In Vivo Accumulation of Cyclin A and Cellular Replication Factors in Autonomous Parvovirus Minute Virus of Mice-Associated Replication Bodies

Abstract

Autonomous parvovirus minute virus of mice (MVM) DNA replication is strictly dependent on cellular factors expressed during the S phase of the cell cycle. Here we report that MVM DNA replication proceeds in specific nuclear structures termed autonomous parvovirus-associated replication bodies, where components of the basic cellular replication machinery accumulate. The presence of DNA polymerases α and δ in these bodies suggests that MVM utilizes partially preformed cellular replication complexes for its replication. The recruitment of cyclin A points to a role for this cell cycle factor in MVM DNA replication beyond its involvement in activating the conversion of virion single-stranded DNA to the duplex replicative form.

Autonomous parvoviruses and other members of the Parvoviridae family are unique among animal viruses in having linear single-stranded DNA genomes. The termini of their approximately 5,000-nucleotide genomes contain palindromic sequences which fold into stable hairpin structures and serve as primers for viral DNA synthesis. The replicative cycle, which takes place in the nucleus, is initiated by the synthesis of the cDNA strand leading to the formation of double-stranded replicative-form DNA. This reaction, also known as conversion, is exclusively dependent on cellular factors. In addition, the subsequent amplification reactions require the activity of the virus-encoded major nonstructural protein NS1 (7).

Several cellular factors which are essential for DNA replication of the prototype autonomous parvovirus minute virus of mice (MVM) have been identified through cell fractionation and complementation assays in vitro replication systems (5, 8, 20). However, very little is known about the subnuclear organization of MVM DNA replication in vivo. In contrast to viruses with double-stranded DNA genomes that replicate in close proximity to preformed nuclear structures known as promyelocytic leukemia (PML) bodies (for review see reference 18), autonomous parvovirus H-1 was shown to induce characteristic nuclear structures, termed H-1 parvovirus-associated replication (PAR) bodies, which are unrelated to PML bodies (9). Similar structures were also characterized in Aleutian mink disease virus-infected cells (21, 22). In this study, we aimed to determine whether MVM also establishes PAR body-like structures in the nuclei of infected cells and to analyze the subnuclear distribution of cellular factors assumed to be involved in MVM DNA replication in vivo. To this end, A9 mouse cells were infected with MVMp at a multiplicity of infection of 10 PFU per cell. At 15 h postinfection, cells were labeled for 20 min with bromodeoxyuridine (BrdU) at a concentration of 10 μM and then immediately fixed in 1% formaldehyde for 10 min at room temperature. BrdU incorporation into replicated viral DNA was detected with a BrdU-specific antibody (Becton Dickinson) without prior denaturation, thus excluding the detection of chromosomal DNA replication (21). Simultaneously, NS1 was detected with the NS1-specific SP8 antibody (11). Analysis by confocal microscopy (LSM510 UV; Zeiss, Jena, Germany) revealed the accumulation of NS1 in specific nuclear bodies (Fig. 1a) which were also found to be the sites of ongoing viral DNA replication, as indicated by BrdU incorporation (Fig. 1b and c). At this time postinfection, no signs of virus-induced cytotoxicity were visible (Fig. 1d). From these data, we concluded that MVM DNA replication proceeds in specific nuclear structures similar to the ones previously described for other autonomous parvoviruses (9, 21, 22). We would therefore like to propose the more general term autonomous PAR (APAR) bodies for these virus-induced structures.

FIG. 1.

MVM DNA replication colocalizes with NS1 in APAR bodies in the nuclei of infected A9 cells. Representative confocal optical sections through the nuclei of infected cells are shown. NS1 was localized with the SP8 polyclonal antiserum and a fluorescein isothiocyanate (FITC)-conjugated secondary antibody (a). Replication was monitored by incorporation of BrdU and indirect immunofluorescence using a BrdU-specific antibody and a tetramethyl rhodamine isothiocyanate (TRITC)-conjugated secondary antibody (b). In a merged image, colocalized structures from panels a and b appear yellow (c). By phase-contrast microscopy (Nomarski), the cells show no obvious sign of NS1-induced cytotoxicity at the time of fixation (15 h postinfection) (d).

MVM DNA replication starts only after host cells enter the S phase of the cell cycle. Recently, it was shown that in vitro the conversion reaction is activated by the cell cycle factor cyclin A, whose production is induced at the G₁/S transition (1). Cyclin A is involved in the regulation of the S and G₂ phases (25) and has been reported to be required for chromosomal (16) and simian virus 40 (10, 12) DNA replication. In order to test whether cyclin A is present in APAR bodies, we performed double immunofluorescence labeling and confocal microscopical analysis of MVM-infected A9 cells. Cyclin A, which is homogeneously distributed throughout the nuclei of mock-infected cells and absent from nucleoli (reference 2 and data not shown), was indeed found to massively accumulate together with NS1 in the APAR bodies of infected cells (Fig. 2a to c). Even if one assumes that the subnuclear location at which the conversion reaction takes place predetermines the site of APAR body formation, a sole role for cyclin A in the conversion reaction is difficult to reconcile with such a strong accumulation of cyclin A in APAR bodies at 15 h postinfection, when conversion is thought to be complete. This result could indicate that the continuous presence of cyclin A is also required for subsequent NS1-dependent replicative steps. One can furthermore speculate that the sequestration of this factor is responsible for the previously described parvovirus-induced cell cycle arrest (23, 24) because deprivation of cyclin A may lead to a failure of cdk1 activation in the G₂ phase, thus preventing cell cycle progression.

FIG. 2.

Cyclin A, but not cyclin E or cdk2, accumulates in APAR bodies. Representative confocal images of nuclei from double-labeled MVM-infected A9 (a to c) or NBE (d to i) cells are shown. NS1 was detected with FITC using the NS1-specific 3D9 antibody (a generous gift from N. Salomé and D. Pintel) (a, d, and g). Images of cyclin A (b), cyclin E (e), and cdk2 (h) were detected with TRITC in the same confocal plane as shown in the left column using the respective primary antisera for cyclin A (a generous gift from M. Pagano), cyclin E (Ab-1; Neomarkers, Fremont, Calif.), and cdk2 (Ab-3; Neomarkers). The superimposition of both channels from the left and middle columns reveals either the colocalization (yellow signal [c]) or the lack of colocalization (distinct red and green signals [f and i]) of the respective factors in relationship to APAR bodies.

Besides cyclin A, cyclin E is involved in the regulation of the G₁/S transition and is known to play a role in chromosomal DNA replication (14, 16). In contrast to cyclin A, however, cyclin E did not accumulate in APAR bodies (Fig. 2d to f). Since the cyclin E-specific antibody best recognized the human protein, in these experiments we infected human NBE cells and verified that MVMp indeed induces the formation of APAR bodies in this human cell line, as detected by NS1-specific (Fig. 2d) and BrdU-specific (data not shown) antibodies. The failure to detect an accumulation of cyclin E does not exclude the presence of minute amounts, of this factor in APAR bodies. Our data are, however, in agreement with recent in vitro experiments showing no contribution of cyclin E or cdk2 in the initiation of conversion (1).

Cyclin A forms an active complex with cdk2 at the beginning of the S phase. This complex is assumed to be involved in the regulation of DNA replication, since both cyclin A and cdk2 were found to be associated with replication foci in mammalian cells (3) as well as with replicating DNA in the simian virus 40 in vitro replication system (13). We therefore attempted to detect cdk2 in APAR bodies. Using several different cdk2-specific antibodies, cdk2 could not be shown to accumulate in APAR bodies (Fig. 2g to i). The observed accumulation of cyclin A in APAR bodies (Fig. 2b) would then point to a structural role for cyclin A in these bodies, independent of complex formation with cdk2. However, it is still possible that only catalytic amounts too small to be detected by the method used here are required for MVM DNA replication in APAR bodies.

Parvovirus DNA replication proceeds according to a leading-strand-synthesis mechanism and has been shown to be dependent on proliferating cell nuclear antigen (PCNA) in vitro (5). These and other lines of evidence point to DNA polymerase δ as the enzyme responsible for parvovirus DNA replication (6). Here, we show that DNA polymerase δ indeed accumulates in APAR bodies in MVM-infected NBE cells (Fig. 3a to c) and therefore at the sites of ongoing MVM DNA replication. DNA polymerase δ requires for processive DNA synthesis the presence of the cofactor PCNA, which is also involved in other processes, like DNA recombination and repair (15). PCNA is distributed throughout the nucleus except the nucleoli, but it specifically accumulates in the APAR bodies of MVM-infected NBE cells (Fig. 3d to f). This finding corroborates earlier in vitro data demonstrating that PCNA is indispensable for MVM DNA replication (1, 5).

FIG. 3.

Cellular replication factors DNA polymerases δ and α, PCNA, and RPA accumulate in APAR bodies. Representative confocal images of nuclei from double-labeled MVM-infected NBE cells are shown. NS1 was detected with FITC using the SP8 antiserum (a, d, g, and j). Replication factors were detected with TRITC in the same confocal plane as shown in the left column, using the respective antibodies against DNA polymerase δ (Transduction Laboratories) (b), PCNA (PC10; Upstate Biotechnology) (e), RPA (Ab-1; Oncogene) (h), and DNA polymerase α (SJK132-20 [26]) (k), respectively. A merged image of both channels from the left and center columns provides evidence for colocalization of these factors in APAR bodies (c, f, i, and l).

The single-strand-DNA-binding protein replication protein A (RPA) is required for MVM DNA replication in vitro (5) and cannot be replaced by other proteins with single-strand-DNA-binding activity (J. Christensen, personal communication). Furthermore, a direct interaction between the 70-kDa subunit of RPA and NS1 was demonstrated in vitro (4). In agreement with these in vitro findings we have observed that RPA massively accumulates at the sites of ongoing parvovirus DNA replication in infected cells (Fig. 3g to i), which strongly suggests its involvement in this process in vivo.

Surprisingly, DNA polymerase α was also found at the sites of MVM DNA replication (Fig. 3j to l). So far, there is no evidence for a contribution of DNA polymerase α to parvovirus DNA replication, since neutralizing antibodies against this enzyme do not impair viral DNA replication in vitro (5, 19). Furthermore, the presence of preformed hairpin primers makes a DNA polymerase α-dependent primase activity dispensable. However, it was shown that DNA polymerase α can be isolated together with DNA polymerase δ and the PCNA loading factor, replication factor C, from mammalian cells as a stable complex that is replication competent when template DNA, PCNA, and nucleotides are added (17). It could be demonstrated that in this complex, DNA polymerase α is active on single-stranded-DNA templates only when primer synthesis is a precondition for replication. On a primer-template junction, as present in parvovirus DNA, the DNA-binding affinity of replication factor C is higher than that of DNA polymerase α, thereby rendering the latter inactive (17, 27). Since DNA polymerase α is required for chromosomal DNA replication, it is tempting to speculate that MVM utilizes cellular replication complexes that are at least partially preformed, exploiting only the DNA polymerase δ activity and thus evading the need for an energy-consuming dissociation of preformed cellular complexes. This would again highlight the opportunistic and minimalistic strategies used by these viruses.

In summary, the data presented here show for the first time the formation of APAR bodies in MVM-infected cells and the accumulation of various cellular replication factors in these structures. Our data support the idea that the basic cellular replication machinery is used by MVM for viral DNA replication, including DNA polymerase δ, PCNA, RPA, and cyclin A. DNA polymerase α, which is part of the cellular replication machinery, was unexpectedly found to accumulate in APAR bodies but is most likely not active in parvovirus DNA replication. For cyclin A, it will be of interest to identify the role this factor plays in parvovirus DNA replication, in addition to its role in the activation of the conversion reaction in vitro.