Analysis of genomic instability in Li-Fraumeni fibroblasts with germline p53 mutations

Abstract

Germline p53 mutations are frequently observed in the normal DNA of cancer-prone patients with Li-Fraumeni syndrome (LFS). Fibroblasts from LFS patients develop chromosomal aberrations, loss of cell cycle control, and spontaneous immortalization. We transfected four different mutant p53 genes into human skin fibroblasts from normal donors with two copies of wild-type p53 (p53^wt/wt). Each mutant p53 expression-plasmid induced genomic instability equivalent to that seen in LFS cells. To test the role of wild-type and mutant p53 alleles in DNA replication and fidelity in LFS cells, we analysed the replication of the SV40-based shuttle vector pZ189 in four types of cells. We used p53^wt/mut and p53^mut/− LFS fibroblasts, and p53^−/− non-LFS cells. Replication of pZ189 in vivo was significantly reduced by the presence of a p53^wt allele. To show that this was not just due to inhibition of the function of T-antigen in SV40-based replication, we constructed a shuttle vector, pZ402, that contains a mutation in SV40 T-antigen which blocks its ability to interact with p53. Replication of pZ402 in LFS cells was also reduced by the presence of p53^wt, indicating that p53 can inhibit replication by interacting with proteins within the cellular replication machinery. Replicative errors in this shuttle vector are detected as mutations in a marker gene, supF. In addition to supF mutations, we observed deletion of a portion of the SV40 T-antigen gene in 100% of replicated plasmid pZ189 mutants (supF⁻) from the p53^wt/mut fibroblasts and in 88% of the supF⁻ mutants from the p53^mut/− (amino acid 175 arg to his) LFS cells. In one cell strain of immortal LFS cells, p53^mut/−, containing a p53 frameshift mutation at amino acid 184, pZ189 replication yielded very few of these deleted shuttle vector plasmids (15%). These large deletions were not detected in plasmids replicated in p53^−/− non-LFS cells, Saos-2 cells. Replicated plasmids with a normal supF gene were never found to have this large deletion regardless of the cell from which they were derived. Because the supF gene is not in the same region of the shuttle vector as the T-antigen gene it appears that second, independent gene deletions are frequent when replicative errors in supF occur in cells with a mutant p53. We conclude, therefore, that p53^wt/mut LFS cells contain an activity that promotes mutations. Such an activity, which is likely to be due to the p53^mut, could result in the high rate of chromosomal instability and allelic loss of the wild-type p53 observed as these cells spontaneously immortalize.

Introduction

Genetic damage in protooncogenes or tumor suppressor genes affects cell growth, tissue invasion, and metastasis. One mutation in a protooncogene is generally required for oncogenic activation, whereas mutations in both alleles of a tumor suppressor gene are required for cellular transformation (Knudson, 1971; Comings, 1973; Bishop, 1986). Germline mutations in the tumor suppressor gene p53 in individuals with Li-Fraumeni syndrome (LFS) predispose them to cancer (Malkin et al., 1990; Srivastava et al., 1990; Strong et al., 1992). LFS is characterized by multiple primary tumors and an early age of onset of a variety of tumor types, including soft tissue sarcoma, brain tumors, osteosarcoma, leukemias, adrenocortical carcinoma, melanoma, tumors of gonadal germ cells, and carcinoma of the breast, lung, pancreas, and prostate. Moreover, the presence of p53 mutations in the sporadic forms of many tumors suggests that these mutant p53 proteins participate in the transformation of diverse cell types. The physiological function of the p53 gene is that of a transcription factor (Fields and Jang, 1990; Raycroft et al., 1990) involved in cell cycle control (Yin et al., 1992) by inducing the expression of the cyclin dependent kinase inhibitor p21 (Harper et al., 1993; El-Deiry et al., 1993) in response to DNA damage (Kasten et al., 1991). This growth arrest pathway requires other p53-regulated genes such as gadd45 and is defective in fibroblasts from ataxia telangiectasia (Kasten et al., 1992). Overexpression of the wild-type p53 gene into carcinoma cells inhibits their growth (Baker et al., 1990).

The p53 protein interacts with components of the transcription/DNA repair complex TFIIH, XPB and XPD inhibiting their helicase activity (Wang et al., 1995). In yeast, TFIIH interacts with a DNA damage-recognition factor, RAD14, through the DNA repair protein RAD23 (Guzdar et al., 1995) and in mammalian cells through XPA (Park et al., 1995). It interacts with a replication factor, RP-A (Dutta et al., 1993; Li and Botchan, 1993), also reflecting an overlap between transcription and replication/repair functions of p53. p53 in able to inhibit SV40 DNA replication in vitro (Friedman et al., 1990) by inhibiting the helicase activity of SV40 T-antigen which is required for SV40 DNA replication. p53-mediated inhibition of replication in vitro requires the DNA binding function of p53, cannot be blocked by excess RP-A (Miller et al., 1995), and is independent of transcription (Cox et al., 1995). p21 can inhibit replication and excision repair through an interaction with proliferating cell nuclear antigen (PCNA) (Waga et al., 1994; Pan et al., 1995).

In culture, fibroblasts with germline p53 mutations from LFS familial cancer patients are initially normal and then spontaneously show changes in morphology, chromosomal aberrations (deletions, amplifications, breaks, translocations, aneuploidy) and lose their wild-type p53 allele. The loss of a wild-type p53 allele is necessary but not sufficient for the spontaneous cellular immortalization observed in LFS human fibroblasts (Bischoff et al., 1990, 1991; Rogan et al., 1995), LFS human mammary epithelial cells (Shay et al., 1995), and murine fibroblasts from p53 knockout mice (Harvey et al., 1993). The high frequency of spontaneous chromosomal aberrations in these LFS cells suggests that they may have a mutator function which inactivates genes that induce cellular senescence. To test this hypothesis, we transfected mutant p53 genes into normal skin fibroblasts (p53^wt/wt) and could induce chromosomal instability. We also analysed the fidelity of in vivo DNA replication of the shuttle vector pZ189 (Seidman et al., 1985) in LFS fibroblasts and found that the rate of replication and the frequency and types of mutations are affected by their complement of p53 genes.

Results

Effect of mutant p53 genes on chromosomal instability in normal skin fibroblasts

Because soft-tissue sarcomas are frequently found in LFS patients (Malkin et al., 1990), LFS fibroblasts provide an appropriate at-risk tissue for investigation of the mechanism of mutator activity that may be associated with genomic instability and cancer predisposition. Chromosome analysis of LFS fibroblasts with germline p53 mutations revealed dramatic genomic instability (Bischoff et al., 1990). Early after their establishment in culture these cells become hypodiploid. Later, by population doubling, (pd) 25–35, >90% of the cells have nondiploid numbers of chromosomes often containing homogeneous-staining regions, double minute chromosomes, telomeric associations, and dicentric and tricentric chromosomes.

In order to evaluate whether the presence of mutant p53 protein in a cell makes it prone to genomic instability, we transfected normal diploid human fibroblasts at 8 or 12 pd with plasmids containing a G418-resistance gene and a cytomegalovirus (CMV) promoter driving expression of human mutant p53 cDNAs. Some of these mutations are identical to those found in the germline of LFS patients. We have used 2 independent G418-resistant transfectant cell lines from each normal human fibroblast cell line expressing mutant p53 cDNA. p53 expression was confirmed by Western blot analysis (data not shown). Cell lines tended to grow more rapidly if they expressed the transfected mutant p53 relative to control cell lines that were transfected with pSV2neo. Those that expressed the p53 protein with a mutation at amino acid 273 made the most p53 protein and those that expressed the p53 protein with a mutation at amino acid 248 made the least. The transfected cell lines were subjected to chromosome analysis within 5 – 6 pd after expansion of the G418-resistant colony. Retesting of cell lines did not reveal dramatic changes in chromosome numbers. Growth from a single transfected cell to a full 100 mm tissue culture plate of G418-resistant cells requires approximately 22 pd, and therefore the cells were analysed at a cumulative pd of 35 to 40.

Normal fibroblasts transfected with mutant p53 genes developed significant chromosomal aberrations (Figure 1). For each transfected cell line, 20 metaphase chromosome spreads were analysed. Cells expressing p53 with a mutation at amino acid 248 (arg to trp) developed the least instability. At pd 5, one cell line was mostly hypodiploid whereas another had seven of 20 metaphases with hyperdiploid chromosome numbers. Aneuploidy was not enhanced by the level of transfected mutant p53 expression as this cell line with mostly hypodiploid metaphases (Figure 1) expressed far more of the mutant p53 than the other p53 mutant 248-expressing cell line, which had some aneuploid cells. Hypodiploidy was frequently observed in low passage LFS fibroblasts (Bischoff et al., 1990).

Figure 1.

Plasmid DNA was transfected into normal human fibroblasts. G418-resistant colonies expressing the mutant p53 proteins (determined by Western blotting) were tested for chromosome abnormalities by standard techniques. Twenty metaphase chromosome spreads were analysed for each cell line. The bar graphs indicate the percent of metaphase spreads with the indicated number of chromosomes. G418-resistant cell lines derived after transfection with pSV2neo were used as controls

Cells expressing a transfected p53 with a mutation at amino acid 175 (arg to his) contained no normal metaphases. The chromosomes in some of the metaphases analysed resembled the most aneuploid LFS germline p53 fibroblasts, immortal MDAH041. About half of the metaphases were hypodiploid, and the remainder contained subtetraploid chromosome numbers. Two of 20 metaphases contained greater than 100 chromosomes (Figure 1).

Cells expressing a transfected p53 with a mutation at amino acid 273 (arg to his) or the amino acid 143 mutant (val to ala) contained almost no normal metaphases (Figure 1). These cells also resembled aneuploid immortal LFS fibroblasts with germline p53 mutations. The amino acid 143 mutant p53 expression plasmid was also transfected into a strain of normal skin fibroblasts from another donor, and a similar pattern of chromosomal aberrations was found (data not shown), indicating that our results are not dependent on a specific fibroblast strain.

Normal human fibroblasts were also transfected with a plasmid vector only, pSV2neo, and chromosome analysis revealed no genomic instability. These colony-derived cells were near diploid with approximately 25% of the metaphase spreads containing loss of 1–2 chromosomes.

In summary, it appears that expression of a mutant p53 in normal skin fibroblasts is sufficient to inititate genomic instability.

Shuttle vector DNA replication in LFS fibroblasts with germline p53 mutations

We have developed a set of LFS fibroblasts with a p53^wt/mut genotype at less than 18 cumulative pd (pre-crisis) or with a p53^mut/− genotype at greater than 100 cumulative pd (post-crisis/immortal) from three individuals (Table 1). Non-LFS Saos-2 osteosarcoma cells were chosen because the cell line is also mesodermal in origin and contains no p53 protein representing a control for p53-null cells (p53^−/− genotype). The fibroblast cell strains have no significant difference in their growth rates at the pd used for these experiments. The difference in the genotypes of the same cell line provides us with a unique and powerful system in which to study the effect of p53 mutation on shuttle vector replication rate and fidelity.

Table 1.

Mutation fequency in shuttle vector pZ189

Cell line	Mutation at p53 locus	Total coloniesa	Sup F− mutants (frequency)
p53wt/mut Li-Fraumeni fibroblasts:
MDAH 041	one base deletion at AA 184 (GAT→GAA, ter 244)	323	8 (2.5 × 10−2)
MDAH 087	one base substitution at AA 248 (GGG→TGG or Arg→Trp)	10	0
MDAH 172	one base substitution at AA 175 (CGC→CAC or Arg→His)	430	21 (4.9 × 10−2)
p53 mut/− Li-Fraumeni fibroblasts:
MDAH 041	one base deletion at AA 184 (GAT→GAA, ter 244)	46268	13 (2.8 ×10−4)
MDAH 087	one base substitution at AA 248 (GGG→TGG or Arg→Trp)	3356	1 (3 × 10−4)
MDAH 172	one base substitution at AA 175 (CGC→CAC or Arg→His)	6931	8 (12 × 10−4)
p53−/− cells:
Saos-2 Cells	human primary osteogenic sarcoma cells (ATCC HTB 85) both p53 alleles deleted	30332	15 (5 × 10−4)
Background*		134287	1 (7 × 10−6)

Five micrograms of pZ189 plasmid was transfected. The replicated pZ189 was recovered by the Hirt method and DpnI digestion. All supF⁻ mutants were verified by isolation of pZ189 and by molecular analyses for the presence of genes that encoding supF tRNA and SV40 T-antigen.

^aThe total colonies obtained from four independent transfections of the human cells with the exception of MDAH087 p53^wt/mut which were performed twice

^*Indicates the routine control of transforming the bacteria with the plasmid DNA preparation without any transfection into mammalian cells. This provides the lower limit of the frequency of supF ⁻ colonies

The pZ189 vector we used is a 5.5-kb plasmid containing an SV40 T-antigen coding region that is essential for replication in primate cells (Carty et al., 1990, 1993). The plasmids that have replicated in mammalian cells can be selected by resistance to Dpn I endonuclease which digests plasmids that have retained the pattern of methylation found in bacteria. Moreover, the pZ189 shuttle vector mutation system contains a tyrosine suppressor tRNA gene (supF). Mutations within this gene are detected by white colonies on X-gal agar plates due to a lack of suppression of a β-galactosidase amber mutation in the host bacteria. Plasmid pZ189 was transfected and recovered after 22, 57, 78 and 102 h, the optimal recovery time was 78 h and was used in this study unless otherwise indicated.

We noticed that significantly less replicated pZ189 plasmid DNA was recovered from LFS cells pre-crisis (p53^wt/mut) than from the same strains of cells post-crisis (p53^mut/−), a result which suggests that the presence of the p53^wt allele affects replication and recovery of pZ189 (Table 1). To show that there was no difference in the transfection efficiency, we recovered DNA from the pre-crisis MDAH087 cells at 6 h post-transfection and found that the total plasmid DNA was the same within a factor of 2. The amount of replicated pZ189 was reduced by at least one order of magnitude in the cells with a p53^wt/^mut as compared to p53^−/mut cells (Table 1). These data suggest that the replication of pZ189 in cells with a p53^wt allele was less than in cells without the p53^wt allele. To further delineate the relationship between the replication of pZ189 and the presence of the wild-type p53 allele, we cotransfected p53^mut/− or p53^−/− cells with pZ189 and pCMVp53-SN₃, which carries a wild-type p53 cDNA whose expression is under the control of a CMV promoter (Karasuyama et al., 1989). To ensure that pZ189 was transfected into LFS cells with pCMVp53-SN₃, we transfected 212 ng of pZ189 with 2 μg of pCMVp53-SN₃. Cotransfection with the p53^wt reduced the amount of replicated pZ189 by 85–98% (Table 2). Therefore, the reduction in pZ189 replication was not due to intrinsic differences among these cell strains but rather the replication of pZ189 was inhibited by the presence of the wild-type p53 gene.

Table 2.

Replication of pZ189 in LFS cells or Saos-2 cells with or without cotransfection of wild-type p53 gene in pCMVp53-SN₃

Cell strain		pZ189 + vector DNA	pZ189 +pCMVp53-SN3
MDAH041	p53mut/−	8159	178 (2%)
MDAH087	p53mut/−	268	40 (15%)
MDAH 172	p53mut/−	976	86 (9%)
Saos-2	p53−/−	9177	631 (7%)

The total number of ampicillin-resistant colonies from one of two transfections is shown. The transformation efficiency was 8 transformants per picogram of pZ189 in this particular experiment. Restriction endonuclease digestion analysis using HaeIII on 39 plasmids of white colonies from cotransfection revealed that they were all pCMVp53-SN₃ plasmid

The reduction in pZ189 replication by cotransfection of wild-type p53 might be due to the interaction with T-antigen or to an effect on the replication machinery. We analysed this hypothesis by constructing a new shuttle vector with an amino acid 402 substitution from aspartate to asparagine in T-antigen (Lin and Simmons, 1991). This mutation blocks the binding to p53 without affecting its ability to support the replication of SV40 DNA. This new shuttle vector, pZ402, was transfected into MDAH172 cells and found to replicate with a similar efficiency to the pZ189 vector (Table 3). As with pZ189, cotransfection with CMV-plasmids driving the expression of wild-type p53 reduced plasmid replication of pZ402. Therefore, the inhibition of plasmid replication by p53 is not due to its inactivation of the replicative function of T-antigen but rather to some effect on the replication/repair machinery of the cell.

Table 3.

Comparison of the replication of pZ189 and pZ402 in p53^mut/− LFS MDAH 172 cells with or without cotransfection of wild-type p53 gene in pCMVp53-SN₃

Shuttle vector plasmid	pCMV-vector	SupF−	PCMVp53-SN3	SupF−
pZ189	104000	380	33000 (32%)	150
pZ402	88000	1550	8500 (9.8%)	1150

pZ402 contains a mutant SV40 T-antigen with an amino acid substitution from aspartate to asparagine at position 402 that cannot associate with p53 but is capable of mediating the replicative functions of the SV40 T-antigen (Lin and Simmons, 1991). The total number of ampicillin-resistant colonies per 200 ng of shuttle vector transfected is shown from one of two transfections. The percentage reduction due to cotransfection of p53^wt is shown in parentheses. The transformation efficiency was 200 transformants per picogram of plasmid in this particular experiment due to improvement in the efficiency of bacterial transformation using electroporation

Enhanced mutation frequency in shuttle vector DNA replicated in LFS fibroblasts with germline p53 mutations

The appearance of supF⁻ mutants as white colonies in the MBM7070 bacterial transformants was used to determine the frequency of mutations in the supF⁻ marker gene in replicated plasmids. The background supF⁻ mutant frequency for the untransfected pZ189 was 7 × 10⁻⁶ (Table 1). SupF⁻ mutants were generated during pZ189 replication in p53^wt/mut and p53^mut/− LFS fibroblasts and in p53^−/− Saos-2 cells. The mutation frequency in plasmids replicated through p53^mut/− or p53^−/− cells was 2 to 10 × 10⁻⁴. This frequency was similar to spontaneous supF⁻ mutant frequencies reported by others working with the pZ189 system in Epstein-Barr virus-, SV40 virus-, or adenovirus 5 DNA-transformed human cells (Bredberg and Nachmansson, 1987; Cleaver et al., 1989; Bubley et al., 1991; Clarke and Clements, 1991; Jaberaboansari et al., 1991; Muriel et al., 1991; Seetharam and Seidman, 1991; Parris and Kraemer, 1992; Runger et al., 1992; Sikpi et al., 1992; Wang et al., 1990; Yagi et al., 1992; Green et al., 1993). The frequency of supF⁻ mutants from cells with a _P53^wt/mut genotype was 2.5 × 10⁻² or 4.9 × 10⁻². This frequency was either 125 or 50 times greater than the frequency of supF⁻ mutants in the respective p53^mut/− cells which might somehow reflect the lower level of pZ189 replication in p53^wt/mut cells.

To determine the spectrum of mutations in the apparent supF⁻ mutants, we analysed these pZ189 plasmids by restriction enzyme digestion with HaeIII, which produces a characteristic fragment of 1.66 kb, and several smaller fragments of 0.8, 0.6, 0.4, 0.3, 0.25 kb (and some less than 0.25 kb) in the untransfected pZ189. There were three HaeIII patterns among the supF⁻ mutants (Figure 2): one had the 1.66-kb fragment (as in untransfected pZ189, a normal pattern) and the other two did not (variant patterns).

Figure 2.

Restriction analysis of supF⁻ mutants. The pZ189 DNA from nine representative supF⁻ mutants was digested with Hae III (5U/μg DNA, overnight at 37°C). Lanes 1 and 21 contain the 1-kb DNA ladder as a size marker (m), and lane 6 contains untransfected pZ189 DNA. The 1.66-kb fragment (arrow, equivalent to map unit 4862-3201 in the SV40 genome) is shown on the left panel. Cell strain: Saos = Saos-2; 041 =MDAH041; 172 = MDAH 172. pZ = untransfected pZ189 vector

pZ189 contains a unique EcoRI site (Seidman et al., 1985), so that the fragment generated by EcoRI digestion of supF⁻ mutants reveal the overall size of the recovered plasmids. Whereas most supF⁻ mutants had a 5.5-kb fragment, as did untransfected pZ189, the variant plasmids had fragments of approximately 3.2 kb. These variant plasmids hybridized in Southern blot analysis to an Sp2.2 probe containing the SV40 T-antigen region (Figure 3), or a probe of the supF gene (data not shown). These results indicated that these pZ189 supF⁻ mutants may have had deletions and/or rearrangements. Hybridization to the Sp2.2 probe showed that mutation(s) in these variant supF⁻ mutants included the HaeIII fragment from the SV40 T-antigen region (data not shown).

Figure 3.

Presence of DNA coding for SV40 T-antigen in variant supF⁻ mutants. EcoRI-digested plasmid DNA was resolved by agarose electrophoresis, blot-transferred to a positive-charged nylon membrane and hybridized with the Sp2.2 DNA probe labeled with ³²P-dCTP. The blot was washed in 0.5 × SSC and 1% SDS at room temperature (six times for l0min each), and then at 68°C (twice for 15min each). No DNA was recovered in clone No 19 after endonuclease digestions

All of the sup⁻ mutants with variant patterns were from plasmids replicated in LFS cells (Table 4). In fact, variations in the pZ189 plasmid were found in all 29 supF⁻ mutants from p53^wt/^mut LFS cells and in seven of eight supF⁻ mutants (88%) from the MDAH 172 cells with a p53^mut/− genotype. However, only two of 13 for the frameshift mutant p53^mut/− cell MDAH 041 (15%) and none of 15 supF⁻ mutants (0%) from Saos-2 cells p53^−/− were of this variant type. LFS cells with a mutant p53 yielded more of these variant, deleted plasmids than the null cell (Saos-2) or the MDAH 041 cells bearing only the p53 fragment due to the amino acid 184 frameshift mutation. Replicated plasmids (>4000) with a wild-type supF gene (blue colonies) were tested in pools of 300 to 500 colonies by Southern blotting the DNA. These DNA preparations were never found to have a subpopulation of plasmids with this large deletion regardless of the cell from which there were derived (data not shown).

Table 4.

Summary of restriction analysis in supF⁻ mutants

Cell strain	Total supF−	Normal DNA pattern	Variant DNA pattern
Background*
Untransfected pZ189	1	1	0
p53wt/mut Li-Fraumeni fibroblasts
MDAH041	8	0	8
MDAH172	21	0	21
p53mut/− Li-Fraumeni fibroblasts
MDAH041	13	11	2
MDAH087	1	1	0
MDAH172	8	1	7
P53−/− Control osteosarcoma cells
Saos-2	15	15	0

supF⁻ mutants from Table 1 were treated with HaeIII and then resolved in 2% agarose electrophoresis. The DNA fragments that contained a 1.66 kb fragments (Figure 1) were classified as normal pattern and those that did not were variants.

^*Indicates the routine control of transforming the bacteria with the plasmid DNA preparation without any transfection into mammalian cells

When pZ402 was replicated in MDAH172, variant plasmids were observed with a high frequency just as was observed with pZ189. For example, restriction endonuclease digestion analysis of 40 replicated plasmids from white colonies with or without p53 cotransfection revealed that 100% were the deleted, variant plasmid (Table 3). This analysis was possible because in this experiment the pCMVp53-SN₃ was grown in the M7070 strain of E. coli which is dam⁻ and any residual pCMVp53-SN₃ DNA was degraded when other unreplicated DNA was digested with Dpn I. As with pZ189, the position of the DNA deletion within the replicated plasmid appeared reproducibly within the SV40 T-antigen region. Moreover, we observed that these deletions occurred frequently in both p53^wt/mut MDAH172 cells, p53^mut/− MDAH172 cells, and with p53^wt cotransfection into p53^wt/− MDAH172 cells. This new shuttle vector produces a T-antigen protein that is incapable of interacting with p53. Therefore, in replicated plasmids there should be no selective pressure for the loss of T-antigen to avoid the interaction with p53 even though T-antigen is necessary for SV40-based replication.

We determined the sequence of the tRNA gene among a number of supF⁻ mutants that had a normal (nonvariant) Hae III digestion pattern. There were small deletions, multiple base substitutions, insertions, and interestingly combinations of all three mutations within a single supF gene (Figure 4). In LFS MDAH041 p53^mut/− cells 50% of the supF⁻ mutant plasmids sequenced (three of six) had simple point mutations while 50% had deletions of the supF gene. In Saos-2 non-LFS cells, the proportion of supF⁻ mutants with mutations (substitution, deletion, or insertion) in less than 4 bp was 30% (two in seven supF⁻ mutants sequenced), and the remaining 70% had mutations of greater than 9 bp. The mutations within the supF did not appear to be significantly different between MDAH041 and Saos-2. Although the number of supF⁻ mutants from non-LFS cells was low, the result was similar to the spontaneous supF⁻ mutants in transformed human cells reported by Mah et al, in adenovirus-transformed cells (Yang et al., 1988; Mah et al., 1989; Maher et al., 1989). In summary, our study thus provides molecular data supporting the previous cytogenetic finding that these LFS fibroblasts with germline p53 mutations contain an apparent mutator activity. This mutator activity renders plasmids that have a mutation in one part of the episome, supF, prone to a second mutation, deletion, elsewhere on the plasmid.

Figure 4.

The supF tRNA gene sequence in each supF⁻ mutant that contained a normal nonvariant pattern after HaeIII digestion

Discussion

We have previously reported that dermal fibroblasts from LFS patients have germline mutations in one of their p53 genes, develop extraordinary genomic instability, and spontaneously immortalize. To demonstrate that genomic instability was due to the presence of germline p53 mutations, we transfected normal human fibroblasts from normal donors with plasmids expressing mutated p53 genes. We found that aneuploidy and chromosome damage were present in the cells transfected with a mutated p53 gene but not in cells transfected with a control plasmid. If genomic instability is dependent on loss of the G1 check-point (Yin et al., 1992) then these data suggest that expression of a mutant p53 acts dominantly to abrogate that cell cycle checkpoint in many of the cells in such a population. However, if genomic instability results from inappropriate DNA repair by the TFIIH complex then the presence of the mutant p53 might interact with the XPB or XPD but fail to affect its helicase activity (Wang et al., 1995).

Evidence has suggested that the SV40 T-antigen is involved in the initiation of SV40 replication by binding to a palindromic sequence in the origin of DNA replication (ori) (Bargonetti et al., 1991; Li and Botchan, 1993). As a result, the helicase activity of the SV40 T-antigen in this pre-initiation complex, coupled with a cellular single-strand DNA-binding protein (RF-A/RP-A), unwinds the double-strand DNA at the SV40 ori (Sturzbecher et al., 1988; Hay and Russel 1989; Kienzle et al., 1989; Stillman, 1989; Tack et al., 1989; Borowiec et al., 1990). The pre-initiation complex at SV40 ori may facilitate its further binding to the DNA polymerase α-primase complex and result in primer synthesis and DNA replication. In vivo evidence supporting the interaction between SV40 T-antigen and DNA synthesis comes from transfection studies in which expression of antisence SV40 T-antigen DNA inhibits DNA synthesis (Jennings and Molley, 1987). In vitro DNA replication of SV40-based plasmids by DNA polymerase α requires SV40 T-antigen (Ariga and Sugano, 1983; Li and Kelly, 1984; Stillman and Gluzman, 1985; Wobbe et al., 1985; Yamagucchi and DePamphilis, 1986; Carty et al, 1990, 1993; Murakami et al., 1992). Efficient replication requires both the DNA polymerase α-primase complex and DNA polymerase δ-PCNA (Weinberg and Kelly, 1989).

The wild-type p53 protein, but not the mutant p53, is known to inhibit SV40 ori-dependent DNA replication (Braithwaite et al., 1987; Wang et al., 1989; Friedman et al., 1990). In vitro studies show that the wild-type p53 protein forms a complex with large T-antigen and that the p53 protein and DNA polymerase α compete for binding to SV40 T-antigen (DeLeo et al., 1979; Kress et al., 1979; Lane and Crawford, 1979; Linzer and Levine, 1979; McCormick and Harlow, 1980; Harlow et al., 1981; Ruscetti and Scolnick, 1983; Eliyahu et al., 1984; Jenkins et al., 1984; Parada et al., 1984; Meek and Eckhart, 1988; Dornreiter et al., 1990). However, the reduction in pZ189 replication by cotransfection of wild-type p53 is not simply due to the interaction with T-antigen but rather a general effect on plasmid replication. We determined this by constructing a new shuttle vector with an amino acid 402 mutation in T-antigen (Lin and Simmons, 1991). This mutation blocks the binding to p53 without affecting its ability to support the replication of SV40 DNA. This shuttle vector is also inhibited by overexpression of wild-type p53.

The wild-type p53 protein was shown to bind to a region of DNA composed of several specific GC-boxes adjacent to the SV40 ori (Bargonetti et al., 1991; Li and Botchan, 1993) and to inhibit initiation of SV40-based DNA replication by the DNA polymerase α-primase complex with consequent elongation by the DNA polymerase δ-PCNA complex. Mutant p53 proteins do not display sequence-specific DNA binding and for that reason may fail to inhibit replication. As DNA binding is necessary for p53-mediated inhibition of replication in vitro, (Miller et al., 1995), this function of wild-type p53 may be abrogated by a dominant negative effect of mutant p53 in p53^wt/mut LFS cells.

As a transcription factor the wild-type p53 has at least two activities: DNA binding (Kern et al., 199la, b; Mack et al., 1993) and transcriptional activation (Fields and Jang, 1990; Raycroft et al., 1990; Agoff et al., 1993). Through these activities, the wild-type p53 may regulate the transcription of growth-regulated genes whose products suppress cell proliferation (Ginsberg et al., 1991) at the G₁-S boundary in the cell cycle (Mercer et al., 1982; Diller et al., 1990; Yin et al., 1992). Wild-type p53 is necessary for the synthesis of p21, the regulator of cyclin dependent cdk2 kinase (Harper et al., 1993; El-Deiry et al., 1993). The absence p53^wt results in failure of the cyclin/cdk2 kinase/p21 complex to associate with PCNA in immortal LFS fibroblasts (Xiong et al., 1993). Possibly the failure to form this complex with the proper ratios of proteins underlies the genomic instability in LFS cells. In addition, p21 can inhibit replication by its interaction with PCNA.

The wild-type p53 may regulate the transcription of genes whose products inhibit replication in human cells. The frameshift germline mutation (in the pre-crisis cells of MDAH041 strain) resulting from the one-base deletion at one p53 allele produces a truncated p53 protein (see Table 2). Loss of the wild-type p53 allele in MDAH041 immortal cells results in a p53^mut/− genotype and in that cell line, shuttle vector replication is similar to that of p53^−/− Saos-2 cells (see Tables 1 and 2). Alternatively, p53^mut/− MDAH087 and MDAH172 express p53 proteins with missense mutations, and pZ189 replication is lower than in p53^mut/− MDAH041 and p53^−/− Saos-2 cells. These p53 proteins with missense mutations appear to maintain some wild-type p53 conformation and thereby retain the ability to reduce pZ189 replication. As one would expect, the truncated protein in MDAH041 p53^mut/− cells could not develop the partial wild-type p53 conformation that inhibits replication.

It is not clear what mechanism causes the 50 to 125-fold increase in the mutation frequency (supF⁻ plasmids) in p53^wt/mut MDAH041 and MDAH172 cells over their p53^mut/− LFS cell counterparts. The overall frequency of supF⁻ mutants in p53^mut/− LFS cells was similar to that of p53^−/− Saos-2 cells. MDAH172 p53^mut/− cells produce a full-length p53 with a mutation at amino acid 175 and no wild-type p53 protein. Most of the plasmids replicated through these cells contain a 2.3 kb deletion seen in the low passage p53^wt/mut MDAH172 cells. This may be related to the lower amount of pZ189 replication we observed in these cells in that some ‘wild-type-like’ conformation of the amino acid 175 mutant p53 in replication/repair may underlie this mechanism of deletion mutagenesis as deleted plasmids were also produced by replication through p53^mut/− LFS MDAH172 cells. It is possible that this mutator activity, which gives rise to frequent deletion of the pZ189 DNA, directly or indirectly results from interactions of the replication/repair complexes with mutant p53 proteins either as p53^wt-p53^mut or p53^mut-p53^mut oligomers. The new shuttle vector pZ402 produces a T-antigen protein which is incapable of interacting with p53. Therefore, the loss of T-antigen should not be selected for in replicated plasmids because it is necessary for SV40-based replication, which may indicate that this region is deleted late in replication. Alternatively, if multiple plasmids are retained in any one cell, then T-antigen could be supplied in trans from another plasmid genome, and the deletion in the T-antigen region could occur any time during replication.

Because supF gene and T-antigen gene are separated on the pZ189 plasmid, these mutations represent two independent events during plasmid replication. We have not observed deletions in >4000 replicated plasmids containing a wild-type supF gene. Thus the mutator activity in LFS cells may result from a subset of replication or DNA repair complexes that are error prone and thereby generate multiple independent mutations in a single replicated plasmid. Gene amplification in the form of double minute chromosomes is the first indication of genomic instability observed after p53^wt/mut LFS cells are placed in culture (Bischoff and Tainsky, unpublished data). Deletion mutations and DNA amplification are believed to arise via the same mechanism (Windle et al., 1991).

A mutator phenotype, as defined by an increase in cytogenetic detection of chromosomal aberrations and in mutant frequencies of several reporter genes, has been reported in several rodent cell lines (Chang et al., 1981; Liu et al., 1983, 1993; Murakami et al., 1985; Aizawa et al., 1987; Fukuchi et al., 1989). Whether p53 mutations in LFS cells render DNA replication or repair defective has yet to be determined. Nevertheless, an increase in spontaneous mutagenesis reported in these LFS strains is consistent with the hypothesis that DNA replication inaccuracy is the initial step toward carcinogenesis (Speyer, 1965; Hartwell, 1967; Springgate and Loeb, 1973; Loeb et al., 1974; Chan and Becker, 1979; Dresler et al., 1982; Loeb, 1991; Aaltonen et al., 1993; Peltomake et al., 1993; Thibodeau et al., 1993). Mutation of one allele of p53 may be sufficient to begin the process. Because the replication of pZ189 in vitro and in vivo requires complex enzymes, another possible candidate for the LFS mutator activity may be in the replication complex of LFS cells. In light of the physical interaction of p53 with the cellular replication protein A, RP-A, a component of the replication/repair complex (Bargonetti et al., 1991; Li and Botchan, 1993), and that p53 binding sites have been found in origins of replication, a direct role for p53 in replication fidelity must be considered. Mutations in p53 have been shown to alter the ability of the protein to bind DNA (Kern et al., 1991a, b; Mack et al., 1993). Mutant p53 proteins do not bind DNA specifically, but retain the ability to interact with a replication complex and this may underlie the mechanism of genomic instability in these LFS cells. Alternatively, stable binding of p53 to insertion/deletion mismatches in DNA is mediated by the terminus of the protein binding as a tetramer (Lee et al., 1995). Mutation of p53 may reduce the ability of mismatch repair systems to function properly (Kolodner and Alani, 1994; Modrich, 1991).

The associations of p53 with SV40 T-antigen, adenovirus E1b (Sarnow et al., 1982; Yew and Berk, 1992), human papilloma virus E6 (Werness et al., 1990), and human hepatitis B viral X protein (Bennett et al., 1994) repress the antiproliferative effect of the wild-type p53 by rapid degradation (Scheffner et al., 1991) or by inactivation (Oren et al., 1981; Reich et al., 1983). p53 mutations and hepatitis B virus infection are associated with cancer of the liver (Hollstein et al., 1991). Both wild-type and mutant p53 can interact with the TFIIH-associated proteins, XPB and XPD (Rad3), in transcription/DNA repair complexes by direct protein/protein interaction but only the wild-type can inhibit their DNA helicase activities (Wang et al., 1995). The hepatitis B virus X protein inhibits p53 binding to XPB, which has an effect similar to p53 mutation on the p53/XPB interaction: both result in a failure of p53 to regulate XPB helicase. In fact, in p53^wt/mut LFS cells the repair of UV-induced dimers is slower than in wild-type fibroblasts (Wang et al., 1995) and appears to affect the efficiency of DNA repair (Ford and Hanawalt, 1995). The mechanism by which p53 inhibits replication through regulation of the cell cycle or possibly some more direct effect on the replication machinery may be independent from the effect of a mutant p53 on increased mutations and genomic instability.

In summary, p53 gene products regulate a complex web of DNA replication and/or DNA repair, gene expression, and cell cycle control. Such a critical role in control of cell growth is consistent with the high cancer incidence observed in patients with LFS and mutant p53 gene transgenic mice (Lavigueur et al., 1991). This critical role is also consistent with the frequent chromosomal abnormalities we observed in p53^wt/mut LFS cells or when we expressed mutant p53 proteins in normal fibroblasts. Taken together, these results support the notion that a functional loss of the p53 allele is a step in carcinogenesis leading to increased genomic instability and mutations in oncogenes and tumor suppressors.

Materials and methods

Cell culture

The growth and maintenance of LFS fibroblasts was performed as previously described (Bischoff et al., 1990). Saos-2 human osteosarcoma cells were obtained from the American Type Culture Collection, and we confirmed that they produce no p53 protein. A strain of normal human diploid fibroblasts, AFB-1, was produced from a fresh skin biopsy from a normal female and used as the recipient cells for transfection of the mutant p53 plasmids at pd 3 to 8. All cells were grown in modified Eagle’s medium with 10% fetal calf serum and antibiotics.

Plasmids

p53 expression vectors in the pCMV-NEO-BAM vector whose expression is under the control of a CMV promoter were kindly provided by Drs C Findley and A Levine (Karasuyama et al., 1989).

Shuttle vector plasmid pZ189 contains the SV40 early region for replication, the supF tyrosine suppressor tRNA gene for a mutation marker, the pBR327 origin, and the β-lactamase gene for propagation in Escherichia coli and transformant selection. Plasmid pCMVp53-SN₃ is wild-type p53 gene in the expression vector pCMV-NEO-BAM, which contains a CMV promoter. Plasmid grown in E coli NR9099 (a rec^–strain from Dr Shaafer, NIEHS) was purified by the alkaline/lysozyme lysis method and was resuspended in Tris-EDTA buffer (10 mM Tris-HCl, pH 7.4 and 1 mM EDTA) (TE). Plasmid DNA in TE was treated with heat-inactivated RNase (100 μg/ml) for 10 min at 37°C and digested with proteinase K (200 μg/ml) in 0.5% SDS and then was extracted with phenol and chloroform. Plasmid pZ189 DNA in human cells was isolated by the Hirt method. Briefly, DNA was treated with lysis buffer (1.0 M NaCl and 0.6% SDS in TE buffer) as 4°C for 16 h and then was centrifuged at 12 000 g for 20 min at 4°C. One milliliter of supernatant was mixed with 3.0 ml of CsCl (125 grams in 80 ml of TE, refractory index 1.410) for 24 h. The CsCl-DNA mixture was then filtered through a Millex HA filter disk (0.45 μm; Millipore Corp, Bedford, MA). The filtrate was dialyzed by centrifugation in a Centricon 30 (Amicon Division, WR Grace and Co, Danvers, MA) with three TE buffer changes.

The Sp2.2 probe was a 2.2-kb fragment from SphI and PstI-digested pZ189; this probe is equivalent to bp 3204–5243/0–128 in the SV40 genome. At 68°C in 0.1 ×SCC and 1% SDS, Sp2.2 does not hybridize to pUC13 DNA but hybridizes to the 1-kb ladder (Bethesda Research Labs Gaithersburg, MD). The plasmid DNA used in this DNA replication study was not exposed to u.v. or ethidium bromide. All experiments were the average of 2–4 determinations.

The pZ402 vector was constructed on the pZ189 backbone by replacing the T-antigen sequences with those from a SV40 clone in pBR322 kindly provided by Dr Dan Simmons, University of Delaware. pZ189 was cut with SfiI and BelI. The T-antigen fragment, 2471 bp, was liberated from the pBR322 vector with the same enzymes and gel purified on a 0.8% agarose gel. The fragment was ligated and amino acid 402 mutant clones isolated by standard techniques. The presence of the amino acid 402 substitution from aspartate to asparagine was confirmed by DNA sequence analysis. The junctions of the cloning ligation were also confirmed by recutting with the appropriate restriction enzymes.

Stable transfection of mutant p53 expression vectors into human fibroblasts

Mutant p53 plasmids were transfected by a standard Ca-phosphate procedure (Wigler et al., 1979). Normal human adult skin fibroblasts at pd 8 or 12 were seeded in 100 mm culture dishes at 5 × 10⁵ cells per dish in duplicate. Five μg of plasmid DNA were transfected per dish. At 48 h post transfection the cells were treated with 200 μg/ml G418. The medium containing G418 was changed twice per week and colonies appeared among remnants of dying cells after 2 to 3 weeks. The colonies were picked after 4 weeks and expanded for chromosome analysis (Bischoff et al., 1990) and Western blot analysis for p53 protein at 5 pds after cloning.

Transient transfection of shuttle vector

Human cells were transfected with 212 ng of pZ189 + 2 μg of carrier DNA from E coli MBM7070 or +2 μg pCMV p53-SN₃ DNA using the Ca-phosphate method (Wigler et al., 1979; Liu and Loeb, 1984) followed by glycerol shock 4–6 h later. Seventy-eight hours after transfection, the plasmid DNA was recovered, followed by digestion with DpnI to remove bacterially methylated and unreplicated pZ189. The DNA was transformed into E coli MBM7070 using the calcium-chloride procedure. E coli MBM7070 has the genotype F⁻ lacA(AM)CA7020, lacY1 hsdR hsdM d(araABC-leu)7679 galU galK rpsL thi, which has an amber mutation in the lacZ gene. In the presence of isopropyl-β-D-thiogalactoside (IPTG), an inducer of the lac operon, and 5-bromo-4-chloro-3-indoyl-β-D-galactoside (X-gal), untransfected E coli MBM7070 forms white colonies in the absence of ampicillin but forms blue colonies if transformed with an active supF suppressor tRNA gene. Transformants were selected with ampicillin (100 μg/ml), X-gal (2%, 100 μl) and IPTG (20%, 4 μl) to determine the total number of transformants (blue and white colonies). Transformation efficiency of untransfected pZ189 was one to 40 ampicillin-resistant cells per pg DNA. The white colonies were isolated for the analysis of the spectrum of mutations in pZ189 DNA. The supF tRNA sequence in each supF⁻ mutant that contains a normal fragment pattern after Hae III digestion was determined using universal pBR322 EcoRI primer (100 ng), dideoxy NTP and ³⁵S-dATP with T7 Sequenase.