JOURNAL OF VIROLOGY, Mar. 1990, p. 1304-1313 0022-538X/90/031304-10$02.00/0 Copyright C) 1990, American Society for Microbiology

Vol. 64, No. 3

Low Probability of Double Integration in Transformation of Nonpermissive Cells by Polyomavirus SAW YIN OH,t ANDREA AMALFITANO, KAREN FRIDERICI,t MING CHU CHEN, AND

MICHELE M. FLUCK*

Department of Microbiology, Interdepartmental Program in Molecular Biology, Michigan State University, East Lansing, Michigan 48824-1101 Received 24 July 1989/Accepted 22 November 1989

The fate of polyomavirus genomes in stable transformants was examined in experiments in which rat or hamster cells were infected with mixed viral populations containing either two distinguishable wild types or a wild type plus a transformation-deficient mutant. The results demonstrate that in 90% of the cases in either situation, a single polyomavirus parental genome became integrated into the host genome. Both parents in the mixed infection were present in the same cells and both persisted in the infected cells until transformants appeared, eliminating the possibility that one virus would exclude the other in the early steps of the infection process. We believe that these results can be generalized to transformation events derived from normal single infections. Thus, contrary to previous results for the number of integration sites determined by restriction endonuclease and blot hybridization analysis, it appears that the probability for more than one integration event in transformation by polyomavirus is low (1 in 10). A reinterpretation of previous data is proposed.

The patterns of viral integration in cells neoplastically transformed by papovaviruses, polyomavirus, and simian virus 40 have been extensively studied. Blot hybridization of cellular DNA from transformants has indicated that the viral genome is apparently inserted at multiple sites in the host chromosome and that each site may contain tandem copies of the genome (4, 5). The variations in pattern from transformant to transformant imply great diversity in chromosomal integration sites. A review of the published literature concerning the number of integration sites for the polyomavirus genome in rat transformants suggests that there are multiple sites in 80% of the cases reported, with the number of sites per transformant averaging over three (1, 4, 9, 15, 20, 30). In experiments with gels which have more resolving power, this number appears to be even higher (i.e., 8 to 10) (18; K. Friderici and M. Fluck, unpublished data; see below). Thus, it has generally been assumed that multiple viral integrations occur in each transformation event and that the number of chromosomal sites available for integration is very high, if not unlimited. The simplest model based on these results would assume (i) that each of the integration events in any transformant represents a random interaction between available sites on the host chromosome and available members of the pool of integratable viral genomes and (ii) that each event happens independently of the events occurring at other sites. Assuming that all genomes present in the parental infection mix have an equal probability of becoming integrated, we expected the integrated genome population to be a faithful representation of the viral genomes present in the cell at the time of integration. To test this prediction, we have carried out mixed infections of nonpermissive rat and hamster cells

with parental genomes whose integration can be followed by restriction endonuclease analysis. We chose wild-type and viable-deletion mutants which are transformation defective (hr-t mutants) as well as pairs of distinguishable wild types, and we used various total multiplicities and ratios of the two parents. We isolated transformants from such infections and studied the integration patterns of the two parental genomes. In the first case, it was expected that in addition to the selected transforming wild-type parent, the deletion mutant would be recovered in the transformants at a frequency and dosage which reflect the ratio of the two genomes in the infection. Similarly, in the second case, it was expected that both wild-type genomes would be integrated. The results of these experiments are described in the present report. The results failed to support the prediction of the model and consequently have generated an alternate model. MATERIALS AND METHODS Cells and viruses. Fischer F-111 (8, 11) and FR-3T3 (26) rat cells and baby hamster kidney (BHK) cells (29) were grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% newborn calf serum. Strains A2 and A3 are widely used wild-type strains which were isolated from the same parental stock and which have been sequenced (16, 24). A3 differs from A2 by a deletion of 11 base pairs, which eliminates one of the large-T-antigenbinding sites. The pseudo-wild-type strain NG59RA (RA) was obtained by marker rescue (7) of the middle-T-antigen defect of hr-t mutant NG59 (3). Differences between RA and A2 in the enhancer-origin region (24), as well as a largeT-antigen mutation in RA (19), have been noted previously. RA and A2 also differ in plaque morphology: the former is a small-plaque virus with a heat-stable hemagglutination pattern (6), while the latter is a large-plaque virus. Transformation-defective hr-t mutants B2, 30'B, and 3A3 were derived from the Pasadena small-plaque strain by mutagenesis with ICR-191 (28). All hr-t mutants used contained rearranged enhancer regions (7; K. Higgins, K. Friderici, A. Amalfi-

* Corresponding author. t Present address: School of Biological Sciences, Flinders University, Bedford Park, South Australia 5042. t Present address: Department of Pathology, Michigan State University, East Lansing, MI 48824.

1304

INTEGRATION OF UNSELECTED POLYOMAVIRUS GENOMES

VOL. 64, 1990

1305

TABLE 1. Frequency of double integration in mixed infections of rat and hamster cells' Viral strain Mutan and no. expt and expt no.

Wild type plus transformation defective 1 2a 2b 3 4a 4b Sa Sb Sc 6 7

8 9 Two transformationcompetent strains 10 11 12 13

WT Parent" WT1 ~~~ ~~~WT

Cell lntegrations`bl Mold Host' ~~~~~~~~~divisionsf Dbl WT2Muat

F-111 F-111

2-3' 2-3'

0/10 0/4 0/3

ND 400 1,700

RA RA

S/5 1/1 5/1 20/0.1 20/1

F-111 F-111

8',j

RA

10/10

F-111

1/14 2/6 1/6 2/12 4/12 1/6 0/12 1/5 0/5 1/13

200 ND ND ND ND

B2 B2

RA RA

B2 B2 B2

Pk 4-5'-i 0'

6-7'' B2 30'b 3A3 B2

% WT transformationh

RA(eB2) A2 A2 RA

dl45 dl45 A2(eRA) A2(eB2)

A3 A3 A2 A2

2/2 20/4 5/1 20/1 5/5

F-111 F-111 F-111 BHK

10/10 10/10 10/10 10/10

F-111 FR-3T3 FR-3T3 FR-3T3

6-7' 2-3' 2-3'

2-3' 2-3'

2-3' om om

o0"

150 30 18 ND

2/12 4/22 0/10 0/13

a Summary of the transformants analyzed in this report. Rat or hamster cells were coinfected with the parental strains indicated, as described in Materials and Methods. Experiments 1 through 9 were carried out with a transforming parent plus a nontransforming deletion parent (td transformants). Experiments 10 through 13 were carried out with two wild-type strains (tf transformants). b Deletion mutants (see Materials and Methods) can be distinguished from the wild type by the presence of deletions in their genomes. B2, 30'b, and 3A3 are transformation-defective hr-t mutants, which are not selected in the transformation process. ' WT, Wild type; source of the transforming parents RA, RA(eB2), and A2. WT1 and WT2, Wild-type or mutant (d145) parents which are both transforming (see Materials and Methods). d MOI, Ratios of the multiplicities of infection of nontransforming to transforming mutants and of the two wild types which were used to obtain the transformants. e F-111 and FR-3T3, Fischer rat cell lines; BHK, hamster cell line. f Number of cell divisions allowed postinfection until infected cells formed a confluent monolayer. This number depended on the cell density at time of infection (see Materials and Methods). " Mutant, Summary of the number of transformants containing the unselected viral genome over the number of transformants analyzed as determined by Southern blotting analysis. Double, Summary of the number of transformants containing both parental genomes over the number of transformants analyzed. h Ratio (in percent) of transformation frequency in the mixed infection over transformation frequency in the infection with the transforming parent alone. ND, Not done. i2 x 105 cells per 60-mm dish. Cells were passed to 10 plates within 24 h of infection. k5 x 105 cells per 60-mm dish. ' 1 x 106 cells per 60-mm dish. m Cells were passed to agar immediately after infection.

tano, M. C. Chen, and M. M. Fluck, unpublished data) as well as a deletion in the middle-T-antigen gene located in MspI-HpaII fragment 4 (3, 7, 16). This deletion is also easily detectable in PvuII digestions. RA(eB2) was constructed by ligating the small BglI-BamHI fragment encompassing the enhancer-origin region of hr-t mutant B2 to the large genomic Bgll-BamHI fragment of pseudo-wild-type RA. In wild-type A2(eRA) and A2(eB2), the small BgII-BamHI and BglI-BclI fragments, respectively, were ligated to the large BglI-BamHI and BglI-BclI genomic fragments, respectively, of wild-type A2. Both A2(eRA) and A2(eB2) contain a duplication in the enhancer (transferred from the parental RA and B2 mutants) which can be detected by an increase in size of MspI-HpaII fragment 3. Mutant d145, derived from wild-type A3 (2), has a deletion of 66 nucleotides in the overlapping large-T-antigen and middle-T-antigen reading frames. The deletion can be visualized by digestion with SstI. This mutant is reported to have no transformation or growth defects (2). Infections. F-111, FR-3T3, or BHK cells were seeded at a density of between 105 and 1.5 x 106 cells per 60-mm plate (see below; Table 1) and infected at the multiplicities and

ratios of mutant to wild-type virus indicated below and in Table 1. In most instances, the relative ratio of the two parental genomes in the infection mix, established from the titer of the two parental viral stocks, was confirmed by hybridization analysis of the viral sequences in the input. Cells were fed with DMEM supplemented with 5% newborn calf serum and became confluent within one to eight generations postinfection (Table 1). In experiments 4b and 5b (Table 1), each infected 60-mm dish was passed to 10 100-mm dishes (leading to approximate cell confluencies of 1.5 and 3%, respectively) and fed with DMEM supplemented with 10% serum until confluent. In most cases, transformants from infections carried out under the various conditions described were isolated as foci overgrowing a cell monolayer. In some cases, transformants were isolated directly (Table 1, experiments 11 through 13) or recloned by growth in agar. DNA analysis. Total cellular DNA was isolated from infected or transformed cells as described previously (12, 21). For gel electrophoresis, each sample containing 10 to 20 ,ug of DNA was digested with the restriction enzymes indicated below and in the figure legends. The endonucleases

1306

J. VIROL.

OH ET AL. M

1

2

3

4

5 6 7 8

9 10

D

23A

M

1 2 3 4

5 6

-

4m

w

8

9 10

-W ;..

9.5-

5.3-

...:.,

. 40 'A-1

fsl.,.

.o

b.-v .iw

c ar

..;.

r iK-.. .g ..4zfs,

....

I

i

...

FIG. 1. Integration patterns of td transformants derived from mixed infections of F-111 rat cells. F-111 rat cells (2 x 105/60-mm dish) were infected at a multiplicity of 5 PFU each of transforming strains RA and transformation-defective deletion mutant B2 (Table 1, experiment 1). Independent transformed foci were selected and cloned by anchorage independence in soft agar. A 10- to 20->Lg portion of highmolecular-weight DNA isolated from these clones was digested with BglII (which does not cut the viral genome [A]) or EcoRI (which cuts the genome once [B]) and was fractionated on 0.6% agarose gels. Hybridizations were as described in Materials and Methods, with a probe (pPy-1) (106 cpm/ml) representing the whole genome. M, Position of marker wild-type viral DNA digested with EcoRI; SC, position of supercoiled form I unintegrated viral DNA. Lanes 1 through 10 contain DNA from individual independent transformants.

chosen for their ability to generate fragments that distinguish between the two parental genomes. MspI-HpaII or PvuII was chosen for the analyses of all infections with hr-t mutants (Table 1, experiments 1 through 9), SstI was chosen to distinguish between A3 and d145 (Table 1, experiments 10 and 11), and MspI was chosen to distinguish between parents with differences in the enhancer regions (Table 1, experiments 12 and 13). After being transferred to nitrocellulose, (21, 27), the blots were hybridized to polyomavirus probes (10"9 cpm/lpg) synthesized by nick translation or random priming. Hybridization was carried out at 65°C for 3 days in 2x SSC (lx SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-lx Denhardt solution (0.1 mlIcm2 with 1 x 106 to 2 x 106 cpm of the labeled probe per milliliter of hybridization solution). In many of the hybridizations, a probe consisting of a specific subfragment of the genome was used, e.g., HpaII fragment 4 (designated pPyH4) or HpaII fragment 3 (pPyH3). were

RESULTS Assay for integration of unselected viral genomes. To analyze the frequency of integration of viral genomes, we first established transformants (td transformants) from nonpermissive cells infected with mixtures of transforming and nontransforming viral strains containing deletions in their genomes (hr-t mutants) (see Materials and Methods). Such td transformants are expected to contain an integrated transforming parental genome which was selected for in the transformation process. Integration of the unselected transformation-defective deletion genome present in the infection mix was monitored by analysis with restriction endonucleases which distinguish between the undeleted and the deleted parental genomes (see Materials and Methods). A total of 108 td transformants were analyzed (Table 1). These were derived from nine different infections involving three

different transforming strains and three transformation-defective deletion mutants. Integration patterns in td transformants. The viral integration patterns of td transformants were analyzed by restriction endonuclease digestion, electrophoresis, Southern blotting, and hybridization to viral probes. The patterns obtained for transformants derived from infections of Fischer F-111 rat cells, with transforming strain RA and hr-t mutant B2 used at a multiplicity of infection of 5 each, are shown in Fig. 1 (Table 1, experiment 1). In 8 of 10 transformants, a typical integration pattern was observed: (i) multiple, apparently independent sites of integration, as evidenced by multiple bands in digestions with BglII, which does not cut the viral genome (for example, the transformant analyzed in lane 2 of Fig. 1A shows seven bands, including an unresolved band at limit entry which may include more than a single band); (ii) free viral genomes, as evidenced by a band comigrating with supercoiled viral DNA (Fig. 1A); and (iii) head-to-tail tandem repeats of the viral genome at most sites, as revealed by a genome size 5.3-kilobase-pair band in digestions with EcoRI, which cuts the polyomavirus genome at a single site (Fig. 1B). Only 2 transformants in 10 showed a single site of integration containing an incomplete viral genome (Fig. 2) and no free viral DNA (Fig. 1, lanes 1 and 8). Similar normal integration patterns were observed with a second set of 63 td transformants derived from independent mixed infections with RA and B2 (Table 1, experiments 2 through 5), with 12 td transformants derived from infections with B2 and RA(eB2), a transforming strain with an RA coding region and B2 enhancer region (see Materials and Methods and Table 1, experiment 6) and with 10 td transformants derived from coinfections of hr-t mutants 30'b or 3A3 with wild-type A2 (Table 1, experiments 7 and 8). In summary, 95% of the Fischer rat cell transformants analyzed contained evidence for multiple integration sites when analyzed by restriction

INTEGRATION OF UNSELECTED POLYOMAVIRUS GENOMES

VOL. 64, 1990 NK

1

2

HI

614

43

H 2

61

*

3

4

5

6

7

4.

9

A

10

'*r

2

43

3

t'.

61 !>,

H 4

B

2

3*_

0

H 3

8

1307

*

09 4

> >

_

_

_ 9

M.3..... H 5/6

H 7

FIG. 2. Low frequency of integration of the unselected parent integrated viral sequences in rat transformants. The F-111 rat transformants described in the legend to Fig. 1 were further analyzed for the presence of the unselected parental genome. A 10- to 20-,ug portion of DNA was digested with MspI and fractionated on 2% agarose gels. Blotting and hybridizations were as described in Materials and Methods. H, Positions of wild-type A2 fragments 1 through 7. The deleted fragment 4 of mutant B2 is marked with an among

arrow.

endonucleases which do not cut the viral genome. Each independent integration event should give rise to two junction fragments by EcoRI digestion. One unexpected finding is that the apparent number of junction fragments was low (compare, for example, lanes 2 in Fig. 1A and B: 14 bands might be expected in Fig. 1B). We address this apparent paradox in the discussion. Low frequency of integration of the unselected parent. The td transformants were further analyzed for the presence of the two parental genomes. Each parent can be distinguished by the presence of a normal or a deleted restriction endonuclease fragment. Figure 2 shows the results of MspI digestion of the td transformants shown in Fig. 1. Note that as expected, all transformants contained all fragments of the viral genome except transformants 1 and 8, which lacked the capsid-encoding fragment 1. All transformants contained a normal-size fragment 4 corresponding to the presence of the transforming parent. This too was expected, since fragment 4 is in the coding region for the middle T antigen and is required for transformation. In contrast, none of the transformants contained the deleted fragment 4 corresponding to the B2 parent. This was also true when the MspI-HpaII fragment 4 (within which the B2 deletion [nucleotides 491 to 730] is located [6]) was used as a probe (results not shown). Since 8 of the 10 transformants showed evidence of multiple integration events (Fig. 1A), this result demonstrates that no mutant is integrated at any of these sites. Similar underrepresentation of the B2 mutant was observed in td transformants from other experiments in which the RA and B2 parental genomes were used in equal ratios while total multiplicity of infection varied (Table 1): the B2 mutant was recovered in 0/4 transformants from experiment 2a and 2/12, 4/12, and 1/6 from experiments 5a, b, and c, respectively. To increase the probability of mutant recovery, the ratio of the unselected to the selected parent was varied in the

FIG. 3. Ratio of wild-type and mutant genomes. F-111 rat cells (5 x 105/60-mm dish) were infected with a mixture of pseudo-wild-type RA and hr-t mutant B2 (at multiplicities of 1 and 20, respectively) (Table 1, experiment 4). Transformants were obtained as described in the legend to Fig. 1. Total DNA was extracted, digestion was performed with PvuII, and hybridization was done with pPyH4 containing MspI fragment 4 (in which the mutant deletion is located). The 1.2-kilobase wild-type fragment and the 0.94-kilobase mutant fragment are shown. (A) Viral DNA isolated 3 days postinfection. Lane 1, Confluent cells (Table 1, experiment 4a); lane 2, cells from lane 1 passed 1:10 at 10 h postinfection and allowed to grow to confluency. (B) DNA from the three mixed transformants (Table 1, experiment 4) which contain integrated wild-type and mutant genomes.

infection mix. Figure 3A illustrates that the ratio of mutant to wild-type genome in the infected cells in experiment 4 (Table 1) was indeed as high as expected for the 20/1 ratio established in the infection mix from the titer of the viral stocks. The DNA was extracted 3 days postinfection (a time at which there had been little or no replication of the input genomes [17]), digested with PvuII, and hybridized with the pPyH4 probe, which spans the region deleted in mutant B2 (6). The 0.94-kilobase deleted PvuII fragment is clearly in high excess over the wild-type 1.2-kilobase fragment. The high ratio of mutant to wild-type genome did not alter the results: only 1 of 14 transformants analyzed from experiment 3 performed at a 200/1 mutant/wild type ratio contained the unselected parent; 0 of 3 in experiment 2b performed at a 5/1 ratio; and 3 of 12 from experiment 4 performed at a 20/1 ratio (Table 1). The recovery of the unselected parent seemed to be independent of the total multiplicity or ratio of infection, which ranged from 1/1 (Table 1, experiment 2a) to 10/10 (experiments 5a and b) and 20/1 (experiments 3 and 4). The low probability of recovery of the two parental genomes was also demonstrated in transformants derived from mixed infections with the transformation-defective MOP1033 and transformation-competent ts3 strains. In these, only 9 of 69 transformants analyzed contained both parental genomes (18). Underrepresentation of mutant sequences in td transformants. In the few instances in which the unselected parent was integrated in transformed cells, the parent was present at a level lower than its level in the infection. Figure 3A illustrates the high ratio of nontransforming to transforming parents in the infected cells at early times postinfection in experiment 4 (Table 1). In contrast, the 3 of 12 transformants analyzed which do carry the nontransforming mutant showed ratios of deleted to nondeleted sequences which varied from approximately 1/1 (Fig. 3B, lane 1) to approxi-

1308

OH ET AL.

J. VIROL. 1

2 34

5

6 7

1

2 3 4 5 6 7 8 9

WT

B2 dpi

*

152.

**-

WT *r

* *-

B2

10 15

FIG. 4. Persistence of the viral genome in the infected-cell population. Confluent F-111 rat cells were infected as described in Materials and Methods (Table 1, experiment 5), with transforming parent RA and nontransforming parent B2 at multiplicities of 10 for each parent. One-half of the plates were maintained at confluency (lanes 1 to 3), while the other half (lanes 4 to 7) were diluted 1:30 to allow the cells to divide. Total cellular DNA was isolated at the times indicated (dpi, days postinfection), digested with PvuII, electrophoresed, and analyzed by Southern blotting and hybridization as described in Materials and Methods and in the legends to Fig. 1 to 3. Hybridization was carried out with pPyH4, which contains MspI fragment 4, in which the hr-t deletion is located. The positions of the RA and B2 PvuII fragment 2 are designated WT and B2.

mately 1/10 (Fig. 3B, lane 3). Similar results were obtained in the other experiments involving high ratios of mutant to wild type (Table 1A, experiments 3 and 4). The presence of free viral DNA in those transformants does not alter these conclusions, since the unintegrated DNA is produced by in situ replication from the integrated sequence (15) and the composition of unintegrated copies is a faithful copy of the integrated sequences (15, 18). Loss of deletion mutant is not due to underreplication. The persistence of the two viral genomes in infected cells was compared from early times after infection until the appearance of transformed foci in the monolayer. For this purpose, F-111 cells were infected at equal multiplicities (10/10) of the parents RA and B2 to allow the detection of both parents throughout the course of infection (Table 1, experiment 5). This experiment was carried out at two cell densities: one-half of the plates were maintained at confluency, while the other half were diluted 1:30 following infection to allow cells to divide. Total DNA was isolated from infected cells at various times and analyzed by Southern blotting (Fig. 4). It is clear that the signal from the deletion mutant was retained in the population of infected cells throughout the course of the experiment, up to the time of selection of transformed foci (15 days). As is evident from the infection of confluent cells, the ratio of wild-type and mutant viruses was not altered during the course of infection. In the population of dividing cells, the signals of both wild-type and mutant DNA decreased relative to constant amounts of total cellular DNA. We have recently demonstrated that a low level of viral DNA synthesis occurs in F-111 cells maintained at 37°C (17). Clearly, the amount of replication does not ensure efficient synthesis and segregation of the viral genomes in daughter cells, so that under conditions of low cell density, the signal was lost after a few generations. However, there was no preferential loss of the deletion parent under these conditions. If anything, the B2 genome was preferentially retained, which may be due to differences in the enhancer region between the two parents (M. C. Chen and M. M. Fluck, manuscript in preparation).

FIG. 5. Recovery of an unselected viral genome from transformants obtained from nondividing and dividing infected F-111 rat cells. Infections of F-111 rat cells were carried out as described in the legend to Fig. 4 and in Materials and Methods. Transformants were isolated and high-molecular-weight DNA was extracted and analyzed as described in the legends to Fig. 1 to 3. DNA digestion with PvuII and hybridization were as described in the legend to Fig. 4. Lanes 1 to 4 contain DNA from cells maintained at confluency postinfection. Lanes 5 to 9 were diluted 1 30 postinfection. WT, Wild type.

Loss of the nontransforming deletion mutant is not due to a segregation problem. Most of the experiments described above were carried out under conditions designed to optimize transformation. These conditions usually allowed a few cell divisions postinfection before the monolayer became confluent. Since the times at which the integration events are fixed is not known and may not be during the first generation, we considered the possibility that the transformationdefective hr-t mutant and wild type might segregate to different cells during cell division, resulting in the loss of the hr-t mutant from cells which can potentially become transformed because they contain wild-type virus. To test this hypothesis, we carried out infections with monolayers derived from exponentially growing cells which were reseeded at confluency prior to infection (Table 1, experiments 4a and Sa). For nine transformants from experiment 5 (Table 1), this procedure did not increase recovery of the hr-t mutant (Fig. 5). Of 42 transformants analyzed in experiments 4 and 5 (Table 1), 4 in 18 contained the hr-t mutant in the infections with nondividing cells, versus 6 in 24 from dividing cells. This point was further confirmed by the analysis of transformants selected by anchorage-independent growth in agar (Table 1, experiments 11 to 13). As noted above, the ratios of the intensity of deleted to undeleted fragments in mixed transformants from experiments 4 and 5 (Table 1) were considerably in favor of the undeleted fragment, even for those transformants which arose in the nondividing condition. Evidence that selected and unselected parents are in the same cells. One possible explanation for the underrepresentation of the unselected parent in transformed cells is a low frequency of coinfection, which may be linked to an exclusion phenomenon. This possibility is not ruled out by the persistence of viral genomes of both types in the infected monolayer, since these genomes might reside in different cells. However, a physiological aspect of the coinfection suggests that this is not the case. When cells are coinfected with RA and hr-t mutants such as B2, the yield of transformants in the mixed infection is higher than that observed in the infection with RA alone at the same multiplicity. This effect is dependent on the dose of B2 mutant. The yield of transformants was increased 4-fold at a 1/1 ratio and 17-fold

VOL. 64, 1990

INTEGRATION OF UNSELECTED POLYOMAVIRUS GENOMES

at a 5/1 ratio (Table 1, experiment 2). We assume that this effect is due to complementation by the B2 parent of a large-T-antigen mutation which was very recently discovered in strain RA (19). Similar examples of complementation have been documented previously between large-T-antigen and middle-T-antigen mutants (10). It is worth emphasizing that even when a high level of complementation is observed, the complementing parent is underrepresented among the transformants. A 17-fold complementation factor was observed in coinfections with B2 at a 5/1 ratio, yet the mutant was not recovered in three transformants analyzed (Table 1, experiment 2b). In contrast, when coinfections were carried out with hr-t mutants plus wild-type A2, the yield of transformants was depressed proportionally to the increasing dose of hr-t mutant. The depression of transformation was recorded in experiments 7 and 8 (Table 1A). For example, in experiment 8, the yield of transformants obtained in coinfection with wild-type A2 and hr-t mutant 3A3 was reduced five times compared with the yield obtained with the same dose of wild-type A2 in a single infection. This effect has been noted previously and is referred to as a dominant lethal effect (10). We believe that its potential origin lies in competition for a limiting factor required for replication or transcription or both (Chen and Fluck, in preparation). At very high levels of hr-t mutant B2 (beyond the complementing range), a depression of transformation is also observed in mixed infections between RA and B2 (Chen and Fluck, in preparation). The inhibitory effect (competition) of hr-t mutants on wild-type transformation and the positive effect (complementation) of these same hr-t mutants on strain RA transforming potential reflect a complex biological system with multiple levels of interaction. It is worthwhile to point out that in mixed infections between hr-t and ts-a (large-Tantigen-deficient) mutants, similar complex interactions are observed (10). When the mixed infections are carried out at 33°C, a situation in which the ts-a mutant can transform, a depression of transformation in the mixed (ts-a plus hr-t) infection (compared with infection with ts-a alone) is observed, indicative of what we believe to be competition for a limiting factor. In contrast, when the mixed infections are carried out at 39°C, a situation in which the ts-a mutant cannot transform, an enhancement (complementation) is observed in the mixed infection (ts-a plus hr-t) compared with infection with ts-a alone (10). In this case, complementation of the large-T-antigen defect of the ts-a mutant by the functional large T antigen of the hr-t mutant is stronger than the competition between the two strains for a limiting factor(s), and net complementation is observed. Experiments in progress suggest that the differences in competition between A2-plus-B2 and RA-plus-B2 coinfections are due to differences in the enhancer region between these two strains (Chen and Fluck, unpublished data) in addition to the complementable RA defect in large T antigen noted above. In summary, we believe that both the cooperation and the inhibitory effect in the coinfection reflect the fact that the two parental genomes are coinfecting the same cells. The reductive effect of mutant B2 on wild-type transformation opens the possibility that in the population of infected cells, only those cells with relatively high ratios of wild type to mutant can become transformed, thus providing an explanation for the rare recovery of the transformationdefective mutant genome in transformed cells. However, this explanation is not applicable to the infections performed with strain RA.

1309

Underrepresentation of an unselectable genome is also observed in hamster cell transformants. To analyze whether the low frequency of cointegration of two parental genomes in transformed cells can be generalized to different hosts, we repeated the experiment with the other nonpermissive host for polyomavirus transformation, i.e., the hamster. BHK cells were infected with mixtures of strains RA and B2 (Table 1, experime;it 9). Thirteen transformants were analyzed in detail. Only one contained the B2 parent in addition to the wild type (results not shown). Underrepresentation of unselected genomes in td transformants is a general phenomenon. We considered the possibility that the underrepresentation of hr-t mutant B2 in transformants derived from mixed infections with strain RA is specific to this particular pair of genomes. This possibility needs to be considered, since RA and B2 differ at multiple sites: the origin-enhancer (K. Higgins, K. Friderici, M. C. Chen, and M. M. Fluck, unpublished data), the large-Tantigen coding region (between EcoRI and NsiI) (19), and the capsid region (6; Friderici and Fluck, unpublished data). To analyze the role of differences in the origin-enhancer region (encompassing binding sites for large T antigen and transcription factors), strain RA(eB2) was constructed, in which the small BglI-BclI fragment comprising the originenhancer region from strain RA was replaced by the equivalent region from mutant B2. This exchange did not relieve the underrepresentation of B2, since B2 was not recovered in any of 12 transformants analyzed (Table 1, experiment 6). To further test the generality of the underrepresentation, other pairs of parental strains were used, including wild-type A2, as well as other hr-t mutants, such as 30'b and 3A3 (Table 1, experiments 7 and 8). The unselected parent was also underrepresented in these infections. Low probability of double integration events in transformants derived from mixed infections with two selectable genomes. All the experiments described above used hr-t mutants whose integration is not selected for in the transformation process, raising the possibility that the underrepresentation of hr-t genomes represents a problem specific and limited to hr-t mutants. Differences in capsid phosphorylation (14) and in chromatin acetylation between the hr-t mutant and wild type (25) have been documented, and it is conceivable that these differences might lead to differences in processing by the host. Therefore, mixed infections were carried out with an isogenic pair of strains which can also be distinguished by restriction endonuclease analysis and in which both parents can transform at equal frequencies. Two such situations were designed. In the first situation, wild-type A3 and d145 were chosen, since they have identical sequences, except for the d145 deletion of 66 base pairs in the overlapping middle-T-antigen and large-T-antigen reading frames (2). The d145 mutation has no apparent phenotype (2), and the deletion can be detected by digestion with SstI and observation of fragment 3 (Fig. 6A). Experimental conditions were varied in these experiments (Table 1, experiments 10 and 11). Another independently derived Fischer rat cell line, FR-3T3, was used. In experiment 11 (Table 1), FR-3T3 cells were synchronized by starvation and released from Go by trypsin treatment and serum addition prior to infection, a procedure which we have recently shown to optimize transformation (H. H. Chen and M. M. Fluck, manuscript in preparation). Cells were infected at a multiplicity of 10 each of the two parental viruses; the analysis of the viral genomes in the input shows equal levels of each viral DNA (results not shown). Infected cells were passed to agar after infection to

1310

J. VIROL.

OH ET AL.

A i,

__B

*~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~..... ..- fi . -. -. . .- ...

A3~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

FIG. 6. Low probability of double integration events in transformants derived from mixed infections with two transforming strains (tf transformants). Fischer FR-3T3 rat cells were infected with wild-type A3 and transformation-competent deletion mutant d145 at a multiplicity of infection of 10 PFU of each parent (Table 1, experiment 11). Transformants were isolated and high-molecular-weight cellular DNA was analyzed as described in the legend to Fig. 1. Digestion was with SstI (A) or BgIII (B), and hybridization was to a probe for the whole viral genome (pPy-1). The same lane-numbering system was used in panels A and B. bp, Base pairs; Kb, kilobase pairs. L.E., Limit entry.

prevent cell division and segregation of the two parental genomes to different cells prior to transformation. Transformation frequencies in the mixed infection and in the controls were very close (1.3% for the infection with A3, 0.6% for that with d145, and 1.8% for the mixed infection), supporting the notion that A3 and d145 transform with similar efficiencies. A total of 22 transformants were analyzed in detail; all showed normal patterns of integration (Fig. 6). In 8 transformants, the A3 wild-type genome alone was present; in 10 transformants, the d145 genome alone was present. The equal recoveries of A3 and d145 genomes among the transformants suggest that these two strains have equal abilities to integrate and transform. Only 4 transformants of the 22 analyzed contained both parental genomes. Results are shown for 14 transformants derived from the mixed infections, including 3 transformants in which both parental genomes are integrated (Fig. 6A, lanes 3, 13, and 15). One clone each was derived from a single infection with either A3 or d145 (Fig. 6A, lanes 1 and 2). Finally, note that all the fragments expected after digestion with SstI are present (except for one fragment missing in Fig. 6A, lane 12). This pattern supports the finding that the viral genome is integrated in head-to-tail tandem arrays in the transformants (confirmed in digestions with EcoRI; results not shown). The vast majority of the transformants (12 of 14) contained multiple apparent sites of viral integration and produced free viral DNA (Fig. 6B). There was one exception, with a single integration site (Fig. 6B, lane 10). Transformants 9, 11, and 12 displayed a large number of integration sites; however, only the d145 parent is present in those transformants. Similarly, many transformants display multiple integration sites with only the A3 parent (results not shown). The second situation concerned mixed infections carried out with two wild types with identical coding sequences (those of the A2 strain) and differences involving a simple duplication in the BclI-BglI enhancer region (Table 1, exper-

iments 12 and 13). Detection of the two parental genomes was carried out by digestions with MspI and observations of the enhancer-containing fragment 3. Double parental integration was observed in 0 of 10 transformants derived in mixed infections between wild-type A2 and wild-type A2(eRA) and 0 of 13 transformants derived with A2 plus A2(eB2). In agreement with these results, double parental integration was detected in only 1 of 23 and 0 of 8 transformants derived from mixed infections with two transforming strains, Py3-33 plus ts3 (D. L. Hacker and M. M. Fluck, manuscript in preparation). We conclude that the low probability of double parental integration is a general phenomenon which also applies to mixed infections in which both parental genomes can transform. DISCUSSION The experiments reported above were designed to analyze the number of independent viral genomes which integrate into the host chromosome during the process of neoplastic transformation of rat and hamster cells by polyomavirus, i.e., a context in which the integration of at least one viral genome is known to occur. For this purpose, 108 transformants (td transformants) were derived from mixed infections with a transforming strain (a parental genome whose integration is selected) and a viable nontransforming deletion mutant (hr-t mutant) whose integration is not selected, and 57 transformants (tf transformants) were selected from mixed infections with two selectable transforming or wildtype parents. The recovery of the two parental genomes in the transformants was analyzed by restriction endonuclease digestion of cellular DNA followed by electrophoresis and hybridization, by using appropriate endonucleases which resolved the two parental genomes. The overall integration patterns of transformants derived from such infections in F-111 or FR-3T3 rat cells were typical, i.e., each transfor-

VOL . 64 7 1990

INTEGRATION OF UNSELECTED POLYOMAVIRUS GENOMES

displayed head-to-tail tandem arrays of the viral genome, apparently integrated at multiple sites in the host genome. This integration pattern is compatible with the occurrence of normal integration events in these mixed mant

infections. As expected, all td transformants contained the transforming parent, since retention of that parent was selected by the transformation process. Surprisingly, the unselected hr-t parent (or at least that fraction of the viral genome encompassing the deletion marker mutation) was vastly underrepresented-there were 12 cases among 108 td transformants. A total of 60% of the td transformants analyzed were obtained from infections in which the hr-t parent was used in considerable excess (up to 200-fold) over the transforming parent, yet this increase in the ratio did not increase the probability of recovering the hr-t parent. Furthermore, in those td transformants in which the hr-t parent was recovered, the ratio of mutant to wild-type sequences was at most equal, even those derived from infections carried out at an extreme ratio of mutant to wild type. Control experiments demonstrated that the hr-t mutant genomes persisted as well as the wild type did for up to 15 days in the infected cells, suggesting that the mutants have no major adsorption-decapsidation-replication-persistence problem. Previous experiments have demonstrated that an hr-t-like mutant integrates normally (20), as do the recombinants generated in crosses between two nonoverlapping hr-t mutants (S. Kalvonjian, C. Priehs, and M. M. Fluck, manuscript in preparation). The presence of both mutant and wild-type genomes in the same cells early in infection is a prerequisite for any conclusion regarding the integration of

the two genomes in the same cell. This presence was ensured

by infecting confluent cells which were not allowed to divide

between the time of infection and transformation or by passing the infected cells into a semisolid agar suspension. Furthermore, biological features of the mixed infection (in the case of td transformants) indicated that coinfection was taking place, since mixed infections with transforming and hr-t mutants led to an increase in transformation frequency when the transforming parent was RA or, conversely, a decrease in transformation when the transforming parent was wild-type A2. This strongly suggests that both parental genomes were in the same cell and, in the case of strain RA, complemented each other. The apparently opposite effect of the hrt-t parent in mixed infections with wild-type A2 compared with pseudo-wild-type RA has been addressed in detail (see Results). The possibility that cells cannot become transformed when both the wild-type and the transformation-defective parents have integrated appears highly unlikely, since multiple examples of such transformants have been described previously (10, 23). In summary, it appears that both the transforming and nontransforming parents infect the same cells and persist equally in those cells during the course of infection. Yet the frequency of recovery of a second parent in cells selected for the presence of the transforming parent (i.e., transformed cells) is low, and the nontransforming parent is highly underrepresented in those cells. The generality of the phenomenon of low frequency of double parental integration was established by the analysis of 57 tf transformants derived from mixed infections with two transforming parents; 23 of these were derived with two wild-type strains with identical coding regions (that of wildtype A2) and a duplication difference in the enhancer region, and the remaining 34 were derived from two transforming strains with identical enhancer regions and a 66-nucleotidedeletion difference in the early coding region. In each case,

1311

the equality of integration and transformation potential of both parents was demonstrated by similar transformation frequencies in the single and double infections as well as by the equal (-50%) probability of their recovery in tf transformants. However, as in the case of td transformants, the probability of double parental recovery was low, i.e., 6 cases in 57 transformants analyzed. Interestingly, the frequency at which a second parental genome is recovered is not affected by the number of times the infected cells are allowed to divide after infection, before the selection for transformation is applied. This suggests (as do other unpublished results from our laboratory and previous results [13]) that the integration events are fixed early after infection, before potential parental genomes segregation. The results presented in this paper are not consistent with the expectations from the simplest model of integration reviewed in the introduction. The analysis of integration patterns of viral genomes in a large collection of transformants derived from both polyomavirus and simian virus 40 infections of nonpermissive cells has suggested that multiple independent integration events have usually occurred within a single transformant (8 to 10 events; see Introduction) (4, 5). As judged by the usual criteria of multiple bands in digests with restriction enzymes which do not cut the viral genome, multiple integration sites are apparently present in both td and tf transformants (an example with seven countable apparent sites is shown in Fig. 1A, lane 2; also see Fig. 6). Thus, in transformants derived from mixed infections, we expected to find both parental genomes integrated at independent sites and represented within a single transformant in proportion to their ratio in the infected mix. However, as best demonstrated for td transformants, the nonselected nontransforming parent is very strongly underrepresented in the population of integrated viral genomes in the transformants. Our interpretation of the results is as follows. Viral integration is in fact infrequent, and in most transformants, a single integration event has occurred. Thus, in td transformants, only the transforming parent will be integrated, even when the nontransforming parent was present in large excess in the same cell at the beginning of the infection. The proposal that integration is a rate-limiting step in transformation is not new; it was first suggested (29) on the basis of the low frequency of stable transformants compared with the frequency of abortive transformants. However, such an interpretation contradicts the picture based on Southern blot analysis, and this issue is discussed in more detail below. The low frequency of integration could be due to a limited amount of host factor(s) involved in integration or to limited numbers of sites in the host chromosome available for integration. Some indirect support for the former hypothesis is available. The progressive inhibition of transformation in response to increasing doses of transformation-deficient mutant in coinfection with wild-type A2 is compatible with competition for a limiting factor or a limited site or both. The factor could be an enhancer-binding factor required for the transcription and replication of the viral genome (22). We are preparing data which support a role for the enhancer in integration (Chen and Fluck, in preparation). The fact that coinfections with wild-type A2 or pseudo-wild-type RA have sufficient levels of factors to allow the replication but not integration of the nonselected B2 parent points to a limit specific for integration but not replication. Possibly, integration sites (or a structure linked to them) are limiting or not easily reached. If the explanation above is correct, we must reinterpret

1312

OH ET AL.

J. VIROL.

the apparent number of integration sites revealed by previous restriction endonuclease analyses reviewed above. Preliminary results from our laboratory suggest that at least some of the multiple bands obtained in digests with enzymes which do not cleave the viral genome may consist of multiple forms of the same integration site present in different cells (K. Friderici, L. J. Syu, and M. M. Fluck, unpublished data), and we are testing the hypothesis that these are produced by in situ replication of the integrated viral genomes. Recent experiments in our laboratory (18) have demonstrated a high level of interviral recombination among the viral genomes which have become integrated in the transformed cells derived from mixed infections. Thus, recovery of sequences from the deletion mutant cointegrated with those of the wild-type parent among such recombinant transformants might have been expected in the present experiments. However, interviral recombination usually generates genomes which, in any interval, contain information from one parent but not both (18). In this particular case, it is predicted that most recombinants would not contain both the wild-type and deleted middle-T-antigen sequences but only the selected wild-type sequences. Furthermore, the frequency of double integration in the present experiments (11%) is in the same order of magnitude as that obtained for recombination and suggests that at least a fraction of double parental integration events are in fact generated by interviral recombination. Interestingly, in a few cases analyzed so far in which both wild-type and hr-t mutant sequences were recovered in the same transformants, the sequences were present at the same site, supporting this point (9, 12). In conclusion, the experiments described above suggest that a limited number of viral genomes (one, in the majority of the cases) become integrated in the host genome concomitantly with the neoplastic transformation of rat or hamster cells by polyomavirus. This number appears to be more restricted than the number of viral genomes which can be transcribed (since complementation or competition can be observed) or the number which can be replicated, since both parents persist. The intriguing restriction in integration of the viral genome to only one site in the genome of individual transformants remains to be understood. ACKNOWLEDGMENTS We gratefully acknowledge the contributions of Mike Thompson, Kay Higgins, Chuck Reidy, and Karen McWilliams (undergraduate students) and Sung Yeol Son (rotating graduate student) at various phases of these experiments. We thank Susan Kalvonjian for helpful discussions and Susan Conrad for a critical reading of the manuscript. This work was supported by Public Health Service grant CA29270 from the National Cancer Institute. LITERATURE CITED 1. Basilico, C., S. Gattoni, D. Zouzias, and G. Della Valle. 1979. Loss of integrated viral DNA sequences in polyoma-transformed cells is associated with an active viral A function. Cell

17:645-659. 2. Bendig, M. M., T. Thomas, and W. R. Folk. 1980. Viable deletion mutant in the medium and large T-antigen-coding sequences of the polyoma virus genome. J. Virol. 33:1215-1220. 3. Benjamin, T. L. 1970. Host range mutants of polyoma virus. Proc. Natl. Acad. Sci. USA 67:394-399. 4. Birg, F., R. Dulbecco, M. Fried, and R. Kamen. 1979. State and organization of polyoma virus DNA sequences in transformed

rat cell lines. J. Virol. 29:633-648. 5. Botchan, M., W. Topp, and J. Sambrook. 1976. The arrangement of simian virus 40 sequences in DNA of transformed cells.

Cell 9:269-287. 6. Carmichael, G. G., and T. L. Benjamin. 1980. Identification of DNA sequence changes leading to loss of transforming ability in polyoma virus. J. Biol. Chem. 255:230-235. 7. Feunteun, J., L. Sompayrac, M. M. Fluck, and T. L. Benjamin. 1976. Localization of gene function in polyoma virus DNA. Proc. Natl. Acad. Sci. USA 73:4169-4173. 8. Fluck, M. M., and T. L. Benjamin. 1979. Comparisons of two early gene functions essential for transformation in polyoma virus and SV-40. Virology 96:205-228. 9. Fluck, M. M., R. Shaikh, and T. L. Benjamin. 1983. An analysis of transformed clones obtained by coinfections with Hr-t and Ts-a mutants of polyoma virus. Virology 130:29-43. 10. Fluck, M. M., R. J. Staneloni, and T. Benjamin. 1977. Hr-t and ts-a: two early gene functions of polyoma virus. Virology 77:610-624. 11. Freeman, A. E., R. V. Gilden, M. L. Vernon, R. G. Wolford, P. E. Hugunin, and R. J. Huebner. 1973. 5-Bromo-2-deoxyuridine potentiation of transformation of rat embryo cells induced in vitro by 3-methyl-cholanthrene: induction of rat leukemia virus gs antigen in transformed cells. Proc. Natl. Acad. Sci. USA 70:2415-2419. 12. Friderici, K., S. Y. Oh, R. Ellis, V. Guacci, and M. M. Fluck. 1984. Recombination induces tandem repeats of integrated viral sequences in polyoma-transformed cells. Virology 137:67-73. 13. Fried, M. 1965. Cell transforming ability of a temperaturesensitive mutant of polyoma virus. Proc. Natl. Acad. Sci. USA 53:486-491. 14. Garcea, R. L., K. Ballmer-Hofer, and T. L. Benjamin. 1985. Virion assembly defect of polyomavirus hr-t mutants: underphosphorylation of major capsid protein VP1 before viral DNA encapsidation. J. Virol. 54:311-316. 15. Gattoni, S., V. Colantuoni, and C. Basilico. 1980. Relationship between integrated and nonintegrated viral DNA in rat cells transformed by polyoma virus. J. Virol. 34:615-626. 16. Griffin, B. E., M. Fried, and A. Cowie. 1974. Polyoma DNA: a physical map. Proc. Natl. Acad. Sci. USA 71:2077-2081. 17. Hacker, D. L., and M. M. Fluck. 1989. Studies of viral DNA synthesis in nonpermissive rat F-111 cells and its role in neoplastic transformation by polyomavirus. Mol. Cell. Biol. 9:648-658. 18. Hacker, D. L., and M. M. Fluck. 1989. Hyperelevated and segregated recombination in the integration pathway of polyomavirus. Mol. Cell. Biol. 9:995-1004. 19. Hacker, D. L., K. Friderici, and M. M. Fluck. 1989. A nonlethal mutation in large T antigen of polyomavirus which affects viral DNA synthesis. J. Virol. 63:776-781. 20. Lania, L., M. Griffiths, B. Cooke, Y. Ito, and M. Fried. 1979. Untransformed rat cells containing free and integrated DNA of a polyoma nontransforming (Hr-t) mutant. Cell 18:793-802. 21. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 22. Muller, W. J., D. Dufort, and J. A. Hassell. 1988. Multiple subelements within the polyomavirus enhancer function synergistically to activate DNA replication. Mol. Cell. Biol. 8: 5000-5015. 23. Priehs, C., K. Friderici, L. Winberry, and M. M. Fluck. 1986. Properties of cells transformed by the middle T-antigen-coding region of polyomavirus. J. Virol. 57:211-218. 24. Ruley, H. E., and M. Fried. 1983. Sequence repeats in a polyoma virus DNA region important for gene expression. J. Virol. 47:233-237. 25. Schaffhausen, B. A., and T. L. Benjamin. 1976. Deficiency in histone acetylation in nontransforming host range mutants of polyoma virus. Proc. Natl. Acad. Sci. USA 73:1092-1096. 26. Seif, R., and F. Cuzin. 1977. Temperature-sensitive growth regulation in one type of transformed rat cells induced by the tsa mutant of polyoma virus. J. Virol. 24:721-728. 27. Southern, E. M. 1975. Detection of specific sequences among

VOL. 64, 1990

INTEGRATION OF UNSELECTED POLYOMAVIRUS GENOMES

DNA fragments separated by gel electrophoresis. J. Mol. Biol. 98:503-517.

28. Staneloni, R. J., M. M. Fluck, and T. L. Benjamin. 1977. Host range selection of transformation defective hr-t mutants of polyoma virus. Virology 77:598-609. 29. Stoker, M., and R. Dulbecco. 1969. Abortive transformation of

1313

the TsA mutant of polyoma virus. Nature (London) 223:397398.

30. Winberry, L. K., C. J. Stewart, B. S. Schaffhausen, and M. M. Fluck. 1985. Transformation by polyoma ts-a mutants. I. Characterization of the transformed phenotype. Virology 144:433447.