INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Precise maintenance of the cellular genome requires exact doubling of the genome once and only once during the S phase and proper partitioning of the chromosomes between the daughter cells during the M phase of the cell cycle (26). Some DNA viruses, like papillomaviruses and Epstein-Barr virus (EBV), replicate as episomal multicopy nuclear plasmids in the host cells' nuclei during a latent infection (11, 13). In order to be successfully maintained in host cells during latency, these viruses have to possess certain control mechanisms that couple multiplication of the viral genome and partitioning to the host genome maintenance cycle. The relatively small size of the papillomavirus genome puts certain limits on the use of these maintenance mechanisms. It is clear, for example, that episomal DNA viruses, unlike the cellular chromosomes, cannot afford to possess long and complex centromeric regions in their genomes that could ensure the proper partitioning and nuclear retention functions during mitosis. Therefore, some other strategy has to be used instead.

Papillomaviruses infect basal epithelial and mucosal cells in a wide range of different hosts. The infection can cause benign or malignant lesions; the most known example is common skin warts. Papillomavirus genome replication can be generally described as a three-step process (11). After entry into the basal cells, the viral genomes are quickly amplified in the host cell nucleus. Initial amplification is followed by a viral latency period, during which the viral genomes are maintained extrachromosomally at a constant copy number in the proliferating host cells. The final, vegetative amplification stage, where the formation of new infectious particles occurs, takes place only after the host cells have terminally differentiated into keratinocytes.

The process of initiation of papillomavirus DNA replication has been extensively studied, focusing mainly on bovine papillomavirus type 1 (BPV1) as a model. Only two viral proteinsE1 and E2are required for this process, and all other necessary components are derived from the host replication machinery (5, 16, 38-40). E1 has been shown to be a viral origin recognition factor and helicase (12, 33, 41). E2, apart from being a central viral transcription regulator (9, 23), also acts as an auxiliary factor that binds to E1 and to the replication origin in a cooperative manner, thus facilitating the formation of replication initiation complex (2, 21, 24, 32, 35). The origin of papillomavirus replication has been located to the noncoding upstream regulatory region (URR). The minimal part of the URR, sufficient for the initiation of viral replication (minimal origin of replication), is composed of an A/T-rich region, binding site for E1, and one binding site for E2 (37, 39). URR sequences of different papillomaviruses contain a different number of E2 binding sites that also play an important role in viral latency. The URR of BPV1 contains 12 E2 binding sites that together form a BPV1 minichromosome maintenance element (MME). This element, in addition to the minimal origin of replication, is required for long-term episomal maintenance of BPV1 replicator in cells expressing the E1 and E2 proteins. A sufficient number of high-affinity E2 binding sites is critical for proper MME function (27). However, the function of E2 binding sites in the stable maintenance of the viral genome has been unclear until lately. Two recent publications provided the first insights, showing that BPV1 genomes, as well as E2 protein, are localized to host cell mitotic chromatin in C127 mouse fibroblasts and that mutations in E2 and E1 coding regions are able to affect such localization (18, 34).

In this study, we demonstrate that MME is likely to exert its role in episomal minichromosome maintenance of the BPV1 genome through the viral E2 protein-mediated association with the host cell nuclear chromatin. Viral E2 protein in trans and MME, comprised of multiple E2 binding sites, in cis, are both necessary and sufficient for chromatin attachment of the plasmids in our model system. On the other hand, the E1 protein or its binding site as well as the plasmid replication function can be removed without affecting the plasmid association with the chromatin. These data suggest that E2-mediated association of the viral genomes with nuclear chromatin is likely to guarantee the proper partitioning and nuclear retention of papillomavirus genomes in dividing cells as well as the optimal exposure of papillomavirus replicon to cellular replication control mechanisms during S phase. Therefore, the BPV1 stable episomal maintenance consists of two main functionschromatin attachment, which provides proper partitioning and nuclear retention to the viral genomes, and replication function, which is responsible for compensation of the plasmid loss during host cell division.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid constructs. Plasmid pNeo10E2BS9 contains 10 oligomerized head-to-tail copies of high-affinity E2 binding site 9 and was constructed by inserting the BamHI-Ecl136II fragment (containing the synthetic E2 binding site oligomer) from plasmid Msp/15+10×BS9 (37) between BamHI and HpaI sites of pNeo vector. All other BPV1 URR plasmids (27) and BPV-1 E1 expression vector pCGEag (38) were described previously.

Cells and transfections. Chinese hamster ovary cell line (CHO) derivatives CHO49 (expressing BPV1 E2 protein), CHO4.15 (expressing BPV1 E1 and E2), and CHOBgl40 (CHO4.15 cells that maintain BPV-1 full-length URR plasmid pNeoBgl40 episomally) (27) were maintained in Ham's F12 medium supplemented with 10% fetal calf serum. Electroporation experiments were performed with a Bio-Rad Gene Pulser with capacitance and voltage settings of 975 µF and 230 V, respectively. The transfection efficiencies were determined by in situ staining of the cells transfected in parallel with a -galactosidase-expressing plasmid pON260 (38). The extraction of episomal DNA from cells and its analysis by Southern blotting were performed as described previously (38).

Cytogenetic analysis. Chromosome preparations were done by standard methods. Briefly, cells were exposed to Colcemid added at a final concentration of 0.1 µg/ml for 1 to 4 h to enrich the mitotic fraction. Colcemid-treated cells were harvested by trypsin treatment and suspended in a 0.075 M KCl solution, incubated at room temperature for 15 min, and fixed in ice-cold methanol-glacial acetic acid (3:1 [vol/vol]). The spread-out chromosomes at metaphase and nuclei at interphase for cytogenetic or fluorescence in situ hybridization analysis were prepared by dropping the cell suspension on wet slides. Chromosome analysis was performed by standard staining methods. CHO cells were karyotyped by G-banding analysis as described previously (4).

FISH. Cells were harvested and prepared for analysis as described above. Hybridization probes were generated by nick translation, using biotin-16-dUTP as a label and pNeoBgl40 plasmid as a template. The final size of probe fragments was adjusted to 100 to 300 bp by DNase I digestion in all cases. Fluorescence in situ hybridization (FISH) was performed essentially by the protocol of Tucker and coauthors (36). Briefly, chromosome preparations were denatured at 70°C in 70% formamide (pH 7.0 to 7.3) for 5 min, then immediately dehydrated in a series of washes (70, 85, and 96% ice-cold ethanol washes [for 3 min each]), and air dried. The hybridization mixture (18 µl per slide) was composed of 50% formamide in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 10% dextran sulfate, 160 ng of a biotinylated plasmid probe DNA, and 10 µg of herring sperm carrier DNA. After 5 min of denaturation at 70°C, probe DNA was applied to each slide, sealed under a coverslip, and hybridized for 2 days at 37°C in a moist chamber. The slides were washed in three changes of 2× SSC containing 50% formamide, 2× SSC, and 2× SSC containing 0.1% IGEPAL CA-630 (Sigma Chemical Co.) at 45°C. Prior to immunofluorescence detection, slides were preincubated for 5 min in PNM buffer (PN buffer [25.2 g Na₂HPO₄ · 7H₂O, 0.83 g NaH₂PO₄ · H₂O, and 0.6 ml of IGEPAL CA-630 in 1 liter of H₂O] with 5% nonfat dried milk and 0.02% sodium azide). After that, the probe was detected with fluorescein isothiocyanate (FITC)-conjugated extravidin. The signal was amplified with biotinylated antiavidin antibody and a second round of extravidin-FITC treatment. Between each of the steps, the slides were washed in PN buffer containing 0.05% IGEPAL CA-630 at room temperature for 15 min. Chromosomes were counterstained with propidium iodide and mounted in p-phenylenediamine antifade mounting medium. Slides were analyzed with a Olympos VANOX-S fluorescence microscope equipped with appropriate filter set. All FISH experiments were coded, and the chromosomes from at least 50 cells at metaphase and at least 200 interphase nuclei were analyzed on each slide. In addition, two slides from each sample were prepared, hybridized, and scored on different dates. Fuji Fujicolor and Agfa Agfacolor films for color prints were used for photomicrographs.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

BPV1 URR-containing plasmids are associated with the host cell chromatin. It has been shown that the full-length BPV1 genomes are attached to the chromatin in C127 cells (18, 34). First we decided to examine whether the plasmids, which contain only the BPV1 URR sequences, possess the same ability in the Chinese hamster ovary (CHO) cell line-based model system, developed by us for the study of transient replication and stable maintenance of BPV1 (27, 39). We first analyzed the CHOBgl40 cell line, which expresses the BPV1 E1 and E2 proteins from integrated cassettes and maintains extrachromosomally the full-length BPV1 URR (Fig. 1) containing plasmid pNeoBgl40. A FISH analysis of both, prefixed mitotic metaphase spreads and interphase nuclei was performed with biotin-labelled BPV1 URR plasmid-specific DNA probe. The signals from hybridized probe were detected and amplified with FITC-conjugated extravidin and antiavidin antibodies, as described in Materials and Methods. The representative data are shown in Fig. 2A. The discrete yellow dots corresponding to plasmid-specific signals appeared as a merged yellow signal of the green FITC fluorescence on the red background of nuclear DNA stained with propidium iodide. The BPV1 URR plasmid signals were localized on the metaphase chromosomes with obviously random pattern distribution. Random distribution of plasmid signals was also observed in the interphase nuclei. Almost all (~90%) interphase nuclei and mitotic metaphase chromosomes from 180 analyzed cells contained BPV1 URR plasmid-specific signals, with overall number of plasmid signals from around 10 to 50 per nucleus in the majority of individual nuclei analyzed. This number is close to the estimated average of episomal plasmid copy number in CHOBgl40 cells (27), suggesting that FISH analysis was sensitive enough to detect every single plasmid copy in fixed nuclei. A small proportion of the cells from the total population contained higher number of signals (2% of cells containing more than 40 signals). However, the fractions of both high-copy-number and plasmid-negative phenotype nuclei varied significantly (up to 20% in both fractions) in several other CHOBgl40 subclones that were derived from the same long-term passage cell population. This apparent heterogeneity supports the earlier suggestion based on the similar phenotype in the case of long-term stable maintenance of full-length BPV1 genomes (28, 29, 31), that the papillomavirus partitioning and replication processes are not subjected to very strict control mechanisms.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Schematic representation and some relevant properties of the BPV1 URR constructs used in this study. The presence (+) or absence () of intact replication origin (replication), intact MME (sufficient number of E2 binding sites), competence for stable episomal maintenance in the long-term assay (stable maintenance), and competence of the construct for attachment to the host cell chromatin as determined by FISH analysis are indicated to the left of the schematic representations. The numbers in the schematically represented DNA sequences are the nucleotide positions in the BPV1 genome.

View larger version (40K): [in this window] [in a new window]   FIG. 2.   Multiple E2 binding sites determine the competence for chromatin attachment, but a functional replication origin is not necessary for this activity. The results of FISH analysis in the CHOBgl40 cell line that stably maintains a full-length BPV1 URR plasmid pNeoBgl40 (A) and of CHO4.15 cells transfected with plasmids (1 µg) containing different BPV1 URR inserts (B to G) that are depicted schematically in Fig. 1. Panel G shows the control experiment with plasmid containing no URR sequences E1 and E2 expression cassettes integrated into genome give cross-hybridization signals represented by double dots (indicated by arrowheads).

The multimerized E2 protein binding sites determine the chromatin attachment of the plasmids in the CHO4.15 cells. The data presented above showed clearly that chromatin attachment of the BPV1 URR plasmids could be studied in transient-transfection assay. A panel of different BPV1 URR-derived constructs (Fig. 1) in the same plasmid context, pNeo5', was transfected into the BPV1 E1- and E2-expressing cell line CHO4.15. Half of the cells were fixed after 48 h, and FISH analysis of the plasmid localization with specific DNA probe was performed. Low-molecular-weight DNA was extracted from the other half of the cells and analyzed by Southern blotting to estimate the overall level and replication competence of the transfected plasmid DNA in cells. FISH analysis indicated that in addition to the intact full-length URR plasmid (Bgl40) (Fig. 2B), the plasmid containing URR with disrupted E1 binding site (XhoHpa) also displays the localization to mitotic chromatin (Fig. 2C). The fraction of nuclei considered plasmid positive was smaller (usually 10 to 20%) than the transfection efficiencies estimated in parallel with -galactosidase expression plasmids (60 to 70%). These differences may be explained by different sensitivities of -galactosidase staining and FISH protocols. Alternatively, the lower percentage of the positive cells by FISH analysis may indicate that not all plasmid molecules that reach the nucleus after transfection will be able to attach to the chromatin or perhaps they will require longer incubation periods. For example, the successful establishment of the chromatin association may be dependent on passage of the cells through the particular cell cycle phase. One possible explanation could also be that chromatin attachment requires higher E2 levels than in some of the cells, but previous immunofluorescence analysis of the status of the E2 protein has not revealed any significant heterogeneity in the used subclones of CHO4.15 cell line (data not shown).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Southern blot analysis of the extrachromosomal DNA from cells used for the parallel FISH analysis (see Fig. 2 for FISH results). Lane M contains 100 pg of linearized plasmid marker (pNeoBgl40). Lanes 1 and 2 contain extrachromosomal DNA from CHOBgl40 cells maintaining the BPV1 URR plasmid pNeoBgl40 episomally, and all the other lanes correspond to different transfections with BPV1 URR constructs (1 µg of each plasmid) in CHO4.15 cells. The mock-transfected control cells (carrier) are indicated. DNA preparations were digested with the appropriate restriction enzyme to linearize the plasmid DNA and with DpnI where indicated (+ if added,  if not) which digests only bacterially methylated DNA, thus revealing the de novo-replicated plasmid pool.

Multiple E2 binding sites in cis and viral E2 protein in trans are the viral determinants of the chromatin attachment activity of the BPV1 URR-derived plasmids. The fact that oligomerized E2 binding sites were sufficient while functional replication origin and replication ability were unnecessary for chromatin attachment made us hypothesize that the attachment occurs only if viral E2 protein were provided in cells. In order to test that, we transfected the BPV1 E2-expressing CHO49 cell line with the same plasmids. Forty-eight hours after transfection, the cells were processed further for FISH analysis to demonstrate the plasmid localization in the nuclei and for parallel Southern blotting analysis to determine the plasmid levels in cells as described above in the case of CHO4.15 cells. As shown in Fig. 4, the E2 protein alone appeared to be sufficient in trans for providing the chromatin attachment activity for URR plasmids in the host cell nuclear background. Similar to the results obtained with E1- and E2-expressing CHO4.15 cells, all constructs containing a sufficient number of E2 binding sites, Bgl40, XhoHpa, and D234/221+10E2BS9 (Fig. 4A, B, and C), were attached to mitotic chromosomes. Of transfected cells, 10 to 20%, depending on the transfection, were clearly positive for plasmid signals, with transfection efficiencies estimated in parallel around 50%. The percentage of positive chromosomes at metaphase was again approximately equal to the percentage of positive nuclei at interphase in analyzed individual transfections, and the plasmid-specific hybridization signals followed an apparently random pattern. Control analysis with construct that contains only two E2 binding sites did not reveal any chromosomal localization of the plasmid (Fig. 4D), even though according to the Southern blotting analysis, the plasmid DNA was present in cells in case of this and other constructs used (Fig. 5, compare lanes 1 to 8). Cross-hybridization with chromosomally integrated E2 expression cassettes (one site on two different chromosomes) served as an internal control for the success of the FISH analysis.

View larger version (68K): [in this window] [in a new window]   FIG. 4.   The chromatin attachment of URR plasmids occurs also in the absence of E1 expression; E2 protein determines the attachment activity. The results of FISH analysis in the transient assay using cell line CHO49 that expresses only BPV1 E2 protein are shown. Cells were transfected with 1 µg of plasmids containing BPV1 URR inserts depicted schematically in Fig. 1. E2 expression cassette integrated into genome gives cross-hybridization signal represented by double dots (indicated by arrowheads).

View larger version (30K): [in this window] [in a new window]   FIG. 5.   Southern blot analysis of the extrachromosomal DNA from cells used for the parallel FISH analysis (see Fig. 4 for FISH results). Lane M contains 100 pg of linearized plasmid markers (pNeoBgl40 and pNeo). Lanes 1 to 8 correspond to different transfections with BPV1 URR constructs in CHO49 cells. BPV1 E1-expressing plasmid pCGEag (250 ng) was cotransfected on the panel Bgl40+E1 (lanes 1 and 2) as a control for E2 expression (ori plasmid replicates only if both E1 and E2 proteins are present). DNA preparations were digested with the appropriate restriction enzyme to linearize the plasmid DNA and with DpnI where indicated (+ if added,  if not) which cuts only bacterially methylated DNA, thus revealing the de novo-replicated plasmid pool.

The competence of BPV1 URR plasmids for stable episomal maintenance correlates with their ability to associate with host cell chromatin. A sufficient number of E2 binding sites form the MME which together with the viral minimal replication origin provides the long-term episomal maintenance property for the BPV1 replicator (27). Because of our results indicating that MME also determines the chromosomal attachment activity for BPV1 URR, we decided to further analyze the possible connection between the stable maintenance and chromosomal attachment activities. A panel of BPV1 URR deletion constructs with known stable maintenance properties was transfected into CHO4.15 cells (Fig. 6). Cells were processed 48 h after transfection for the FISH analysis to demonstrate the chromatin attachment of plasmids (Fig. 7) and for parallel Southern blotting analysis to detect the plasmid levels in cells (Fig. 8). The results of FISH analysis demonstrated clear correlation between the competence of each plasmid for stable episomal maintenance and its ability to associate with host cell chromatin. Only constructs with functional MME (DCla/234, DCla/41, and D221/134 [Fig. 7C, D, and E]), previously shown to be capable of stable maintenance (27), were tightly associated with the mitotic chromosomes at metaphase and in the nuclei at interphase. Constructs lacking functional MME and incapable of stable replication (D221/234 and D134/234 [Fig. 7A and B]), also failed to localize to chromosomes.

View larger version (19K): [in this window] [in a new window]   FIG. 6.   Schematic representation of the structure and some relevant properties of the second panel of BPV1 URR deletion variants used in this study. See the legend to Fig. 1 for explanation of abbreviations, designations, etc.

View larger version (54K): [in this window] [in a new window]   FIG. 7.   The competence of BPV1 URR constructs for stable episomal maintenance correlates with their ability to associate with host chromatin. The results of FISH analysis in the transient assay using cell line CHO4.15 are shown. Cells were transfected with 1 µg of plasmids containing BPV1 URR inserts depicted schematically in Fig. 6. E1 and E2 expression cassettes integrated into the chromosomal DNA give cross-hybridization signals represented by double dots (indicated by arrowheads).

View larger version (27K): [in this window] [in a new window]   FIG. 8.   Southern blot analysis of the extrachromosomal DNA from cells used for the parallel FISH experiments (see Fig. 7 for FISH results). Lane M contains 100 pg of linearized plasmid markers (pNeoBgl40 and pNeo). Lanes 1 to 10 correspond to transfections with BPV1 URR constructs in CHO4.15 cells in the transient assay. DNA preparations were digested with HindIII to linearize the plasmid DNA and with DpnI (+ if added,  if not) which cuts only bacterially methylated DNA, thus revealing the de novo-replicated plasmid pool.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Two recent articles dealing with BPV1 chromatin attachment have reviewed the results of studies of full-length viral genome DNA in mouse fibroblasts (18, 34). This system has the disadvantage of being relatively complicated, because all viral early genes, including oncogenes, are expressed from episomal viral genome in these transformed cells. Therefore, the complex interplay between E1 and different E2 transactivation and repressor forms in the processes of regulation of viral transcription, transformation, replication, and genome copy number complicates unambiguous interpretation of the involvement of different viral gene products in chromatin attachment. We have used a different approach, trying to simplify the system as much as possible, looking for the minimal viral determinants for the chromosome attachment activity. In this regard, E1- and/or E2-expressing stable cell lines serve as good model systems. These cells do not express any other papillomavirus proteins and express constant E1 and E2 levels from the integrated constructs, thus providing a more defined system for comparative studies on the behavior of different BPV1-derived constructs.

Previous studies have pointed toward the E2 protein as being the best candidate for viral trans factor required for chromatin attachment. The genetic analysis in the full-length viral genome context by Lehman and Botchan (18) has suggested that in addition to E2, the viral E1 protein seems to participate in the tethering of viral genomes to chromosomes. Our data show that E2, in the absence of E1, can be sufficient for the chromatin attachment of BPV1 URR plasmids. It is possible that E1, as well as other viral (and cellular) proteins, does contribute, indirectly or directly through interaction with the E2 protein to the attachment process. However, E2 protein clearly appears to be the central viral trans determinant for this process.

This is also the first study of the viral cis elements that determine the chromatin attachment. We show that MME, which is composed of E2 binding sites and is necessary for stable episomal maintenance of BPV1 replicon, is also necessary and sufficient for chromatin attachment activity. The experiments with BPV1 URR deletion constructs demonstrated clear correlation between the competence for stable maintenance and chromosome association. Thus, MME is likely to exert its role in the stable maintenance of BPV1 episomes by providing access to necessary cellular control mechanisms through association with host cell chromatin, presumably providing access to those cellular mechanisms that grant the partitioning and nuclear retention functions to the viral genome.

It is interesting to note that chromatin attachment occurs both in the short-term transient-transfection and long-term stable-maintenance assays. This fact supports the idea that the attachment of the viral genome to the chromatin may occur soon after sufficient levels of the E2 protein have been achieved in the cell and is not a result of the long-term selection process. It is possible that the establishment of the chromatin association is linked specifically to a certain stage of the host cell cycle as has been shown, for example, in the case of the formation of the preinitiation complex on the chromosomal replication origins. However, no experimental data are available at the moment to clarify this point. The initial viral amplification during S phase probably creates and maintains the starting population of viral genomes large enough for subsequent finding and occupying of the optimal attachment sites on the chromatin. On the other hand, the analysis of different BPV1 URR deletion constructs demonstrated clearly that the replication and chromatin attachment functions are separate E2-dependent activities of the BPV1 replicator. Thus, the plasmid replication process itself is not directly linked to the chromatin attachment process. Resulting chromatin attachment is very likely to guarantee the viral genome partitioning and nuclear retention functions during host cell division, as was suggested previously (3, 18). In addition, chromatin association can also provide the cellular replication control function to the viral origin through the optimal exposure to chromatin-associated regulatory complexes. The latter may be needed in order to avoid undesired viral overreplication and therefore can provide the copy number control mechanism for the virus during latency.

According to results of our FISH analysis, the plasmids that failed to show any attachment to metaphase chromosomes also failed to show any staining in the interphase nuclei of the transfected cells. Replication-competent plasmids that failed to give any FISH signal were capable of replicating in the same cells according to Southern blotting analysis (e.g., pNeoAlu), confirming that these plasmids had to be present in the nucleus before FISH analysis was performed. Thus, it seems most likely that plasmids which were not attached to the chromatin were simply washed away both from metaphase chromosomes and interphase nuclei during fixation and following steps of the FISH procedure. On the other hand, in the case of attachment-competent plasmids, we could not observe any considerable difference in the percentage of plasmid-specific staining if the interphase nuclei and the mitotic chromosomes at metaphase in the same transfected populations were compared. These data suggest that the MME-dependent association with host chromatin could be maintained throughout the cell cycle, including S phase. It can also be speculated that the replication of stably maintained BPV1 replicator in S phase could take place on the host chromatin, where these genomes are well exposed to the replication control mechanisms that are utilized during host genome multiplication. However, additional and more-detailed studies are necessary to examine these possibilities.

The above-proposed possible access of chromatin-attached papillomavirus genomes to chromatin-associated cellular control mechanisms cannot be sufficient to grant the viral genome with very precise replication control. It is known that the papillomavirus genome is not replicating in a strict once per cell cycle mode during the viral latency that is used by host genome but rather follows a random-choice statistical initiation mechanism (7, 27, 29). On the other hand, an example of EBV indicates that once per cell cycle replication mode can still be achieved by episomal DNA viruses (42). EBV genome plasmids and viral latent replication origin (oriP) binding protein EBNA1 are associated with the host chromosomes (8, 10), and EBNA1 is able to provide nuclear retention function to the plasmids containing multiple EBNA1 binding sites (15). It is very likely that EBNA1, similar to E2 in the case of BPV1, mediates the attachment of viral genome to chromatin. Thus, chromatin attachment as a tool to exploit cellular control mechanisms for coupling the viral partitioning and replication to the host cell genome maintenance cycle may represent a more general feature for nonlytic episomal DNA viruses. The similar functional role for DNA binding proteins and their binding sites in partitioning function has also been reported for bacterial plasmids, bacterial chromosomes (19, 20), and Saccharomyces cerevisiae plasmids (1). These data seem to point toward general evolutionary similarities in different mechanisms of partitioning of the chromosomal and extrachromosomal elements.

E2 protein appears to be necessary and sufficient for linking of the MME-containing plasmids to the chromatin. As was discussed above, E2 protein has previously been shown to be capable of associating with the chromatin (18, 34). Two previous studies have indicated that the N-terminal transcription and replication activation domain of the E2 protein is crucial for the chromatin attachment activity of the protein itself. In addition, Lehman and Botchan (18) suggest that the hinge region between N- and C-terminal domains, which includes the major phosphorylation sites of the E2 protein, is also important for the attachment. Based on these and our data, it seems reasonable to assume that both the N-terminal chromatin-bound transactivation domain and the C-terminal MME-bound DNA binding domain, serve as necessary linkers for tethering MME-containing plasmids to the host chromatin. The E2 protein binding affinity to multiple oligomeric binding sites in MME would be remarkably high due to the cooperative interaction of the bound E2 molecules with DNA (14, 25). This would provide a tightly bound proteinaceous surface formed by multiple E2 N-terminal activation domains, which is responsible for the high efficiency of the interaction with the host chromatin. Efficient multicontact interaction with chromatin might explain why this survives a relatively harsh treatment, including DNA denaturation step, during the FISH procedures. The interaction with chromatin is sufficiently strong only in the case of E2 transactivation domain, because replacing it with the respective VP16 or p53 domain inactivates the hybrid protein's ability to tether the plasmids to the chromatin in CHO and human cells (22). The chromatin binding and the DNA binding, replication, and transcription activities of the E2 protein are possibly modulated through its phosphorylation and other posttranslational modifications. This could also explain the effect of the E2 protein linker region between N- and C-terminal domains in the regulation of the chromatin binding, as the modifications in hinge region may alter the placement of the protein domains in regard to each other (18). Also, the regulation of the full-length E2 protein by its repressor forms through heterodimer formation should be considered (6, 17). Altogether it could provide a complex regulatory mechanism to control the BPV1 genome multiplication and maintenance during viral latency.

It is still hard to guess which cellular factors from the chromatin side are required for the papillomavirus genome attachment. In the case of BPV1, the minimal number of E2 binding sites sufficient to provide the minichromosome maintenance function exceeds the number of these sites generally found in upstream regulatory region of different human papillomavirus (HPV) types. Thus, in the case of the stable maintenance of the HPV genome in the transformed cells, some additional viral or cellular factors are probably necessary to provide the chromatin attachment activity. HPV URR sequences carry a so-called enhancer region, which contains numerous binding sites for different cellular transcription factors. It is tempting to speculate that certain cellular transcription activators or specific combinations of these activators, through some feature common with E2 protein, may compensate for the lack of sufficient contribution from HPV E2 binding sites. On the other hand, HPV E2 protein may provide some organizing function to these enhancer binding proteins. Interestingly, the EBNA1 protein, which is believed to be a possible mediator of the chromatin attachment of EBV genome, is also a viral transcription activator (30). It is possible that the target from the nuclear chromatin side, which allows the viral genome anchoring, may be identical in all these cases. However, the existence of such an attractive common mechanism in the case of different episomal DNA viruses remains to be proven in the future.