TEXT

Top Abstract Text References

Parvoviruses are small nuclear-replicating viruses that infect a wide variety of animal species, including humans (37). Their linear, single-stranded DNA (ssDNA) genome is characterized by its terminal hairpins, small size (about 5,000 nucleotides), and low level of genetic complexity (22). As a result of their limited coding capacity, parvoviruses rely strongly on cellular factors for their ongoing life cycle. It is known that the expression of at least some of these factors is not constitutive but rather is modulated by the physiological state of the host cell. Parvoviruses multiply exclusively in proliferating cells, due to the dependence of virus DNA amplification and expression upon the cell being in S phase (19). Furthermore, cell permissiveness to parvoviruses is conditional on an appropriate differentiation state and can be enhanced by oncogenic transformation (34, 35). This holds true in particular for the autonomous parvovirus minute virus of mice (MVM). Transformation of a number of rodent and human cells by ionizing radiation, chemical carcinogens, tumor viruses, or oncogenes of cellular or viral origin correlates with an increase in susceptibility to MVM infection (7, 10, 29, 36). In the systems tested so far, cell transformation has no significant effect on MVM uptake (10) but does lead to increased parvoviral genome replication (2, 10), viral gene transcription (11, 12), infectious virus production (10), and cell killing (24).

The mechanism underlying the transformation-associated enhancement of the cell's capacity for amplifying parvoviral DNA is unknown. MVM DNA replication involves successive steps (22) depicted schematically in Fig. 1. The single-stranded virion DNA gets converted into a closed, double-stranded, monomeric replicative form (cRF) by extension of the 3'-terminal hairpin (left-hand terminus) and ligation of the growing strand to the folded-back 5' terminus (right-hand terminus) (step 1) (3, 16). Further processing of cRF DNA requires the activity of the major viral nonstructural protein NS1. NS1 is involved in the formation of a strand- and sequence-specific nick at the cRF 5' telomere, followed by initiation of displacement synthesis and terminal extension, giving rise to an extended molecule (5'-terminally extended monomeric replicative form [5'eRF]) (step 2) (3, 15, 46). Hairpin refolding at the extended terminus, supported by host cell nuclear factors (3, 13, 14, 47) and efficiently stimulated by NS1 (47), creates a so-called rabbit-ear structure (5'-rabbit-eared monomeric replicative form [5'reRF]) (step 3). This structure provides a primer for strand displacement synthesis and dimeric replicative-form (dRF) formation (step 4) (3, 5, 19, 46, 47). Subsequent NS1-dependent resolution of concatemer junctions results in the formation of two types of monomeric replicative-form (RF) DNA, with covalently closed (5'eRF) or extended (3'-5'eRF) left-hand termini, respectively (steps 5a and 5b) (17, 18, 20, 21). The 5'eRF molecule generated in this way reenters the cycle as in step 3, while duplex-to-hairpin transition at the right-hand palindrome of the 3'-5'eRF molecule (step 6) is thought to lead to the displacement of single-stranded genomic DNA, which is then immediately packaged into the preformed empty capsid (22).

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Modified rolling hairpin model for MVM DNA replication (according to reference 19). NS1 is depicted as a small filled circle. Small arrowheads indicate DNA 3' ends. ss, ssDNA; eRF, extended monomeric RF; v, viral strand; c, complementary strand. The open polygon in step 6 represents the capsid.

In order to identify MVM DNA replication steps sensitive to host cell transformation, we compared the temporal accumulation of viral DNA replicative intermediates in a pair of normal and transformed cells infected with MVM (prototype strain MVMp) and studied individual replication reactions in an in vitro system (3) using different MVM DNA templates and extracts from normal and transformed cells. A system of choice for studying the effect of transformation on MVMp DNA replication consists of pairs of normal and transformed human fibroblasts (10). Independently of the multiplicity of infection (MOI) used, normal human fibroblasts proved to be quite resistant to the parvovirus, undergoing an abortive infection characterized by a low level of viral DNA replication. In contrast, a number of transformed human fibroblasts show increased permissiveness to the parvovirus. Thus, the human lung fibroblasts MRC-5V1, transformed with simian virus 40 (SV40), sustain a 10-fold-higher level of MVMp DNA replication than MRC-5 cells, their untransformed progenitors (10). In keeping with this observation, production of RF intermediates clearly takes place in infected MRC-5V1 cells, while RF DNA is hardly detectable in normal cells (2, 10).

In order to investigate the differential capacities of MRC-5 and MRC-5V1 cells for supporting MVM DNA amplification, we performed time course analyses of the accumulation of MVM DNA replicative intermediates in this pair of cell lines after infection with MVMp. MRC-5 and MRC-5V1 cultures (2.5 × 10⁵ cells) were seeded in 35-mm-diameter petri dishes and inoculated with MVMp at an MOI of 5 PFU per cell. At various times postinfection (p.i.), intracellular parvoviral DNA was extracted by using the Hirt procedure, including proteinase K digestion (27). DNA replicative intermediates were size fractionated via 0.8% agarose gel electrophoresis, blotted, and hybridized with an MVM-specific ³²P-labeled DNA probe synthesized from a SalI-digested MVM DNA clone, p98 (1). ssDNA isolated from MVM virions by sodium dodecyl sulfate-proteinase K treatment and phenol extraction (45) was used as a size marker (Fig. 2A, lane 1). The marker DNA minor band corresponding to monomeric RF (mRF) can be assigned to the spontaneous reannealing of the small fraction of packaged plus strands with the major virion DNA species (which are of minus polarity) (4). Viral DNA extracted 2 h p.i. was mostly present in the unreplicated single-stranded form (Fig. 2A, lanes 2 and 6). The difference in the amounts of input ssDNA associated with MRC-5 and MRC-5V1 cells at this early time reflects experimental fluctuations, as indicated by the data from a series of experiments. This is in keeping with the similar competence of both cell types for MVMp uptake as reported earlier (10). Double-stranded RF DNA was also detected in minute amounts 2 h p.i. and is thought to arise from plus and minus input viral DNA strand reannealing, as mentioned above. The amplification of MVM DNA in infected cells, as measured by Southern blotting (Fig. 2A, lanes 2 to 9), was quantitated by densitometric scanning. RF values are plotted as a function of time p.i. in Fig. 2B after subtraction of the mRF background signal detected 2 h p.i. MRC-5 and MRC-5V1 cells could be distinguished by their differing capacities for replicating MVM DNA over the course of time. MRC-5 cells sustained the formation of a limited but significant amount of monomeric RF species up to 18 h p.i. but subsequently failed to further amplify this DNA (Fig. 2A, lanes 3 to 5, and Fig. 2B). In contrast, mRF formation in MRC-5V1 cells was followed by the amplification of this species, as revealed by the time-dependent increase in the corresponding signal (Fig. 2A, lanes 7 to 9, and Fig. 2B). In agreement with the involvement of multimeric DNA species in RF replication (19), mRF amplification in transformed MRC-5V1 cells was accompanied by the clear-cut appearance of dRF molecules (Fig. 2A, lanes 8 and 9) at a position at which a band was hardly detectable in extracts from infected normal cells (Fig. 2A, lanes 4 and 5). Given the overlap between the conversion of input ssDNA into mRF and the amplification of mRF in infected asynchronous cells, these experiments do not allow the assessment of a possible quantitative effect of cell transformation on the conversion step. However, the data clearly point to a qualitative difference between normal and transformed cells at the level of double-stranded RF DNA amplification. The additional band detected in both MRC-5 and MRC-5V1 cells, designated X (Fig. 2A), has been previously identified by others (43) as an RNA-DNA duplex whose biological significance remains unclear. MVM DNA replication has been reported to depend on the S phase of the cell cycle (19). Since MRC-5 and MRC-5V1 cells do not grow at the same rate, a difference in their distribution through the mitotic cycle may conceivably contribute to the observed variation in replication efficiency. To test this possibility, we conducted fluorescence-activated cell sorter analyses with exponentially growing MRC-5 and MRC-5V1 cells. These studies revealed that both cultures contained approximately the same fraction (25 to 30%) of S-phase cells (data not shown). This argues against accumulation in S phase being responsible for the greater capacity of MRC-5V1 cultures for MVM DNA amplification.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Southern blot analysis of in vivo-produced MVM DNA replicative intermediates. (A) DNA was extracted from MVM-infected MRC-5 (lanes 2 to 5) or MRC-5V1 (lanes 6 to 9) cells at 2, 18, 26, and 48 h p.i., fractionated by 0.8% agarose gel electrophoresis, transferred to a nitrocellulose membrane, and hybridized with an MVM-specific DNA probe. BamHI- and EcoRI-digested phage  DNA and MVM genomic DNA were used as molecular weight markers. DNA samples were matched for the number (105) of originally infected cells. Positions of the major MVM RF species are indicated (see the text for details). X, additional band detected in both MRC-5 and MRC-5V1 cells. (B) The accumulation of MVM mRF DNA as a function of time p.i. (see panel A) was quantitated by densitometric scanning and was plotted after subtraction of the background signal detected at 2 h p.i.

In order to understand the molecular mechanisms underlying the increased amplification of mRF DNA in transformed fibroblasts, MRC-5 and MRC-5V1 cells were then compared for their ability to support the individual steps of MVM DNA replication by an in vitro assay. DNA replication reactions were carried out as previously reported (3) with cytosolic extracts (10 to 100 µg of proteins) from MRC-5 and MRC-5V1 cells. About 200 ng of purified baculovirus-produced NS1 (32) was added where required. The reaction was started by addition of natural or cloned viral DNA template (20 or 100 ng, respectively) and performed for 1 h at 37°C. After the reaction was stopped, replication products were purified as described elsewhere (3) and analyzed either directly or after restriction digestion by 0.8% agarose or 5% polyacrylamide gel electrophoresis, as indicated. Natural MVM mRF DNA templates, i.e., cRF and 5'eRF (Fig. 1), were obtained from MVM-infected A9 cells by Hirt extraction, centrifugation through 5 to 30% neutral sucrose gradients, and further purification on 10 to 30% alkaline sucrose gradients (3, 39).

We have previously shown that the MVM RF DNA intermediate that is covalently closed at both termini (cRF DNA) is processed in vitro in a reaction involving NS1-specific nicking of the right (genomic 5') telomere, followed by initiation of displacement synthesis, copying of the right-end hairpin sequence, and formation of a terminally extended molecule (5'eRF) (3, 46) (Fig. 1, step 2). Using an MVM DNA preparation containing predominantly cRF DNA as a source of template, we tested the ability of MRC-5 and MRC-5V1 cell extracts to sustain resolution of the right-end telomere into an extended open structure in the presence of NS1. For the analysis of short terminal fragments, replication products were digested with PshAI and fractionated by native 5% polyacrylamide gel electrophoresis. PshAI cleaves MVM DNA at nucleotide 4916, giving rise to right-end fragments of 130 or 254 bp, respectively, depending on whether the telomere is in the hairpin or extended configuration (3, 46). While the hairpin fragment is only weakly labeled due to nonspecific repair synthesis, the extended species becomes heavily labeled as the result of the nicking and extension reactions leading to the copying of the terminal hairpin sequence. As illustrated in Fig. 3A (lanes 1 and 5), similar competences for sustaining the NS1-induced extension reaction were observed in the presence of small amounts of extract from either MRC-5 or MRC-5V1 cells. By increasing the concentration of MRC-5V1 cell extract, the efficiency of right-end resolution could be further enhanced (Fig. 3A, lanes 6 to 8), whereas the same increase in the amount of MRC-5 cell extract had no significant effect (Fig. 3A, lanes 2 to 4). This suggests that transformation is associated with the production or activation of a cellular factor(s) that stimulates nicking and/or extension of the 5'-terminal hairpin. As a consequence, MRC-5 cells showed a relative, but not absolute, impairment in this reaction, which may contribute to the lack of MVM DNA amplification found in normal cells under in vivo conditions but appears insufficient to account for this defect alone.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   In vitro processing of MVM RF DNA. (A) NS1-dependent processing of cRF in normal and transformed cells. cRF DNA was incubated with NS1 and increasing amounts of extracts from MRC-5 (lanes 1 to 4) or MRC-5V1 (lanes 5 to 8) cells. Reaction products were digested with PshAI and analyzed by 5% polyacrylamide gel electrophoresis. EcoRI-digested pBR322 DNA was used as a molecular weight marker. 5'H, 5' hairpin fragment; 5'E, 5' extended terminal fragment. (B) Synthesis of MVM dRF DNA from 5'eRF template DNA (mRF). 5'eRF DNA was incubated with increasing amounts of extracts from MRC-5 (lanes 1 to 5) or MRC-5V1 (lanes 6 to 10) cells. Reaction products were separated by 0.8% agarose gel electrophoresis. EcoRI-digested phage  DNA was used as a molecular weight marker.

The 5'eRF product resulting from the right-end resolution of cRF can be further processed in vitro, leading to the formation of dRF DNA in an NS1-stimulated reaction (3) (Fig. 1, steps 3 and 4). To compare normal and transformed cells for their ability to initiate replication on 5'-terminally extended DNA, in vitro replication reactions were conducted in the presence of NS1 by using 5'eRF template DNA purified from MVM-infected mouse fibroblasts. Figure 3B shows the analysis of the undigested replication products by 0.8% agarose gel electrophoresis. When incubated with extracts from human fibroblasts, the 5'eRF monomer template (mRF band) was specifically labeled, presumably as a result of re-nicking and extension events taking place at the RF right end (3, 5, 46). In addition, the formation of a distinct DNA species was observed at a position corresponding to an apparent molecular size of about 10 kbp. The identity of this species with MVM dRF DNA has been demonstrated in previous experiments (3). As is apparent from Fig. 3B, extracts from both normal and transformed fibroblasts were able to drive the formation of dRF molecules from 5'eRF DNA in the presence of NS1. Extracts from the transformants were two- to threefold more efficient in this respect than extracts from normal cells. However, as this impairment is only moderate under in vitro conditions, limitations to RF multimerization may contribute to, but cannot alone account for, the lack of viral DNA amplification in the normal cells.

Most dimer duplex intermediates formed in MVM-infected cells contain monomeric subunits connected in a left-to-left-end (viral 3'-to-3'-end) manner, while a small minority (about 5%) contain a right-to-right-end (viral 5'-to-5'-end) junction (22). These palindromic junction regions of MVM dRF are resolved to generate monomeric forms in the presence of NS1 both in vivo and in vitro (17, 18, 20, 21, 28). The MVM terminal palindromes are not perfectly symmetrical; each consists of two arms with slightly different sequences designated A and B (18). Resolution of the 3'-to-3' junction is asymmetrical and involves an initial nick in the B arm, giving rise to resolved A and B arms that predominantly terminate in the extended or turnaround configuration, respectively (18, 21). In contrast, 5'-to-5' junction fragments are resolved symmetrically and give rise to predominantly extended-form structures from both arms of the palindrome (17). We examined the resolution of MVM DNA concatemeric junctions (Fig. 1, step 5) in extracts from MRC-5 and MRC-5V1 cells. As templates, we used the pUC18-based plasmid pLEB711, containing the palindromic 711-bp PstI fragment derived from the left-to-left-end junction of dRF DNA (20), and the pUC19-based plasmid pREB1412, spanning the right-to-right-end bridge (20). Resolution reactions were carried out for 1 h in the presence of cytosolic extracts. Reaction products were digested with ScaI, which cuts once within the vector sequence and generates two fragments of differing sizes from resolved molecules. Digestion products were fractionated by 0.8% agarose gel electrophoresis.

As shown in Fig. 4A (lanes 1 to 4), bands corresponding to the linearized forms of unresolved pLEB711 and pREB1412 plasmids were obtained, migrating at the anticipated 3,430- and 4,340-bp positions, respectively. From the results of immunoprecipitation reactions, Cotmore et al. (17, 18) concluded that these forms consist in part of NS1-bound superstructures arising from incomplete resolution (i.e., NS1-mediated nicking and initiation of rolling-circle replication) and in part of unresolved NS1-free molecules probably labeled by repair synthesis. This is in line with control reactions carried out in the absence of NS1 in the present study (data not shown). In the presence of NS1, linear forms of pLEB711 became labeled to similar extents in extracts from MRC-5V1 and MRC-5 cells (Fig. 4, lanes 1 and 2). In contrast, the labeling of linearized pREB1412 was much lower in extracts from MRC-5 than in extracts from MRC-5V1 cells (Fig. 4, lanes 3 and 4). We interpret this result in favor of the involvement of transformation-dependent cellular factors in NS1-induced resolution of the right-to-right-end junction, as supported by the analysis of the junction resolution products shown below. The band migrating between the 3'-to-3' linear form and the 3'-to-3' B replication product (Fig. 4A, lanes 1 and 2) results from nonspecific labeling of the pUC18 vector DNA.

View larger version (44K): [in this window] [in a new window]   FIG. 4.   Resolution of 3'-to-3' and 5'-to-5' concatemer junctions in extracts from normal and transformed cells. Plasmids pLEB711 (lanes 1 and 2) and pREB1412 (lanes 3 and 4), containing the MVM 3'-to-3' or 5'-to-5' junction fragments, respectively, were incubated with NS1 in extracts from MRC-5 (lanes 2 and 4) or MRC-5V1 (lanes 1 and 3) cells. (A) Reaction products were digested with ScaI prior to fractionation on a 0.8% agarose gel. BamHI- and EcoRI-digested phage  DNA was used as a molecular weight marker. (B) Resolution products were digested with PstI (pLEB711; lanes 1 and 2) or XbaI (pREB1412; lanes 3 and 4) and analyzed on a 5% polyacrylamide gel. For comparison, cRF template DNA was incubated in the presence of NS1 with the same MRC-5 (lane 6) or MRC-5V1 (lane 5) cell extracts, digested with PshAI, and analyzed on the same gel. EcoRI-digested pBR322 DNA was used as a molecular weight marker. Only the major bands of interest are labeled.

Specific resolution products corresponding to the A and B arms of the palindromes were detected for both pLEB711 and pREB1412 clones when reactions were performed with extracts from MRC-5V1 cells (Fig. 4A, lanes 1 and 3). Doublet bands were visible at the positions of the A and B arms of resolved pREB1412 (Fig. 4, lane 3), in agreement with the above-mentioned occurrence of each arm in both turnaround and extended configurations. In contrast, no resolved fragments could be detected when reactions were performed in the presence of MRC-5 cell extracts (Fig. 4A, lanes 2 and 4).

To ascertain that the failure of MRC-5 cell extracts to support concatemer resolution was specific for this reaction, we reexamined the capacity of the same MRC-5 extracts for supporting the nicking and extension reaction described above for cRF. DNA products from replication reactions carried out in the presence of cRF template and either MRC-5V1 or MRC-5 cell extracts were digested with PshAI and electrophoresed through a 5% polyacrylamide gel. As illustrated in Fig. 4B (lanes 5 and 6), this analysis confirmed the ability of both extracts to support nicking and extension of the cRF right-hand telomere. In contrast, resolution of 3'-to-3' or 5'-to-5' concatemeric junctions was supported only by extracts from transformed cells, as revealed by the analysis of PstI- and XbaI-digested resolution products in the same gel (Fig. 4B, lanes 1 to 4). This comparison suggests that the concatemer resolution reaction plays a key role in restricting the efficiency of parvovirus DNA replication in normal human fibroblasts.

In order to gain information as to whether the differential capacity of MRC-5 and MRC-5V1 cells for supporting concatemer resolution was due to an activator(s) present in transformed cells or an inhibitor(s) present in normal cells, we performed in vitro replication reactions in the presence of various combinations of cell extracts. As shown in Fig. 5 (lanes 2 to 5), the efficiency of the 3'-to-3' junction resolution in MRC-5V1 cell extracts was not affected by the addition of increasing amounts of MRC-5 cell extracts. In contrast, increasing the amount of MRC-5V1 cell proteins led to a dose-dependent stimulation of the resolution reaction in the absence (Fig. 5, lanes 6 to 9) or presence (data not shown) of MRC-5 cell extracts. These results argue for the presence of a factor(s) able to activate 3'-to-3' concatemer resolution in transformed cells and against the presence of a resolution inhibitor in normal cells.

View larger version (63K): [in this window] [in a new window]   FIG. 5.   Resolution of the 3'-to-3' junction in mixtures of extracts from MRC-5 and MRC-5V1 cells. Plasmid pLEB711 was incubated with NS1 and indicated amounts of proteins from either MRC-5V1 cell extract alone (lane 1 and lanes 6 to 9) or a mixture of MRC-5V1 and MRC-5 cell extracts (lanes 2 to 5). Bovine serum albumin was included to adjust the final protein amount to 100 µg. Reaction products were digested with ScaI and analyzed by 0.8% agarose gel electrophoresis. BamHI- and EcoRI-digested phage  DNA was used as a molecular weight marker.

MVM DNA replication is known to depend upon the entry of host cells into S phase. As mentioned above, MRC-5 and MRC-5V1 cultures comprise similar fractions of S-phase cells; hence, the distinctive capacity of the latter for in vitro concatemer resolution cannot be assigned to their enrichment in S-phase cells. Furthermore, extracts prepared from MRC-5 cells that were highly (99%) synchronized in S phase still proved to be unable to resolve concatemeric junctions in vitro (data not shown), again arguing against a major contribution of cell cycle-related variations to the differential capacity of normal and transformed cells for sustaining this reaction. It should also be stated that MRC-5V1 cells are transformed by SV40 and express the viral large T antigen (40) (data not shown) which is essential for the initiation of SV40 DNA replication (38) and might therefore potentially facilitate the resolution of MVM DNA concatemers. However, the human fibroblast cell line KMST-6, which was transformed by gamma-irradiation, was found to efficiently amplify MVM DNA (10), and extracts from these cells proved to be as active as extracts from MRC-5V1 cells in sustaining the 3'-to-3' junction resolution in the present assay (data not shown). This shows that the stimulation of MVM DNA amplification is not restricted to cells expressing the SV40 large T antigen, arguing for the involvement of transformation-dependent cellular factors in this modulation.

In this study, we provide evidence that the transformation dependency of MVM DNA amplification results from the stimulated replication of double-stranded intermediates. In particular, the resolution of MVM DNA concatemer junctions is strongly enhanced in transformed versus normal cells. This holds true for the resolution of both 3'-to-3' and 5'-to-5' junctions. Interestingly, normal cells appeared to be deficient in the first stage of 5'-to-5' resolution, presumably including NS1-mediated nicking and initiation of strand displacement synthesis (17, 18). In contrast, MRC-5 cells were competent for this initial stage of the 3'-to-3' resolution, although they were unable to complete the process. This indicates that the 3'-to-3' and 5'-to-5' junctions differ, at least in part, in their requirements for cellular resolution factors and that the full resolution of both types of junctions is subject to a transformation-sensitive restriction. Furthermore, our data suggest that the failure of normal cells to support dRF internal resolution can be traced back to the absence of a transformation-specific activator(s). Given the failure of dimer resolution in extracts from untransformed cells, one would predict an accumulation of dRF intermediates in MVM-infected MRC-5 cells. This cannot be assessed from Fig. 2, due to the low RF yields achieved with these cells. However, quantitation of the RF signals present on the original autoradiograph showed that the dRF/mRF ratio was indeed markedly higher for infected MRC-5 cells than for MRC-5V1 cells. This is in keeping with the hypothesis that MVM DNA replication becomes blocked at the concatemeric resolution stage during infection of normal MRC-5 cells. It is worth noting that the original host cells for MVM replication are of mouse origin, while human fibroblasts were used in the present study. Therefore, the possibility that at least a partly different limitation(s) to MVM DNA amplification takes place in the mouse system needs to be considered, although no qualitative difference could be observed so far between mouse and human cell extracts in their capacity for MVM DNA replication in vitro (3, 18). Furthermore, transformation of mouse fibroblasts, as induced in particular by SV40, was also shown to correlate with an increase in their ability to amplify MVM DNA (10).

Little is known about the cellular factors required for MVM DNA replication. DNA polymerase delta (pol-) appears to be involved (8, 13), implying that the auxiliary factors replication protein A (RPA), replication factor C, and proliferating cell nuclear antigen (PCNA) are also likely to take part in parvovirus DNA replication. RPA and PCNA have indeed been shown to participate in the in vitro resolution of both 3'-to-3' and 5'-to-5' concatemers (8). In the case of 3'-to-3' concatemer resolution, a host cell component designated parvovirus initiation factor (8, 9) was shown to activate the endonuclease function of NS1 by binding to a consensus motif specific for the activated transcription factor (8, 9, 28). Furthermore, MVM replication factors were found to bind to cis-regulatory elements located inboard of the MVM 5' palindrome (5, 42), and interaction of the MVM 5' hairpin with members of the high-mobility-group (HMG) proteins has recently been reported (23). HMG protein was actually shown to be the only cellular protein required to allow the viral initiator protein NS1 to nick the right-hand (5') MVM hairpin in a sequence-specific fashion.

Neoplastic transformation is characterized by deregulation of cell growth and division, two complex processes that are controlled by multiple factors in normal cells (32). Certain cell regulatory pathways have been shown to be particularly altered in a majority of cancers (32), in association with changes in the levels of various cellular proteins, including DNA replication factors. The level of PCNA was found to be 10-fold higher in MRC-5V1 cells than in MRC-5 cells (data not shown) (6, 25). Conversely, we detected higher quantities of the cyclin-dependent kinase inhibitor p21^WAF1/CIP1 (26) in MRC-5 cells than in MRC-5V1 cells (data not shown). p21^WAF1/CIP1 has been shown to interact specifically with PCNA, leading to a block in pol--dependent DNA synthesis (31, 40, 44). Furthermore, quaternary complexes between p21^WAF1/CIP1, cyclins, cyclin-dependent kinases, and PCNA are disrupted in transformed cells (30, 49), resulting in the release of p21^WAF1/CIP1-free PCNA that is able to activate pol-. Thus, the increase in PCNA levels and the decrease in p21^WAF1/CIP1 levels, both of which are associated with the MRC-5 cell's transformation, deserve to be considered as possible contributors to the overall enhancement of MVM DNA amplification. However, the observed quantitative differences in these factors are unlikely by themselves to account for the inability of MRC-5 cell extracts to support detectable resolution of MVM DNA concatemers. This leads us to postulate the involvement of a transformation-specific factor(s), the nature of which remains to be established, in the resolution of MVM concatemeric intermediates. The above-mentioned HMG protein, which participates in the resolution of the MVM DNA right end (23), is known to be differentially expressed as a function of neoplastic transformation (48). This raises the intriguing possibility that the difference observed between MRC-5 and MRC-5V1 cells in efficiency in resolving cRF right-hand and tail-to-tail concatemers may be traced back to their different HMG protein contents.