacetic acid (PAA; 100 mg/mL; Sigma), incubated for 24 h, and then induced for EBV's lytic phase using 200 nM 4hydroxytamoxifen (4-OHT; Sigma). The concentrations of GCV and PAA used were based experimentally on their inhibiting DNA synthesis optimally, as measured by qPCR and incorporation of EdU, and not being toxic (see also reference 43). At 24 h postinduction (hpi), cells were pulsed with 10 mM 5-ethynyl-29-deoxyuridine (EdU; Millipore Sigma; catalog no. 900584) for 30 min, at 37°C, and then washed three times with D10F. D10F was added back to the wells along with the appropriate 4-OHT and GCV or PAA concentration. The EdU pulse was repeated at 48 h postinduction, following which cells were fixed for click chemistry and immunofluorescence assays. The effects of GCV and PAA on the viability of cells were measured, and they were found to not affect cell growth in induced populations (data not shown). Additional experiments were performed using a DAPI exclusion assay to measure the viability of cells upon treatment with PAA. Briefly, cells were treated with PAA and induced for EBV's lytic phase as described above. At 48 hpi, cells were washed twice with 1Â Dulbecco's phosphate-buffered saline (DPBS), stained with 1.5 mg/mL DAPI (Invitrogen) solution, incubated for 10 min at room temperature, and washed twice more with 1Â DPBS. Live-cell fluorescence microscopy was used to assess DAPI signals from these cells. Ninety-seven percent of these cells excluded DAPI, indicating that they were viable at 48 hpi, when they were assessed for their ROCC phenotypes (see Fig. S4).
Quantification of viral DNA copy number by qPCR. Cells were trypsinized, counted, pelleted, and resuspended in 1Â DPBS at a concentration of 10 7 cells/mL. Cell suspensions were sonicated using a Qsonica Q700 sonicator at 100 mA for at least 1 min. Sonicated cell suspensions were incubated with RNase A (Roche; final concentration, 100 mg/mL) for 30 min at room temperature. Following RNase A treatment, the cell lysates were incubated with proteinase K (Roche; final concentration of 100 mg/mL), 0.1% sodium dodecyl sulfate (SDS), and 1 mM EDTA for 2 to 3 h at 50°C. DNA was isolated from these samples by phenol-chloroform extraction. Briefly, 20 mg of linear acrylamide carrier was added to the samples, they were mixed with an equal volume of 1:1 phenol-chloroform solution and vortexed, and the aqueous phase was again extracted with an equal volume of chloroform. DNA was precipitated from the resulting aqueous solution by ethanol precipitation and resuspended in a small volume of Tris-EDTA (TE) buffer. Fifty nanograms of each purified DNA sample, in triplicate, was subjected to quantitative PCR (qPCR) in a 384-well plate as previously described (61). Briefly, each 20-mL reaction mixture contains 50 ng of the purified sample DNA in a 2-mL volume (diluted in water where necessary), 0.5 mM (each) forward and reverse primers, 0.2 mM probe, 1Â ROX reference dye (Invitrogen), and 1Â AmpliTaq Gold 360 master mix (Applied Biosystems). The reaction mixtures were incubated at 50°C for 2 min and then at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Measurements were collected on a 7900HT Fast real-time PCR system (Applied Biosciences) and analyzed with the Sequence Detection Systems (SDS) 2.4 from Applied Biosystems. The following primer and probe sequences, specific to EBV oriP and cellular rhodopsin, were used: oriP (forward, 59-AGAAGCAGGCGAAGATTCAG-39; reverse, 59-CCATTTTAGTCACAAGGGCAG-39; probe, 59-/56-FAM/AAGATCAAG/ZEN/GAGCGGGCAGTGAA/ 3IABkFQ/-39), rhodopsin (forward, 59-ATCAGGAACCATTGCCACGTCCTA-39; reverse, 59-AGGCCAAAGATGG ACACACAGAGT-39; probe, 59-/56-FAM/AGCCTCTAG/ZEN/TTTCCAGAAGCTGCACA/3IABkFA/-39). Standard curves were generated for each primer/probe set at a range of 10 3 to 10 9 molecules per well, in triplicate. The following plasmids were used for standard curves: for EBV oriP, p152.22 (previously described as pHEBo2 [62]), linearized with BamHI-HF; for rhodopsin, p3934 (a plasmid containing a rhodopsin amplicon), linearized with PvuI. p2134, a plasmid not containing oriP or rhodopsin amplicons, was used as carrier. Three technical replicates were used per condition per transfection replicate, with a total of three transfection replicates measured. Statistical analysis was performed using Welch's two-sample t test comparing each condition to the induced (without transcomplementation) condition.
Click chemistry and immunofluorescence. Cells grown on 22-by 22-mm coverslips were washed twice with 1Â DPBS and fixed with 4% paraformaldehyde (PFA) solution in 1Â DPBS (10-min incubation at room temperature). The cells were then washed three times with 1Â DPBS, permeabilized with 0.2% Triton X-100 solution in 1Â DPBS (10-min incubation at room temperature), and washed three times with 1Â DPBS. For EdU detection, coverslips were incubated in freshly prepared click chemistry solution (4 mM CuSO 4 , 50 mM ascorbic acid, 5 mM Cy5-azide [Sigma-Aldrich; catalog no. 777323] in 1Â DPBS). Click reaction mixtures were incubated for 30 min, in a dark humidified chamber, at room temperature. Coverslips were then washed three times with 1Â DPBS. For immunofluorescence, samples were blocked with a 2% bovine serum albumin (BSA) solution and incubated for 30 min with gentle rocking. Primary antibodies were prepared in in 2% BSA solution, coverslips were incubated with this solution for 1 h at room temperature, and they were washed three times with 1Â DPBS. Secondary antibodies were prepared in 2% BSA solution, and coverslips were incubated with this solution for 1 h at room temperature and then were washed three times with 1Â DPBS. All antibody incubations were done in a dark, humidified chamber. Coverslips were then carefully dried and mounted with Vectashield (Vector Laboratories) plus 1.5 mg/mL DAPI (Invitrogen). Immunofluorescence images were acquired using a Nikon A1RS HD confocal microscope. Primary antibodies were mouse anti-BMRF1 (EMD Millipore; MAB8186; 1:1,000) and rabbit anti-SV40 large T antigen (GeneTex; GTX134378; 1:250). Secondary antibodies were Alexa Fluor 568 goat anti-mouse (Molecular Probes; A11019; 1:1,000) and Alexa Fluor 647 donkey anti-rabbit (Invitrogen; A31573; 1:500).
SV40 cotransfection assay. iD98/HR1 cells were plated on coverslips in 6-well plates and treated with 100 mg/mL of PAA, 24 h prior to being mock transfected or transfected with isolated SV40 DNA. SV40 DNA isolation was carried out as previously described (64). The transfection mix was incubated with the cells for 4 h in the 37°C cell incubator (humidified, with 5% CO 2 ), and then the cells were washed with 2 mL of DMEM with 10% FBS (D10F) before another 2 mL of D10F was added as growth medium. The cells were grown for 1 h, and then 100 mg/mL PAA was added back to the medium. Cells were grown for another 2 h and then were induced to enter EBV's lytic phase by the addition of 200 nM 4-OHT. At 48 h after transfection, the cells were incubated with 10 mM EdU for 1 h and then fixed for click chemistry and immunofluorescence assays.
Statistical analysis. For EBV transcomplementation, GCV/PAA treatment, and SV40 assays, the total number of cells counted per condition exceeded a predetermined number generated using power calculations to obtain a statistical power of 0.9, with a of 0.05, based on preliminary experiments with a smaller sample size. The program Mstat v7.0 was used for power calculations and statistical analyses with Fisher's exact test (N. Drinkwater, McArdle Laboratory for Cancer Research, School of Medicine and Public Health, University of Wisconsin) and is available for download (https://oncology.wisc.edu/mstat/). R version 4.1.3 was used for statistical analyses using 2 Â 3 Fisher's exact test and Welch's t test and is available for download (https://www.r-project.org/).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.01 MB.